Impact of Stem Cell Therapy on Brain Metabolic Profile in Cerebral Palsy Assessed by Magnetic Resonance Spectroscopy in a Randomized Clinical Trial

Major Points and Findings:

This paper is a further report from one of the few double-blind, sham-controlled trials of intrathecal cell therapy in cerebral palsy. The Tehran trial randomized 108 children to a single spinal injection of umbilical cord mesenchymal stem cells (UC-MSCs), a single injection of umbilical cord blood mononuclear cells (UCB-MNCs), or a sham procedure. The secondary outcome reported here is whether the cells changed the chemistry of the brain’s white matter, as measured by magnetic resonance spectroscopy.

The main result is negative. Twelve months after one injection, neither cell type changed any measured brain metabolite more than the sham procedure did. Motor scores rose more in the treated groups, but in this reduced sample the difference from sham was not significant. The authors themselves say their subgroup signals need cautious interpretation.

Aim:

To test whether a single intrathecal injection of UC-MSCs or UCB-MNCs alters metabolites in the periventricular white matter of children with spastic cerebral palsy at 12 months. Exploratory questions were whether sex, type of cerebral palsy or gestational age influence the response, and whether metabolite changes track with motor gains.

Methods:

The study was a multicentre, randomized, double-blind, sham-controlled trial at three academic hospitals in Tehran and Bandar Abbas, Iran, running from August 2017 to December 2019 (IRCT201706176907N13, NCT03795974). The primary endpoint was change in GMFM-66 at 12 months, and spectroscopy was a pre-specified secondary endpoint.

Eligible children were aged 4 to 14 with cerebral palsy at GMFCS level II to V and white matter lesions on MRI, such as periventricular leukomalacia. Exclusion criteria were:

• Non-spastic cerebral palsy (ataxic, athetoid or mixed)
• Other neurological conditions such as untreated epilepsy, or congenital infection
• Severe anaemia, a coagulation disorder, malignancy, kidney or liver failure
• Any previous cell infusion

Both cell products were allogeneic and came from healthy mothers after uncomplicated full-term vaginal delivery, with donor blood screened for HIV, hepatitis B and C and cytomegalovirus. UC-MSCs from Wharton’s jelly were cultured, checked for sterility, endotoxin, mycoplasma, chromosomal abnormalities and surface markers, and given as a fixed dose of 20 × 10⁶ cells. UCB-MNCs from the Royan Cord Blood Bank were HLA-matched, thawed, washed and given at 5 × 10⁶ cells per kg. No immunosuppression was used.

The injection was given under sedation with the child lying on one side, through a lumbar puncture at L3-L4 or L4-L5. Two millilitres of cerebrospinal fluid were withdrawn, of which 1 mL was kept for analysis. Then 2 mL of cell suspension was injected slowly over 2 minutes and the remaining 1 mL of fluid was returned. The sham group received a needle prick to the lower back with a similar-looking needle. All children were monitored in hospital for 24 hours, lying 10 degrees head-down. All three groups received Bobath-based rehabilitation, 75 minutes three times a week.

Measurements: Spectroscopy of the periventricular white matter was performed on a 1.5T scanner before and 12 months after the intervention. Metabolites were N-acetyl aspartate (NAA, neuronal health), choline (Cho, membrane turnover), creatine (Cr, energy metabolism), myo-inositol (mI, glial activity), and the ratios NAA/Cho and NAA/Cr. GMFM-66 was scored at 1, 3, 6 and 12 months by blinded paediatric neurologists.

Results:

About a third of the scans could not be used. Of 108 randomized children, 35 were excluded for poor-quality spectra, leaving 73 (UCB-MNC 27, UC-MSC 26, sham 20). Mean ages were 7.2 to 8.6 years, and 77% to 85% of each group had quadriplegia. The sham group contained far more term-born children (65%, against 22% and 27%, p = 0.005).

The spectroscopy analysis showed no treatment effect. NAA, Cho and Cr rose significantly over the year in all children regardless of group, which the authors attribute to ordinary brain maturation. There was no significant time-by-treatment interaction for any metabolite or ratio, and no difference between groups at baseline or at 12 months.

Exploratory subgroup analyses, adjusted for baseline values and age, gave the following:

• Children with diplegia given UC-MSCs had a higher NAA/Cho than sham (p = 0.02).
• Preterm-born children given UC-MSCs had lower Cho than sham (p = 0.009).
• Term-born children given UCB-MNCs had lower mI than sham (p = 0.03).
• Sex made no difference.

These come from many comparisons in small subgroups (only 13 children had diplegia).

Motor function: In these 73 children GMFM-66 rose significantly from baseline to 12 months within each cell group, while the change in the sham group was not significant. The median 12-month gain was 11 points with UC-MSC (IQR 4 to 15), 9 with UCB-MNC (IQR 3 to 17) and 6 with sham (IQR 5 to 8). Between groups this did not reach significance: p = 0.06 for UC-MSC versus sham and p = 0.44 for UCB-MNC versus sham. The authors suggest that losing 35 children left the comparison underpowered, and note that the full 108-child dataset had shown significant motor benefit for both cell types.

Metabolite changes were also compared with motor gains. Rising NAA/Cho went with larger GMFM-66 gains in the UC-MSC group (p = 0.02), and rising NAA/Cr with larger gains in the UCB-MNC group (p = 0.01). Rising choline went with smaller gains in the UCB-MNC group (p = 0.03). No such associations appeared in the sham group.

Safety: This paper does not report adverse events. They were covered in the trial’s earlier publications.

Conclusions:

A single intrathecal injection of UC-MSCs or UCB-MNCs had no significant overall effect on white matter metabolites at 12 months. The authors read their exploratory results as a reason to look for “responder subgroups”, for example children with diplegic as opposed to quadriplegic cerebral palsy, and to match cell type to patient. They do not present these results as proof of benefit.

The limitations they state are small groups and the exclusion of 35 participants, which cut statistical power for both the imaging and the motor comparisons. There was only one follow-up scan, and only one injection was given, which may not be enough to shift brain metabolites. They propose larger trials with repeated injections and longer follow-up.

The baseline table shows one further imbalance that the authors do not list. The sham group started with a much lower median GMFM-66 (25, against 64 and 75), so the three groups in this 73-child sample were not well balanced.

Background Information:

Metabolites: Low NAA suggests neuronal loss or dysfunction. Choline rises when cell membranes are being broken down or rebuilt, as in gliosis or demyelination. Myo-inositol reflects glial activation and has been linked to persistent neuroinflammation. Earlier studies have linked a higher NAA/Cho ratio to better outcomes in preterm infants and children with cerebral palsy.

Previous imaging reports from the same trial found improved white matter integrity on diffusion tensor imaging in the corticospinal tract and posterior thalamic radiation after both cell types, while conventional MRI showed no change in visible lesions or atrophy.

On severity and response, the authors note that recent reviews and trials consistently report better responses in children with less severe motor impairment, which is more typical of diplegia than quadriplegia. They find this biologically plausible, since more surviving neural tissue leaves more capacity for recovery.