Clinical Evidence of Mesenchymal Stromal Cells for Cerebral Palsy: Scoping Review With Meta-Analysis of Efficacy in Gross Motor Outcomes

Major Points and Findings:

This review collects what is publicly known about mesenchymal stromal cell (MSC) treatment for cerebral palsy, drawing on published reports, registered trials and declared compassionate access programmes. It is relevant to families because umbilical cord tissue MSCs, given intrathecally or intravenously and usually more than once, were found to be the most common way this treatment has been delivered worldwide.

The main results are favourable. Across six controlled studies MSCs improved gross motor function with a large effect size, and no serious adverse events were reported. The authors, led from the Cerebral Palsy Alliance Research Institute in Sydney, also set out the weaknesses. The studies disagree with each other a great deal, and two of five assessed trials were at high risk of bias. No phase 3 trial has been published, and nearly a third of treated patients received cells outside a trial.

Aim:

To describe who has been treated with MSCs for cerebral palsy, with which cells, route and dose and in what kind of study, and to estimate the effect on gross motor function by meta-analysis of controlled studies.

Methods:

The authors carried out a scoping review following PRISMA-ScR, with a published protocol. MEDLINE, Cochrane Central and Embase were searched on 20 May 2024, and twelve trial registries plus the grey literature up to 13 February 2025.

The review included any published clinical report or registered trial, of any phase and language, that gave MSCs from any source to treat cerebral palsy. Conference abstracts and registered compassionate access programmes were also eligible. Secretome products such as extracellular vesicles, mixed cell products, and cord blood or bone marrow cells not expanded or enriched for MSCs were excluded.

For the meta-analysis, controlled studies reporting any version of the Gross Motor Function Measure (GMFM) were pooled in RevMan 5.4 using a random-effects model and standardized mean differences (SMD). Risk of bias was rated independently by two authors with the Cochrane RoB 2 tool.

Results:

In total 30 published reports and 10 registered trials were included, covering 1,292 people with cerebral palsy aged 6 months to 45 years. There were 16 case reports or series, 8 phase 1 trials, 12 phase 2 trials, 2 phase 3 trials (both registered as complete, neither published) and 1 compassionate access programme.

• 31% of all participants (401 people) were treated through compassionate or expanded access, the largest single share. Phase 2 trials accounted for 27%.
• Most cells were allogeneic, and the most frequent source was umbilical cord tissue (75% according to the abstract). For 21.9% of participants the cell source was not reported.
• The commonest routes were intrathecal (40%) and intravenous (38%), followed by both together. Eleven routes or combinations were used in all, including injection into the brain or its ventricles in about 5% of participants.
• 72% of participants received two or more doses, ranging from 2 to 32 administrations. 69% received a fixed dose regardless of body weight, ten studies gave no dosing information, and one study gave up to 750 million cells over 8 treatments by 4 routes. Severity and type of cerebral palsy were often not reported.

Effect on gross motor function: Eight controlled studies were considered and six had usable data. Compared with controls:

• 1 month: no effect (SMD 0.00, 95% CI -0.65 to 0.66, p = 1.00; only two studies)
• 3 months: SMD 1.05 (0.19 to 1.92, p = 0.02)
• 6 months: SMD 0.97 (0.30 to 1.64, p = 0.005)
• 12 months: SMD 0.99 (0.30 to 1.67, p = 0.005)

By Cohen’s convention these are large effects. Heterogeneity was substantial throughout (I² 64% to 90%), meaning that the individual studies produced quite different results.

Subgroups: At 12 months there was no significant difference between one dose and more than one dose (p = 0.38), or between intrathecal and intravenous delivery (p = 0.47). At 3 months there was no significant difference between bone marrow MSCs and cord tissue MSCs (p = 0.33). Grouping the studies this way removed the residual heterogeneity, so the authors regard schedule, route and source as plausible reasons the trials disagree, without being able to say which option is better. The trial with the largest effect was also the one that gave the most doses (eight). There were too few data to examine age, severity or cell dose.

Risk of bias: Of five studies assessed, one raised some concerns and two were high risk (selective reporting in one, deviations from the intended intervention in the other).

Safety: No serious adverse events were reported in any study that reported on safety. However, 7 of the 30 published reports contained no information on safety or side effects.

• After intravenous MSCs: fever, diarrhoea, vomiting and breathlessness, mostly mild and short-lived infusion reactions.
• After intrathecal MSCs: fever, irritability, headache, low back pain, vomiting, nausea and meningism, consistent with changes in cerebrospinal fluid pressure.
• All events were treatable and resolved within days. In the one study that looked, two participants developed donor-specific anti-HLA class I antibodies, without symptoms.

Conclusions:

MSCs appear safe and improve GMFM scores with a large pooled effect, but the variety of products, doses, routes and schedules makes the data hard to combine. The authors call for head-to-head studies, powered later-stage trials with common outcomes, and sharing of child-level data so that an individual participant data meta-analysis, like the one done for cord blood, can identify the best source, route, dose and responders.

The authors state as limitations that registry entries, case reports and compassionate access data are lower-quality evidence, and that missing data were widespread. Follow-up rarely went beyond 12 to 24 months, so there are no long-term results. This is a particular gap for young children who have not yet reached their motor potential. Gross motor function was the only outcome pooled.

Background Information:

MSCs are thought to work through anti-inflammatory and immune-modulating signals. They are not thought to turn into brain cells. Cells given into a vein do not cross the blood-brain barrier in meaningful numbers. They are trapped in the lungs and cleared quickly by macrophages, and laboratory work suggests this clearance is required for their effect.

On the comparison with cord blood, an earlier phase 2 trial (Sun and colleagues, 2022) suggested cord blood may produce a larger and earlier effect than cord tissue MSCs, although it was not powered to compare them. The authors argue their pooled result keeps MSCs in contention, and add that MSCs need no donor matching and allow repeat dosing.

The authors find it striking that treatment has moved into compassionate access when no MSC product or regimen has been taken through to phase 3, which they attribute partly to a lack of commercial interest. They write that people with cerebral palsy should not have to rely on compassionate access in the long term. In their view the priority is research establishing a full safety and efficacy profile.