A Phase II Randomized Clinical Trial of Intravenous Umbilical Cord Blood Infusion for Children With Autism Spectrum Disorder:

Major Points and Findings:

This is the largest placebo-controlled, double-blind trial of cord blood in autism, run at Duke University. The primary outcome was negative. The findings usually quoted from it come from exploratory analyses in children without intellectual disability, where type I error was not controlled. The product was cord blood mononuclear cells, not umbilical cord tissue mesenchymal stem cells.

Aim:

To evaluate whether a single intravenous infusion of autologous or allogeneic umbilical cord blood is safe and improves social and communication abilities in children with autism spectrum disorder against placebo.

Methods:

The study was a single-site, randomized, double-blind, placebo-controlled trial (NCT02847182). Of 531 children screened, 180 were randomized.

Children were aged 2 to 7 years (mean 5.47) with DSM-5 autism confirmed by ADOS-2 and ADI-R, negative Fragile X and microarray testing, a qualifying cord blood unit and stable medication. Exclusions covered psychiatric comorbidity, known genetic syndromes or pathogenic copy number variants, CNS infection or HIV, metabolic, mitochondrial or seizure disorders, immunodeficiency, autoimmune cytopenias, chemotherapy-treated malignancy, major sensory impairment, prior cell therapy, and prior IVIG, anti-inflammatory or immunosuppressive treatment.

Children with a banked autologous unit received their own cells (n = 56). The rest received an unrelated allogeneic unit matched at 4 of 6 or more HLA loci (n = 63). Randomization was 2:1 against placebo (n = 61), stratified by age, nonverbal IQ (NVIQ) and cord blood type. Placebo children received cord blood after the 6-month primary endpoint. The target dose was at least 2.5 × 10⁷ total nucleated cells per kg; the actual median was 26.88 × 10⁶/kg for autologous and 38.45 × 10⁶/kg for allogeneic units. The placebo, TC199 with 1% DMSO, looked and smelled like a thawed unit.

The primary outcome was change in the Vineland Adaptive Behavior Scales-3 (VABS-3) Socialization standard score at 6 months. Key secondary outcomes were VABS-3 Communication, the PDDBI Autism Composite, clinician-rated CGI-S and CGI-I, and EOWPVT-4 vocabulary; eye-tracking and EEG were exploratory. Safety was followed to 12 months. Only 101 children rather than the planned 144 met the higher IQ threshold, which the authors say weakened the primary analysis.

Results:

The modified intention-to-treat population was 176 children (119 cord blood, 57 placebo); all 180 were analysed for safety.

The mean 6-month change in VABS-3 Socialization was 3.13 (SD 8.76) with cord blood and 1.98 (SD 8.41) with placebo, a difference of 1.02 points (95% CI −1.67 to 3.71; P = .458). There was no difference by cord blood type (P = .872) and no key secondary outcome differed; every FDR-adjusted P value was .924 or higher. Expectancy effects were large in both arms. CGI-I improvement was 50.1% on placebo and 54.7% on cord blood.

In the post hoc subgroup with NVIQ of 70 or above, 6-month VABS-3 Communication scores were 5.45 points higher with cord blood (95% CI −0.08 to 10.23). CGI-I improvement in this subgroup was 76.9% with cord blood versus 57.1% with placebo, odds ratio 2.36 (95% CI 0.80 to 6.96). By cell source, overall CGI-I improvement was 75.9% with allogeneic cells, 57.1% with autologous and 57.1% with placebo. Allogeneic units delivered a higher dose, confounding source and dose.

On eye-tracking, cord blood recipients had higher odds of looking at toys rather than the actress (OR 1.43; 95% CI 1.15 to 1.78). Sustained attention to moving toys rose more with cord blood, mainly in children without intellectual disability (0.59 seconds; 95% CI −0.14 to 1.32), and on EEG that subgroup showed increased relative alpha (P = .02) and beta1 power (P = .02).

There were 6 serious adverse events, none product-related: viral gastroenteritis, dehydration and aggression on placebo, a concussion in the autologous arm, and PANDAS and dehydration in the allogeneic arm. No deaths, graft-versus-host disease, alloimmunization or product-related infections occurred. In 6 months 535 non-serious events were recorded (485 mild, 45 moderate, 5 severe), at similar rates in both arms (81.5% cord blood, 83.6% placebo). Infusion reactions occurred in 4 placebo children (6.6%, all mild) and 12 cord blood children (10%: 5 mild, 3 moderate, 4 severe). They were more frequent with allogeneic units (9 of 63, 14.3%) than autologous (3 of 56, 5.4%), and all four severe reactions, with bronchospasm and/or facial flushing and swelling, were in allogeneic recipients. One child developed donor-specific HLA antibodies. Worsening autism-related behaviours were recorded in 47.5% of placebo and 46.2% of cord blood children.

Conclusions:

The authors’ wording is that “a single infusion of CB was not associated with improved socialization skills or reduced autism symptoms,” that “more research is warranted to determine whether CB infusion is an effective treatment for some children with ASD,” and that the results “do not currently support the use of CB as a treatment for autism outside a formal or expanded access IND-sponsored clinical trial.”

Their limitations are a much higher placebo response than expected, too few children without intellectual disability, an uneven share of intellectual disability between autologous (35.7%) and allogeneic (53.4%) recipients confounded with the higher allogeneic dose, no control for multiple testing in the exploratory analyses, and high variability between children.

Background Information:

The hypothesis was that CD14+ monocytes in cord blood reduce brain inflammation through paracrine signalling. The group’s open-label phase I study of 25 children (NCT02176317) had reported better social communication at 6 months, greater in children with higher IQ. The Marcus Foundation funded the trial; several authors hold related pending patents.