An Updated Meta-Analysis of Umbilical Cord Blood to Treat Cerebral Palsy: Distinguishing Cord Blood Infusions From Mesenchymal Stromal Cell Therapy

Major Points and Findings:

This review comes from the Ottawa Hospital Research Institute and the University of Ottawa in Canada. It updates the team’s earlier systematic review of controlled trials using umbilical cord blood for new indications, and asks whether two different cord-based products perform differently in cerebral palsy. One is unmanipulated cord blood, given as mononuclear cells (MNCs). The other is mesenchymal stromal cells (MSCs), which are grown in culture before infusion.

Ten controlled trials were identified and seven of them were pooled for gross motor function at 12 months. The overall result favoured treatment. Split by product, the MSC trials showed a significant improvement and the cord blood MNC trials did not. The authors’ own test of the difference between the two subgroups was not significant, and the “cord blood MSC” group mostly contains MSCs grown from cord tissue. Both points matter when reading the headline result and are covered below.

Aim:

To update the meta-analysis of controlled trials of umbilical cord blood-derived cells in cerebral palsy and to compare unmanipulated cord blood mononuclear cell infusions with cord blood-derived MSC therapy.

Methods:

The search was aligned with the group’s earlier review registered on PROSPERO (CRD42016040157). Cochrane, Embase and Medline were searched on 5 November 2024 by an information specialist, and the search was peer reviewed against the PRESS guideline. The updated search was limited to systematic reviews and meta-analyses. The authors describe it as an umbrella review, used to find controlled trials published since their 2018 review, which had identified four trials. Trials were not searched for directly.

Risk of bias was assessed in duplicate with the Cochrane RoB 2 tool. The primary analysis was the change in Gross Motor Function Measure (GMFM) score from baseline to 12 months, pooled as a standardized mean difference with a random-effects model and split into MNC and MSC subgroups. Baseline GMFM scores of treated and control groups were compared with two-tailed t-tests.

Results:

The ten trials were published between 2013 and 2022 and came from South Korea (three), China (two), the United States (three) and Iran (two). Trial size ranged from 20 to 96 participants (mean 62), with 621 patients in total, and ages ranged from 6 months to 20 years. Eight trials used allogeneic cells and two used autologous cells. Follow-up ran from six months to two years.

MNC trials, as classified by the authors:

• Min 2013, South Korea: allogeneic cord blood with or without erythropoietin (500 IU/kg), intravenous once, mean 8.33 × 10⁷ cells, 96 children, 6 months
• Kang 2015, South Korea: allogeneic cord blood, intravenous and intra-arterial, median 5.46 × 10⁷ cells/kg, 34 participants, 6 months
• Sun 2017, USA: autologous cord blood, intravenous once, 1 to 5 × 10⁷ cells/kg, 63 children, crossover design
• Min 2020, South Korea: 2 × 2 factorial trial of allogeneic cord blood and erythropoietin, intravenous once, mean 4.8 × 10⁷ cells, 88 children, 12 months
• Cox 2022, USA: autologous cells, intravenous once, 2 to 10 × 10⁶ cells/kg, 20 children, crossover design
• Zarrabi 2022, Iran: allogeneic cord blood mononuclear cells, intrathecal once, 5 × 10⁶ cells/kg, 63 participants, sham-controlled, 12 months

MSC trials:

• Huang 2018, China: allogeneic cord blood MSCs, 5 × 10⁷ cells every 7 days × 4, repeated after a 3-month break, 54 children
• Gu 2020, China: allogeneic umbilical cord-derived MSCs, intravenous on days 0, 7, 14 and 21, 4.5 to 5.5 × 10⁷ cells, 40 children
• Amanat 2021, Iran: umbilical cord tissue MSCs, intrathecal once, 2 × 10⁷ cells, 72 participants, sham-controlled
• Sun 2022, USA: open-label trial randomising children to one intravenous dose of allogeneic cord blood (10 × 10⁷ cells) or three intravenous doses of cord tissue-derived MSCs (2 × 10⁶ cells) at 0, 3 and 6 months, 91 children

Of the 621 patients, 440 (71%) were in trials that gave cord blood MNCs.

Risk of bias: Five trials were rated low risk overall and five had some concerns. Most concerns were in domain 5 (selection of the reported result), which affected all five. No trial was rated high risk.

Outcomes reported: All trials reported GMFM change, but only six reported follow-up of 12 months or longer. Five used the Pediatric Evaluation of Disability Inventory and nine of ten used at least one imaging method. Three of the seven trials reporting diffusion tensor imaging found an increase in fractional anisotropy in treated children. Imaging and non-GMFM outcomes were too varied to pool.

GMFM at 12 months: Nine groups across seven trials were pooled, with 239 patients in intervention groups and 197 in control arms. Min 2013 and Kang 2015 were excluded because they followed patients only to six months, and one further study was excluded for lacking raw control data. The Min 2020 arm given erythropoietin alone was also left out. Baseline GMFM scores did not differ significantly between treated and control groups in the included trials.

• All trials: standardized mean difference 0.5828 (95% CI 0.19 to 0.98), p = 0.004
• Cord blood MNC subgroup: 0.3225 (95% CI -0.19 to 0.84), p = 0.22
• MSC subgroup: 0.9192 (95% CI 0.33 to 1.50), p = 0.002

The test comparing the two subgroups gave p = 0.13, which is not significant. The text gives no heterogeneity statistic for the pooled analysis. Looking at the individual trials, the authors report that 1 of the MNC trials and 3 of the 4 MSC trials found a significant benefit on their own.

Dose and age: No statistically significant relationship between cell dose and outcome was found, and the authors say the small number of trials made this analysis difficult. Route of administration could not be analysed. Of the two trials that reported results by age, one found better outcomes in children under 3 years.

Safety: The review did not collect or analyse adverse events.

Conclusions:

The authors conclude that UCB-derived cells, and UCB-MSCs in particular, may benefit children with cerebral palsy based on GMFM change at 12 months, and that cord blood MNC benefit is less clear. They call for larger randomized trials with consistent designs, standard outcome measures beyond motor function, and follow-up well past 12 months. They suggest MSCs from cord blood and other tissues deserve priority.

The authors list these limitations: trials missed by the search or not published in English, Huang 2018 being a significant outlier that may have skewed the MSC result, variation in patient age, most trials being run outside North America under different regulation, and no distinction between cord blood and cord tissue-derived cells.

Further limitations are apparent from the full text:

• The subgroup comparison was not significant (p = 0.13), so the data do not show that MSCs outperformed cord blood MNCs. The text nonetheless reads this result as suggesting the two may have different effects.
• Despite the “UCB-MSC” label, Gu 2020 used umbilical cord-derived MSCs and Amanat 2021 and Sun 2022 used cord tissue MSCs. Only Huang 2018 used MSCs from cord blood, and that is the trial the authors call an outlier.
• The pooled analysis rests on seven trials. Sun 2022 had no untreated control group, and its two treated arms were compared against the same comparison group (25 children in Table 3).
• Searching only reviews for new trials may have missed controlled trials that no review had yet included.
• The paper is inconsistent in places. It says six trials reported 12-month follow-up and later says eight. The reference given for the study excluded for missing control data points to an unrelated review on MSCs. The previous review is dated 2018 in one place and 2019 in another.
• The trials differ widely in dose, route, number of infusions, age range and use of rehabilitation, and safety was not assessed.

Dr. Allan declares that he is a paid consultant with Stem Cells, Canadian Blood Services. No other conflicts are declared.

Background Information:

Cord blood MNCs are given without expansion in the laboratory. MSCs are expanded in culture to reach enough cells of sufficient purity, and consensus on how such products should be characterised is still being developed. The authors cite work showing cord blood MSCs proliferate faster in vitro and are less immunogenic than bone marrow or adipose MSCs, as they express little HLA-DR when activated. Both products are thought to act mainly through cytokines and growth factors that support repair and reduce inflammation in the injured brain.

The authors compare their findings with the 2025 individual participant data meta-analysis of cord blood MNC infusions, which reported a mean gain of 1.42 GMFM-66 points at 12 months from a baseline of 41.9, along with a dose response. They also compare it with a 2025 scoping review of MSC therapy that found a significant effect in five controlled studies. They regard both as consistent with their own analysis. Both are summarised elsewhere on this site.