Neil H. Riordan, Maria Luisa Hincapié, Isabela Morales, Giselle Fernandez, Nicole Allen, Cindy Leu, Marialaura Madrigal, Jorge Paz Rodriguez, Nelson Navarro

First published: 11 June 2019

Data Availability Statement: The data that support the findings of this study are available from the corresponding author upon reasonable request.


Individuals with autism spectrum disorder (ASD) suffer from developmental disabilities that impact communication, behavior, and social interaction. Immune dysregulation and inflammation have been linked to children with ASD, the latter manifesting in serum levels of macrophage‐derived chemokine (MDC) and thymus, and activation‐regulated chemokine (TARC). Mesenchymal stem cells derived from umbilical cord tissue (UC‐MSCs) have immune‐modulatory and anti‐inflammatory properties, and have been safely used to treat a variety of conditions. This study investigated the safety and efficacy of UC‐MSCs administered to children diagnosed with ASD. Efficacy was evaluated with the Autism Treatment Evaluation Checklist (ATEC) and the Childhood Autism Rating Scale (CARS), and with measurements of MDC and TARC serum levels. Twenty subjects received a dose of 36 million intravenous UC‐MSCs every 12 weeks (four times over a 9‐month period), and were followed up at 3 and 12 months after treatment completion. Adverse events related to treatment were mild or moderate and short in duration. The CARS and ATEC scores of eight subjects decreased over the course of treatment, placing them in a lower ASD symptom category when compared with baseline. MDC and TARC inflammatory cytokine levels also decreased for five of these eight subjects. The mean MDC, TARC, ATEC, and CARS values attained their lowest levels 3 months after the last administration. UC‐MSC administration in children with ASD was therefore determined to be safe. Although some signals of efficacy were observed in a small group of children, possible links between inflammation levels and ASD symptoms should be further investigated.

Significance Statement
To the authors’ knowledge, this is the first single‐arm phase I/II clinical trial of repeated dose umbilical cord mesenchymal stem cells administration in children diagnosed with autism spectrum disorder (ASD). Umbilical cord mesenchymal stem cell infusions were safe and generally well tolerated. Forty percent of children showed notable improvements of symptoms as measured by standardized autism diagnosis tools. Whereas other studies have reported links between inflammatory cytokine levels and ASD, this study only observed a possible link in a small group of children, which merits further investigation.


Individuals with autism spectrum disorder (ASD) suffer from developmental disabilities that impact communication, behavior, and social interaction. Although the clinical presentation of this disorder varies in the presence and intensity of the signs and symptoms displayed, children with ASD typically present repetitive behavior and speech patterns, as well as deficits in social interactions and verbal/nonverbal communication. Additionally, anxiety, attention‐deficit/hyperactivity disorder, motor impairments (e.g., hypotonia, clumsiness, toe‐walking), sleep disorders (e.g., insomnia), intellectual disability, and gastrointestinal problems (e.g., chronic constipation, diarrhea, abdominal pain) are also associated with ASD1.

The prevalence of autism, which is approximately four times more frequent in boys than girls, has increased in recent years2, causing a significant economic burden in special education, healthcare costs, and parental productivity loss3,4. Current management of the condition is limited to psychological interventions and other alternative therapies (behavioral, cognitive, and speech therapy)5, and management of symptoms with pharmacotherapy (e.g., selective serotonin reuptake inhibitors [SSRIs], antipsychotic medications6 known for causing adverse effects such as extrapyramidal symptoms, sedation, weight gain, among others1,7,8). However, despite the growing number of cases and the financial and social impact of this condition, the benefits of these interventions may be limited, prompting the need for biologic approaches targeting the etiology of ASD at the cellular and molecular level.

Immune dysregulation has been linked to children with ASD, manifesting in the form of altered T‐cell responses9, elevated plasma cytokine levels10, and significantly lower plasma levels of transforming growth factor β‐111, among others12,13. In particular, intestinal immune dysregulation14 and gastrointestinal symptoms have been observed in children with ASD15-19. Furthermore, brain inflammation may be linked to the pathogenesis of neuropsychiatric disorders such as ASD20, as observed through findings that indicate neurological inflammation, including neural fiber formation21, enhanced oxidative stress22, apoptosis23, and high secretion of amyloid protein breakdown products24. The relationship between inflammation and autism was further evidenced in a study by Al‐Ayadhi and Mostafa, in which children with ASD were found to score higher than neurotypical children in measures of macrophage‐derived chemokine (MDC) and thymus and activation‐regulated chemokine (TARC). Additionally, those with severe autism based on the Childhood Autism Rating Scale (CARS) had significantly higher serum levels than those with mild to moderate autism25.

Mesenchymal stem cells (MSCs) have immune‐modulatory and anti‐inflammatory properties and have been safely used in the treatment of a variety of neurological and autoimmune conditions15, 26-33. In particular, MSCs derived from the Wharton’s jelly of umbilical cord tissue (UC‐MSCs) may possess greater immune‐modulatory activity34 and proliferative capacity compared with other MSCs35,36. The rationale for MSC therapy to treat ASD has been discussed over the past decade37,38; our group proposed the use of stem cell therapy to treat ASD in 200739. Some studies to date have demonstrated the safety of treatment that included MSCs40: of note, the results of a study by Sharma et al. showed that the majority (96%) of children with ASD treated with bone marrow‐derived cells including MSCs showed global improvements including behavior patterns (66%), social relationships (90.6%), and speech, language, and communication (78%)41. In another study, children with ASD treated with UC cells, including MSCs, showed significant differences in nonverbal communication and visual, emotional, and intellectual responses, among other measures42.

In this context, the purpose of this study was to analyze the safety and signals of therapeutic effects of a 9‐month intervention of intravenously administered UC‐MSCs in 20 children diagnosed with ASD.

Materials and Methods

Study Design
In this single‐arm phase I/II clinical trial of 20 subjects with ASD, enrolled subjects received one treatment series every 12 weeks for a total of four treatment series over the course of 9 months (treatment phase). Subjects were then followed for 1 year, with evaluations 3 and 12 months after the last treatment (12‐month and 21‐month visits, respectively). Complete medical and psychiatric evaluations, complete blood count, complete metabolic panel, and infectious disease tests, serum cytokine levels (MDC and TARC), and autism‐specific questionnaires (CARS and Autism Treatment Evaluation Checklist [ATEC]) were administered at each time point during the treatment and follow‐up phases.

During the first visit, in week 1, participants were evaluated for safety and efficacy baseline values, and received 36 million UC‐MSCs intravenously over the course of 1 week, in four intravenous infusions of 9 million viable UC‐MSCs in each infusion. Twelve weeks later, at week 13, the subjects received the same dose of UC‐MSCs and were evaluated for safety and efficacy endpoints. This procedure was repeated at week 25 and week 37 after the start of treatment. The total dose received over the course of treatment was 144 million UC‐MSCs (4 × 36 million). In the follow‐up phase, visits occurred at week 49 (12 months after the start of treatment, 3 months after the last dose) and week 89 (21 months after the start of treatment, 12 months after the last dose).

The study was approved by the Panamanian Institutional Review Board (Comité Nacional de Bioética de la Investigación) and registered with the National Institutes of Health U.S. National Library of Medicine database ( identifier NCT02192749). The study was sponsored by Translational Biosciences. All treatments were administered at the Stem Cell Institute in Panama City, Republic of Panama, under protocol number TBS‐UCMSC‐ASD001. Written informed consent was obtained for all study participants and cord donors.

Subject Population
Stem cell therapy‐naïve children aged 6–16 years were considered for this study if they had a prior diagnosis of autism per the Diagnostic and Statistical Manual of Mental Disorders, 4th edition, as confirmed by the Autism Diagnostic Observation Schedule or the Autism Diagnostic Interview—Revised. Eligible candidates were required to be ambulatory, able to sit still for at least 5 minutes, and have adequate vision, hearing, and arm‐hand‐finger coordination (i.e., be able to point), as well as have normal serum lead and mercury levels at screening and no other uncontrolled medical disorders. Children who were premature (<32 weeks gestation) or significantly small for gestational age were excluded from the trial, as were those with intellectual disability, seizure disorders, auto‐immune conditions, or a history of head trauma. Additionally, candidates were not considered eligible if they had recently made or anticipated any changes in their routine treatment or diet.

UC‐MSC Preparation and Culture
UC‐MSCs used in this study were isolated from human UC tissue from voluntarily donated UCs, obtained from normal healthy births after a rigorous screening process. In brief, a standard risk assessment questionnaire was given to the mothers aged 18–35 years old at the time of delivery, and the donor was screened for infectious diseases (including Human Immunodeficiency Virus (HIV) [1+2]‐Ab and HIV [1+2] Ag‐Ab, V.D.R.L., Hepatitis B (HB)sAg and HB‐anti‐core/IgG‐IgM, cytomegalovirus IgM, Hepatitis A Virus‐IgM, Hepatitis C Virus‐Ab, Chagas‐Ab, Human T‐lymphotropic Virus [1+2]‐Ab, and toxoplasmosis IgM [this disease endemic to Panama is routinely scanned as part of standard of care]). The cells used for this study were manufactured by MediStem Panama, a biotechnology laboratory located in the International Science and Technology Park, City of Knowledge, Panama, following good manufacturing and laboratory practices.

UC‐MSCs were obtained through the enzymatic digestion of UC Wharton’s jelly using collagenase 1.67% (Sigma, C9891, Saint Louis, MO, USA) at 37°C. After isolation, cells were expanded up to passage 5 using α‐MEM (Gibco, 32561‐102, Carlsbad, CA, USA) supplemented with 4 mM GlutaMax (Gibco, 35050‐079, Carlsbad, CA, USA) and 10% inactivated fetal bovine serum (Gibco, 16000044, Grand Island, NY, USA). Vials containing only UC‐MSCs were cryopreserved using 6% hydroxyethyl starch (Claris, G/LVP‐5, Ellisbridge, Ahmedabad, India) containing 10% Dimethyl sulfoxide (Sigma, D2650, Irvine, United Kingdom), first cooled in the −80°C ultra‐freezer at approximately 1°C/minute from 25°C to −80°C in a freezing container (Nalgene, 5100‐0001, Rochester, NY, USA), and then plunged directly into the gas phase of liquid nitrogen. They were kept in quarantine until it was confirmed that they met the requirements for viability (before freezing and after thawing), sterility, mycoplasma, endotoxin, characterization, and differentiation, by testing random vials of the same lot. Before each treatment, cell vials were selected according to the number of total viable cells as obtained by quality controls after freezing and thawing (following syringe preparation procedure), to have the dose of cells required by the protocol. After post‐thaw washing, the dose was adjusted to attain the treatment target of 9 million cells per infusion as closely as possible.

Vials were thawed under controlled conditions and prepared into the corresponding treatment dose of 2.25 million cells per milliliter, suspended in a 4‐ml solution (1 ml 5% dextrose and 3 ml sterile 0.85% saline), for a total of 9 million viable cells per infusion. The procedures were done under strict adherence to aseptic technique to ensure sterility of the prepared syringe and following the results of quality control vials to ensure viability of the cells. Each syringe was inspected for the absence of cell clumps, integrity of the containers, and correct volume. Labels were checked to verify traceability against the provided documents of certificate of analysis of the lot and chain of custody. Viability, characterization, and differentiation methodologies were validated both internally and by a third‐party independent laboratory. Cells were counted and viability was measured using flow cytometer with the Guava ViaCount Reagent (MerckMillipore, 4000‐0041, Hayward, CA, USA) from time 0 to 4 hours at room temperature (20–24°C). Once syringes were prepared, the cells were infused in less than 2 hours, as this was determined by a post‐thaw stability study (data not shown) to be an optimal threshold to preserve stability. Only cells with a post‐thaw viability ≥75% (mean viability 86.5%, SD 3.63%, coefficient of variance 4.20%, median 88.0%, minimum 76.8%, maximum 93.6%); negative for aerobes, anaerobes, and mycoplasma; with an endotoxin level ≤ 3.0 EU/ml; ≥95% positive for CD90, CD73, and CD105 cell surface markers; negative for CD34 and CD45 cell surface markers according to the International Society for Cellular Therapy criteria for MSC43; and with the ability to differentiate into adipocytes, chondrocytes, and osteocytes were used clinically.

Study Endpoints
Safety, the primary endpoint of this study, was assessed at six different time points during the study through complete psychiatric and medical evaluations, safety laboratory exams (complete blood count, complete metabolic panel, and infectious disease tests), occurrence of adverse events and serious adverse events, and their relatedness to the study product.

Signals of efficacy were evaluated by parent‐reported outcomes via the CARS and ATEC tools44, in collaboration with the study pediatric psychiatrist, who evaluated the appearance, behavior, mood, speech, and intellectual functioning of the subjects to supplement parental reports. The second set of efficacy measures, MDC and TARC serum levels, were measured using enzyme‐linked immunosorbent assay in duplicate by RayBiotech, Inc. Service division. Optical density was measured to determine average concentration per milliliter.

Statistical Analysis
Data were analyzed using IBM SPSS software version 25. Mean, SD, minimum, and maximum values were calculated for MDC and TARC levels, and CARS and ATEC scores at six different time points: week 1 (T1, baseline), week 13 (T2, second treatment series), week 25 (T3, third treatment series), week 37 (T4, fourth treatment series), week 49 (12‐month visit), and week 89 (21‐month visit). Missing data were analyzed in order to determine whether data were missing completely at random using Little’s MCAR test. An EM algorithm with a maximum of 25 iterations was used to attempt to replace missing values. To determine whether the treatment had a significant therapeutic effect, a test of difference of repeated measures multivariate analysis of variance (MANOVA) was conducted to determine whether there were significant changes in the mean MDC, TARC, ATEC, and CARS values at any of the six time points for participants who had a complete data set. A level of significance of p < .05 was used for all analyses.


Twenty subjects of diverse ethnicities were enrolled into this study between March 2015 and December 2015. Of these, most (95%) were male, and the average age of enrollees was 10.25 years (Table 1). Average baseline CARS and ATEC scores were 37.48 and 61.10, respectively, and average pretreatment serum MDC and TARC levels were 949.60 and 212.35, respectively. Of the enrolled subjects, 16 completed all four treatment series specified in the study protocol; 296 infusions were administered in total. Subjects received a total dose of 36 million UC‐MSCs at each treatment time point (mean 36.1 million, SD 0.06, coefficient of variance 0.16%, median 36.1, minimum 36.03, maximum 36.16), for a total of 144 million over the course of treatment (mean 144.3 million, SD 0.19, coefficient of variance 0.13%, median 144.3, minimum 144.07, maximum 144.73) for those who completed the treatment series (n = 16). Fifteen subjects were followed to the end of the study period (five did not complete it: two subjects discontinued after receiving two treatment series due to their parents being significantly ill and unable to comply with the study visits, two children discontinued for personal reasons after completing three treatment series, and one was lost to follow‐up after completing the entire treatment phase). Missing data were found to be missing completely at random under Little’s MCAR test (χ2 [131] = 121.60, p = .71). The number of subjects who received treatment and had a fully complete set of efficacy endpoints (all CARS scores, ATEC scores, MDC, and TARC serum levels) at all time points of the study was 10.

Table 1. Demographics of the study population (n = 20)

Image of Demographics Table

No treatment‐related serious adverse events (SAEs) were observed during the course of this trial. There was one instance of an aggression crisis that required hospitalization in a patient with a documented history of severe aggression prior to entering the study. In total, 133 adverse events were recorded for 296 infusions, of which 58 (19.6%) were considered to be related to treatment with UC‐MSCs (Table 2). Most of the adverse events (AEs) observed during the study (56.4%) were qualified as “not related” or “not likely related” to treatment with UC‐MSCs. Mild inflammation, swelling, and/or redness at the infusion site were reported by two subjects as short in duration and self‐resolved, and were determined to be “definitely related” to treatment. AEs that were considered “possibly related” to treatment included moderate increases in tics, obsessive–compulsive behaviors, and aggression reported by six subjects, as well as mild fatigue, headache, fever, and increase in hyperactivity or anxiety. No clinically significant changes in basic hematologic and chemistry laboratory tests were observed throughout the duration of the study.

Table 2. Number and severity of adverse events related to treatment over the course of treatment, for a total of 296 UC‐MSC intravenous infusions administered
Image of Adverse Events Table

Abbreviation: UC‐MSC, mesenchymal stem cells derived from umbilical cord tissue.

The repeated measures MANOVA (Table 3, n = 10) showed that the mean MDC (p = .003, ηp2 = 0.63), TARC (p = .001, ηp2 = 0.70), ATEC (p = .005, ηp2 = 0.60), and CARS (p < .001, ηp2 = 0.77) values were significantly different at the six different time points.

Table 3. Repeated measures MANOVA of differences in mean MDC, TARC, ATEC, and CARS values (n = 10)

Image of MANOVA table

a p < .05.

Abbreviations: ATEC, Autism Treatment Evaluation Checklist; CARS, Childhood Autism Rating Scale; MANOVA, multivariate analysis of variance; MDC, macrophage‐derived chemokine; TARC, thymus and activation‐regulated chemokine.

MDC levels (Fig. 1A, n = 20, 20, 18, 16, 13, and 10 at each time point) remained relatively stable from T1 (mean = 949.60; SD = 165.30) to T3 (mean = 951.33; SD = 200.73), after which a decreasing trend was observed at T4 (mean = 801.23; SD = 383.81) until the 12‐month visit, where the values were halved (mean = 483.46; SD = 343.66) from those measured at T1. Serum MDC levels increased again up to the 21‐month visit (mean = 899.25; SD = 342.78) to levels similar to or lower than those measured at T1.

Image of MDC and TARC levels

Figure 1.

(A): Mean serum macrophage‐derived chemokine levels at the four treatment points and the 12‐month and 21‐month visits (n = 20, 20, 18, 16, 13, and 10, respectively). (B): Mean serum thymus and activation‐regulated chemokine levels at the four treatment points and the 12‐month and 21‐month visits (n = 20, 20, 18, 16, 13, and 10, respectively).

Although showing a decreasing trend when compared with baseline (mean = 212.35; SD = 115.82), TARC serum levels increased slightly between T2 (mean = 186.21; SD = 115.95), and T4 (mean = 195.93; SD = 104.34), after which a decrease was observed at the 12‐month visit (mean = 134.05; SD = 70.09). TARC levels continued to decrease thereafter, with the lowest results seen at the 21‐month visit (mean = 130.90; SD = 63.32) compared with those measured at baseline (Fig. 1B, n = 20, 20, 18, 16, 13, and 10 at each time point).

Scores for ATEC (Fig. 2A, n = 20, 20, 18, 17, 14, and 14 at each time point) and CARS (Fig. 2B, n = 20, 20, 18, 17, 15, and 14 at each time point) followed a decreasing trend during treatment, with the lowest scores observed at the 12‐month visit (ATEC: mean = 39.14, SD = 22.85; CARS: mean = 31.17, SD = 8.79). The values increased at the 21‐month visit, reaching levels similar to or lower than those observed before treatment.