Evaluation of Stem/Stromal Cell Transplantation Safety and Efficacy in Children Diagnosed With Cerebral Palsy: A Systematic Review and Meta-Analysis of Randomized Controlled Trials

Major Points and Findings:

This meta-analysis from Kermanshah University of Medical Sciences in Iran pools 13 randomized controlled trials covering every cell type tested in children with cerebral palsy. It reports results separately by route (intrathecal versus intravenous) and by cell type (mesenchymal stem cells versus cord blood), which are the choices that differ between clinics.

The pooled result favours cell therapy, with a moderate improvement in gross motor function by 6 to 12 months and no more fever, vomiting or irritability than in controls. The review calls for more caution than the individual participant data meta-analysis summarised elsewhere on this site. The low heterogeneity was reached only after three trials were removed from the analysis. The subgroups rest on very few trials each, and parts of the reporting are hard to follow. These problems are described below.

Aim:

To evaluate the efficacy (Gross Motor Function Measure, GMFM) and safety (adverse events) of stem or stromal cell therapy in children with cerebral palsy.

Methods:

The authors performed a systematic review and meta-analysis registered on PROSPERO (CRD42024588170). PubMed, Web of Science, Scopus and Embase were searched to 23 August 2024. Risk of bias was rated with the Cochrane RoB 2 tool.

Randomized controlled trials in people with cerebral palsy were included if they gave a cell infusion and reported the GMFM.

The review excluded non-randomized and observational studies, reviews, animal and laboratory work, studies that did not report the GMFM, and studies whose full text the authors could not access. Of 1,064 records, 13 trials were included. Fourteen studies were dropped for missing control group data or an unavailable full text, a considerable gap for a review of this size.

GMFM results were pooled as standardized mean differences (SMD) with a random-effects model. Adverse events were pooled as risk ratios (RR) with a fixed-effects model. Subgroups were follow-up time, cell type and route. Because heterogeneity was very high, the authors ran a sensitivity analysis. All published forest plots and subgroup results reflect the data after that sensitivity analysis.

Results:

Of the 13 trials, five were from China, four from South Korea, two from Iran and two from the United States, published between 2012 and 2023, with 25 to 105 children each and ages from about 5 months to 20 years. Products and routes were:

• Allogeneic umbilical cord blood, intravenous, about 3 to 10 × 10⁷ cells per kg, with or without erythropoietin (four trials), and autologous cord blood (one trial)
• Umbilical cord or cord blood MSCs, intravenous, around 5 × 10⁷ cells per infusion (two Chinese trials), or three doses of 2 × 10⁶ in one arm of a US trial
• A single intrathecal injection of 2 × 10⁷ umbilical cord MSCs, or of 5 × 10⁶ cord blood mononuclear cells per kg (two Iranian trials)
• Autologous bone marrow MSCs or bone marrow mononuclear cells, intrathecal (China)
• Neural stem or progenitor cells, intranasal or into the lateral ventricle (two Chinese trials), and mobilised autologous peripheral blood cells, intravenous (South Korea)

Controls received placebo, a sham procedure or rehabilitation alone.

Overall effect on GMFM: With all 13 trials the SMD was 0.82 (95% CI 0.32 to 1.31), but heterogeneity was extreme (I² = 91.37%), meaning the trials disagreed strongly. The authors identified three trials as the main source. These were the Iranian intrathecal cord blood mononuclear cell trial, a Chinese trial of intravenous cord blood MSCs, and the Korean peripheral blood cell trial. After removing them the estimate fell to SMD 0.50 (0.33 to 0.67), I² = 8.08%. An SMD of 0.5 is conventionally a moderate effect.

By time since treatment:

• 3 months: SMD 0.27 (0.05 to 0.48)
• 6 months: SMD 0.53 (0.30 to 0.76)
• 12 months: SMD 0.52 (0.34 to 0.70)

By cell type:

• Mesenchymal stem cells (four trials: three from umbilical cord or Wharton’s jelly, one from bone marrow): SMD 0.63 (0.35 to 0.91)
• Umbilical cord blood (five trials, four allogeneic and one autologous): SMD 0.30 (0.04 to 0.55)
• Neural stem or progenitor cells (two trials): SMD 0.88 (0.28 to 1.49)

By route, at 12 months:

• Intrathecal (two trials after the sensitivity analysis: Iranian umbilical cord MSCs and Chinese bone marrow cells): SMD 0.56 (0.26 to 0.87)
• Intravenous (seven trials): SMD 0.29 (0.04 to 0.53)

The paper reports no statistical test of intrathecal against intravenous delivery, or of MSCs against cord blood, and the confidence intervals overlap. The intrathecal figure comes from two trials that used different cells, so it does not prove that the route itself made the difference.

Safety: Three adverse events were pooled: fever (RR 0.66, 95% CI 0.18 to 1.14), nausea or vomiting (RR 0.98, 0.21 to 1.75) and irritability (RR 0.99, 0.13 to 1.84). All three intervals include 1, so none differed significantly between treated and control children. The authors note that several trials gave pre-medication (diphenhydramine, methylprednisolone), which probably reduced infusion reactions. Serious, rare and long-term events were not analysed.

Risk of bias: All trials were rated low risk for randomization. One (the Korean peripheral blood cell study) was rated high risk for missing outcome data, and three others raised some concerns on the same point.

Conclusions:

The authors conclude that stem cell transplantation was safe and effective for improving GMFM scores, while calling for larger, high-quality trials with standardized protocols, cell types and doses.

The paper has no limitations section. It says only that the small number of studies prevented separate analysis by cell origin. The following limitations are apparent from the full text:

• The low heterogeneity was produced by excluding three trials after seeing the results. One was a double-blind, sham-controlled trial of intrathecal cord blood cells. It is directly relevant to the route comparison and is missing from it.
• One included “trial” is the Korean open-label extension in which every child received cord blood, so it does not compare treated with untreated children.
• The MSC subgroup merges bone marrow and umbilical cord cells, and the paper does not explain how the single control group of the two-arm US trial was handled.
• The safety section is inconsistent. The discussion speaks of “a higher tendency to fever” while reporting a ratio of 0.66, which would mean fewer fevers in treated children. Given the confidence interval, the data support only the statement that no difference was shown.
• Only gross motor function was pooled, and the paper does not report which children (by age or severity) benefit.

Background Information:

The cells are thought to work through paracrine signalling. They release factors that reduce inflammation and promote blood vessel growth, and may help the brain’s own stem cells migrate and multiply.

The authors also give the argument for intrathecal delivery. When MSCs are given into a vein, most lodge in the lungs, and the authors describe the proportion reaching injured brain tissue as insignificant. Their own data, however, do not settle whether route matters.