Ataxias with Oculomotor Apraxia (AOA1, AOA2, AOA4) and PNKP-Related Disorders

The class of DNA damage response ataxias: AOA1 (APTX), AOA2 (SETX), AOA3 (PIK3R5, disputed), AOA4 (PNKP) and AOA5 (XRCC1), with the allelic disorders of PNKP and SETX

Literature and regulatory information updated: 28 October 2026.

0. Initial Reference Table

The following table brings together the identifiers and essential features of the five numbered forms of ataxia with oculomotor apraxia (AOA). The prevalence rows indicate the entity and the source, because the available figures measure different things (population prevalence, proportion in an ataxia cohort or number of families described).

The allelic disorders of PNKP and SETX complete the scope of this article. The following table presents them with the same summary structure.

1. Introduction

1.1. Definition of the class

The ataxias with oculomotor apraxia (AOA) are progressive autosomal recessive cerebellar ataxias that begin in childhood or adolescence and combine three components: cerebellar degeneration visible on magnetic resonance imaging (MRI), axonal peripheral neuropathy that eventually dominates the motor disability, and impaired initiation of voluntary saccades that forces the patient to use head movements to shift gaze. The current group comprises five numbered subtypes, AOA1 to AOA5, caused by biallelic variants in APTX, SETX, PIK3R5, PNKP and XRCC1 [4]. Four of these genes encode proteins that repair DNA breaks or resolve RNA:DNA hybrids; the fifth, PIK3R5, rests on a single family and its relationship with the disease was classified as disputed by ClinGen in 2025 [36]. Ataxia-telangiectasia (A-T, gene ATM) and ataxia-telangiectasia-like disorder (ATLD, genes MRE11 and PCNA) share the mechanism and the oculomotor semiology, but are classified separately because they have immunodeficiency, chromosomal instability, clinical radiosensitivity or cancer predisposition [4, 47].

The name of the class comes from a sign that is not constant. Oculomotor apraxia is absent in one third to two thirds of patients depending on the subtype, varies with the stage of the disease and with the genotype, and when present lies on a continuum of severity; the 2026 review by Novis and colleagues therefore proposes treating it as a supportive rather than a mandatory criterion [4]. The combination of cerebellar ataxia with sensorimotor neuropathy alone justifies measurement of AFP and albumin and genetic testing, even without evident oculomotor abnormalities [4].

1.2. Historical overview

The term "oculomotor apraxia" was introduced by Cogan in 1952, with publication in 1953, to describe children with a defect in the initiation of voluntary horizontal saccades compensated by head thrusts; this congenital, isolated and non-progressive form still has its own entry (OMIM 257550) [4, 39]. In 1988, Aicardi, Barbosa, Andermann and colleagues described 14 patients from three continents with a slowly progressive syndrome of ataxia, choreoathetosis and horizontal and vertical oculomotor apraxia, with neurological signs indistinguishable from A-T but with later onset, without recurrent infections and with normal immunoglobulins, AFP, T- and B-lymphocyte markers and chromosomes 7 and 14; the title of the article, "a syndrome mimicking ataxia-telangiectasia", summed up the clinical question that the class still poses today [48]. The 14 patients belonged to 10 families and none had a tendency to frequent infections [5].

In Japan, Uekawa and colleagues described in 1992 families with early-onset ataxia, hypoalbuminaemia and hypercholesterolaemia, who would turn out to share the APTX variant c.689dup [1]. In Portugal, the systematic national survey of hereditary ataxias begun in the 1990s identified a large group of families with ataxia, ocular apraxia and neuropathy; Barbot and colleagues reviewed in 2001 22 patients from 11 families, with onset between 1 and 15 years (mean 4.7), symptom duration at last observation between 5 and 58 years and cerebellar atrophy in the 16 patients studied by MRI, and proposed five clinical criteria that still guide suspicion (subsection 5.1) [49]. This series predates the separation of AOA1 and AOA4, and it probably included patients with both genotypes. In 2001, the same survey, begun in 1993 in a country of 9.8 million inhabitants, counted 107 patients with recessive ataxia: Friedreich ataxia accounted for 38% and ataxia with oculomotor apraxia for 21%, the second most frequent recessive form [5]. Moreira and colleagues showed that year linkage to 9p13 in the two largest Portuguese families and, in five families from northern Portugal, homozygosity and haplotype sharing in a 2 cM region, indicative of a founder effect; three other families were excluded from this locus, revealing non-allelic heterogeneity [5].

In 2001, two groups published the identification of the AOA1 gene in the same issue of Nature Genetics. The Japanese group of Date and colleagues described "early-onset ataxia with oculomotor apraxia and hypoalbuminaemia" linked to 9p13, identified a new gene of the HIT (histidine triad) superfamily and named its product aprataxin [2]. The Portuguese-French group of Moreira and colleagues identified the main Portuguese and Japanese mutations and described the three domains of aprataxin, one of them homologous to the N-terminal domain of polynucleotide kinase 3′-phosphatase (PNKP); at that time, AOA1 was considered the most frequent cause of recessive ataxia in Japan and the second, after Friedreich ataxia, in Portugal [47].

The French-Canadian family described by Bouchard and colleagues in 1980, with seven siblings affected by progressive ataxia, tremor and severe distal amyotrophy, was recognised in 2005 as carrying SETX variants [9]. In 2000, Németh and colleagues linked to 9q34 a British family with ataxia and oculomotor apraxia without chromosomal instability or radiosensitivity, and Bomont and colleagues mapped to the same region the disease of four Japanese siblings with ataxia and raised serum creatine kinase, gamma globulins and AFP [5, 9]. In 2001 Koenig proposed separating the 9p13 form, usually with onset between 2 and 6 years, constant oculomotor apraxia and hypoalbuminaemia after a long course (AOA1), from the 9q34 form, with onset between 11 and 22 years, inconstant oculomotor apraxia and sometimes raised AFP, gamma globulins and creatine kinase (AOA2), and considered the AOA designation poorly suited to the latter [5].

In 2004, Moreira and colleagues showed that AOA2 results from variants in SETX, the human orthologue of the yeast RNA helicase Sen1p, with causal mutations in 15 families; they defined the disease by onset between 10 and 22 years, cerebellar atrophy, axonal sensorimotor neuropathy, oculomotor apraxia and raised AFP [50]. In the same year, Chen and colleagues identified dominant missense SETX variants in juvenile amyotrophic lateral sclerosis type 4 (ALS4), establishing the first allelic disorder of the class [46].

Shen and colleagues linked biallelic PNKP variants in 2010 to a syndrome of microcephaly, early intractable seizures and developmental delay (MCSZ) in consanguineous families [40]. In 2012, Al Tassan and colleagues described in a Saudi family with an AOA2 phenotype, without SETX variants, a homozygous missense variant in PIK3R5, designating the entity AOA3 [11]. In 2015, Bras and colleagues studied nine Portuguese families from the national survey that remained undiagnosed after exclusion of APTX, SETX and PIK3R5, combined homozygosity mapping with exome sequencing and found biallelic PNKP variants in eight, defining AOA4 [7]. Hoch and colleagues associated biallelic XRCC1 variants in 2017 with progressive cerebellar ataxia with oculomotor apraxia and axonal neuropathy, and showed hyperactivation of poly(ADP-ribose) polymerase 1 (PARP1) as the mechanism [33]. In 2018, Leal and colleagues reassigned the CMT2B2 of a Costa Rican family, previously linked to MED25, to PNKP variants [43]. In 2026, an Egyptian cohort of 27 patients proposed microlissencephaly as the fourth phenotype of the PNKP spectrum [41].

1.3. Nomenclature and reclassifications

The numbered nomenclature (AOA1 to AOA5) coexists with three other systems. OMIM uses its own designations: AOA1 is recorded as "early-onset ataxia with oculomotor apraxia and hypoalbuminaemia" (EAOH, 208920), a name proposed by Date and colleagues, in an entry that also includes adult-onset ataxia with oculomotor apraxia [1, 5]; AOA2 is recorded as spinocerebellar ataxia, autosomal recessive, with axonal neuropathy 2 (SCAN2, 606002) [6]; AOA5 corresponds to SCAR26 (617633) [12, 19]. The AOA2 entry 606002 previously had the designation SCAR1, which OMIM retains as a former designation and which Bras and colleagues still used in 2015 [7, 9]; the term SCAN2, proposed by Duquette and colleagues in 2005 because oculomotor apraxia is not universal and to distinguish the disease from SCAN1, is the one adopted by OMIM, GeneReviews, Orphanet and the 2025 review by Fogel and colleagues [6, 8, 9, 26]. OMIM groups under the SCAN designation SCAN1 (TDP1, 607250), SCAN2 (SETX) and SCAN3 (COA7, 618387) [51]. The table in the 2026 Novis review assigns the SCAR1 designation to APTX ataxia; this assignment has no counterpart in OMIM, whose entry 208920 records no SCAR designation, and it is not adopted in this article [4, 5, 7]. Dawson noted in 2001 that AOA was sometimes called "Aicardi syndrome", a name that is confused with Aicardi syndrome proper (agenesis of the corpus callosum with chorioretinal lacunae, OMIM 304050) [5].

The Movement Disorder Society (MDS) task force on the nomenclature of genetic movement disorders proposed in 2018 the prefix ATX followed by the gene symbol for 62 recessive ataxias with predominant ataxia, with double prefixes for 30 mixed entities [3]; the 2025 epidemiological review uses ATX-APTX, ATX-SETX and ATX-PNKP [10]. The Society for Research on the Cerebellum and Ataxias published in 2019 a consensus that classifies 59 primary recessive ataxias by clinical and pathophysiological criteria, plus 48 complex multisystem disorders with ataxia [52]. In 2026, Novis and colleagues proposed replacing the numbers with gene-based names, ordered by prevalence ("SETX-related ataxia", "PNKP-related", "APTX-related", "XRCC1-related"), and grouping these forms, A-T and the ATLDs under the collective designation of DNA damage response ataxias; AOA3 is excluded as a candidate association pending replication [4]. This is an expert proposal, not an official classification.

1.4. Entities covered and related entities

This article covers the AOA class uniformly: AOA1, AOA2 and AOA4 with in-depth subsections, AOA3 and AOA5 with the available evidence, and the allelic disorders of PNKP (MCSZ/DEE10, microlissencephaly, CMT2B2) and SETX (ALS4 and dominant ataxia with an allele in cis). Early-onset ataxia in general and ataxia-telangiectasia are included here only as differential diagnoses. Spinocerebellar ataxia with axonal neuropathy type 1 (SCAN1, gene TDP1) belongs to the same repair pathway and is treated as a close differential.

2. Molecular Genetics and Pathophysiology

2.1. Common mechanism: single-strand break repair and R-loop resolution

The proteins affected in the AOA class act at different points in the DNA damage response. ATM and the MRE11-RAD50-NBS1 complex detect and signal double-strand breaks; aprataxin, PNKP and XRCC1 process the ends of single-strand breaks; PCNA coordinates replication-coupled repair; senataxin resolves R-loops, RNA:DNA hybrids that form during transcription and threaten genome stability [4]. PNKP interacts with polymerase β, ligase III and XRCC1 to form the single-strand break repair (SSBR) complex that acts after ionising radiation or reactive oxygen species [1, 47].

The AOAs differ from A-T in having no defect in double-strand break signalling and no clinical radiosensitivity: AOA1 does not share the immunodeficiency, chromosomal instability and X-ray hypersensitivity of A-T [47, 53]. The almost exclusive restriction of the phenotype to the nervous system is explained by three factors described by Yoon and Caldecott: the brain depends on oxidative metabolism, which continuously generates single-strand breaks; post-mitotic neurons cannot use homologous recombination, which requires a sister chromatid; and unrepaired breaks block transcription in cells that must last a lifetime [54]. Non-syndromic ataxias due to SSBR defects affect the cerebellum almost exclusively [54].

2.2. APTX and aprataxin (AOA1)

The APTX gene has seven coding exons and produces isoforms by alternative splicing of exon 3; the long isoform, of 342 amino acids, is the main one in human cell lines [1]. The short isoform, of 168 amino acids, begins at the first ATG codon of exon 3; the long one (39.1 kDa) begins in exon 1 and has an additional 174 N-terminal amino acids, predominates in brain, heart, lung, kidney, liver and trachea, and in the liver the two exist in similar amounts [13]. The 2001 articles numbered the variants according to one isoform or the other, which gives different names to the same variant: 167insT of the short form is 689insT of the long form (c.689dup), P32L is P206L and V89G is V263G [13]. The current MANE Select (Matched Annotation from NCBI and EMBL-EBI) reference transcript is NM_001195248.2, whose numbering coincides, for the founder variant p.Trp279Ter, with the historical nomenclature based on NM_175073.1 [55, 56]. Aprataxin has three domains: an N-terminal domain designated PANT (PNKP-AOA1 N-terminal) or FHA (forkhead-associated), with 41% identity to the N-terminus of PNKP, which binds phosphorylated proteins; a HIT domain, characteristic of nucleotide hydrolases and transferases; and a C-terminal zinc finger motif that binds DNA and RNA [1, 47].

Aprataxin is nuclear, distributed in the nucleoplasm and the nucleolus, and resolves abortive ligation intermediates: when ligase attempts to seal a break whose ends are not ready, an adenylate group remains attached to the 5′-phosphate of the DNA, and aprataxin removes it, regenerating an end that can be ligated [57, 58]. It interacts with XRCC1, through the 20 N-terminal amino acids of the FHA domain, with XRCC4 and with PARP1, and keeps the level of XRCC1 stable [1]. Binding to XRCC1 depends on phosphorylation of the latter by casein kinase 2 (CK2) at residues Ser518, Thr519 and Thr523; in dividing cells two XRCC1 complexes coexist, one with ligase III, PNKP and polymerase β and the other with ligase III and aprataxin [13, 20]. PARP1 is required to recruit aprataxin to break sites, and AOA1 cells have lower expression of PARP1, apurinic endonuclease 1 (APE1) and OGG1, more oxidative DNA damage and less efficient base excision repair [5]. Tumbale and colleagues showed in 2014 that aprataxin also removes adenylate from RNA-DNA junctions formed in RNase H2-dependent ribonucleotide excision repair, and that AOA1 variants affect recognition of these junctions [13]. Pathogenic variants, including missense variants, destabilise the protein; patient cells are sensitive to agents that cause single-strand breaks, without a gross SSBR defect, without radioresistant DNA synthesis and without defects in ATM-dependent checkpoints [1]. In cells from a patient homozygous for a nonsense variant there was an increase in induced chromosomal aberrations, without hypersensitivity to ionising radiation [1]. In fibroblasts, Hannan and colleagues observed in 1994 moderate cellular radiosensitivity in two patients and in most relatives, of a degree very different from A-T, and Clements and colleagues described in 2004 only mild sensitivity to ionising radiation, hydrogen peroxide and methyl methanesulfonate [5, 13].

Secondary coenzyme Q10 (CoQ10) deficiency in muscle is a biochemical feature of AOA1: Quinzii and colleagues found it in a family with ataxia and an APTX variant [59], and Le Ber and colleagues measured reduced muscle CoQ10 in 5 of 6 patients, with lower values in homozygotes for p.Trp279Ter (p = 0.003) [60]. The deficiency did not correlate with duration, severity or blood parameters, and mitochondrial morphology and respiratory function were normal [5]. In the Italian series of Castellotti and colleagues (2011), by contrast, CoQ10 in muscle, fibroblasts and plasma was normal in 5 of 6 patients [5]. This is not a primary defect of a CoQ10 biosynthesis gene [61]: Garcia-Diaz and colleagues showed in 2015 that most, but not all, AOA1 fibroblast lines have lower expression of PDSS1, the first specific enzyme of biosynthesis, owing to reduction of a transcriptional pathway that includes APE1, NRF1 and NRF2, and that the mitochondrial dysfunction results from a role of aprataxin in transcriptional regulation and not in mitochondrial DNA repair [5].

2.3. SETX and senataxin (AOA2 and ALS4)

The SETX gene has 24 coding exons and two non-coding exons; the MANE Select transcript is NM_015046.7 [6, 55]. Senataxin, of 2677 amino acids and ubiquitously expressed, is an ATP-dependent 5′→3′ helicase with a C-terminal domain homologous to Sen1p, RENT1 (UPF1) and IGHMBP2 [45, 46]. It preferentially unwinds RNA:DNA hybrids and resolves R-loops at guanine-rich pause sites downstream of the polyadenylation signal, allowing recruitment of the exonuclease XRN2 and transcription termination; it also participates in splicing, chromosomal stability, autophagy, RNA processing and degradation, completion of meiosis and the innate immune response [8, 45]. Unresolved R-loops damage DNA and are toxic to neurons [45]. Senataxin binds RNA polymerase II and the poly(A)-binding proteins PABP1 and PABP2, and its depletion reduces polymerase binding to target genes, impairs transcription termination and alters splice-site selection; with SMN and ZPR1 it forms an R-loop resolution complex [14].

More than 100 pathogenic variants associated with AOA2 have been described: mainly nonsense, missense, splicing and small insertions and deletions, but also large deletions, insertions, duplications and a complex allele [6]. Rearrangements account for 8–20% of alleles depending on the study, 8% (4 of 51) in the largest series [6]; a deletion of exons 7 to 10 was described in a Chinese family [32], and Arning and colleagues described in 2008 an out-of-frame tandem duplication of the same exons, generated by unequal homologous recombination between AluY sequences [14]. Airoldi and colleagues described in 2010, in two sisters, an in-frame deletion of an N-terminal leucine residue (designated L144del in the publication) with expressed protein and cellular hypersensitivity to DNA-damaging agents, corrected by silencing of the mutant allele, whereas cells without protein owing to a nonsense variant did not have this hypersensitivity; the authors proposed a toxic effect of this variant [14]. Missense variants cluster in the N-terminus and the helicase domain, and the existence of complete deletions confirms loss of function as the mechanism of AOA2 [6]. In spermatogenesis, loss of senataxin causes DNA damage-mediated germ cell apoptosis, as in the Setx knockout mouse [62].

ALS4 results from a different mechanism: specific heterozygous missense variants (p.Thr3Ile, p.Leu389Ser, p.Arg2136His) cause dominant disease, whereas heterozygotes for AOA2 loss-of-function alleles are asymptomatic [6, 46]. In cells from patients with ALS4, p.Leu389Ser disrupts the interaction of senataxin with ZPR1, mislocalises both proteins and reduces R-loop levels, suggesting accelerated resolution (gain of function), as opposed to the loss of function of AOA2; expression of the variants p.(Leu1976Arg), from AOA2, and p.(Arg2136His), from ALS4, produced different transcriptional profiles [14]. SETX is thus a gene with two inheritance patterns, both confirmed: AR for AOA2 and AD for ALS4/distal hereditary motor neuropathy, the latter with definitive validity in ClinGen (2022) [36].

2.4. PNKP (AOA4, MCSZ/DEE10 and CMT2B2)

PNKP is a bifunctional DNA end-processing enzyme: the 3′-phosphatase activity removes phosphates from the 3′ end and the 5′-kinase activity phosphorylates the 5′-hydroxyl end, preparing them for polymerase and ligase in SSBR, base excision repair and non-homologous end joining [15]. The protein is modular: an N-terminal FHA domain binds XRCC1 and XRCC4, followed by the phosphatase and kinase domains; ATM phosphorylates PNKP [7]. The MANE Select transcript is NM_007254.4 [55]. The gene has 17 exons and encodes a protein of 521 amino acids (57.1 kDa), expressed in dividing neuronal precursors of the ventricular zone and in post-mitotic neurons of the cortical plate; XRCC1 stimulates the kinase and phosphatase activities of PNKP at damaged ends [16, 20]. In cells from patients with MCSZ, repair of damage induced by hydrogen peroxide and camptothecin was impaired, and RNA interference against Pnkp in mouse neurons increased apoptosis of precursors and differentiated neurons [16].

The study of fibroblasts from patients across the spectrum (MCSZ, AOA4 and CMT2B2) by Kalasova and colleagues showed a marked reduction in SSBR without a significant defect in double-strand break repair [63]. Restoration of SSBR required the kinase and phosphatase activities and the FHA domain; reduced phosphatase activity was associated with neurodevelopmental dysfunction, reduced kinase activity with neurodegeneration, and the extent and nature of the SSBR reduction were the main determinant of severity [63]. This model explains the distribution of variants: those of AOA4 are located in or near the kinase domain, whereas missense variants in the FHA and phosphatase domains were associated, as of 2015, only with neurodevelopmental syndromes [7, 64]. The Portuguese founder variant p.(Gly375Trp) lies in a conserved potential ATP-binding motif (amino acids 372–379) [7]. In CMT2B2, residues Gln517 to Gly521, removed by the variant p.(Gln517Ter), interact with the phosphatase domain (Cys308, Leu312, Leu315) and with the kinase domain (Arg458, Arg462) [43].

2.5. XRCC1 and PARP1 hyperactivation (AOA5)

XRCC1 is the scaffold protein of SSBR and base excision repair: it binds the poly(ADP-ribose) synthesised by activated PARP1, assembles ligase III, polymerase β and PNKP, and restrains PARP1 activity [19]. In cells from patients with biallelic XRCC1 variants, Hoch and colleagues observed slow SSBR and raised protein ADP-ribosylation, a cellular phenotype also present in PNKP-related disease [33]. In mice with an Xrcc1 defect, genetic deletion of Parp1 normalised cerebellar ADP-ribose and reduced neuronal loss and ataxia, identifying PARP1 as a therapeutic target; no PARP inhibitor has been tested in patients [19, 33]. Germline deletion of Xrcc1 is lethal in the mouse embryo, and conditional deletion in the brain causes ataxia with apoptosis of cerebellar granule neurons, reduced interneurons and lower electrical activity of Purkinje cells; deletion of Parp1 improved motor performance without correcting break repair [24]. The cells of the patient described had about 5% residual protein, reduced nuclear recruitment of XRCC1 after damage, slow repair of single-strand breaks without a defect in double-strand break repair, and increased sister chromatid exchanges [20]. Common XRCC1 polymorphisms (such as p.Arg399Gln, rs25487) do not confer an established cancer predisposition [19].

2.6. PIK3R5 (AOA3)

PIK3R5 encodes the p101 regulatory subunit of class γ phosphatidylinositol 3-kinase, and the Saudubray textbook includes AOA3 among the disorders of phosphoinositide phosphorylation [17, 65]. AOA3 rests on a consanguineous Saudi family with four affected siblings, homozygous for the variant p.(Pro629Ser) in a linkage interval at 17p12-p13, absent in 477 controls [11, 17]. This variant has a global allele frequency of about 0.15% in gnomAD v4.1, a maximum filtering frequency of about 2.6% in alleles of African ancestry and 30 homozygotes recorded [17]. No second family has been described, none of the 52 Brazilian patients with an AOA phenotype had PIK3R5 variants [12], and the ClinGen cerebellar ataxia expert panel classified the gene–disease relationship as disputed on 9 April 2025 [36]. OMIM and Orphanet maintain the entity with their own identifiers, which constitutes a divergence between databases [17]. The OMIM entry, last edited in 2017, classifies the relationship as confirmed on the basis of one family and records that the variant (c.1885C>T in NM_001142633.1, rs61761068), without functional studies, had frequencies of 0.011 to 0.021 in other control populations [18, 23].

2.7. Inheritance, penetrance and expressivity

All AOAs are autosomal recessive. The parents of a patient are usually asymptomatic heterozygotes, and there is no evidence of disease in carriers of AOA1, AOA2 or AOA4 alleles [1, 6]. A patient with AOA2 has been described whose second SETX variant (c.6106+1G>T) arose de novo, showing that both parents are not always carriers [66]. The penetrance of biallelic pathogenic genotypes is, in the cohorts described, complete for ataxia and neuropathy: none of the 90 patients with AOA2 reported by Anheim and colleagues was without neuropathy and without cerebellar atrophy [35]. Expressivity is variable within each form: oculomotor apraxia, chorea, dystonia and cognitive impairment have very different frequencies between cohorts and between genotypes (sections 3 and 2.9). Consanguinity is frequent in families with AOA4 (4 of 8 Portuguese families) [7].

2.8. Variant spectrum, nomenclature and founder variants

The following table presents the variants of clinical or population relevance, described according to Human Genome Variation Society (HGVS) nomenclature, normalised in VariantValidator (GRCh38) and checked in ClinVar on 28 October 2026 [55, 56]. The historical nomenclature column allows the variants to be recognised in older articles.

Three notational discrepancies in the sources deserve to be recorded. Bras and colleagues write that MCSZ is caused by the variant "p.Glu326Lys (c.1123G>T)"; the variant c.1123G>T corresponds to p.(Gly375Trp) of AOA4, and p.(Glu326Lys) of MCSZ is c.976G>A [7, 55]. Paucar and colleagues write the 17 base pair duplication as c.1250_1266dup, a notation not aligned with the 3′-most position rule; the normalised form is c.1253_1269dup [55, 64]. The GeneReviews chapter on AOA1 uses transcript NM_175073.1, whose positions were not converted to the MANE transcript in the case of c.689dup [1]. OMIM describes p.(Glu326Lys) as c.975G>A; in NM_007254.4, c.975G>A is a synonymous variant, p.(Glu325=), and the correct notation is c.976G>A [16, 55]. OMIM also records the 17 bp duplication as c.1250_1266dup, the notation of Shen and colleagues (2010) [16]. Despite the Saudi, Turkish, North American of European ancestry, Dutch and Portuguese origins of the carriers, the 2010 haplotype analysis suggested a common founder for this duplication, absent in 1080 control chromosomes from the Middle East and Europe [16].

In the variant table of the founder cohort, the Portuguese genotypes were: homozygosity for p.(Gly375Trp) in four families; compound heterozygosity for p.(Gly375Trp) with p.(Thr424GlyfsTer49) in one family and with p.(Gly442AlafsTer27) in another (phase not determined); and compound heterozygosity for p.(Thr408del) with p.(Gln517LeufsTer24) or with a complex allele p.(Arg439GlyfsTer51) in two families [7]. In Belarus, the variant c.1123G>T occurred in compound heterozygosity with two novel variants, c.1270_1283dup and c.1029+2T>C, in the first Slavic families described [67]. A variant p.Pro20Ser (c.58G>A), reported as pathogenic in homozygosity in an epileptic encephalopathy, had a frequency of 0.9% in the Exome Variant Server and a benign classification in ClinVar in 2015, illustrating the risk of attributing causality to common variants [7].

In AOA2, GeneReviews states that sequencing detects about 80–92% of pathogenic variants and that targeted deletion and duplication analysis detects 8–20% [6]. These percentages refer to the proportion of variants detectable by each method and not to the clinical sensitivity of the test.

2.9. Genotype-phenotype correlations

In AOA1, missense variants are associated with later onset (about 9 years) than homozygous truncating genotypes (2–12 years, mean 4.6) [1]. Date and colleagues had already observed in 2001 that insertions and deletions were associated with severe disease beginning in childhood and missense variants with milder and later disease; in a family compound heterozygous for p.Val89Gly and p.Pro32Leu (short isoform numbering), onset was at 25 years [13]. The variant p.Leu223Pro in homozygosity was associated with onset at 40 years with normal eye movements and albumin, and two French siblings compound heterozygous for p.(Trp279Ter) and a missense variant had late-onset ataxia without hypoalbuminaemia or oculomotor apraxia [13]. In the Italian series of Castellotti and colleagues there was no genotype-phenotype correlation [5]. Deletion of the seven exons of APTX in a Tunisian family and a variant at the exon 7 acceptor site in two other families caused the complete phenotype, showing that loss of the terminal zinc finger is sufficient to cause the disease [13]. The variant p.(Ala198Val) was associated with predominant, severe and persistent chorea [29]. Two Italian patients homozygous for p.(Pro206Leu) and p.(His201Gln) had disease onset at 28 and 29 years, whereas in Japan p.(Pro206Leu) was associated with onset at about 10 years [1]. The variant p.(Val263Gly) was associated with onset at 15 years and p.(Lys197Gln) with onset at 25 years, and two compound heterozygotes for p.(Arg199His) and an unidentified second variant had marked dystonia and mask-like facies [1]. The archived GeneReviews chapter brought together, in 2015, 16 pathogenic APTX variants in 37 families from three continents, including splicing variants (c.770+1G>A, c.875-1G>A) and a whole-gene deletion [1]. Oculomotor apraxia tracks severity and variant type: in the series of 80 patients reported by Renaud and colleagues it correlated with more severe disease, and the Novis review reports it in 83% of patients with truncating variants versus 17% of patients with missense variants [4, 69]. The same variant p.(Trp279Ter) may be associated with cognitive impairment or normal cognition, and severe cognitive impairment has been described in compound heterozygotes p.(Glu232GlyfsTer38)/p.(Pro206Leu), with mild impairment in the respective homozygotes, without a known explanation [1]. No correlation between variant and survival is known [1].

In AOA2, in a cohort of 90 patients, missense variants in the helicase domain were associated with less severe disease and with more pyramidal signs and dystonia than missense variants outside the domain, deletions and nonsense variants (p = 0.001, 0.008 and 0.01) [35]. The lower frequency of pyramidal signs with truncating genotypes may reflect masking by the severe motor neuropathy [35]. Two missense variants in the helicase domain, p.(Leu1976Arg) and p.(Leu1977Phe), were described in compound heterozygosity in a man with onset at 16 years, oculomotor apraxia, axonal sensory neuropathy and tremor [14].

In the PNKP spectrum, phenotypes do not relate simply to the type or location of the variant: the duplication c.1253_1269dup was observed in homozygosity in MCSZ, in two Dutch siblings with progressive cerebellar atrophy and polyneuropathy, and in compound heterozygosity with p.(Gly375Trp) in AOA4 [7]. In the Egyptian cohort of 27 patients there was no genotype-phenotype correlation [41]. Kalasova's functional model (phosphatase and neurodevelopment, kinase and neurodegeneration) is the most consistent explanation for the continuum [63].

2.10. Allelic disorders and modifiers

The allelic disorders of SETX are ALS4/distal hereditary motor neuropathy with pyramidal signs, which is dominant, and a dominant ataxia described in a mother and daughter with two missense variants in cis, p.(Asn603Asp) and p.(Gln653Lys) [6, 46]. The allelic disorders of PNKP form a continuum with four overlapping subgroups, according to the 2026 proposal: MCSZ/DEE10, microlissencephaly, AOA4 and CMT2B2 [41]. The validity of MCSZ is definitive in ClinGen; CMT2B2 is flagged in OMIM as a provisional association and has no ClinGen curation [15, 36]. For APTX no allelic disorder had been described up to 2015, but secondary CoQ10 deficiency acts as a potentially treatable biochemical modifier [1, 28].

3. Clinical Spectrum and Manifestations by System

3.1. Common phenotype of the class

The AOAs share a recognisable clinical sequence. After normal early motor development, cerebellar gait ataxia appears, followed by dysarthria and upper-limb dysmetria; later an axonal neuropathy sets in with generalised areflexia, distal amyotrophy and weakness, which becomes the main cause of disability [1, 6]. The oculomotor abnormalities include oculomotor apraxia, saccadic pursuit, gaze-evoked nystagmus and hypometric saccades, and may progress to ophthalmoplegia [1, 4]. Hyperkinetic movement disorders (chorea, dystonia, tremor) are frequent at onset, especially in AOA1 and AOA4. MRI shows cerebellar atrophy in practically all patients. Unlike A-T, there are no telangiectasias and no clinically relevant immunodeficiency, and cancer predisposition is not an established feature of any of the forms [48, 72].

Oculomotor apraxia is defined by head-eye dissociation in the head-free condition: when looking at a lateral target, the patient turns the head first, the eyes are initially deviated in the opposite direction by the vestibulo-ocular reflex, and the target is reached with several small saccades or with head thrusts [1, 6]. Saccade initiation may be preserved and duration increased by the succession of hypometric saccades, which may be mistaken for slow saccades [29]. In patients without head thrusts, "staircase" hypometric saccades were a more reliable sign than thrusts [6].

3.2. AOA1 (APTX)

Onset and first decade. In GeneReviews, which brings together Portuguese, Japanese and French cohorts published between 2001 and 2015, the disease begins between 2 and 10 years (mean 4.3), with about half of patients before 7 years and rare cases in adults (two Italians with onset at 28 and 29 years) [1]. In the French series of 14 patients from 9 families reported by Le Ber and colleagues, onset occurred at 6.8 ± 4.8 years (range 2–18) [29], and in the series of 80 patients from 51 families reported by Renaud and colleagues, diagnosed in a French laboratory between 2002 and 2014, at 7.7 ± 7.4 years [69]. The first manifestation is slowly progressive gait imbalance, followed by dysarthria and upper-limb dysmetria with mild intention tremor [1].

Other series. In Japan, Shimazaki and colleagues described in 2002 five patients from three families and an isolated case, with onset between 3 and 12 years, external ophthalmoplegia, choreiform movements, facial grimacing, mental deterioration, hypoalbuminaemia and hypercholesterolaemia [5]. In Tunisia, three families had a mean onset at 5 years; in one of them, oculomotor apraxia was absent in all but one patient and knee reflexes were preserved, and no patient had cognitive impairment [5]. In Italy, Castellotti and colleagues identified APTX variants in 13 of 204 probands with progressive cerebellar ataxia (6.4%), with onset between 3 and 7 years; six were in a wheelchair in early adulthood, six had intellectual disability from childhood and five later cognitive decline, many had distal weakness and atrophy of all four limbs, hypoalbuminaemia was present in 58%, hypercholesterolaemia in 69% and raised AFP in three [5]. Also in Italy, onset at 28 and 29 years has been described, as has a patient with onset at 40 years, normal eye movements, tongue and limb fasciculations, areflexia and normal albumin, showing the existence of mild adult forms [5].

Eye movements. GeneReviews described oculomotor apraxia in all patients, usually noticed some years after the ataxia, with exaggerated blinking, inability to move the eyes with the head immobilised, oculocephalic reflexes preserved until advanced stages and loss of balance when turning the head while standing [1]. Pursuit is normal in the first years, and then a progressive external ophthalmoplegia sets in, beginning with upgaze [1]. Later series show lower frequencies: 86% in France in 2003 [29] and 61% in the largest AOA1 series, in which oculomotor apraxia was considered an optional finding [4, 69]. The GeneReviews chapter on SCAN1 reports 80% [71]. The difference mainly reflects the mode of recruitment: the first cohorts were identified by the sign itself, and series based on molecular diagnosis include patients with missense variants and milder disease.

Neuropathy. All patients have axonal peripheral neuropathy with early areflexia, which dominates the advanced stages, with severe weakness and atrophy, short and atrophic hands and feet, pes cavus in about 30% and scoliosis in some [1]. Vibration and position sense are affected only in patients with a very long course, and pain and touch are preserved [1]. Independent walking is lost about 7 to 10 years after onset, and most need a wheelchair between 15 and 20 years of age [1]. In young adults, the severe neuropathy masks the initial ataxia and chorea and is the most typical symptom [29].

Movement disorders. Chorea appears at onset in up to 79–80% of patients and disappears in about half over the course of the disease; it persists in about 45% even after decades [1, 29]. Upper-limb dystonia occurs in about 50% [1]. AOA1 should be considered in a child with chorea without an HTT expansion [29].

Cognition. Cognitive impairment is variable: GeneReviews reports different degrees, with severe impairment in only one family [1]; Le Ber and colleagues found it in all patients in their series [29]; Fogel and colleagues report intellectual disability in about 50% [8].

Biochemical abnormalities. Hypoalbuminaemia and hypercholesterolaemia appear with disease duration: GeneReviews indicates reduced serum albumin in 83% and total cholesterol above 5.6 mmol/L in 68% of patients with more than 10 to 15 years of disease [1]; Le Ber and colleagues found hypoalbuminaemia in 83% and hypercholesterolaemia in 75% [29]; Renaud and colleagues, in a series with patients at more diverse stages, found hypoalbuminaemia in 63% and mildly raised AFP in 41% [69]; in the Italian series of 13 patients, the proportions were 58% for hypoalbuminaemia and 69% for hypercholesterolaemia [5]. The albumin threshold printed in the archived GeneReviews chapter ("< 3.8 g/L"), which remains in the online version of the retired chapter, contains a unit error: the physiologically compatible value is 3.8 g/dL, that is, 38 g/L [1]. Hypoalbuminaemia may cause lower-limb oedema [1, 68].

Survival. In the Portuguese families, age at last observation ranged from 17 to 68 years and disease duration from 12 to 58 years (mean 27.5); there were two deaths, one of unknown cause and the other in an 11-year-old girl, from a thalamic tumour [1]. A Japanese patient died at 71 years [1]. There is no extraneurological involvement beyond the biochemical abnormalities [1].

3.3. AOA2 (SETX)

Onset. Ataxia is the first sign in all patients, after normal development [6]. Age at onset is between 7 and 25 years (mean 14.6) in the large series, between 3 and 30 years (mean 20.3) in an Italian series of 10 patients [6], 15.1 ± 3.8 years in a French series of 18 patients [31] and a median of 15 years (interquartile range 13–18) in a systematic review of 216 published patients, 77.1% with onset between 10 and 20 years [32].

Frequency of signs. In the international cohort of 90 patients from 15 countries reported by Anheim and colleagues, polyneuropathy was present in 97.5%, cerebellar atrophy in 96%, oculomotor apraxia in 51%, pyramidal signs in 20.5%, head tremor in 14%, dystonia in 13.5%, strabismus in 12.3% and chorea in 9.5% [35]. In the review of 216 patients, ataxia was present in 100%, dysarthria in 99.2%, cerebellar atrophy in 99.4%, raised AFP in 97.6% and oculomotor apraxia in 34.2% [32]. In other series, oculomotor apraxia ranged from 20% in Italy and 32% in 19 Algerian patients to 56% in France [6, 31], and was absent in the 24 patients from 10 families of the Quebec cluster, who had dysarthria, saccadic pursuit, distal amyotrophy, sensorimotor neuropathy and raised AFP [70]. Convergent strabismus was described in 37% of Algerian patients [6]. In the French series of 18 patients from 6 families (France, the Antilles, Turkey, Morocco and Portugal), all had impaired slow pursuit, 89% gaze-evoked nystagmus, 11% strabismus, 28% dystonic posturing of the hands, 22% chorea and 17% head or postural tremor, and six had dysphagia [9]. In the Quebec patients, AFP was around five times the upper limit, gaze nystagmus was present in 91%, strabismus in 30% and tactile hypoaesthesia in 57%; motor conduction velocities were mildly reduced and sensory potentials absent, 7 patients had isolated cerebellar atrophy and 5 pontocerebellar atrophy [9]. In Tunisia, 13 patients from 5 consanguineous families, aged 16 to 30 years, had constant ataxia and dysarthria and oculomotor apraxia in 9; two became wheelchair-dependent between 20 and 29 years of age [9]. In a series of 7 patients from 4 families, all with onset in adolescence and areflexia, only one had oculomotor apraxia and two had mild lower-limb dystonia [9]. In three Japanese siblings of consanguineous parents, ataxia began in adolescence and distal weakness after 30 years of age, without oculomotor apraxia, with a wheelchair between 39 and 47 years [9].

Neuropathy. Sensorimotor neuropathy, mainly of the lower limbs, is present in 90–100% and comes to determine progression [6]. Sural nerve biopsy shows chronic axonal neuropathy with preferential loss of large myelinated fibres [6].

Movement disorders. Dystonic posturing of the hands, chorea and head or postural tremor occur in about 14% and, unlike the chorea of AOA1, tend to persist [6]. In the 2004 French series, choreic or dystonic movements reached 44% [31], and in the French-Canadian patients tremor reached 57% [6]. Rarely, a movement disorder is the presentation (upper-limb dystonia, truncal and facial chorea, isolated head tremor at 9 years) [6].

Cognition. Some patients have mild cognitive impairment, but none develops severe intellectual disability or dementia even after a long course; in the Italian series, 3 of 10 patients had mild intellectual impairment at around 50 years of age [6]. In this series of 4 Italian families, no patient walked without support at the time of examination (27 to 67 years), and the patient who underwent autopsy died of heart failure at 79 years [9].

Endocrine and reproductive system. Early menopause, ovarian failure, polycystic ovary syndrome and amenorrhoea due to hypogonadotrophic hypogonadism have been described [6]. In the most severely affected French family described in 2004, with pes cavus, scoliosis and wheelchair use in all 7 affected members, the 3 women had menopause between 21 and 24 years of age [9]. In men, Becherel and colleagues found impaired spermatogenesis in the 3 patients assessed, with testicular histology similar to that of the mouse model, and oligospermia or azoospermia may precede the neurological manifestations [62]. Low gonadotrophins are part of the laboratory profile described for AOA2 [10]. A dermatofibrosarcoma protuberans has been reported in one patient, without cancer predisposition having been established [6].

Other signs. Loss of deep sensation, Babinski sign, dysphagia and sphincter disturbances occur in some patients [6]. Immunoglobulin G and immunoglobulin A were raised in several families, and creatine kinase (CK) may be increased [6].

Neuropathology and imaging. Autopsy of a 79-year-old Italian woman showed atrophy of the cerebellar folia, maximal in the vermis and anterior lobe, marked loss of Purkinje cells without inclusions, chromatolysis of the oculomotor nuclei, severe demyelination of the gracile and cuneate fasciculi and degeneration of Clarke's columns [6]. 3 Tesla MRI in five patients showed atrophy of all cerebellar lobules, maximal in the vermis and anterior lobe, and absence of iron-related hypointensity in the dentate nucleus [6]. Proton spectroscopy showed a correlation between cerebellar N-acetylaspartate and the Friedreich Ataxia Rating Scale (FARS) [6].

Course. Disease duration in the series ranges from 2 to 53 years, with a maximum age at last observation of 79 years; a wheelchair is usually needed by about 30 years of age [6]. In the review of 216 patients, wheelchair dependence occurred on average 15.9 ± 8.6 years after onset, with no difference between sexes or between juvenile and adult onset [32]. Later onset and oculomotor apraxia were associated with slower progression (p = 0.03 and p = 0.009), and strabismus with faster progression (p = 0.03) [35].

3.4. AOA4 (PNKP)

Portuguese founder cohort. The 11 patients from 8 families described by Bras and colleagues (4 men and 7 women; consanguinity in 4 families) had disease onset between 1 and 9 years (mean 4.3 ± 2.3) [7]. The first sign was dystonia in 6, ataxia in 3 and oculomotor apraxia in 2; dystonia was so prominent that several patients were investigated for extrapyramidal disease, and it subsided spontaneously in all [7]. Ataxia and oculomotor apraxia followed, and all developed polyneuropathy with early generalised areflexia, distal muscle wasting, weakness that progressed to tetraplegia, and short and atrophic hands and feet [7]. Walking was lost between 7 and 21 years after onset, with a wheelchair between 14 and 25 years of age [7]. No patient had microcephaly or epilepsy [7].

Source: [7]. The biochemical abnormalities were determined at late stages of the disease.

Comparison with the other forms. The mean onset of AOA4 (4.3 years) is earlier than that of AOA1 (6.9 ± 4.7 in the series compared by Bras) and AOA2 (14.6 ± 3.4); the marked extrapyramidal component and rapid progression bring it close to AOA1, and the combination of low or normal albumin with normal or high cholesterol resembles late AOA1 [7]. Initial dystonia that regresses and obesity in some patients are features that distinguish AOA4 from the other AOAs, in which hyperkinesias usually persist [64]. Paucar and colleagues calculated a mean time to wheelchair use of about 13 years in the Portuguese cohort [64].

Variability outside Portugal. Later reports have broadened the phenotype. A Swedish patient had chorea and dystonia from 5 years of age, without clinical oculomotor apraxia and with slow progression (section 10) [64]. A Norwegian patient, the first non-Portuguese case described, had cognitive impairment, neuropathy, dysarthria, nystagmus, oculomotor apraxia, severe ataxia and hypoalbuminaemia with massive lower-limb oedema, which partially improved with treatment [68]. Two siblings described by Schiess and colleagues had hypoalbuminaemia, hypercholesterolaemia with raised LDL cholesterol, raised immunoglobulin E and normal AFP, with a marked defect in initiation of very hypometric horizontal saccades and head thrusts that exploited the intact vestibulo-ocular reflex [73]. A 25-year-old Italian woman began with chorea at 2 years of age, stable until adulthood, which mimicked benign hereditary chorea before a sensorimotor neuropathy appeared [74]. In Brazil, a patient presented first with Charcot-Marie-Tooth-type axonal neuropathy, and only years later with ataxia without oculomotor apraxia [21]. In Portugal, two sisters aged 52 and 58 years with cerebellar dysarthria, oculomotor apraxia, dystonia, gait ataxia and epilepsy still walked without support in adulthood [75]. In Belarus, two siblings had AOA4 with microcephaly but without severe disease [67]. In the Dutch cohort of 9 patients reported by Garrelfs and colleagues, AFP was raised regardless of phenotype and increased with age, and progressive cerebellar atrophy was universal [72]. In Italy, a patient with adult onset, among the oldest described, had polyneuropathy and cerebellar signs [76].

3.5. Allelic spectrum of PNKP: MCSZ/DEE10, microlissencephaly and CMT2B2

MCSZ/DEE10. The neonatal form of the spectrum is characterised by congenital or progressive primary microcephaly without a migration anomaly, early seizures that are usually refractory, severe intellectual disability with absent or minimal speech, and hyperactivity in most [15, 40]. In the original description of six families (three Palestinian, one Arab, one Turkish and one of mixed European ancestry), the 11 patients were aged between 1 month and 21 years; all had microcephaly at birth, five had a mildly simplified gyral pattern, five enlarged ventricles and six a thin corpus callosum; seizures, mainly focal with impaired awareness, began before 6 months of age, and two patients underwent surgery and four received a vagus nerve stimulator [38]. In a seventh family, with a less severe phenotype, the oldest patient walked at 14 months, spoke at 18 months and had moderate seizure control [38]. Two Dutch siblings of consanguineous parents, homozygous for the 17 bp duplication, had seizures from 12 months and 2.5 years of age, with decreasing frequency, but progressive microcephaly (from −3.25 to −6 standard deviations), short stature (−4 standard deviations), severe cerebellar atrophy and polyneuropathy, with loss of independent walking in childhood or adolescence [16, 38]. In the original description, a splicing variant was associated with more moderate symptoms, the cells were radiosensitive and no patient had cancer or immunodeficiency [40]. In the Egyptian cohort of 27 patients reported by Abdel-Salam and colleagues, all had early seizures, congenital microcephaly and intellectual disability; 25 had classic MCSZ and 2 a more severe phenotype [41]. MRI with widening of the frontal gyri and frontal hypoplasia, associated with prominent cerebellar folia owing to atrophy, has been proposed as a diagnostic clue [41]. Prenatal presentation with microcephaly has been described [15, 38]. The long-term natural history was documented in two Italian patients followed for 20 and 36 years, with focal epilepsy from childhood, variable seizure control and MRI with diffuse lissencephaly, ventriculomegaly, thin corpus callosum, dysmorphic hippocampi and cerebellar atrophy [77]. In Italy, a patient homozygous for p.(Thr424GlyfsTer49) with early MCSZ had cerebellar ataxia and neuropathy recognised late, and another, homozygous for p.(Ala429ThrfsTer42), had microcephaly and developmental delay without seizures [76]. In the Dutch cohort, cerebellar atrophy and raised AFP were early findings across the spectrum, showing the neurodegenerative component that links MCSZ to AOA4 [72].

Microlissencephaly. The 2 most severely affected patients in the Egyptian cohort had white matter loss and pontocerebellar hypoplasia that met criteria for microlissencephaly, and the authors proposed four overlapping subgroups on a continuum: MCSZ, microlissencephaly, AOA4 and CMT2B2 [41].

CMT2B2. The consanguineous Costa Rican family described in 2001 had a mean onset at 33.8 years (28–42), with symmetrical weakness and atrophy of the ankles and then the hands, hyporeflexia, "stocking and glove" sensory loss and normal or slightly reduced motor conduction velocities; at follow-up of 8 patients, 3 needed a cane and 2 had developed claw toes [42]. In the same family reassessed by Leal and colleagues, the 4 patients (26–59 years) had disease onset in the third decade with severe axonal sensorimotor polyneuropathy as the first and main sign; all had dysarthria and cerebellar atrophy on MRI, 2 of 4 had oculomotor apraxia (one with complete inability to initiate voluntary eye movements, with preserved vestibulo-ocular reflex), all had claw hands, and none had cognitive impairment, microcephaly, seizures, dystonia or pyramidal signs [43]. One patient had been in a wheelchair for 5 years [43]. Five additional Costa Rican probands were compound heterozygous for p.(Gln517Ter) and p.(Thr408del) and shared the ancestral haplotype; the gnomAD frequencies cited by the authors were 18 in 245,644 alleles for p.(Gln517Ter) and 15 in 214,338 for p.(Thr408del) [16]. The disease in this family had been mistakenly attributed to a MED25 variant located on the same haplotype, and the two OMIM entries describe this MED25 variant differently (p.Arg140Trp and p.Ala335Val) [16, 42]. The Brazilian patient reported by Pedroso and colleagues, homozygous for p.(Thr408del), had gait abnormalities and foot deformities (pes cavus and hammer toes) in the first year of life, slow progression after 9 years of age, steppage gait, mild ataxia, dysarthria, areflexia, mild cerebellar atrophy and never had oculomotor apraxia [42]. OMIM maintains CMT2B2 as a provisional association [42]. CMT2B2 is thus, clinically, an adult-onset form of AOA4 dominated by neuropathy.

Immunity and cancer. The Dutch cohort found no clinical signs of immunodeficiency or immunological abnormalities, including in the B-lymphocyte repertoire, and the only cancer observed was attributed to a concomitant VHL variant [72]. By contrast, two Japanese siblings with PNKP deficiency, microcephaly and ocular abnormalities had hypogammaglobulinaemia, with reduced total and class-switched memory B lymphocytes and lower somatic hypermutation [78]. A child with a PNKP variant developed acute myeloid leukaemia at 5 years of age and underwent haematopoietic cell transplantation, in the first report of a neoplasm in this context [79]. These isolated reports do not establish immunodeficiency or cancer predisposition as features of the disease [15].

3.6. AOA5 (XRCC1) and AOA3 (PIK3R5)

Patients with biallelic XRCC1 variants have progressive cerebellar ataxia leading to loss of independent walking, oculomotor apraxia and axonal neuropathy with distal weakness and areflexia [19, 33]. The Fogel review describes adult onset [8], and a second British series confirmed the association with ataxia and peripheral neuropathy [80]. OMIM flags the association as provisional, based on a single patient [19, 24]. This woman of Indian ancestry, with normal development, noticed balance and gait difficulties at 28 years of age; at 41 years she had gait and limb ataxia, dysarthria, dysphagia, oculomotor apraxia with nystagmus and hypermetric saccades, length-dependent axonal sensorimotor neuropathy and progressive cerebellar atrophy, and could no longer walk without support [24]. She was compound heterozygous for p.(Lys431Asn) and p.(Gln465Ter), both with nonsense-mediated messenger RNA decay [20]. In AOA3, the four Saudi siblings had ataxia, oculomotor apraxia, sensory axonal neuropathy, progressive distal muscle atrophy and cerebellar atrophy, with onset between 14 and 18 years (mean 15.6), early raised AFP and a more benign course [1, 11]. According to OMIM, the proband had falls from 14 years of age and was in a wheelchair at 23 years, with slow saccades, head-eye lag with head thrusts and normal pursuit [23]. The phenotype is indistinguishable from AOA2, and the gene–disease relationship is disputed (subsection 2.6).

3.7. ALS4 and SETX-related dominant ataxia

ALS4 usually manifests before 25 years of age with severe distal weakness and atrophy, pyramidal signs and normal sensation; it progresses slowly, spares the bulbar and respiratory muscles and does not reduce life expectancy [6, 46]. In the 11-generation Maryland family, with 17th-century English ancestors and 55 affected members in 2004 (variant p.Leu389Ser), the mean onset was at 17 years, with gait difficulty followed by weakness and atrophy of the hands and distal lower limbs; in the fourth or fifth decade there was proximal weakness and frequent wheelchair use, and in the sixth loss of useful hand function [14, 44]. Among 49 affected members, 86% had pathological hyperreflexia and 17% a Babinski sign, and 44 had normal sensation; autopsy of two patients showed loss of anterior horn neurons, degeneration of the corticospinal tracts and also loss of dorsal root ganglion neurons with degeneration of the posterior columns [44]. De Jonghe and colleagues described in 2002 three families with a diagnosis of distal hereditary motor neuropathy linked to the same locus, with onset usually before 6 years of age in one of them, brisk reflexes and Babinski sign in 8 of 18 patients; in the family with p.Thr3Ile, mean onset was at 8 years [14, 44]. In the dominant ataxia described in a mother and daughter with two SETX variants in cis, onset was at 13 and 3 years, with cerebellar atrophy, dysarthria, saccadic pursuit, gaze-evoked nystagmus and tremor, and with normal AFP and CK [6]. Both had raised immunoglobulin G and neither had peripheral neuropathy or head thrusts; OMIM classifies this cis allele as a SCAN2 variant and records the authors' hypothesis of a dominant-negative effect with partial function [9, 14]. Rare SETX variants in patients with late-onset sporadic ALS, such as c.2750T>C in a 72-year-old man, are isolated reports without established causality [81].

3.8. Comparison by system

Sources: [1, 5–7, 29, 32, 35, 43, 62, 67, 69, 72, 73, 75, 78].

4. Clinical Recognition and Warning Signs

4.1. When to suspect, by age

Early childhood (1–5 years). A child who, after normal motor development, begins to fall, develops chorea or dystonia and has difficulty directing gaze without turning the head should raise the possibility of AOA1, AOA4 and A-T [1, 7, 28]. In AOA4, dystonia may be the first sign and lead to investigation for an extrapyramidal disorder [7]; childhood chorea without an HTT expansion, or resembling benign hereditary chorea, justifies including APTX and PNKP in the work-up [29, 74]. Congenital microcephaly with neonatal or first-year seizures points to the MCSZ form of the PNKP spectrum [40, 41].

Later childhood and adolescence (5–20 years). Progressive ataxia with areflexia and axonal neuropathy, with or without oculomotor apraxia, raises the differential between Friedreich ataxia, AOA1 and AOA2; onset before 20 years is common to all three disorders [28]. In adolescence, AOA2 is the most likely form when AFP is raised [35]. A boy with azoospermia or oligospermia and mild neurological signs, or a girl with amenorrhoea due to hypogonadotrophic hypogonadism and ataxia, should raise the possibility of AOA2 [6, 62].

Adulthood. AOA2 and AOA5 may begin or be recognised in adulthood [6, 8]. A recessive axonal sensorimotor neuropathy with dysarthria and cerebellar atrophy on MRI, beginning in the third decade, suggests PNKP-related CMT2B2 [21, 43]. In adults with probable recessive ataxia, ERN-RND recommends first excluding Friedreich ataxia and RFC1 disease and then using next-generation sequencing [82].

4.2. Guiding warning signs

4.3. Urgent signs

The AOAs do not have acute metabolic crises. Situations requiring rapid assessment are ataxia of acute or subacute onset, which requires exclusion of acquired causes before the picture is attributed to a genetic disorder, dysphagia with weight loss or risk of aspiration, the refractory seizures of MCSZ, and massive oedema due to hypoalbuminaemia [37, 68, 77]. The Ataxia UK guideline recommends urgent referral of any child with ataxia for paediatric assessment [37]. Before general anaesthesia, respiratory weakness and bulbar dysfunction should be assessed (subsection 7.13) [84].

4.4. Newborn screening

There is no newborn screening for any of the AOAs. AFP cannot be used as a newborn screening test because it is raised at birth in all children and only reaches normal values at around 2 years of age [85]. Diagnosis depends on clinical recognition and genetic testing.

5. Diagnosis: Criteria, Tests and Algorithms

5.1. Clinical criteria

There are no formal validated diagnostic criteria for any AOA; the diagnosis is established by identification of biallelic pathogenic or likely pathogenic variants in the corresponding gene [1, 6]. The clinical criteria proposed by Barbot and colleagues in 2001 from the Portuguese series remain useful for suspicion: autosomal recessive transmission; early onset, mostly in early childhood; cerebellar ataxia with ocular apraxia and early areflexia, with full peripheral neuropathy later; absence of intellectual disability, telangiectasias and immunodeficiency; and possibility of long survival with severe motor disability [49]. The criterion of absence of intellectual disability no longer applies since AOA4 was individualised, in which cognitive impairment is frequent [7].

GeneReviews lists the suggestive findings of AOA1: cerebellar ataxia, oculomotor apraxia and areflexia followed by severe neuropathy; onset in childhood; slow progression to severe motor disability; absence of telangiectasias and immunodeficiency; cerebellar atrophy on MRI in all; axonal neuropathy in 100% on electromyography (EMG), which is normal only at very early stages; late hypoalbuminaemia and hypercholesterolaemia; and normal AFP [1]. For AOA2 it lists: cerebellar ataxia; reduced reflexes and then axonal sensorimotor neuropathy in more than 90%; oculomotor apraxia in about half; pyramidal signs; dystonic posturing, chorea or tremor; onset between 3 and 30 years; absence of cardiac involvement, cancer predisposition, immunodeficiency and severe intellectual disability; rare or absent telangiectasias; and raised AFP [6]. The Novis review proposes that oculomotor apraxia become a supportive criterion and that the combination of cerebellar ataxia with sensorimotor neuropathy suffice to request AFP, albumin and genetic testing [4].

5.2. Serum biomarkers

AFP, albumin and cholesterol guide the choice of gene and help interpret variants of uncertain significance; none of them confirms the diagnosis [45, 86]. The units ng/mL and µg/L are equivalent for AFP.

In the cohort of 90 patients with AOA2, the median AFP at diagnosis was 31.0 µg/L, versus 13.8 µg/L in patients with ataxia who tested negative and 3.4 µg/L (0.5–17.2) in normal individuals (p = 0.0004); of the two patients with normal AFP at diagnosis, one had raised values four years later and the other borderline values [35]. The authors calculated that sequencing SETX only in non-Friedreich, non-A-T ataxias with AFP ≥7 µg/L would miss 0.23% of AOA2 cases, and that the probability of AOA2 in this group was 46% [35]. Because raised AFP was one of the selection criteria for the patients studied, these figures overestimate the performance of the marker. In a study of 40 consecutively referred patients between 2008 and 2015, AFP thresholds of 7–15 µg/L for AOA1, 15–65 µg/L for AOA2 and above 65 µg/L for A-T made it possible to distinguish the three disorders, unlike video-oculography [83]. The normal AFP value varies between laboratories and over time, and in AOA2 it is lower than usual in A-T [6]. A first normal measurement does not exclude AOA2: in three Japanese siblings, AFP was described as normal, but a stored serum sample from one of them showed raised values [9]. AFP may also rise in heterozygotes: in a series of 4 families, all 7 patients and 5 of 8 asymptomatic heterozygotes had increased AFP, whereas the three heterozygous siblings of a Japanese family had normal AFP [9, 14]. A mildly raised AFP in a relative without signs is therefore not sufficient to consider them affected. Raised AFP may be an epiphenomenon of hepatic transcriptional dysregulation and also occurs, mildly, in SCAN1 (TDP1) and ARV1-related ataxia [10, 86].

The albumin threshold published in the archived GeneReviews chapter on AOA1 ("< 3.8 g/L") is in an impossible unit; the threshold consistent with the reference intervals is 3.8 g/dL (38 g/L) [1]. Albumin and cholesterol change only after 10 to 15 years of disease, so normal values in a child do not rule out AOA1 or AOA4 [1, 7].

5.3. Imaging, neurophysiology and eye movements

MRI shows cerebellar atrophy in all patients with AOA1 and AOA4 and in 96–99% of patients with AOA2, with relative sparing of the brainstem; brainstem atrophy exists in a minority [1, 7, 32, 35]. In AOA2, cerebellar atrophy is present from onset and remains a stable finding [35]. In MCSZ, the combination of widened frontal gyri with frontal hypoplasia and prominent cerebellar folia suggests PNKP [41]. Nerve conduction studies and EMG show axonal sensorimotor neuropathy (or predominantly motor in AOA1) in 90–100% [1, 6]; in one case of AOA4 the neuropathy was described as axonal-demyelinating [74]. Nerve biopsy confirms the axonal neuropathy, but the diagnosis rests on genetic testing [1].

Video-oculography detects abnormalities in practically all patients with AOA1, AOA2 and A-T: cerebellar signs, unstable fixation, impaired pursuit, hypometric saccades and abnormal antisaccades; very prolonged horizontal saccadic latencies, corresponding to oculomotor apraxia, appeared in about one third of patients [83]. The examination is not mandatory, but careful bedside oculomotor examination is essential: one looks for delayed initiation, staircase hypometric saccades and compensatory head movements [4, 83].

5.4. Diagnostic algorithms

Early-onset ataxia (ERN-RND, 2020). The ERN-RND flowchart begins with the history, including family history, clinical assessment with the Inventory of Non-Ataxia Signs (INAS) and a severity scale, such as the Scale for the Assessment and Rating of Ataxia (SARA), and brain MRI and neurophysiology, with exclusion of acquired causes and mimics [34]. Without specific abnormalities, testing for repeat disorders is considered, followed by exome or genome; with a family history, distinctive phenotype or dysmorphism, targeted laboratory investigation follows, possibly a chromosomal array, and single-gene or panel sequencing; if negative, exome or genome; if still negative, mitochondrial DNA sequencing, array, karyotype and reanalysis of the exome or genome [34]. Exome or genome may be started in parallel with the other investigations after MRI [34].

Adult ataxia (ERN-RND, 2024). When recessive ataxia is suspected, Friedreich ataxia and RFC1 disease are tested first; if negative and if next-generation sequencing is available, a panel, exome or genome is performed [82]. Without this access, the flowchart recommends AFP, vitamin E, albumin, cholesterol, lactate, caeruloplasmin, phytanic acid, very long-chain fatty acids, CK, urinary organic acids, plasma amino acids, lysosomal enzymes, cholestanol and oxysterols, a search for warning signs on MRI and ophthalmological examination and, in the presence of ocular telangiectasias, testing of a candidate gene [82].

EFNS/ENS (2014) and Ataxia UK (2019). The consensus of the European Federation of Neurological Societies and the European Neurological Society recommends, for recessive disease in adults, a stepwise assessment with biochemical markers and targeted genetic tests, in particular for Friedreich ataxia, A-T, ataxia with vitamin E deficiency (AVED), POLG, autosomal recessive spastic ataxia of Charlevoix-Saguenay (ARSACS), AOA1 and AOA2, and a full assessment for recessive forms when the family history is negative and onset is before 45 years of age [87]. The Ataxia UK guideline includes AFP among the first-line investigations and CoQ10, nerve conduction studies and next-generation sequencing among the second-line investigations [37].

Practical algorithm for suspected AOA. In the presence of progressive ataxia with axonal neuropathy and cerebellar atrophy: (1) measurement of AFP, albumin, total cholesterol, vitamin E, immunoglobulins and, in a young man with probable AOA2, gonadal assessment; (2) exclusion of Friedreich ataxia and, in adults, RFC1 disease, by specific expansion tests; (3) an ataxia panel including APTX, SETX, PNKP, XRCC1, ATM, MRE11 and TDP1, with copy number variation (CNV) analysis, or exome/genome; (4) when only one variant is found in a compatible gene, targeted CNV analysis by multiplex ligation-dependent probe amplification (MLPA) or genome and RNA studies; (5) if negative, periodic reanalysis [6, 10, 34, 82]. The order of genes may be adjusted to the population: in Portugal and Brazil, PNKP and SETX are the most frequent [7, 12].

5.5. Genetic testing and yield

The GeneReviews chapter on AOA2 proposes two equivalent approaches: single-gene testing (sequencing followed by deletion and duplication analysis) or a multigene ataxia panel that includes deletion and duplication analysis; alternatively, exome or genome, with an exome array if the exome is not diagnostic [6]. Sequencing of APTX is straightforward because the gene is small [88]. MDSGene maintains a worldwide resource on genetic ataxias (2932 articles, 548 genes and 122 countries) that can support interpretation [10, 89].

The yield depends on the population and on selection. In Brazil, a panel of APTX, SETX, PIK3R5, PNKP and XRCC1 applied to 52 patients with an AOA phenotype identified SETX in 15, PNKP in 12 and APTX in 5 (32 of 52 diagnosed, 62%) [12]. In Portugal, before the discovery of AOA4, testing of APTX and SETX had identified AOA1 or AOA2 in only 20 of 43 patients (46.5%) with an AOA phenotype [1]. In Alsace, among 102 patients with suspected recessive ataxia studied between 2002 and 2008, a molecular diagnosis was made in 57: Friedreich ataxia 36, AOA2 7, A-T 4, AOA1 3, Marinesco-Sjögren 3, ARSACS 2, AVED 1 and primary CoQ10 deficiency (ARCA2) 1 [30].

5.6. Limitations of testing

Exome sequencing and panels do not detect repeat expansions (Friedreich ataxia, spinocerebellar ataxias, RFC1), which require specific assays [28]. Intragenic deletions and duplications are a relevant fraction of SETX alleles (8–20%) and include multi-exon deletions, and a complete deletion of APTX has been described; they require MLPA, high-resolution array or validated CNV analysis on the exome or genome [1, 6, 32]. The frequency of partial or whole-gene deletions of APTX is unknown [1]. Deep intronic variants, such as the deletion in intron 15 of PNKP that causes complete skipping of exon 15 in the RNA, escape the exome and may require genome and RNA studies [77]. A negative result does not exclude the diagnosis, and the cellular tests available for A-T (radiosensitivity assay, ATM Western blot) do not identify the AOAs [58, 64]. For PNKP-related disease there is no biochemical marker or service functional assay [15]. There is no confirmatory non-molecular test for AOA2 [45].

5.7. Report, variants of uncertain significance and segregation

The report should state the gene, the versioned reference transcript (APTX NM_001195248.2, SETX NM_015046.7, PNKP NM_007254.4, XRCC1 NM_006297.3), the variants in HGVS nomenclature (c. and p.), zygosity, phase when determined and the class according to the criteria of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (ACMG/AMP) [55]. Two variants of uncertain significance, or one pathogenic variant and one of uncertain significance, neither establish nor exclude the diagnosis [6]. Segregation in the parents confirms the trans position and may reveal a de novo variant [66]. The high number of variants of uncertain significance submitted to ClinVar (483 in PNKP and 927 in SETX) and the conflicting classifications of variants published as pathogenic (PNKP p.(Gln517Ter) and p.(Arg462Pro)) show that interpretation must use the phenotype, AFP, albumin and, when available, functional studies [15, 45, 56]. A PIK3R5 variant should not be reported as a cause of AOA [17, 36]. When a molecular diagnosis is not obtained, DNA banking from the proband for future reanalysis is recommended [1].

5.8. Prenatal and preimplantation diagnosis

Once the familial variants have been identified, prenatal diagnosis by chorionic villus sampling or amniocentesis and preimplantation genetic testing are technically possible for all AOAs [1, 6]. In MCSZ, microcephaly may be detected on prenatal ultrasound, but it is not specific [15].

6. Differential Diagnosis

6.1. Comparative table

6.2. Main differential diagnoses

A-T. This is the most important differential, because it has implications for cancer surveillance, immunisation and radiation exposure that do not apply to the AOAs. A-T usually begins before 5 years of age, has telangiectasias, immunodeficiency and markedly raised AFP; variant A-T may have less raised AFP [28, 85]. The archived GeneReviews chapter on AOA1 stated that the presence of oculomotor apraxia made it possible to exclude A-T; this statement is not valid, because oculomotor apraxia is also a sign of A-T [1, 85]. In practice, AFP above 65 µg/L, telangiectasias and immunoglobulin deficiency point to A-T, and confirmation is genetic [58, 83].

Friedreich ataxia. This is the most frequent recessive ataxia in populations of European origin and should be excluded before any other [61]. Chorea, sensorimotor neuropathy, oculomotor abnormalities, hypoalbuminaemia, hypercholesterolaemia, cerebellar atrophy on MRI and absence of the Babinski sign favour AOA1 [29]. Exome tests do not detect the FXN GAA expansion [28].

Treatable forms. AVED and primary CoQ10 deficiency should be excluded in adolescence because they have effective treatment [1, 6]. The Saudubray textbook includes among the treatable progressive ataxias CoQ10 deficiency, in which the phenotype of ataxia with oculomotor apraxia deserves particular attention, Refsum disease, cerebrotendinous xanthomatosis and Niemann-Pick disease type C [94].

The AOAs among themselves. In childhood, AOA1 and AOA4 are difficult to distinguish clinically; initial dystonia that regresses, obesity and cognitive impairment favour AOA4, and persistent chorea with the p.(Ala198Val) variant favours AOA1 [7, 29, 64]. AOA2 begins later, has higher AFP and more pyramidal signs, and is associated with hypogonadism [6, 83]. AOA3 and AOA2 are clinically indistinguishable [1].

Neuropathies. Areflexia with pes cavus may lead to a diagnosis of Charcot-Marie-Tooth; dysarthria, oculomotor apraxia and cerebellar atrophy on MRI should point to PNKP (CMT2B2), APTX or SETX [1, 21, 43].

Movement disorders. Childhood chorea may mimic benign hereditary chorea and juvenile Huntington disease; progressive neuropathy and cerebellar atrophy point to AOA1 or AOA4 [29, 74].

Adult with an isolated case. In an adult without a family history, GeneReviews recommends excluding SCA2, which has slow saccades [1, 6]. Adult ataxia also includes ARSACS, Refsum disease, late-onset Tay-Sachs disease, Wilson disease, cerebrotendinous xanthomatosis and the POLG spectrum [98].

Other causes of oculomotor apraxia. Oculomotor apraxia also occurs in Joubert syndrome, Gaucher disease types 2 and 3, vitamin E deficiency, pyruvate dehydrogenase deficiency, congenital disorders of glycosylation, Lesch-Nyhan disease, dopamine transporter deficiency, Huntington disease, corticobasal syndrome and bilateral hemispheric lesions [4, 94]. Cogan congenital oculomotor apraxia spares random movements and voluntary vertical gaze, does not progress, and older patients come to compensate with hypermetric saccades instead of head thrusts; in a 1972 review, 23 of 34 cases were male, and the association with 2q13 deletions that include NPHP1 links some cases to nephronophthisis [96]. ATLD1 is distinguished from the AOAs by cellular radiosensitivity and chromosomal instability, and ATLD2 by deafness, photosensitivity and telangiectasias [90, 91]. Orphanet groups under ORPHA:98688 other rare entities with oculomotor apraxia, such as the syndrome of ataxia, intellectual disability, oculomotor apraxia and cerebellar cysts (OMIM 615960) and the syndrome of psychomotor regression, oculomotor apraxia, movement disorder and nephropathy (OMIM 617595) [39].

7. Treatments and Risk Management Strategies

7.1. Principles

None of the AOAs, nor the allelic disorders of PNKP and SETX, has a disease-modifying therapy approved by the Food and Drug Administration (FDA) or the European Medicines Agency (EMA) [8]. Treatment is supportive, multidisciplinary and problem-oriented: rehabilitation of ataxia and neuropathy, correction of metabolic abnormalities, symptomatic treatment of movement disorders, pain and sphincter dysfunction, prevention of orthopaedic and respiratory complications, and educational and social support [28, 37]. The Ataxia UK guideline, published in 2019 using the Guidelines International Network method, brings together 128 recommendations (6 grade B, 7 grade C, 10 grade D and 105 good practice points, GPP), almost all generic for the progressive ataxias; it is the guideline disseminated by ERN-RND [37, 61]. The 2014 EFNS/ENS consensus adds that, in most heredodegenerative ataxias, riluzole (level B) and amantadine (level C) may provide symptomatic relief regardless of aetiology, and that physiotherapy has class II evidence [87]. The 2018 American Academy of Neurology (AAN) systematic review on the treatment of cerebellar motor dysfunction was retired by the AAN itself and is no longer a current guideline; its data are cited only as historical context [100].

7.2. Rehabilitation and non-pharmacological measures

Physiotherapy is the mainstay of treatment, especially for the disability caused by peripheral neuropathy [1, 6]. Physiotherapy and occupational therapy programmes include balance and gait training, strengthening and assistive devices (canes, walkers, ramps, weighted cutlery, dressing hooks), and inpatient rehabilitation may improve ataxia and function in degenerative ataxias [28]. Weight control and home adaptations to prevent falls are part of the plan [28]. A wheelchair generally becomes necessary between 15 and 20 years of age in AOA1, at around 30 years in AOA2 and between 14 and 25 years in AOA4 [1, 6, 7].

For the neuropathy, the measures used in hereditary neuropathies apply by extension: daily Achilles tendon stretching, grip exercises, ankle-foot orthoses for foot drop and orthopaedic surgery for severe deformities [15]. Scoliosis should be monitored and referred for physiotherapy and spinal surgery; in mild scoliosis observation and possible bracing are proposed, and in severe scoliosis surgery (grade B recommendations formulated for Friedreich ataxia and applied by extension) [37].

Educational support addresses dysarthria (speech), oculomotor apraxia (reading) and upper-limb ataxia and weakness (writing), using a computer with voice recognition and an adapted keyboard [1, 6]. Dysarthria benefits from speech and language therapy and augmentative communication; dysphagia from speech and language therapy assessment, texture modification, supplements and nutritional follow-up if there is weight loss, and percutaneous endoscopic gastrostomy if caloric intake is insufficient [28, 37]. Cognitive rehabilitation has a grade C recommendation, and depression is treated in primary care with drugs, counselling or cognitive behavioural therapy, with referral to psychiatry if severe [37]. For the cerebellar cognitive affective syndrome, the cerebellar cognitive affective syndrome scale (CCAS) can be used in assessment [28].

Cerebello-spinal transcranial direct current stimulation (tDCS) has been studied in neurodegenerative ataxias that included patients with AOA (subsection 7.9) [101, 102]. Cerebellar transcranial magnetic stimulation has been used in several spinocerebellar degenerations [28]. None of these techniques has AOA-specific data.

7.3. Hypoalbuminaemia, hypercholesterolaemia and obesity

In AOA1 and AOA4 with hypoalbuminaemia, GeneReviews proposes a high-protein diet to restore albumin and prevent oedema [1]. In the Norwegian patient with massive lower-limb oedema due to hypoalbuminaemia there was partial improvement with treatment, not specified in the abstract [68]. The hypercholesterolaemia of AOA1, AOA2 and AOA4 is treated with a low-cholesterol diet [1, 6]; in the Swedish patient with AOA4 simvastatin was started [64]. The Ataxia UK guideline states, regarding Friedreich ataxia, that the possibility of statins reducing CoQ10 should not change the choice of treatment for hypercholesterolaemia; application to AOA is an extrapolation [61]. Simvastatin doses are given in table 7.5. The obesity of AOA4 may require intervention: the Swedish patient underwent gastric banding, with a reduction in body mass index from 39.4 to 30.1 kg/m² [64].

7.4. Coenzyme Q10 in AOA1

Secondary muscle CoQ10 deficiency in AOA1 (5 of 6 patients) and the response described in isolated patients underpinned the Ataxia UK recommendation to consider supplementation in AOA1 (grade D) and to treat patients with ataxia and documented CoQ10 deficiency with CoQ10 (grade D) [37, 59, 60]. The full guideline states that some patients with AOA1 and with ANO10 variants responded to 200–500 mg/day, and that in primary deficiency a high dose of 500–1000 mg/day is suggested, without an established dose and formulation [61]. Deficiency can be tested in skeletal muscle or in blood mononuclear cells [61].

The only specific controlled trial, NCT02333305 of the Assistance Publique – Hôpitaux de Paris, randomised 19 adults with molecularly confirmed AOA1 and hypoalbuminaemia to CoQ10 or placebo for one year, followed by one year of CoQ10 for all [103]. CoQ10 was given as a nanoemulsion (ubidecarenone food supplement): 20 drops three times daily (150 mg/day) below 50 kg and 40 drops three times daily (300 mg/day) from 50 kg, fixed doses by weight band without a maximum dose defined in the source; the primary outcome was the change in serum albumin every six months over two years, and secondary outcomes included SARA, prealbumin, cholesterol, AFP and annual oculomotor assessment [103]. The study ended in September 2017 with no results in the registry, and a Europe PMC search in October 2026 found no publication; this absence does not prove that the results do not exist [103].

The reference dose is that of the guideline, 200–500 mg/day orally [61]; the trial dose (150–300 mg/day as a nanoemulsion) is lower because the formulation has different bioavailability, and neither rests on published efficacy results [103]. CoQ10 is a food supplement without a summary of product characteristics (SmPC) or FDA label for this indication. For CoQ10 deficiency in general, a Brazilian handbook indicates 5–50 mg/kg/day and notes that soluble formulations may have better bioavailability, which serves as a paediatric reference by analogy [104]. The French trial excluded patients taking statins or other antioxidants [103].

7.5. Symptomatic pharmacological treatment

The Ataxia UK guideline proposes, for generalised spasticity, physiotherapy and then oral drugs in this order: baclofen, tizanidine, gabapentin, clonazepam, dantrolene or diazepam; focal spasticity is treated with botulinum toxin combined with physiotherapy [37]. For tremor, the order is propranolol, primidone, propranolol with primidone, topiramate, clonazepam and gabapentin, and disabling refractory tremor may be treated by functional neurosurgery (grade D) [37]. Focal dystonia is treated with botulinum toxin and generalised dystonia with oral drugs and, if necessary, surgery [37]. Disabling nystagmus and oscillopsia may be treated with gabapentin or baclofen (grade B), and downbeat nystagmus with baclofen or 3,4-diaminopyridine [37, 61]. Neuropathic pain is treated with amitriptyline, nortriptyline, carbamazepine, pregabalin, gabapentin or duloxetine, and musculoskeletal pain with paracetamol or non-steroidal anti-inflammatory drugs [15, 37]. The following table brings together the reference doses and the label doses for comparison.

The riluzole dose used in ataxia (100 mg/day, 50 mg every 12 hours, in adults) coincides with that of the amyotrophic lateral sclerosis label, the only approved indication, which does not recommend use in children and foresees no additional benefit with higher doses; use in AOA is off-label [28, 107]. For baclofen in adults, the maximum in the North American label (80 mg/day) is lower than that allowed in the Portuguese SmPC under hospital monitoring (100–120 mg/day) [110, 111]. The gabapentin doses for tremor in the Ataxia UK guideline (1200–2400 mg/day maintenance) lie within the range of the Portuguese SmPC for neuropathic pain, but tremor is not an approved indication [61, 112].

7.6. Epilepsy in the PNKP spectrum (MCSZ/DEE10)

The seizures of MCSZ are treated by the paediatric neurologist with the usual antiseizure medications; there is no drug or protocol with demonstrated efficacy for this condition [15]. Published experience consists of case reports. In the original series of 11 patients with refractory seizures, two underwent resective surgery and four received a vagus nerve stimulator, without the results being described in the source consulted [38]. In an Italian patient with focal epilepsy from 6 years of age, valproate had transient benefit, levetiracetam was poorly tolerated owing to psychomotor agitation, and carbamazepine, clobazam and lacosamide gave transient control, with persistent weekly seizures; in another patient, phenobarbital was ineffective, carbamazepine and vigabatrin were stopped for lack of efficacy, and the combination of clobazam with valproate from 12 years of age kept him seizure-free and was well tolerated [77]. With no published doses for MCSZ, the epilepsy label doses are used, with the extrapolation declared.

7.7. Bladder, bowel and sexual function

In 2026, ERN-RND endorsed for the ataxias and hereditary spastic paraparesis the NEUROGED guidelines of the European Academy of Neurology (EAN), the European Federation of Autonomic Societies (EFAS) and the International Neuro-Urology Society (INUS) [121, 122]. For assessment, they recommend asking regularly about urinary symptoms, urinalysis at baseline and in the event of worsening or suspected infection, a three-day bladder diary, measurement of post-void residual at baseline and with new symptoms, and baseline urea and creatinine; urodynamics is reserved for atypical symptoms, high risk of upper urinary tract damage or failure of conservative treatment [121]. For treatment, they recommend individual counselling on fluid intake, pelvic floor exercises for urgency or incontinence (strong evidence), antimuscarinics or β3 agonists for overactive bladder (strong), intermittent catheterisation as first line for retention with a residual persistently above 150 mL (strong), desmopressin selectively for nocturia (consensus), α1 blockers for voiding symptoms (weak evidence) and tibial nerve stimulation if there is failure or intolerance (weak); routine antibiotic prophylaxis in those who catheterise and treatment of asymptomatic bacteriuria are not recommended (strong) [121]. The Ataxia UK guideline adds reducing caffeine, carbonated drinks and alcohol, timed voiding, 1–2 L of fluid per day and, in patients with cognitive impairment, preferring trospium or darifenacin among the antimuscarinics [37]. Constipation is treated with lifestyle measures and then laxatives; sexual dysfunction may be treated with phosphodiesterase 5 inhibitors, with caution in hypotension and heart disease [37].

7.8. Regulatory status

There is no approved medicine for any AOA. The table compares the status of the symptomatic drugs cited in the guidelines.

The search of 28 October 2026 identified no orphan designations, marketing authorisation applications, Committee for Medicinal Products for Human Use (CHMP) opinions or authorisations for these disorders; the FDA and EMA orphan designation registers were not searched by indication, so this absence is not conclusive [8, 124].

7.9. Clinical trials

The limitations of these data are marked. The only trial directed at an AOA (CoQ10) has not published results, so the guideline recommendation rests on reports and a biochemical rationale [37, 103]. The benefit of riluzole was measured in patients with ataxias of other aetiologies; the large relative reduction corresponds to an absolute difference of 39 percentage points in a small sample, with a wide confidence interval, and cannot be transposed to AOA, in which disability is dominated by neuropathy [105]. The neuromodulation studies mixed aetiologies and do not report how many patients had AOA [101]. No negative, suspended or withdrawn trials specific to these disorders were identified. The 2025 Fogel review emphasises that most patients with recessive ataxia available for trials already have more than 10 years of disease and many no longer walk, particularly in AOA2, which makes it difficult to demonstrate benefit and justifies early genetic diagnosis [8].

7.10. Developments in the last 6–12 months (November 2025 to October 2026)

Outside this window, a 2024 study showed in a Huntington disease model that fructose-2,6-bisphosphate restores PNKP repair activity; this is a preclinical proof of concept of pharmacological restoration of the enzyme, with no studies in the AOAs [127].

7.11. Agents and situations to avoid

Cells from patients with PNKP variants are radiosensitive; by extrapolation from cellular data, it is recommended to minimise diagnostic and therapeutic ionising radiation when there is an alternative and to weigh the risk in conditioning regimens for transplantation [15, 40]. In SCAN1, camptothecins (irinotecan, topotecan), bleomycin and radiation are advised against because they may be fatal; application to the AOAs is an extrapolation via the common repair pathway [71]. In patients with neuropathy, neurotoxic drugs should be avoided, especially vincristine [15]. Valproate is contraindicated in POLG mitochondrial disease, a differential to be excluded before using it in an ataxia with epilepsy [116]. None of these warnings corresponds to an AOA-specific guideline recommendation.

7.12. Pregnancy and lactation

The sources consulted do not describe pregnancies in women with AOA, and the targeted Europe PMC search found no reports; this absence does not demonstrate that none exist. In 2018, GeneReviews stated that no offspring of patients with AOA2 were known [6]. The premature ovarian failure, hypogonadotrophic hypogonadism and male infertility described in AOA2 justify fertility assessment before family planning [6, 62]. Symptomatic drugs require review before conception: valproate has teratogenicity flagged in a boxed warning in the North American label [115]. There are no data on obstetric anaesthesia; the risks of respiratory weakness and bulbar dysfunction described in the surgical context apply by extension to delivery planning (subsection 7.13) [84].

7.13. Perioperative care and anaesthesia

The only published description of general anaesthesia for laparoscopic surgery in AOA2 is that of a 46-year-old woman undergoing cholecystectomy: uncertainty about sensitivity to neuromuscular blockers and the risk of respiratory weakness and bulbar dysfunction led to the choice of titratable and reversible drugs — reduced-dose rocuronium with quantitative neuromuscular monitoring and complete reversal with sugammadex before extubation, propofol, sevoflurane and remifentanil by target-controlled infusion —, with recovery without residual weakness [84]. The authors noted failures in the preoperative identification and communication of the disease [84]. The abstract does not state the doses. The preoperative assessment should include swallowing and aspiration risk and, if there are symptoms, pulmonary function tests and a sleep study [28]. In patients with PNKP variants, radiation exposure should be weighed in imaging planning (subsection 7.11) [15].

7.14. Transition to adult care

Most patients with AOA1 and AOA4 reach adolescence already in a wheelchair, and AOA2 often begins in adolescence [1, 6, 7]. Transition should ensure continuity of rehabilitation, surveillance of scoliosis, swallowing and biomarkers, genetic counselling directed at the patient in adulthood and access to a specialist ataxia centre, which the Ataxia UK guideline recommends for all patients with progressive ataxia (GPP) [1, 37]. In AOA2, transition is also the moment to address fertility [62].

7.15. Guidelines and divergences

8. Reproductive Options and Family Counselling

8.1. Recurrence risks in the autosomal recessive forms

In the AOAs, the parents of a patient are usually obligate heterozygotes, asymptomatic and with no known risk of disease [1, 6]. In each pregnancy of the couple, the risk is 25% of an affected child, 50% of an asymptomatic carrier and 25% of a non-carrier; a clinically unaffected sibling has a 2/3 probability of being a carrier [1]. The children of a patient are obligate heterozygotes, and the proband's aunts and uncles have a 50% probability of being carriers [1, 6]. Counselling should take place before testing and include the reproductive implications and the offer of counselling to at-risk relatives (GPP) [37].

8.2. Situations that alter the classic risk

De novo variant. When one of the variants arises de novo, as in the Brazilian patient with AOA2 and de novo c.6106+1G>T, only one parent is a carrier and the recurrence risk for the parents is much lower than 25%, limited by the possibility of germline mosaicism [66]. For this reason, both variants should be tested in both parents.

Populations with founder alleles or consanguinity. When a patient has children with a carrier of the same allele, the risk for the offspring rises to 50% (pseudodominance) [71]. This situation is plausible in Portugal (APTX p.(Trp279Ter), PNKP p.(Gly375Trp) and SETX p.(Arg1363Ter), the latter also described in Cape Verde and Spain [14]), in Quebec (SETX p.(Leu1976Arg), with a carrier frequency of 1 in 28 among Anglo-Normans and 1 in 47 among French Canadians of the Gaspé), in Egypt (PNKP p.(Asn461dup)) and in Costa Rica (PNKP p.(Gln517Ter)) [1, 6, 41, 43]. Consanguinity was present in 4 of the 8 Portuguese families with AOA4 [7].

Two inheritance patterns in SETX. Heterozygotes for AOA2 loss-of-function alleles are asymptomatic according to the sources consulted, whereas certain SETX missense variants cause dominant ALS4, with a 50% risk of transmission [6, 46]. The abstract of Chen and colleagues does not address heterozygotes for AOA2 alleles, so a risk of ALS4 in these carriers can be neither affirmed nor excluded on that basis [46]. A heterozygous SETX missense variant in a patient with motor neuropathy or ataxia requires interpretation in the light of these two mechanisms.

8.3. Cascade testing and carrier testing

Once the familial variants have been identified, carrier testing can be offered to adult siblings and to partners of carriers, particularly in populations with founder alleles [1, 6]. In AOA2, carrier status is determined by molecular testing and not by AFP, which was increased in 5 of 8 asymptomatic heterozygotes in one series and normal in another family [9, 14]. The ideal time for counselling on reproductive risk is before pregnancy, and it should be offered to young adults who are affected, carriers or at risk [1]. If the molecular diagnosis is not confirmed, DNA banking from the proband is recommended [1].

8.4. Fertility

In AOA2, spermatogenesis was impaired in the 3 men assessed, and oligospermia or azoospermia may precede the neurological signs; SETX should be considered in a young man with azoospermia and mild ataxia, and fertility assessment should be part of counselling [62]. Women with AOA2 may have ovarian failure, early menopause or hypogonadotrophic hypogonadism [6].

8.5. Prenatal and preimplantation diagnosis

Prenatal diagnosis and preimplantation genetic testing are possible once both of the family's pathogenic variants have been identified [1, 6]. Perspectives on prenatal diagnosis vary between professionals and families, and the decision rests with the families after non-directive information [1]. In MCSZ, microcephaly may be detected on prenatal ultrasound [15].

8.6. Testing of minors

Predictive testing of asymptomatic minors is not recommended as a rule and should be decided case by case (GPP) [37]. In the AOAs, early onset and the absence of disease-modifying treatment make testing of asymptomatic siblings a decision to be weighed with the family; when a sibling has signs, the test is diagnostic and not predictive.

9. Special Considerations: Population Diversity and Epidemiology

9.1. Nature of the data

The epidemiological data on the AOAs are of three types: population prevalences obtained by systematic surveys (Portugal, Alsace), proportions of each gene in cohorts of patients with ataxia (most studies) and carrier frequencies of founder alleles. These figures are not interchangeable: a proportion in a cohort depends on the selection criteria and the genes tested, and a carrier frequency is not a disease prevalence.

9.2. Portugal

The population-based national survey carried out between 1 January 1994 and 15 April 2004, in a population of 10.322 million, identified 1336 patients with hereditary cerebellar ataxia or hereditary spastic paraparesis (12.9/100,000); hereditary cerebellar ataxia had a prevalence of 8.9/100,000 (5.6 dominant and 3.3 recessive), and the most frequent forms were Machado-Joseph disease (3.1/100,000), Friedreich ataxia (1.0/100,000) and AOA (0.4/100,000), with no genetic cause identified in 39.7% [128]. The archived GeneReviews chapter, based on the earlier phase of the same survey, describes 42 patients with AOA in 20 families (12.6% of recessive ataxias), with AOA1 in 3.6% and AOA2 in 3.3% of recessive ataxias and an AOA prevalence of 0.41/100,000 [1, 129]. The district pilot study had estimated the sensitivity of the survey method at 81.2%, which means that these prevalences are minimum values [129]. At an intermediate stage, in 2001, AOA accounted for 21% of 107 patients with recessive ataxia, after Friedreich ataxia (38%) [5].

AOA4 is the most frequent form of AOA in Portugal: the 8 families with PNKP variants accounted for about 40% of the AOA families identified in the national survey, and AOA4 was described as the most frequent recessive ataxia in Portugal after Friedreich ataxia [7, 64]. All Portuguese families with AOA1 share the p.(Trp279Ter) variant, on a founder haplotype from the north of the country, and 6 of the 8 families with AOA4 have at least one p.(Gly375Trp) allele [1, 5, 7, 13]. In AOA2, the SETX variant p.(Arg1363Ter) was found in homozygosity in families from Portugal, Cape Verde and Spain, suggesting an Iberian founder event [14]. The 2025 review places AOA due to APTX, SETX and PNKP as the most frequent recessive ataxia in Portugal after Friedreich ataxia and records proportions of 1.4% for APTX and 1.2–5.3% for SETX in Portuguese ataxia cohorts [10]. The specific prevalence of each form in Portugal is not published in the sources consulted.

9.3. Rest of Europe

In France, AOA1 was found in 9 of 158 families with progressive non-Friedreich recessive ataxia (5%; 9.1% in the subgroup with onset before 25 years) [1, 29], and AOA2 in 6 of 77 families (about 8%), more frequent than A-T and AOA1 in adults; the authors consider it the most frequent cause of recessive ataxia in adults after Friedreich ataxia [31]. In Alsace, among 102 patients with suspected recessive ataxia studied between 2002 and 2008, a molecular diagnosis was made in 57: Friedreich ataxia in 36, AOA2 in 7 (6.9% of the total), A-T in 4, AOA1 in 3 and Marinesco-Sjögren syndrome in 3 [5, 30]. The prevalence of AOA2 inferred from that study is cited in two ways: slightly less than 1 in 400,000 in GeneReviews and 1 in 900,000 in OMIM, a value close to the point prevalence of 0.11/100,000 in France recorded by Orphanet; the original abstract states only that AOA2 is slightly less frequent than A-T, estimated at about one eighth of Friedreich ataxia (1 in 50,000) [6, 9, 26, 30]. In Italy, the APTX variant p.(Trp279Ter) was the most frequent, in homozygosity in 7 of the 13 patients reported by Castellotti and colleagues and in compound heterozygosity in another [5]. In aggregated European cohorts, APTX accounts for 1.6% and SETX for 4% of ataxias; in Italy APTX reaches 6.4–11.1%, in France 1.1–6.2%, in Germany 1.2–1.8% and in Ireland 0.5%, and SETX 1–15.1% in France, 2.4–8% in Germany, 2.3% in Norway and 1.5–4.1% in Spain [10]. Outside Portugal, AOA4 has been described in Norway, Sweden, the Netherlands, Italy and Belarus [64, 67, 68, 72, 76]. The mean worldwide prevalence of recessive ataxias has been estimated at 3.3/100,000, higher in France, Portugal and Canada, which may reflect methodological differences [10].

9.4. North America

In Quebec, the SETX variant p.(Leu1976Arg) was present on 17 of 20 chromosomes from 24 patients in 10 families with AOA2 without oculomotor apraxia [70]. Among 154 controls from the Gaspé, the carrier frequency was 3.5% (1 in 28) among Anglo-Normans and 2.1% (1 in 47) among French Canadians, with no homozygotes [6]. In the United States and Canada in general, APTX accounts for 0.6% and SETX for 1.1–1.6% of ataxia cohorts [10]. ALS4 was defined in an 11-generation family from southern Maryland and neighbouring states, with 17th-century English ancestors, and the dominant ataxia due to a SETX cis allele in an African American family [9, 44]. The available data do not allow population prevalence in the United States to be estimated.

9.5. Latin America

In Brazil, in the cohort of 52 patients with an AOA phenotype reported by da Costa and colleagues, SETX was the most frequent gene (15 patients), followed by PNKP (12) and APTX (5), with no cases of PIK3R5 or XRCC1; AOA2 and AOA4 are the most common forms in the country [12]. In Brazilian ataxia cohorts, APTX accounts for 1–1.3% and SETX for 1.3–3.2% [10]. The Novis review describes AOA4 as particularly prevalent in Portugal and Brazil [4]. In Costa Rica, CMT2B2 results from a PNKP p.(Gln517Ter) founder haplotype, in homozygosity in the original family and in compound heterozygosity with p.(Thr408del) in five additional probands [16, 43]. In Brazil, the same p.(Thr408del) was found in homozygosity in a boy with CMT2B2 born to non-consanguineous parents [42]. In Argentina, APTX and SETX account for 0.6% each [10].

9.6. Asia, Africa and the Middle East

In Japan, AOA1 was considered the most frequent cause of recessive ataxia in studies from 1992 to 2003, with the founder variant c.689dup [1, 47]; in more recent Japanese cohorts, APTX accounts for 1.3–1.6% and SETX for 0.3–1.6% [10]. In China, SETX accounts for 5.6–15.4%; in India 2.6–6.1% (6.1% of non-Friedreich recessive ataxias); in South Korea APTX accounts for about 3% of recessive ataxias; in Iran APTX 3.1%; in Turkey APTX 0.1–6% and SETX 0.8–4%; in Algeria APTX 3.6% and SETX 10.8% [10]. In Egypt, the PNKP variant p.(Asn461dup) has a founder effect in MCSZ [41]. AOA3 was described in a family from Saudi Arabia, and SCAN1 in one Saudi and two Omani families [11, 71]. In Tunisia, AOA1 was described in three families, one of them with a complete deletion of APTX, and AOA2 in 13 patients from 5 consanguineous families; in Algeria, the SETX variants p.Gln868Ter and p.(Glu343Ter) caused AOA2 in consanguineous families [5, 9, 14]. ATLD1 due to MRE11 was described in 10 patients from 3 Saudi families with the same variant [90]. Patients with AOA1 have also been described in Tunisia, Germany, Italy and the United States [1].

9.7. Table of frequencies by country

Source of the proportions: [10]. The values are proportions in cohorts of patients with ataxia, not population prevalences.

9.8. Gaps

There is no published population prevalence for AOA1 (Orphanet classifies it as unknown), for AOA5 or for the PNKP spectrum outside Portugal [15, 25]. Orphanet records for AOA4 the class <1/1,000,000 and 12 validated cases or families, a number that reflects the 2015 and 2016 articles and is lower than the more than 40 patients with PNKP variants published by 2019 [27, 76]. The relative order of the forms depends on the population: in the Oxford Desk Reference, AOA1 is presented as the least common of A-T, AOA2 and AOA1, which contradicts the Portuguese and Japanese situation [88]. The absence of data in a region does not demonstrate the absence of patients.

9.9. Special populations

In young children, the physiologically raised AFP up to 2 years of age limits the usefulness of the marker [85]. In adults with late onset (AOA1 with missense variants, adult-onset AOA2, CMT2B2, AOA5), diagnosis is often delayed [1, 8, 43]. Women with AOA2 may have premature ovarian insufficiency, and men infertility [6, 62].

9.10. European networks and patient organisations

The relevant European Reference Network is ERN-RND (European Reference Network for Rare Neurological Diseases), ataxia and hereditary spastic paraparesis group, which provides the diagnostic flowcharts, the translated SARA scale and the Ataxia UK and NEUROGED guidelines (https://www.ern-rnd.eu/disease-knowledge-hub/ataxia/); in 2026 the network included 64 centres and 4 affiliated partners in 24 Member States [121]. For MCSZ/DEE10, EpiCARE (rare epilepsies) is relevant, and for CMT2B2 and ALS4, EURO-NMD (neuromuscular diseases). Patient organisations include the Portuguese Association of Hereditary Ataxias (APAHE, https://www.apahe.pt/), the Brazilian Association of Hereditary and Acquired Ataxias (ABAHE, https://abahe.org.br/), Euro-Ataxia, the European federation of associations (https://www.euroataxia.org/), Ataxia UK (https://www.ataxia.org.uk/) and the National Ataxia Foundation in the United States (https://www.ataxia.org/), which offers free membership, regional and virtual support groups, webinars and information on ongoing research [1, 61, 130]. The MDSGene resource on genetic ataxias (https://www.mdsgene.org/ataxia.html) brings together data by gene and by country [89].

10. Illustrative Clinical Case

10.1. AOA4 with chorea, without clinical oculomotor apraxia and with slow progression

Paucar and colleagues described in 2016 a 28-year-old Swedish woman with biallelic PNKP variants [64]. From 5 years of age she had dystonia and chorea of insidious onset, documented on video, which regressed spontaneously over time. Learning difficulties were noted in primary school; at 12 years, an assessment concluded that she met criteria for pervasive developmental disorder, which led to attendance at a special school. From that age a progressive ataxia was documented. At 15 years progressive weakness and areflexia appeared, and nerve conduction studies at 17 years showed diffuse sensorimotor polyneuropathy, more severe in the lower limbs, which progressed with finger contractures, distal amyotrophy and bilateral foot drop requiring orthoses. She had used a wheelchair since 25 years of age and walked only with support. At the most recent examination, the SARA score was 22.5 and the INAS score 5, with lower-limb weakness and complete loss of proprioception and vibration sense in the feet. Oculomotor examination showed saccadic pursuit and nystagmus, without oculomotor apraxia, and video-oculography revealed increased saccadic latency [64].

Four years before publication she had undergone gastric banding for obesity, with a reduction in body mass index from 39.4 to 30.1 kg/m². Before surgery she already had hypoalbuminaemia and hypercholesterolaemia; two years before publication a mild elevation of AFP was found, with normal immunoglobulins. Simvastatin was started and a protein-enriched diet was recommended. Brain MRI at 17 and 28 years showed progressive cerebellar atrophy, without brainstem atrophy (figure 1 of the article: T1-weighted midsagittal MRI at 28 years, https://pmc.ncbi.nlm.nih.gov/articles/PMC4817910/figure/F1/) [64].

The investigation followed a sequence that illustrates the diagnostic difficulties of the class. The polyglutamine expansion spinocerebellar ataxias and Friedreich ataxia were excluded first; the colony radiosensitivity assay and the ATM protein Western blot excluded A-T; sequencing and MLPA analysis of APTX and SETX excluded AOA1 and AOA2. Exome sequencing of the patient and her parents identified two novel compound heterozygous PNKP variants, c.1196T>C p.(Leu399Pro) and c.1385G>C p.(Arg462Pro), both in the kinase domain, confirmed by Sanger sequencing and predicted to be pathogenic by in silico analysis; Epstein-Barr virus-transformed lymphocytes had reduced levels of PNKP protein [64].

The authors highlighted three new features in AOA4: chorea, absence of oculomotor apraxia and slower progression, with 20 years to wheelchair use versus a mean of 13 years in the Portuguese cohort [64]. In ClinVar, in October 2026, the variant p.(Leu399Pro) was classified as pathogenic or likely pathogenic, and p.(Arg462Pro) had conflicting classifications [56]. The metabolic management in the report (simvastatin, high-protein diet, bariatric surgery) corresponds to an individual decision and not to a validated protocol for AOA4; the diets proposed by GeneReviews for hypoalbuminaemia and hypercholesterolaemia have the same level of evidence [1].

10.2. Lessons from the case

The case shows that the absence of oculomotor apraxia does not exclude an AOA, that childhood dystonia and chorea that regress may precede the ataxia by years, that hypoalbuminaemia, hypercholesterolaemia, mildly raised AFP and obesity form a profile suggestive of PNKP, and that the successive exclusion of single genes delayed the diagnosis, which trio exome sequencing resolved. Figure 1 of the article by Bras and colleagues (pedigrees of the eight Portuguese families and position of the AOA4 variants in the kinase domain and of the MCSZ variants in other domains, https://pmc.ncbi.nlm.nih.gov/articles/PMC4375449/figure/fig1/) complements the reading of the case [7].

11. Prognosis and Long-Term Follow-up

11.1. Prognosis

The AOAs are slowly progressive disorders, with long survival and severe motor disability. In AOA1, independent walking is lost 7 to 10 years after onset, a wheelchair becomes necessary between 15 and 20 years of age, and disease duration in the Portuguese families reached 58 years, with a maximum age at last observation of 68 years (and a Japanese patient who died at 71) [1]. In Barbot's Portuguese series, symptom duration reached 58 years, which led the authors to include long survival with severe disability among the criteria for the disease [49]. In AOA2, a wheelchair is needed on average 15.9 ± 8.6 years after onset, at about 30 years of age; the maximum age recorded at last observation was 79 years [6, 32]. Later onset and the presence of oculomotor apraxia are associated with slower progression, and strabismus with faster progression [35]. In AOA4, walking is lost 7 to 21 years after onset and a wheelchair is needed between 14 and 25 years of age, with a mean time of about 13 years [7, 64]; there are slower forms, such as the Swedish patient (20 years) and the Portuguese sisters who walked without support at 52 and 58 years [64, 75]. In CMT2B2, onset is in the third decade [43]. In ALS4, progression is slow and life expectancy normal [46]. In MCSZ, survival with 36 years of follow-up is documented [77].

Survival data come from series of published patients, with publication bias, and there are no comparative survival studies or data on treated disease showing modification of the course. The cause of death is rarely described in the sources consulted; in AOA2, the only published autopsy is that of a 79-year-old woman who died of heart failure [6]. None of the systematically studied PNKP cohorts showed cancer predisposition or immunodeficiency, with isolated published exceptions [40, 72, 78, 79].

11.2. Initial assessment after diagnosis

After diagnosis, the following are recommended: complete neurological examination (cranial nerves, gait, coordination, tone, strength, reflexes and sensation) with SARA as a baseline, cognitive assessment (including the CCAS scale), ophthalmological examination, EMG and nerve conduction studies, MRI, physiotherapy and occupational therapy assessment, speech and language therapy and swallowing assessment, pulmonary function tests and a sleep study if there are symptoms, assessment of scoliosis and pes cavus, measurement of albumin, cholesterol and AFP, and a genetics clinic consultation [1, 6, 28].

11.3. Surveillance table

There are no AOA-specific surveillance guidelines, nor published surveillance guidelines for the hereditary ataxias in general [28]. The intervals in the table follow the generic GeneReviews recommendation for the ataxias (annual assessment) and, for biomarkers, the protocol of the French trial, which is a research model and not a guideline recommendation [28, 103].

12. Summary and Final Considerations

The ataxias with oculomotor apraxia are autosomal recessive cerebellar ataxias caused by defects in DNA single-strand break repair (APTX, PNKP, XRCC1) or in R-loop resolution (SETX). They share onset in childhood or adolescence, cerebellar atrophy, an axonal neuropathy that eventually dominates the disability and, in some patients, the difficulty in initiating voluntary saccades that gives the group its name. They are separated from ataxia-telangiectasia by the absence of telangiectasias, immunodeficiency and established cancer predisposition, and from one another by age at onset and biomarkers: AOA1 and AOA4 begin in early childhood, with initial chorea or dystonia and late hypoalbuminaemia and hypercholesterolaemia; AOA2 begins in adolescence, with raised AFP, pyramidal signs and hypogonadism.

Oculomotor apraxia is absent in a considerable proportion of patients, so cerebellar ataxia with sensorimotor neuropathy suffices to request AFP, albumin, cholesterol and an ataxia panel with CNV analysis or an exome or genome, after Friedreich ataxia has been excluded. AFP is the biomarker with the greatest discriminative value, with thresholds of 7–15 µg/L in AOA1, 15–65 µg/L in AOA2 and above 65 µg/L in A-T. The albumin threshold should be read as 3.8 g/dL (38 g/L), and not in the unit printed in the archived GeneReviews chapter. Reports should use the MANE transcripts (APTX NM_001195248.2, SETX NM_015046.7, PNKP NM_007254.4), and the older notations of some PNKP variants require verified conversion.

AOA4 is the most frequent form of AOA in Portugal, with the founder variant PNKP c.1123G>T p.(Gly375Trp), and AOA2 and AOA4 predominate in Brazil; AOA2 is the most frequent worldwide. The PNKP gene produces a continuum ranging from microcephaly with neonatal epilepsy to adult-onset axonal neuropathy, explained by the degree of loss of the phosphatase and kinase activities. SETX has two inheritance patterns, recessive in AOA2 and dominant in ALS4. AOA3 due to PIK3R5 rests on a single family and has been classified as disputed by ClinGen.

There is no approved disease-modifying treatment. Follow-up combines rehabilitation, educational support, metabolic correction, symptomatic treatment guided by the Ataxia UK guidelines and the NEUROGED recommendations, consideration of CoQ10 in AOA1 with level D evidence, annual surveillance with SARA, caution with ionising radiation and neurotoxic drugs, and genetic counselling that takes into account founder alleles, de novo variants and the infertility of AOA2. Research is at the preclinical stage, with PARP1 as a target in XRCC1 and PNKP defects, and future trials will depend on early genetic diagnosis.

13. References

1. Coutinho P, Barbot C. Ataxia with Oculomotor Apraxia Type 1 – RETIRED CHAPTER, FOR HISTORICAL REFERENCE ONLY. 2002 Jun 11 [updated 2015 Mar 19; retired 2024 May 2]. In: Adam MP, Bick S, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. PMID: 20301629. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1456/
2. Date H, Onodera O, Tanaka H, Iwabuchi K, Uekawa K, Igarashi S, et al. Early-onset ataxia with ocular motor apraxia and hypoalbuminemia is caused by mutations in a new HIT superfamily gene. Nat Genet. 2001;29(2):184-8. doi:10.1038/ng1001-184. PMID: 11586299.
3. Rossi M, Anheim M, Durr A, Klein C, Koenig M, Synofzik M, et al. The genetic nomenclature of recessive cerebellar ataxias. Mov Disord. 2018;33(7):1056-76. doi:10.1002/mds.27415. PMID: 29756227.
4. Novis LE, Rezende Filho FM, Gomes AM, Procaci VR, Barsottini OGP, Pedroso JL. Deconstructing ataxia with oculomotor apraxia: toward a gene-based classification and a reappraisal of the oculomotor phenotype. Mov Disord Clin Pract. 2026 Oct 2 [Epub ahead of print]. doi:10.1002/mdc3.70846. PMID: 42827272. PMCID: PMC13633394.
5. OMIM® Online Mendelian Inheritance in Man. #208920 ATAXIA, EARLY-ONSET, WITH OCULOMOTOR APRAXIA AND HYPOALBUMINEMIA; EAOH. Johns Hopkins University; last edited 12/12/2024. https://omim.org/entry/208920. Accessed 29 October 2026.
6. Moreira MC, Koenig M. Ataxia with Oculomotor Apraxia Type 2. 2004 Nov 15 [updated 2018 Jul 12]. In: Adam MP, Bick S, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1154/
7. Bras J, Alonso I, Barbot C, Costa MM, Darwent L, Orme T, et al. Mutations in PNKP cause recessive ataxia with oculomotor apraxia type 4. Am J Hum Genet. 2015;96(3):474-9. doi:10.1016/j.ajhg.2015.01.005. PMID: 25728773. PMCID: PMC4375449.
8. Fogel BL, Klopstock T, Lynch DR, Maltecca F, Verma M, Minassian BA, et al. Autosomal recessive cerebellar ataxias: translating genes to therapies. Ann Neurol. 2025;98(3):448-70. doi:10.1002/ana.27271. PMID: 40464291. PMCID: PMC12392066.
9. OMIM® Online Mendelian Inheritance in Man. #606002 SPINOCEREBELLAR ATAXIA, AUTOSOMAL RECESSIVE, WITH AXONAL NEUROPATHY 2; SCAN2. Johns Hopkins University; last edited 12/12/2024. https://omim.org/entry/606002. Accessed 29 October 2026.
10. Rossi M, Stephen CD, Damásio J, Pedroso JL, Kuo SH, Lin CR, et al. Unravelling the global tapestry of genetic ataxias: epidemiology and genetic testing approaches. Mov Disord. 2025;40(9):1805-20. doi:10.1002/mds.30302. PMID: 40682316. PMCID: PMC12291621.
11. Al Tassan N, Khalil D, Shinwari J, Al Sharif L, Bavi P, Abduljaleel Z, et al. A missense mutation in PIK3R5 gene in a family with ataxia and oculomotor apraxia. Hum Mutat. 2012;33(2):351-4. doi:10.1002/humu.21650. PMID: 22065524.
12. da Costa SCG, Rezende Filho FM, de Freitas JL, de Assis Pereira Matos PCA, Della-Ripa B, França MC, et al. Clinical and genetic characterization of Brazilian patients with ataxia and oculomotor apraxia. Mov Disord. 2022;37(6):1309-16. doi:10.1002/mds.29015. PMID: 35426160.
13. OMIM® Online Mendelian Inheritance in Man. *606350 APRATAXIN; APTX. Johns Hopkins University; last edited 24/09/2015. https://omim.org/entry/606350. Accessed 29 October 2026.
14. OMIM® Online Mendelian Inheritance in Man. *608465 SENATAXIN; SETX. Johns Hopkins University; last edited 01/03/2023. https://omim.org/entry/608465. Accessed 29 October 2026.
15. GeneGuidance. PNKP: gene and clinical condition curation [internal database; export of 26-09-2026].
16. OMIM® Online Mendelian Inheritance in Man. *605610 POLYNUCLEOTIDE KINASE 3-PRIME PHOSPHATASE; PNKP. Johns Hopkins University; last edited 08/10/2020. https://omim.org/entry/605610. Accessed 29 October 2026.
17. GeneGuidance. PIK3R5: gene and clinical condition curation [internal database; export of 26-09-2026].
18. OMIM® Online Mendelian Inheritance in Man. *611317 PHOSPHATIDYLINOSITOL 3-KINASE, REGULATORY SUBUNIT 5; PIK3R5. Johns Hopkins University; last edited 03/03/2021. https://omim.org/entry/611317. Accessed 29 October 2026.
19. GeneGuidance. XRCC1: gene and clinical condition curation [internal database; export of 26-09-2026].
20. OMIM® Online Mendelian Inheritance in Man. *194360 X-RAY REPAIR CROSS COMPLEMENTING 1; XRCC1. Johns Hopkins University; last edited 27/01/2022. https://omim.org/entry/194360. Accessed 29 October 2026.
21. Pedroso JL, Rocha CR, Macedo-Souza LI, De Mario V, Marques W, Barsottini OG, et al. Mutation in PNKP presenting initially as axonal Charcot-Marie-Tooth disease. Neurol Genet. 2015;1(4):e30. doi:10.1212/NXG.0000000000000030. PMID: 27066567.
22. OMIM® Online Mendelian Inheritance in Man. #616267 ATAXIA-OCULOMOTOR APRAXIA 4; AOA4. Johns Hopkins University; last edited 07/05/2020. https://omim.org/entry/616267. Accessed 29 October 2026.
23. OMIM® Online Mendelian Inheritance in Man. #615217 ATAXIA-OCULOMOTOR APRAXIA 3; AOA3. Johns Hopkins University; last edited 28/09/2017. https://omim.org/entry/615217. Accessed 29 October 2026.
24. OMIM® Online Mendelian Inheritance in Man. #617633 SPINOCEREBELLAR ATAXIA, AUTOSOMAL RECESSIVE 26; SCAR26. Johns Hopkins University; last edited 24/03/2025. https://omim.org/entry/617633. Accessed 29 October 2026.
25. Orphanet. Ataxia-oculomotor apraxia type 1 (ORPHA:1168) [Internet]. Paris: INSERM; Orphadata epidemiological record of 2026-06-23 [accessed 2026-10-28]. Available from: https://www.orpha.net/en/disease/detail/1168
26. Orphanet. Spinocerebellar ataxia with axonal neuropathy type 2 (ORPHA:64753) [Internet]. Paris: INSERM; Orphadata epidemiological record of 2026-06-23 [accessed 2026-10-28]. Available from: https://www.orpha.net/en/disease/detail/64753
27. Orphanet. Ataxia-oculomotor apraxia type 4 (ORPHA:459033) [Internet]. Paris: INSERM; Orphadata epidemiological record of 2026-06-23 [accessed 2026-10-28]. Available from: https://www.orpha.net/en/disease/detail/459033
28. Perlman S. Hereditary Ataxia Overview. 1998 Oct 28 [updated 2022 Jun 16; revised 2026 Jul 9]. In: Adam MP, Bick S, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1138/
29. Le Ber I, Moreira MC, Rivaud-Péchoux S, Chamayou C, Ochsner F, Kuntzer T, et al. Cerebellar ataxia with oculomotor apraxia type 1: clinical and genetic studies. Brain. 2003;126(Pt 12):2761-72. doi:10.1093/brain/awg283. PMID: 14506070.
30. Anheim M, Fleury M, Monga B, Laugel V, Chaigne D, Rodier G, et al. Epidemiological, clinical, paraclinical and molecular study of a cohort of 102 patients affected with autosomal recessive progressive cerebellar ataxia from Alsace, Eastern France: implications for clinical management. Neurogenetics. 2010;11(1):1-12. doi:10.1007/s10048-009-0196-y. PMID: 19440741.
31. Le Ber I, Bouslam N, Rivaud-Péchoux S, Guimarães J, Benomar A, Chamayou C, et al. Frequency and phenotypic spectrum of ataxia with oculomotor apraxia 2: a clinical and genetic study in 18 patients. Brain. 2004;127(Pt 4):759-67. doi:10.1093/brain/awh080. PMID: 14736755.
32. Yang L, Yang S, Liu J, Lei X, He D. Ataxia with oculomotor apraxia type 2: two pedigree studies and a comprehensive review. J Neurol. 2025;272(10):668. doi:10.1007/s00415-025-13277-9. PMID: 41026212.
33. Hoch NC, Hanzlikova H, Rulten SL, Tétreault M, Komulainen E, Ju L, et al. XRCC1 mutation is associated with PARP1 hyperactivation and cerebellar ataxia. Nature. 2017;541(7635):87-91. doi:10.1038/nature20790. PMID: 28002403. PMCID: PMC5218588.
34. European Reference Network for Rare Neurological Diseases (ERN-RND), Disease Group Ataxia and Hereditary Spastic Paraplegias. Diagnostic flowchart for early-onset ataxias. Consented July 2020. Available from: https://www.ern-rnd.eu/wp-content/uploads/2020/11/ERN-RND-Diagnostic-flowchart-for-early-Ataxias_final.pdf
35. Anheim M, Monga B, Fleury M, Charles P, Barbot C, Salih M, et al. Ataxia with oculomotor apraxia type 2: clinical, biological and genotype/phenotype correlation study of a cohort of 90 patients. Brain. 2009;132(Pt 10):2688-98. doi:10.1093/brain/awp211. PMID: 19696032.
36. ClinGen. Gene-Disease Validity curations [CSV download of 2026-10-28]. Available from: https://search.clinicalgenome.org/kb/gene-validity
37. de Silva R, Greenfield J, Cook A, Bonney H, Vallortigara J, Hunt B, et al. Guidelines on the diagnosis and management of the progressive ataxias. Orphanet J Rare Dis. 2019;14(1):51. doi:10.1186/s13023-019-1013-9. PMID: 30786918. PMCID: PMC6381619.
38. OMIM® Online Mendelian Inheritance in Man. #613402 MICROCEPHALY, SEIZURES, AND DEVELOPMENTAL DELAY; MCSZ. Johns Hopkins University; last edited 14/09/2022. https://omim.org/entry/613402. Accessed 29 October 2026.
39. Orphanet. Oculomotor apraxia (ORPHA:98688, group of disorders) [Internet]. Paris: INSERM [accessed 2026-10-28]. Available from: https://www.orpha.net/en/disease/detail/98688
40. Shen J, Gilmore EC, Marshall CA, Haddadin M, Reynolds JJ, Eyaid W, et al. Mutations in PNKP cause microcephaly, seizures and defects in DNA repair. Nat Genet. 2010;42(3):245-9. doi:10.1038/ng.526. PMID: 20118933. PMCID: PMC2835984.
41. Abdel-Salam GMH, Abdel-Hamid MS, Abdel-Ghafar SF, Girgis M, Zaki MS. Refining the phenotypic spectrum of PNKP-related microcephaly: a study of 27 new patients. J Med Genet. 2026;63(4):250-8. doi:10.1136/jmg-2025-111040. PMID: 41436176.
42. OMIM® Online Mendelian Inheritance in Man. #605589 CHARCOT-MARIE-TOOTH DISEASE, AXONAL, TYPE 2B2; CMT2B2. Johns Hopkins University; last edited 01/12/2020. https://omim.org/entry/605589. Accessed 29 October 2026.
43. Leal A, Bogantes-Ledezma S, Ekici AB, Uebe S, Thiel CT, Sticht H, et al. The polynucleotide kinase 3'-phosphatase gene (PNKP) is involved in Charcot-Marie-Tooth disease (CMT2B2) previously related to MED25. Neurogenetics. 2018;19(4):215-25. doi:10.1007/s10048-018-0555-7. PMID: 30039206. PMCID: PMC6280876.
44. OMIM® Online Mendelian Inheritance in Man. #602433 AMYOTROPHIC LATERAL SCLEROSIS 4, JUVENILE; ALS4. Johns Hopkins University; last edited 19/10/2023. https://omim.org/entry/602433. Accessed 29 October 2026.
45. GeneGuidance. SETX: gene and clinical condition curation [internal database; export of 26-09-2026].
46. Chen YZ, Bennett CL, Huynh HM, Blair IP, Puls I, Irobi J, et al. DNA/RNA helicase gene mutations in a form of juvenile amyotrophic lateral sclerosis (ALS4). Am J Hum Genet. 2004;74(6):1128-35. doi:10.1086/421054. PMID: 15106121.
47. Moreira MC, Barbot C, Tachi N, Kozuka N, Uchida E, Gibson T, et al. The gene mutated in ataxia-ocular apraxia 1 encodes the new HIT/Zn-finger protein aprataxin. Nat Genet. 2001;29(2):189-93. doi:10.1038/ng1001-189. PMID: 11586300.
48. Aicardi J, Barbosa C, Andermann E, Andermann F, Morcos R, Ghanem Q, et al. Ataxia-ocular motor apraxia: a syndrome mimicking ataxia-telangiectasia. Ann Neurol. 1988;24(4):497-502. doi:10.1002/ana.410240404. PMID: 3239952.
49. Barbot C, Coutinho P, Chorão R, Ferreira C, Barros J, Fineza I, et al. Recessive ataxia with ocular apraxia: review of 22 Portuguese patients. Arch Neurol. 2001;58(2):201-5. doi:10.1001/archneur.58.2.201. PMID: 11176957.
50. Moreira MC, Klur S, Watanabe M, Németh AH, Le Ber I, Moniz JC, et al. Senataxin, the ortholog of a yeast RNA helicase, is mutant in ataxia-ocular apraxia 2. Nat Genet. 2004;36(3):225-7. doi:10.1038/ng1303. PMID: 14770181.
51. OMIM® Online Mendelian Inheritance in Man. #607250 SPINOCEREBELLAR ATAXIA, AUTOSOMAL RECESSIVE, WITH AXONAL NEUROPATHY 1; SCAN1. Johns Hopkins University; last edited 15/04/2019. https://omim.org/entry/607250. Accessed 29 October 2026.
52. Beaudin M, Matilla-Dueñas A, Soong BW, Pedroso JL, Barsottini OG, Mitoma H, et al. The classification of autosomal recessive cerebellar ataxias: a consensus statement from the Society for Research on the Cerebellum and Ataxias Task Force. Cerebellum. 2019;18(6):1098-125. doi:10.1007/s12311-019-01052-2. PMID: 31267374.
53. Firth HV, Hurst JA. Ataxic child. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 74–77.
54. Yoon G, Caldecott KW. Nonsyndromic cerebellar ataxias associated with disorders of DNA single-strand break repair. Handb Clin Neurol. 2018;155:105-15. doi:10.1016/B978-0-444-64189-2.00007-X. PMID: 29891053.
55. Freeman PJ, Hart RK, Gretton LJ, Brookes AJ, Dalgleish R. VariantValidator: accurate validation, mapping, and formatting of sequence variation descriptions. Hum Mutat. 2018;39(1):61-8. doi:10.1002/humu.23348. PMID: 28967166. PMCID: PMC5765404. Ferramenta: https://variantvalidator.org (consultada 2026-10-28).
56. National Center for Biotechnology Information. ClinVar. Bethesda (MD): NCBI. Available from: https://www.ncbi.nlm.nih.gov/clinvar/ (E-utilities accessed 2026-10-28; VCV 4431, 4426, 430092, 994780, 187766, 4846, 4847, 190219, 854429, 187768, 988277, 206414, 2289, 2290, 2291).
57. GeneGuidance. APTX: gene and clinical condition curation [internal database; export of 26-09-2026].
58. Firth HV, Hurst JA. DNA repair disorders. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 398.
59. Quinzii CM, Kattah AG, Naini A, Akman HO, Mootha VK, DiMauro S, et al. Coenzyme Q deficiency and cerebellar ataxia associated with an aprataxin mutation. Neurology. 2005;64(3):539-41. doi:10.1212/01.WNL.0000150588.75281.58. PMID: 15699391.
60. Le Ber I, Dubourg O, Benoist JF, Jardel C, Mochel F, Koenig M, et al. Muscle coenzyme Q10 deficiencies in ataxia with oculomotor apraxia 1. Neurology. 2007;68(4):295-7. doi:10.1212/01.wnl.0000252366.10731.43. PMID: 17242337.
61. Ataxia UK. Management of the ataxias: towards best clinical practice. 3rd ed. London: Ataxia UK; July 2016. Available from: https://www.ern-rnd.eu/wp-content/uploads/2023/07/Final-guideline-AtaxiaUK_english.pdf
62. Becherel OJ, Fogel BL, Zeitlin SI, Samaratunga H, Greaney J, Homer H, et al. Disruption of spermatogenesis and infertility in ataxia with oculomotor apraxia type 2 (AOA2). Cerebellum. 2019;18(3):448-56. doi:10.1007/s12311-019-01012-w. PMID: 30778901.
63. Kalasova I, Hailstone R, Bublitz J, Bogantes J, Hofmann W, Leal A, et al. Pathological mutations in PNKP trigger defects in DNA single-strand break repair but not DNA double-strand break repair. Nucleic Acids Res. 2020;48(12):6672-84. doi:10.1093/nar/gkaa489. PMID: 32504494. PMCID: PMC7337934.
64. Paucar M, Malmgren H, Taylor M, Reynolds JJ, Svenningsson P, Press R, et al. Expanding the ataxia with oculomotor apraxia type 4 phenotype. Neurol Genet. 2016;2(1):e49. doi:10.1212/NXG.0000000000000049. PMID: 27066586. PMCID: PMC4817910.
65. Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 35, section 35.5: Inborn errors of phosphoinositide phosphorylation (Table 35.1).
66. Fiorini AC, Scorza CA, Scorza FA, Finsterer J. Ataxia with oculomotor apraxia type 2 due to the new compound heterozygous variants c.5825T>C (inherited) and c.6106+1G>T (de novo) in SETX. Clinics (Sao Paulo). 2026;81:101146. doi:10.1016/j.clinsp.2026.101146. PMID: 42715900. PMCID: PMC13573666.
67. Rudenskaya GE, Marakhonov AV, Shchagina OA, Lozier ER, Dadali EL, Akimova IA, et al. Ataxia with oculomotor apraxia type 4 with PNKP common "Portuguese" and novel mutations in two Belarusian families. J Pediatr Genet. 2019;8(2):58-62. doi:10.1055/s-0039-1684008. PMID: 31061747. PMCID: PMC6499616.
68. Tzoulis C, Sztromwasser P, Johansson S, Gjerde IO, Knappskog P, Bindoff LA. PNKP mutations identified by whole-exome sequencing in a Norwegian patient with sporadic ataxia and edema. Cerebellum. 2017;16(1):272-5. doi:10.1007/s12311-016-0784-y. PMID: 27165045. PMCID: PMC5243888.
69. Renaud M, Moreira MC, Ben Monga B, Rodriguez D, Debs R, Charles P, et al. Clinical, biomarker, and molecular delineations and genotype-phenotype correlations of ataxia with oculomotor apraxia type 1. JAMA Neurol. 2018;75(4):495-502. doi:10.1001/jamaneurol.2017.4373. PMID: 29356829. PMCID: PMC5933354.
70. Duquette A, Roddier K, McNabb-Baltar J, Gosselin I, St-Denis A, Dicaire MJ, et al. Mutations in senataxin responsible for Quebec cluster of ataxia with neuropathy. Ann Neurol. 2005;57(3):408-14. doi:10.1002/ana.20408. PMID: 15732101.
71. Salih MAM, Takashima H, Boerkoel CF. Spinocerebellar Ataxia with Axonal Neuropathy Type 1. 2007 Oct 22 [updated 2022 Jun 30]. In: Adam MP, Bick S, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1105/
72. Garrelfs MR, Takada S, Kamsteeg EJ, Pegge S, Mancini G, Engelen M, et al. The phenotypic spectrum of PNKP-associated disease and the absence of immunodeficiency and cancer predisposition in a Dutch cohort. Pediatr Neurol. 2020;113:26-32. doi:10.1016/j.pediatrneurol.2020.07.014. PMID: 32980744.
73. Schiess N, Zee DS, Siddiqui KA, Szolics M, El-Hattab AW. Novel PNKP mutation in siblings with ataxia-oculomotor apraxia type 4. J Neurogenet. 2017;31(1-2):23-5. doi:10.1080/01677063.2017.1322079. PMID: 28552035.
74. Caputi C, Tolve M, Galosi S, Inghilleri M, Carducci C, Angeloni A, et al. PNKP deficiency mimicking a benign hereditary chorea: the misleading presentation of a neurodegenerative disorder. Parkinsonism Relat Disord. 2019;64:342-5. doi:10.1016/j.parkreldis.2019.03.012. PMID: 30956058.
75. Freitas E, Costa O, Rocha S. A new phenotype of ataxia with oculomotor apraxia type 4. Cureus. 2021;13(2):e13601. doi:10.7759/cureus.13601. PMID: 33654647. PMCID: PMC7914377.
76. Gatti M, Magri S, Nanetti L, Sarto E, Di Bella D, Salsano E, et al. From congenital microcephaly to adult onset cerebellar ataxia: distinct and overlapping phenotypes in patients with PNKP gene mutations. Am J Med Genet A. 2019;179(11):2277-83. doi:10.1002/ajmg.a.61339. PMID: 31436889.
77. Ragona F, Messina G, Magri S, Doniselli FM, Freri E, Canafoglia L, et al. Compound heterozygous PNKP variants causing developmental and epileptic encephalopathy with severe microcephaly: natural history of two new cases and literature review. NeuroSci. 2025;6(4):110. doi:10.3390/neurosci6040110. PMID: 41283299. PMCID: PMC12641809.
78. Takada S, Okano T, Tanita K, Tsukada K, Watanabe M, Hijikata A, et al. B-cell immunodeficiency associated with polynucleotide kinase 3'-phosphatase (PNKP) deficiency. J Allergy Clin Immunol Glob. 2025;4(3):100514. doi:10.1016/j.jacig.2025.100514. PMID: 40687947. PMCID: PMC12271857.
79. Bayram N, Yaman Y, Elli M, Ozdilli K, Nepesov S, Dogan MS, et al. HLA-matched related donor hematopoietic stem cell transplantation in a patient with polynucleotide kinase 3-phosphatase mutation developed acute myeloid leukemia. Pediatr Transplant. 2022;26(4):e14255. doi:10.1111/petr.14255. PMID: 35187769.
80. O'Connor E, Vandrovcova J, Bugiardini E, Chelban V, Manole A, Davagnanam I, et al. Mutations in XRCC1 cause cerebellar ataxia and peripheral neuropathy. J Neurol Neurosurg Psychiatry. 2018;89(11):1230-2. doi:10.1136/jnnp-2017-317581. PMID: 29472272. PMCID: PMC6227798.
81. Posa A, Kornhuber M. A very rare SETX gene variant (c.2750T>C) in a 72-year-old man with amyotrophic lateral sclerosis and an unremarkable family history. Should genetic testing be routinely performed in all patients? Eur J Case Rep Intern Med. 2025;12(10):005706. doi:10.12890/2025_005706. PMID: 41064708.
82. European Reference Network for Rare Neurological Diseases (ERN-RND). Diagnostic flowchart – Ataxias (adults). Revision February 2024; consent September 2024. Available from: https://www.ern-rnd.eu/wp-content/uploads/2024/09/Final-Flowchart-adult-ataxias_en-New.pdf
83. Mariani LL, Rivaud-Péchoux S, Charles P, Ewenczyk C, Meneret A, Monga BB, et al. Comparing ataxias with oculomotor apraxia: a multimodal study of AOA1, AOA2 and AT focusing on video-oculography and alpha-fetoprotein. Sci Rep. 2017;7(1):15284. doi:10.1038/s41598-017-15127-9. PMID: 29127364. PMCID: PMC5681651.
84. Steuart L, Millar S. General anaesthesia for laparoscopic cholecystectomy in a patient with ataxia with oculomotor apraxia type 2. Anaesth Rep. 2026;14(1):e70061. doi:10.1002/anr3.70061. PMID: 42459760.
85. Veenhuis S, van Os N, Weemaes C, Kamsteeg EJ, Willemsen M. Ataxia-Telangiectasia. 1999 Mar 19 [updated 2023 Oct 5; revised 2026 Sep 22]. In: Adam MP, Bick S, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK26468/
86. Renaud M, Tranchant C, Koenig M, Anheim M. Autosomal recessive cerebellar ataxias with elevated alpha-fetoprotein: uncommon diseases, common biomarker. Mov Disord. 2020;35(12):2139-49. doi:10.1002/mds.28307. PMID: 33044027.
87. van de Warrenburg BP, van Gaalen J, Boesch S, Burgunder JM, Dürr A, Giunti P, et al. EFNS/ENS consensus on the diagnosis and management of chronic ataxias in adulthood. Eur J Neurol. 2014;21(4):552-62. doi:10.1111/ene.12341. PMID: 24418350.
88. Firth HV, Hurst JA. DNA repair disorders. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 399.
89. MDSGene. Global Genetic Ataxia Resource. Available from: https://www.mdsgene.org/ataxia.html
90. OMIM® Online Mendelian Inheritance in Man. #604391 ATAXIA-TELANGIECTASIA-LIKE DISORDER 1; ATLD1. Johns Hopkins University; last edited 16/12/2019. https://omim.org/entry/604391. Accessed 29 October 2026.
91. OMIM® Online Mendelian Inheritance in Man. #615919 ATAXIA-TELANGIECTASIA-LIKE DISORDER 2; ATLD2. Johns Hopkins University; last edited 10/09/2024. https://omim.org/entry/615919. Accessed 29 October 2026.
92. Firth HV, Hurst JA. DNA repair disorders. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 400.
93. OMIM® Online Mendelian Inheritance in Man. *607198 TYROSYL-DNA PHOSPHODIESTERASE 1; TDP1. Johns Hopkins University; last edited 15/04/2019. https://omim.org/entry/607198. Accessed 29 October 2026.
94. Saudubray JM, García-Cazorla Á. Clinical approach to inborn errors of metabolism in paediatrics. In: Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 1 (section 1.5, Tables 1.29 and 1.30). doi:10.1007/978-3-662-63123-2_1
95. Firth HV, Hurst JA. Cerebellar anomalies. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 93.
96. OMIM® Online Mendelian Inheritance in Man. %257550 OCULAR MOTOR APRAXIA. Johns Hopkins University; last edited 14/10/2016. https://omim.org/entry/257550. Accessed 29 October 2026.
97. Nyhan WL, Hoffmann GF, Al-Aqeel AI, Barshop BA. Gaucher disease. In: Atlas of Inherited Metabolic Diseases. 4th ed. Boca Raton: CRC Press; 2020. p. 678–687 (p. 682).
98. Firth HV, Hurst JA. Ataxic adult. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 70–73.
99. Firth HV, Hurst JA. Epilepsy in infants and children. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 410–415 (table p. 412).
100. Zesiewicz TA, Wilmot G, Kuo SH, Perlman S, Greenstein PE, Ying SH, et al. Comprehensive systematic review summary: treatment of cerebellar motor dysfunction and ataxia: report of the Guideline Development, Dissemination, and Implementation Subcommittee of the American Academy of Neurology. Neurology. 2018;90(10):464-71. doi:10.1212/WNL.0000000000005055. PMID: 29440566. PMCID: PMC5863491. Document retired by the AAN (title marked "RETIRED" in PubMed).
101. Azienda Socio Sanitaria Territoriale degli Spedali Civili di Brescia. Rehabilitative trial with cerebello-spinal tDCS in neurodegenerative ataxia. ClinicalTrials.gov identifier: NCT03120013. Available from: https://clinicaltrials.gov/study/NCT03120013
102. Azienda Socio Sanitaria Territoriale degli Spedali Civili di Brescia. Cerebello-spinal tDCS as rehabilitative intervention in neurodegenerative ataxia. ClinicalTrials.gov identifier: NCT04153110. Available from: https://clinicaltrials.gov/study/NCT04153110
103. Assistance Publique – Hôpitaux de Paris. Evolution of albumin on AOA1 patients supplemented with coenzyme Q10. ClinicalTrials.gov identifier: NCT02333305. Available from: https://clinicaltrials.gov/study/NCT02333305
104. Ribeiro EM, coord. Deficiência de CoQ10. In: Síndromes Genéticas: Reconhecer para Tratar. Fortaleza: EdUnichristus; 2021. p. 127–129.
105. Romano S, Coarelli G, Marcotulli C, Leonardi L, Piccolo F, Spadaro M, et al. Riluzole in patients with hereditary cerebellar ataxia: a randomised, double-blind, placebo-controlled trial. Lancet Neurol. 2015;14(10):985-91. doi:10.1016/S1474-4422(15)00201-X. PMID: 26321318.
106. Sun Pharmaceutical Industries. Riluzole tablets, USP [prescribing information]. ANDA091417. DailyMed set id 4c851be4-f0d0-4b24-baf9-427ddfc6f7b0; version 10 (2020-05-16). Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=4c851be4-f0d0-4b24-baf9-427ddfc6f7b0
107. European Medicines Agency. Rilutek (riluzole): EPAR – product information (SmPC). Available from: https://www.ema.europa.eu/en/documents/product-information/rilutek-epar-product-information_en.pdf
108. INFARMED. Riluzol Aurovitas 50 mg comprimidos revestidos por película: Summary of Product Characteristics (approved 09-02-2023). Infomed med_id 45918. Available from: https://extranet.infarmed.pt/INFOMED-fo/download-ficheiro.xhtml?med_id=45918&tipo_doc=rcm
109. Patrin Pharma. Amantadine hydrochloride oral solution, USP [prescribing information]. ANDA214178. DailyMed set id 0fcf5182-a666-4f38-879f-47f19de30dd5; version 3 (2024-05-20). Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=0fcf5182-a666-4f38-879f-47f19de30dd5
110. INFARMED. Lioresal 10 mg e 25 mg comprimidos (baclofeno): Summary of Product Characteristics. Infomed med_id 5074. Available from: https://extranet.infarmed.pt/INFOMED-fo/download-ficheiro.xhtml?med_id=5074&tipo_doc=rcm
111. Northstar Rx LLC. Baclofen tablets, USP [prescribing information]. ANDA078401. DailyMed set id 0aac77fa-30c2-46c8-a04d-e6b8333ddd81; version 17 (2025-11-05). Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=0aac77fa-30c2-46c8-a04d-e6b8333ddd81
112. INFARMED. Neurontin 100/300/400 mg cápsulas (gabapentina): Summary of Product Characteristics. Infomed med_id 6057. Available from: https://extranet.infarmed.pt/INFOMED-fo/download-ficheiro.xhtml?med_id=6057&tipo_doc=rcm
113. Viatris Specialty LLC. Neurontin (gabapentin) [prescribing information]. NDA020235/NDA020882/NDA021129. DailyMed set id 97935fd9-1d4a-43b6-a5d9-de994591187b; version 48 (2026-05-12). Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=97935fd9-1d4a-43b6-a5d9-de994591187b
114. Organon LLC. Zocor (simvastatin) tablets [prescribing information]. NDA019766. DailyMed set id 8f55d5de-5a4f-4a39-8c84-c53976dd6af9; version 11 (2025-03-28). Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=8f55d5de-5a4f-4a39-8c84-c53976dd6af9
115. AbbVie Inc. Depakote (divalproex sodium) delayed-release tablets [prescribing information]. NDA018723. DailyMed set id 08a65cf4-7749-4ceb-6895-8f4805e2b01f; version 1621 (2026-03-31). Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=08a65cf4-7749-4ceb-6895-8f4805e2b01f
116. INFARMED. Depakine 200 mg/ml solução oral (valproato): Summary of Product Characteristics. Infomed med_id 2413. Available from: https://extranet.infarmed.pt/INFOMED-fo/download-ficheiro.xhtml?med_id=2413&tipo_doc=rcm
117. INFARMED. Castilium 10 mg comprimidos (clobazam): Summary of Product Characteristics (approved 10-01-2024). Infomed med_id 1494. Available from: https://extranet.infarmed.pt/INFOMED-fo/download-ficheiro.xhtml?med_id=1494&tipo_doc=rcm
118. Ascend Laboratories. Clobazam oral suspension [prescribing information]. ANDA213039. DailyMed set id 07e0ae51-cc71-427f-83cb-311b7fcc5993; version 12 (2026-07-09). Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=07e0ae51-cc71-427f-83cb-311b7fcc5993
119. European Medicines Agency. Keppra (levetiracetam): EPAR – product information (SmPC). Available from: https://www.ema.europa.eu/en/documents/product-information/keppra-epar-product-information_en.pdf
120. UCB Inc. Keppra (levetiracetam) tablets and oral solution [prescribing information]. NDA021035/NDA021505. DailyMed set id 3ca9df05-a506-4ec8-a4fe-320f1219ab21; version 36 (2026-01-28). Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=3ca9df05-a506-4ec8-a4fe-320f1219ab21
121. European Reference Network for Rare Neurological Diseases (ERN-RND). Endorsement of the EAN/EFAS/INUS NEUROGED guidelines for ataxias and hereditary spastic paraplegias. Published 28 July 2026. Available from: https://www.ern-rnd.eu/wp-content/uploads/2026/09/Final-Neuroged-Guidelines-ataxias-and-HSP.pdf
122. Panicker JN, Fanciulli A, Skoric MK, Kaplan T, Aleksovska K, Adamec I, et al. European Academy of Neurology (EAN)/European Federation of Autonomic Societies (EFAS)/International Neuro-Urology Society (INUS) guidelines for practising neurologists on the assessment and treatment of neurogenic urinary and sexual symptoms (NEUROGED guidelines). Eur J Neurol. 2025;32(4):e70119. doi:10.1111/ene.70119. PMID: 40208234. PMCID: PMC11984325.
123. Agência Nacional de Vigilância Sanitária (ANVISA). Open data: medicinal products registered in Brazil (DADOS_ABERTOS_MEDICAMENTOS.csv) [Internet]. Brasília: ANVISA [accessed 2026-10-29]. Available from: https://dados.anvisa.gov.br/dados/DADOS_ABERTOS_MEDICAMENTOS.csv
124. National Institute of Neurological Disorders and Stroke (NINDS). Clinical manifestations and biomarkers in amyotrophic lateral sclerosis type 4 and other inherited neurological disorders of RNA processing. ClinicalTrials.gov identifier: NCT04394871. Available from: https://clinicaltrials.gov/study/NCT04394871
125. Benussi A, Dell'Era V, Cantoni V, Bonetta E, Grasso R, Manenti R, et al. Cerebello-spinal tDCS in ataxia: a randomized, double-blind, sham-controlled, crossover trial. Neurology. 2018;91(12):e1090-e1101. doi:10.1212/WNL.0000000000006210. PMID: 30135258.
126. Azienda Socio Sanitaria Territoriale degli Spedali Civili di Brescia. Transcranial alternating current stimulation (tACS) in patients with ataxia. ClinicalTrials.gov identifier: NCT05621200. Available from: https://clinicaltrials.gov/study/NCT05621200
127. Chakraborty A, Sreenivasmurthy SG, Miller W, Huai W, Biswas T, Mandal SM, et al. Fructose-2,6-bisphosphate restores DNA repair activity of PNKP and ameliorates neurodegenerative symptoms in Huntington's disease. Proc Natl Acad Sci U S A. 2024;121(39):e2406308121. doi:10.1073/pnas.2406308121. PMID: 39298485.
128. Coutinho P, Ruano L, Loureiro JL, Cruz VT, Barros J, Tuna A, et al. Hereditary ataxia and spastic paraplegia in Portugal: a population-based prevalence study. JAMA Neurol. 2013;70(6):746-55. doi:10.1001/jamaneurol.2013.1707. PMID: 23609960.
129. Silva MC, Coutinho P, Pinheiro CD, Neves JM, Serrano P. Hereditary ataxias and spastic paraplegias: methodological aspects of a prevalence study in Portugal. J Clin Epidemiol. 1997;50(12):1377-84. doi:10.1016/S0895-4356(97)00202-3. PMID: 9449941.
130. National Ataxia Foundation. Institutional website (mission, membership, support groups and programmes). https://www.ataxia.org/. Accessed 29 October 2026.