Mesenchymal Stem/Stromal Cell-Based Therapies for Autism Spectrum Disorder: Emerging Evidence and Clinical Prospects

Major Points and Findings:

This paper is a review and reports no new patient data. It gives an overview of the current state of the field. The authors count the registered trials of cell therapy in autism, record the stage each has reached and the cells used, and list what none of them has yet shown.

Aim:

To bring together a decade of preclinical work (2015 to 2025) with every registered clinical trial of mesenchymal stem/stromal cell (MSC) therapy in autism, and to set out what stands between the current evidence and routine clinical use.

Methods:

Narrative review of the preclinical literature alongside the trials registered on ClinicalTrials.gov as of March 2026.

Results:

The clinical evidence base consists of nine registered trials. Five are completed, three are recruiting and one has not yet started. Seven of the nine use umbilical cord-derived cells, which the authors consider the most promising source because of a lower risk of immune rejection, high proliferative capacity and fewer ethical constraints. The remainder use adipose-derived, bone marrow or Wharton’s jelly cells. Trials are spread across Panama, Denmark, China and the United States, which introduces differences in genetic background, diagnostic practice and standard of care that make results hard to pool.

The review describes several individual trials:

• NCT04089579: The largest trial, and the only standalone Phase II study. It enrolled 137 children aged 4 to 11 in a double-blind crossover design. In this study one group received 6 × 10⁶ umbilical cord tissue MSCs per kg at baseline and placebo at six months, and the other group received the reverse. Positive mean changes were seen on the Socialization and Communication subscales of the Vineland-3. The listed potential risks are rash, shortness of breath, wheezing, difficulty breathing, low blood pressure, swelling around the mouth, throat or eyes, rapid heartbeat, sweating, transmission of infection, and HLA sensitisation.
• NCT01343511: 37 children aged 3 to 12 were randomized to behavioural therapy alone, cord blood mononuclear cells, or those cells combined with UC-MSCs, delivered both intravenously and intrathecally. The combined group did markedly better than cells alone or therapy alone, with statistically significant improvements on CARS, ABC and CGI. Both routes were well tolerated over 24 weeks, with a few cases of mild self-resolving fever.
• NCT05003960: Children receive a single 100 × 10⁶ dose of cultured allogeneic UC-MSCs, with follow-up planned at 1, 6, 12, 24, 36 and 48 months. Completion is due in December 2027. It is one of the only studies designed with follow-up of that length.
• NCT07243561: A Phase I/II randomized trial of a nasal spray containing UC-MSC-derived exosomes in 40 children aged 3 to 7, given as ten doses over 24 weeks. The product is cell-free.

The authors identify the following gaps in the evidence:

• The vast majority of trials have fewer than 50 participants, leaving them underpowered and vulnerable to chance findings.
• Most are open-label, so neither the placebo effect nor investigator assessment bias can be excluded, and the primary endpoints are usually behavioural scales and parent reports, the measures most exposed to expectation.
• Cell sources, preparation methods, doses and administration protocols differ so much between studies that results cannot meaningfully be combined.
• Although immunomodulation and neurotrophic support are the assumed mechanism, most trials have not validated that mechanism in humans. Symptomatic improvement has been observed, but direct evidence of how it comes about is absent.
• Every trial lacks long-term efficacy and safety data. For a lifelong neurodevelopmental condition, the authors note, effects need to be observed over years. Nothing currently published can say how long any benefit lasts or whether late risks exist.

Conclusions:

The authors describe MSC therapy as a transformative, mechanism-informed strategy for autism, while stating that real hurdles remain in standardisation, dose optimisation and demonstrating long-term efficacy and safety. For future studies they recommend randomized, double-blind, placebo-controlled designs. Rating scales should be supplemented by objective endpoints such as neuroimaging, electrophysiology, eye-tracking and inflammatory biomarkers. Blinded independent raters should be used alongside parent reports, and long-term follow-up should be standardised.

The authors also address ethics. In studies involving children, risk must be minimised and consent processes must leave parents fully aware of both the potential benefits and the unknowns.

Background Information:

The review sets out the mechanisms MSCs are thought to work through in autism: modulating neuroinflammation and microglial activation, supporting neurogenesis and synaptic plasticity, acting on the gut-brain axis, and supporting blood vessel formation. Delivery routes in use span intravenous (much the most common), intrathecal, intranasal and intracerebroventricular, with nasal spray now emerging.

The review summarises the gap in the evidence as follows. Early signals have been consistently positive and results have not been negative. The nine trials are small and mostly unblinded, however, and they use different cells at different doses by different routes in different countries. With no long-term follow-up and no confirmed mechanism in humans, they cannot yet support a confident claim about how well this works or for whom.