Mesenchymal Stem Cell-Based Therapy for Cerebellar Ataxia: From Bench to Bedside

Major Points and Findings:

This is a narrative review from a Korean university group that has tested human mesenchymal stem cells (MSCs) in mouse models of cerebellar ataxia for several years. It covers their animal work, the few human studies, and the one MSC product for ataxia that has reached a placebo-controlled trial.

The review makes three main points. The proposed benefit of MSCs is protective, and the cells are not expected to replace lost Purkinje cells. The authors argue that donor cells are preferable to a patient’s own, based on a laboratory finding about ataxia patients’ MSCs. They also state that the human evidence is still thin. Most of the animal studies reviewed are the authors’ own. The most positive clinical results they cite come from a company conference presentation and trial registry entries, and no peer-reviewed paper is cited for them.

Aim:

To appraise preclinical and clinical evidence for MSC therapy across different causes of cerebellar ataxia, set out a common mechanism, and propose a plan for future trials.

Methods:

The authors searched PubMed/MEDLINE, Embase and the Cochrane Library for 2000 to 2026, pairing MSC terms with “cerebellar ataxia”, “spinocerebellar ataxia”, “Purkinje cell” and “stem cell therapy”. No numbers of records screened, no quality scoring and no pooled analysis are given, so despite the word “systematic” in the methods this is a narrative review. The authors acknowledge that animal models of immune-mediated ataxia (such as anti-GAD65) and Friedreich ataxia were not covered.

Results:

Three mouse models (all preclinical):

• In the inflammatory model, LPS was injected into the cerebellum. Donor human MSCs given three days later improved rotarod performance over four weeks. Purkinje cell loss was about 15% instead of about 40%, TNF-alpha and IL-1 beta fell, and the MSC-derived factor TSG-6 roughly doubled.
• In the toxic/developmental model, newborn mice were given the chemotherapy drug cytarabine. MSCs given to adult mice with established deficits improved rotarod and open-field scores over 12 weeks. BDNF and GDNF rose 1.1 to 2.4-fold in the host cerebellum itself, which the authors call “host tissue reprogramming”.
• In the genetic model (SCA2 transgenic mice), MSCs were started only after symptoms appeared, and decline on the rotarod was slowed out to 50 weeks of age. The review says treated mice kept about 70% of baseline function against 35 to 40% in controls. The original paper, which we have summarised separately, reports rotarod times of 311 seconds (treated) and 206 seconds (untreated) at 50 weeks, from 345 seconds at the start. The direction is the same, but the review’s percentages do not match the source exactly.

Patient cells versus donor cells: A 2020 study from the same group found that MSCs taken from people with cerebellar ataxia secreted markedly less FSTL1, TGFB1, IGFBP3 and GAS6 than MSCs from healthy donors of similar age, although they passed the standard identity tests. The authors use this to argue for allogeneic (donor) cells and potency testing of each batch. Their explanation, epigenetic changes from chronic inflammation, is presented as likely and has not been demonstrated.

Human evidence:

• One case report (Ko 2021) describes a 60-year-old man with sporadic adult-onset ataxia who received donor bone marrow MSCs intrathecally, 3 × 10⁷ cells per infusion, in three cycles four months apart. SARA went from 14 to 10 at six months and was 12 at twelve months. The only adverse finding was a mild, temporary rise in CSF white cells that cleared within 7 days. The authors state this cannot exclude placebo effect, natural fluctuation or observer bias.
• Stemchymal is an allogeneic adipose MSC product given intravenously. According to the review, a randomised, double-blind, placebo-controlled phase II trial in Taiwan (NCT02540655) gave three infusions over 12 months to SCA3 patients with SARA of 9 or more, and reported stabilisation on SARA and functional SARA, with about 1.3 points of improvement in functional SARA at 12 months. Untreated SCA3 typically worsens by 1.4 to 1.6 SARA points a year. A Japanese phase II study in SCA3 and SCA6 is described as consistent, and a US phase IIb trial is registered (NCT06397274). These results are cited from an April 2025 congress presentation. No journal publication is referenced.
• The earlier open-label studies are Tsai 2017 (six SCA3 patients and one MSA-C patient, one intravenous dose of adipose MSCs, tolerated over 12 months) and Jin 2013 (16 genetically confirmed SCA patients, umbilical cord MSCs, improvement on ICARS and Berg Balance Scale).
• Two meta-analyses reached cautious conclusions. Bhartiya 2023 pooled 3 studies with 47 patients and found a non-significant trend toward improvement. Appelt 2021 found no statistically significant functional improvement.
• No study has reported a serious MSC-related adverse event, ectopic tissue or tumour formation.

Other cell therapies: Neural stem cells could in principle replace neurons but carry more manufacturing, immune and tumour risk. Blood stem cell transplantation in ataxia-telangiectasia repairs the immune system but does not slow the neurological decline. For Friedreich ataxia the authors consider gene-corrected blood stem cells the more logical approach, since MSCs do not address the missing frataxin.

Conclusions:

The authors conclude that MSCs act on shared downstream damage (inflammation and loss of growth-factor support) and that the field should move to placebo-controlled trials. They list the following open problems:

• Repeated intrathecal donor-cell infusions carry a theoretical risk of sensitising the immune system. The authors want donor-specific antibodies monitored beyond 12 months.
• The cell source affects the product. Bone marrow, adipose and umbilical cord MSCs secrete different mixtures, and donors differ from each other.
• Dose, interval and route are undefined for ataxia. Intrathecal delivery is said to bring more cells to the back of the brain than intravenous, but no direct comparison exists.
• Injected MSCs survive only briefly. They are detectable for 1 to 4 weeks and are largely gone by 4 to 8 weeks.
• There are no safety data beyond 12 months in ataxia.

They propose a phase I/IIa trial of 20 to 30 patients (SARA 8 to 20, 12 months) and a phase III trial of 80 to 120 patients per arm over 18 months, powered for a 2.4-point SARA difference.

The paper contains one inconsistency. The conclusion says the clinical evidence “currently comprises a single uncontrolled case report”, while the same paragraph calls the Stemchymal results the strongest controlled evidence available.

Background Information:

Cerebellar ataxia affects roughly 26 per 100,000 people and includes more than 50 spinocerebellar ataxia subtypes. The mechanism the review describes is chemical signalling. MSCs release prostaglandin E2 and TSG-6, which shift microglia (the brain’s immune cells) from an attacking state to a calming one, and growth factors (BDNF, GDNF, IGF-1) that support Purkinje and granule cells. Conversion of MSCs into neurons happens at negligible rates. As the cells disappear within weeks, the authors suggest a “hit-and-run” model in which brief exposure sets off longer-lasting protective changes in the host tissue. This model remains a hypothesis.