March 2026 Research Update

This month, small human studies of training in spinocerebellar ataxia (SCA) showed a consistent pattern: practice-based programs can improve certain day-to-day performance measures, but they did not clearly shift core ataxia severity over short follow-up. One gait-focused program improved walking speed and functional mobility without changing overall ataxia scores or showing clear brain structure changes on MRI. A separate digital cognitive training program in SCA2 showed clearer gains in memory than in other thinking skills, with some measures moving in the opposite direction.

After 5 weeks of gait adaptability training, people with early-stage SCA improved walking velocity and performance on functional mobility tasks, but ataxia severity scores and some ataxia-specific gait measures did not change, and structural MRI did not show detectable gray matter change over that period [1] . In plain language, this looks like improved performance with practice rather than a clear signal of slowed disease or reversal of ataxia signs in the short term [1] .

In SCA2, a 6-month home-based digital cognitive training program was linked to the most consistent improvements in memory tests, while executive function showed limited change and some processing-speed measures worsened (longer completion times on certain tasks) [2] . This mixed pattern matters because it suggests different thinking skills may respond differently to repeated training, and it highlights the importance of measuring more than one type of cognitive outcome when studying these programs [2] .

A separate inpatient rehabilitation report did not measure symptom change, but it mapped what therapists actually worked on in routine inpatient care for degenerative cerebellar ataxia, including several SCA types [3] . The program covered many areas of function, and in more severe ataxia it placed greater emphasis on activities like mobility and self-care rather than only body functions, describing the practical focus of rehab when daily tasks become harder [3] .

For patients and caregivers, these studies add a clearer picture of what training studies are capturing right now: improvements may show up first in specific tasks (like walking speed, getting through a mobility test, or certain memory tasks) rather than in a global ataxia severity score. This may matter now as a way research groups choose outcomes for future studies, because it highlights that short programs can detect changes in performance even when core ataxia ratings look stable. What remains uncertain is how durable these changes are, whether they translate to broader daily life across different SCA types, and whether any program can change longer-term progression rather than mainly improving practiced skills.

Studies this month strengthened the idea that measurable body signals can help track and sort different ataxias, but not every measure lines up neatly with symptom severity. Brain imaging picked up subtype-specific patterns in parts of the brain beyond the cerebellum, and one MRI method distinguished SCA6 from SCA31 using a signal related to dentate nucleus iron storage. In presymptomatic SCA2 carriers, wearable-sensor gait measures changed over time in ways that may detect progression earlier or with fewer participants than a standard clinical rating scale.

MRI measures of the hypothalamus (a brain region involved in autonomic and hormonal regulation) showed disease-subtype patterns across polyglutamine SCAs, with clearer volume loss and texture changes in SCA1 and SCA3 than in some other subtypes [4] . Some of these hypothalamic measures moved with disease duration and ataxia severity in SCA1 and SCA3, suggesting they could be tracking parts of the disease process, although the study did not test whether this predicts future symptoms for an individual [4] .

Two other imaging studies focused on distinguishing genetic ataxia subtypes rather than rating how severe symptoms are. Automated measurement of the cerebellar vermis (the midline cerebellum) separated adults with CACNA1A-related ataxia and GAA-FGF14-related ataxia into different atrophy patterns, and vermis loss related to age at onset and genetic mechanism more than to chronic ataxia severity [5] . In another study, an MRI method sensitive to tissue susceptibility found lower dentate nucleus susceptibility in SCA6 than in SCA31 and controls, and autopsy staining in a subset supported that this lower signal reflects ferritin loss in SCA6 [6] .

For tracking very early change, a 4-year study of presymptomatic SCA2 carriers found that average stride-to-stride gait measures progressed more often than gait variability measures, and some progression patterns were influenced by CAG repeat length [7] . The authors also reported that several digital gait measures would need smaller sample sizes than the SARA clinical score to detect change in hypothetical trials, which is mainly important for research planning rather than for individual decision-making right now [7] .

For patients, these studies add to understanding in two practical directions: separating similar-looking ataxias and measuring change more sensitively over time. Some MRI signals (like dentate nucleus susceptibility in SCA6) may help distinguish subtypes, while other measures (like hypothalamus or vermis changes) may capture parts of the disease that do not match symptom severity one-to-one. Wearable gait tracking in presymptomatic SCA2 is early evidence that subtle change can be measured before clear symptoms, but it is still research and does not by itself predict when any one person will develop symptoms or how fast things will change.

This month reinforced how wide the SCA spectrum can be, including very early onset, atypical symptom patterns, and genetic factors that may or may not shift when symptoms begin. A large SCA3 genetics study found no signal for common APOE types changing age at onset, but suggested a possible earlier onset in a rare subgroup with APOE e4 homozygosity in Brazil. Several case reports showed that repeat length alone may not explain childhood onset in SCA17, and that extremely large expansions can present from birth in SCA2.

In over 800 people with SCA3, common APOE haplotypes and single APOE alleles were not linked to age at onset, which helps narrow the list of likely genetic modifiers for most patients [8] . A rare finding stood out in the Brazilian subgroup: APOE e4 homozygosity was linked to onset about 6 years earlier than other APOE haplotypes, suggesting a possible allele-dosage effect that needs confirmation in additional groups [8] .

Case reports highlighted that diagnosis can be complicated when symptoms or imaging do not match expectations. A child with Huntington disease-like SCA17 (HDL-4) had intermediate-range TBP repeats and childhood-onset hyperkinetic movements and seizures, with no cerebellar atrophy on MRI at the time reported, raising the possibility that factors beyond repeat length influence onset and severity in some individuals [9] . Another report described infantile-onset SCA2 beginning at birth in a child with a very large ATXN2 expansion (about 884 repeats), with brain imaging showing cerebellar and pontine atrophy, illustrating the extreme end of the repeat-expansion spectrum [10] .

Other cases expanded what SCA-related genes can look like clinically. In a patient in his 60s with distal axonal polyneuropathy and no cerebellar findings, genetic testing found a truncating AFG3L2 variant, and patient-derived cells showed mitochondrial stress pathway changes consistent with reduced AFG3L2 function [11] . A family report of SCA14 linked a PRKCG variant (p.C142G) with cerebellar atrophy on MRI and autosomal-dominant inheritance, alongside a proposed mechanism involving disrupted PKCgamma regulation and chronic cellular stress [12] .

One diagnostic study in suspected Wilson disease showed how genetic testing can sometimes redirect an ataxia diagnosis. A multigene panel confirmed Wilson disease in many participants, but it also reclassified one person with a Leipzig score of 4 to SCA12 when testing revealed a PPP2R2B CAG expansion along with another gene variant, showing that mimic conditions can overlap clinically [13] . A separate SCA12 case report described a 71-year-old woman with about 69 CAG repeats in PPP2R2B, mixed tremor, and hippocampal as well as cerebellar atrophy on MRI, adding another example of the clinical range reported in SCA12 families [14] .

One additional case report involved ataxia that was not genetic SCA, but can resemble it in symptoms. A man with anti-GAD antibody-associated ataxia had a distinctive eye-movement finding (unilateral loss of certain torsional saccades) and cerebellar vermis atrophy on MRI, with no improvement reported after several immunotherapies in this single case [15] .

For patients, these studies mainly add clarity about uncertainty and variation rather than introducing new treatments. The SCA3 APOE result suggests that for most people, common APOE types do not explain differences in when symptoms start, while a rare subgroup signal may be important later if confirmed. The case reports underline that genetic ataxias can start very early, can sometimes lack classic imaging findings early on, or can present with symptoms outside the usual cerebellar picture, which is why careful genetic and clinical workups remain central in research and specialist settings. What remains uncertain is how often these unusual patterns occur and which additional genes or non-genetic factors drive the big differences seen between individuals.

Preclinical studies this month offered several different angles on possible future therapies, but all are still far from proving benefit in people. Some work focused on reducing toxic proteins or aggregates in repeat-expansion disorders, including SCA8 and SCA3. Other studies developed or refined experimental models and tools that make it easier to test ideas, including gene-editing delivery, controllable RNA foci in cells, and new animal models.

In an SCA8 mouse model, chronic metformin treatment improved motor behavior and reduced RAN protein aggregates and neuroinflammation, while partially correcting RNA mis-splicing without changing repeat length or the levels of repeat-containing RNA [16] . In SCA3 mouse models, CARHSP1 (a stress granule-related protein) was reported to be increased, and lowering Carhsp1 reduced mutant ataxin-3 aggregates and improved motor function and neuropathology in mice, pointing to stress granule pathways as one possible target area [17] .

Gene-targeting and delivery methods were also explored for SCA3. One study engineered extracellular vesicles to carry SpCas9 gene-editing complexes aimed at inactivating ATXN3, and reported ATXN3 knockout in patient-derived cells and in two SCA3 animal models, with a light-triggered release design that increased target engagement in lab testing [18] . This represents a technical step toward delivery strategies, but the summary did not provide information on symptom outcomes or safety in animals within these experiments [18] .

Several studies deepened the biology of specific SCA types in ways that could shape future therapy concepts. Two separate SCA37 studies pointed to neurodevelopmental wiring effects: zebrafish expressing AUUUC repeat RNA had axon and muscle-innervation abnormalities and reduced movement, and adding NOVA2 rescued axonal defects, while patient-derived neuron work linked an ATTTC insertion to overactivation of a DAB1 developmental enhancer and abnormal axonal pathfinding when DAB1 is overexpressed [19] [20] . In a mouse model of SCA1, respiratory dysfunction was driven by mutant ATXN1 in neural tissues rather than skeletal muscle, and removing mutant ATXN1 from neural lineages improved breathing measures, pointing to specific brain circuits involved in breathing control [21] .

New models and research tools were introduced to make future testing more precise. A CRISPR knock-in mouse model of SCA5 showed impaired motor performance and cerebellar Purkinje-cell changes including beta-III-spectrin redistribution and inclusions, along with proteomic signals related to glutamate signaling, supporting its use for studying SCA5 mechanisms and testing interventions in animals [22] . A lentiviral striatal model using expanded ataxin-2 was presented as a controlled platform to produce inclusions and loss of neuronal markers in a specific brain region, aimed at studying region-specific neurodegeneration in polyglutamine SCAs [23] .

Finally, a cell-based tool study showed that RNA foci in SCA31 can be manipulated with light. A photoswitchable RNA-binding ligand (NCTA) made UGGAA repeat RNA foci grow under UV light and shrink back under visible light, with accompanying changes in which RNA-binding proteins were present in the foci [24] . This does not test a therapy in animals or people, but it gives researchers a way to study how these RNA structures form and dissolve over time in living cells [24] .

For patients and caregivers, these studies mainly matter as early signals about what might be possible later, not as evidence for use in people now. Metformin in an SCA8 mouse model and lowering CARHSP1 in SCA3 mice both reduced disease-like changes and improved movement in animals, which helps prioritize directions for future translational work. Gene editing delivery and light-controlled RNA tools are enabling technologies that could support future therapies, but the key uncertainties are how well these approaches can be delivered safely and effectively in the human brain, and whether benefits seen in models will translate to meaningful symptom change in people.

One study this month stepped back from patient symptoms and examined how the ATXN3 gene family evolved across primates, with the goal of understanding why the human gene has repeat patterns that can become disease-associated. It identified several ATXN3-related retrocopies in primates, with some likely non-functional and others possibly preserved by evolution. This kind of work does not change diagnosis or care, but it can clarify the biological context for why certain repeats exist and how they might become unstable.

Across many primate genomes, researchers described four ATXN3 retrocopies (ATXN3L0 to ATXN3L3) and reported that ATXN3 and ATXN3L1 show signs of evolutionary conservation, while other copies may be non-functional or absent in humans [25] . They also analyzed CAG-repeat interruption patterns and discussed mechanisms that could contribute to the long, disease-associated repeat tracts seen in human ATXN3 [25] .

For patients, this is background research that may matter later by sharpening how scientists think about repeat expansions and why some genes are more prone to problematic repeat lengths. It does not offer a new test, a new monitoring tool, or a treatment signal for SCA3 in the near term. The uncertainty is how directly these evolutionary observations can be connected to repeat instability or symptoms in living people, which would require separate human and experimental studies.