Study design
This multi-center, population-based randomized double-blind sham-controlled trial with single intrathecal stem cell injection was conducted in Children’s Medical Center and Imam Reza hospital in Tehran province and Bandar Abbas pediatric hospital in Bandar Abbas province, Iran. The study was divided into four phases: [1] initial screening phase, [2] baseline phase, [3] double-blind treatment phase, and [4] follow-up phase.
Inclusion and exclusion criteria
Males and females aged 4 to 14 years old who were diagnosed with spastic CP according to standard criteria [16], gross motor function classification system (GMFCS) level 2–5, and white matter lesions in brain magnetic resonance imaging (MRI) (e.g., periventricular leukomalacia) were included in the study. Individuals with other types of CP (e.g., athetoid, ataxic, or mixed CP), co-morbid neurological disorders (e.g., untreated epilepsy), or congenital infections (TORCH Syndrome) were excluded. Severe anemia (hemoglobins< 8 mg/dl), coagulation disorders, history of malignancy, prior cell infusion, renal insufficiency, and liver failure were other exclusion criteria.
The ethics committee of Tehran University of Medical Sciences approved the final methods (Number: IR.TUMS.VCRREC.1996.2506). Parents of participants had access to all information of the trial and a printed protocol of the study was given to them. They were informed that participation was optional, and withdrawal was possible whenever they asked for. Written informed consent was achieved from parents before the initiation of any study procedures. We also explained the protocol to children and assent was obtained. The study was registered with ClinicalTrials.gov (NCT03795974) and Iranian registry of clinical trials; irct.ir (IRCT201706176907N13).
Rehabilitation therapy
All individuals were under rehabilitations during the study. One approach and technique of rehabilitation was established and conducted for included participants to reduce confounders. The Bobath concept was used in this study that aimed to affect muscle tone and improve the postural alignment by specific handling techniques [24, 25]. Each session lasted for 75 min and participants attended three times per week for rehabilitation during the study.
Randomization and blinding
All included participants were randomly assigned in 1:1 ratio using permuted block randomization via interactive web response system to receive either UCT-MSC or sham procedure, respectively. The responsible statistician was masked to the clinical data of cases. Personnel staff responsible for cell preparations was not masked, but they had no contacts with participants, parents, or investigators. They also had no information about the clinical and imaging characteristics of patients. All participants, their parents, and investigators were blinded during the study unless serious adverse events occurred that emergent evaluations and treatments by medical staff were essential. A small needle prick to the lower back was performed as a sham procedure. All individuals were sedated to prevent awareness and to reduce spasticity during the procedure.
Procedures
Initial screening and baseline phases
Physical and neurological examinations were performed on children and adolescents with CP. GMFCS was used for initial functional assessment and brain MRI was also performed. Screening tests included blood count (e.g., hemoglobin, white blood cells, and platelets), serum chemistry (e.g., liver function test, creatinine, and urea), prothrombin time, partial thromboplastin time, and electro-encephalography (EEG) (phase 1). Gross motor function measure (GMFM)-66, modified Ashworth scale (MAS), pediatric evaluation of disability inventory (PEDI), and CP quality of life (CP-QoL) were used to evaluate the baseline clinical characteristics of eligible participants. Diffusion tensor imaging (DTI) was also performed to track white matter abnormalities in motor fibers (phase 2).
Motor function assessment
GMFCS is a 5-level clinical classification system to describe the motor function of people with CP [26, 27]. The distinctions between levels are based on functional abilities (Supplement 1). A prior study reported excellent reliability of GMFCS for children aged 2 to 12 years old with CP (kappa: 0.75) [26]. The GMFM-88 was developed to measure changes in gross motor function over time or with treatment in people with CP [28]. The GMFM-66 was found through Rasch analysis to best describe the gross motor function of children with CP of varying abilities and is a 66 item subset of the original 88 items [29]. It has a unidimensional scale providing interval scaling rather than the ordinal scaling of the GMFM-88. It was shown that inter-rater reliability of Farsi version of this scale for all dimensions was between 0.97 and 0.99 and the intra-rater reliability was 0.99 [30]. Cronbach’s alpha coefficient for all dimensions was between 0.78 and 0.94 [30].
Spasticity assessment
The MAS can be determined according to the examination of muscle tone [31] (Supplement 2). The scores are measured based on the level of resistance during the passive movement of the antagonist muscles. The elbow flexor, wrist flexor, knee extensor, hip adductor, and ankle plantar flexor were examined on the spastic side(s), and the mean MAS score was recorded in each individual. Participants were in sitting position to examine hip adductor and supine position to test other muscles. The interpretation of MAS should be with extreme caution. Several limitations to MAS were described previously [32]. Low inter- and intra-rater reliability of MAS in children with CP were estimated [33, 34], and the validity of the scale was poor [35]. The scale, however, quantify muscle tone in ordinal numbers and can be easily used in clinical settings.
Disability assessment
PEDI was developed to assess the performances of children with disabilities in 3 dimensions including self-care, mobility, and social function. The Farsi version of PEDI in children with CP was reported to have high internal consistency (Cronbach’s alpha 0.94 to 0.98). The results of test-retest reliability were excellent in self-care (0.99) and social performance [1], and good in mobility dimension (0.66) [36].
Quality of life assessment
CP-QoL was designed to evaluate the well-being across different domains of life in children and adolescents with CP. The CP-QoL-child form with primary caregiver proxy report was used in this trial. The domains included family and friends, participation in activities, communication, physical health, special equipment, pain and bother, access to services, and family health. Good internal consistency (Cronbach’s alpha 0.61 to 0.87) and moderate to good test-retest reliability (0.47 to 0.84) in all domains were reported in the Farsi version of questionnaire [37].
Brain imaging
MRI was performed on 1.5T scanner (Philips Ingenia, Eindhoven, the Netherlands). All individuals had to be remained still and cover their ears with sponge earplugs for hearing protection. Intravenous propofol (2 mg/kg dose) was used for sedation if participants did not cooperate and moved during imaging process. Intravenous thiopental (5 mg/kg dose) was used if there was sensitivity history to propofol. Possible side effects of propofol or thiopental were explained to the parents, and informed consent was obtained before using the medications. Heart and respiratory rates as well as oxygen saturation were monitored before and during the procedure. The protocol of MRI included 3D T1-weighted imaging (TR: 9.5 ms, TE: 4.6 ms, flip angle: 8 °, FOV: 210 × 210 mm2, voxel size: 1 × 1 × 1 mm3) and 2D T2-weighted sequence (TR: 4000 ms, TE: 110 ms, flip angle: 90 °, FOV: 230 × 230 mm2, voxel size: 0.8 × 0.8 × 3.5 mm3). The 2D T2-weighted sequence was used to acquire DTI data (TR: 4228 ms, TE: 94 ms, flip angle: 90 o, FOV: 224 × 224 mm2, voxel size: 2.5 × 2.5 × 2.5 mm3).
Image post-processing
DTI processing was performed using ExploreDTI software [38]. Post-processing included a cubic interpolation and robust estimation of tensors to correct for subject motion, eddy current, and EPI distortion. Non-rigid registration on the structural images was also performed. A whole-brain white matter tract construction was carried out for each participant using a linear interpolation. Seed point resolution was set at 1 mm × 1 mm × 1 mm with a seed fractional anisotropy threshold of 0.2 and an angle threshold of 50°.
Region of interest (ROI)-based tractography was performed. Predefined tracts such as corticospinal tract (CST) and posterior thalamic radiation (PTR) were isolated in both hemispheres. For segmentation of CST, first ROI “AND” was drawn at the pons level and the second ROI “AND” was drawn at the centrum semi oval level. To reject the fibers that project to the cerebellum via the middle cerebellar peduncle, ROI “NOT” was drawn. For illustration of PTR, first ROI was drawn at retro-lenticular part of the internal capsule and the second ROI was at the thalamus using “AND” operation. All other tracts that not related to the PTR and were outside to ROIs were eliminated by ROI “NOT” (Fig. 1). The mean value of fractional anisotropy (FA) and mean diffusivity (MD) were, then, estimated for each separated tract in both hemispheres. The two hemispheres of each participant were compared to each other, and data of the most affected tracts were used in the analysis.
Cell preparation
Allogenic UCT-MSCs were derived from umbilical cord tissues of unrelated donors. The ethics committee of Tehran University of Medical Sciences approved the methods (Number: IR.TUMS.VCRREC.1996.2506). The donors were selected from full-term healthy mothers who had normal vaginal delivery without complication. The written informed consents were provided to use the umbilical cords for medical research purposes. The blood samples of donors were collected and tested for reactive transmissible infectious agents including human immunodeficiency virus, hepatitis B virus, hepatitis C virus, and cytomegalovirus. After birth, the umbilical cords were detached from the placenta and put in the plates containing phosphate-buffered saline and streptomycin. They were transferred to the laboratories of Royan at 4oC within 24 h. After removal of the umbilical arteries and vein, the umbilical cords were cut into 2 to 4 cm pieces to obtain Wharton’s jelly. The Wharton’s jelly was cut into 1 mm3 fragments. Collagenase and hyaluronidase were, then, added. The UCT-MSCs were isolated by centrifugation and cultivated in DMEM (Dulbecco’s modified Eagle’s medium) supplemented with 10% fetal bovine serum, 100 U/mL penicillin, 100 mg/mL streptomycin, and 2 mmol/L l-glutamine and, then, incubated at 37oC in a humidified tissue culture incubator in 5% CO2 and 95% air. The cells were passaged and collected at the fourth generation. Strict quality controls were performed before any clinical application. The products were analyzed for bacterial contamination using a BACTEC instrument (BD Bactec; BD Diagnostics, Franklin, NJ, http://www.bd.com). Furthermore, the amount of bacterial endotoxin was determined by limulus amebocyte lysate (LAL) kit (Lonza, Switzerland) and an ELISA reader (Amersham, USA). The cultivated cells were also tested to detect mycoplasma species by nested PCR, and karyotyping was performed to identify any chromosomal abnormalities. These procedures were conducted according to recommendations for cell and tissue therapy promotion and validation tests of the Iranian Health Ministry Pharmacopoeia Commission and the Department of Health and Human Services Food and Drug Administration.
The monoclonal antibodies against the cell surface markers CD11b, CD29, CD31, CD34, CD45, CD73, CD90, and CD105, as well as isotype control antibodies (eBioscience) were used to stain the UCT-MSCs for 1 h, and the analysis was performed using FACSCalibur (BD Biosciences). The flow cytometry analysis of UCT-MSCs was shown in Fig. 2.
Double-blind treatment and follow-up phases
Single dose of 2 × 107 cells were injected via intrathecal route in the experimental group. Cell suspensions were filtered through a cell strainer and transferred into storage vials. The cells were suspended in normal saline by unmasked personnel staff and then transferred for injection. Participants lied down on their sides (lateral decubitus position) with their knees drawn up to their chests. All participants were sedated and lumbar puncture was performed with 21 or 22 gauge spinal needles between lumbar level 2 and 5 intervertebral space after washing the back with iodine. The specific lumbar level was determined individually for each participant based on anatomical consideration. After assurance of placing the spinal needle in sub-arachnoid space, 2 mL of cerebrospinal fluid (CSF) was collected and 1 mL was sent to determine baseline CSF characteristics. The 2 mL of cells were, then, injected slowly within 2 min. The 1 mL of remained CSF was infused in the last step and after cell injection. Sham procedure was performed in the control group. The puncture site was covered with a band-aid in all individuals (phase 3). All participants were hospitalized in child neurology departments. They were placed in 10o trendelenburg position to maximize the distribution of cells in CSF. No immunosuppressive medications were used. Heart rates, temperature, blood pressure, and respiratory rates were monitored for 24 h after the intrathecal injection, and individuals were discharged if there were no side-effects. The GMFM-66 and MAS were assessed 1, 3, 6, and 12 months after the intervention. The PEDI and CP-QoL were evaluated 6 and 12 months after the procedure. MRI and DTI were also performed 12 months after treatment (phase 4).
Outcomes
Primary endpoints were the mean changes in GMFM-66 scores from baseline to 12 months after double-blind treatment phase. The mean changes in MAS, PEDI, and CP-QoL domains were also evaluated. The secondary endpoints were the mean changes in FA and MD of CST and PTR fibers from baseline to 12 months after the double-blind treatment phase. The differences between groups were also assessed.
Safety endpoints were the adverse events. All participants and their parents were asked to report any side effects during the follow-up visits. A phone number was given to the parents so adverse events could be reported as soon as possible. Parents were asked to bring their children to the emergency department if any serious complication occurred.
Statistical analysis
Sample size was estimated based on the mean changes in GMFM-66 scores. It was calculated using repeated measures analysis of variance (ANOVA) by G*Power 3.1 software (University of Kiel, Germany). Effect size of 0.25, two-sided α (the probability of type I error) of 0.05, and β (the probability of type II error) of 0.20 were considered, and total sample size of 72 individuals was estimated after considering 10% drop-out rate to provide at least 80% power (Supplement 3).
The statistician was blinded to study groups. Numeric variables were reported as means with standard deviation (SD) or standard error of the mean (SEM). Categorical variables were presented as percentages and were compared between groups using Pearson’s chi-squared test (gender, type of CP, and GMFCS). Kolmogrov-Smirnov test was performed to report the distribution of variables. Two sided significance (P value) lower than 0.05 showed the non-normal distribution (GMFM-66, PEDI, and CP-QoL) and higher than 0.05 represented normal distribution of data (MAS and ROI-based data). Intention to treat approach was used and all participants who were randomized were included in the statistical analysis. Multiple imputation was conducted using Markov chain Monte Carlo to handle missing data. Generalized estimating equations (GEE) model was used to compare GMFM-66, PEDI, CP-QoL, and MAS mean scores between groups [39]. It was assumed that the interaction was between the intervention groups and time measurements. Exchangeable structure was considered for working correlation matrix and linear model was used. The model was adjusted to covariates including type of CP, GMFCS, gender, age, and weight of participants. Independent sample t test was conducted to compare numeric variables in baseline and DTI data between groups. Statistical analyses were performed using the IBM SPSS Software, version 25.0 (SPSS Inc., Chicago, IL) and GraphPad Prism version 7.04. Two-sided significance testing was conducted, and P values< 0.05 were considered statistically significant. Cohen’s d test with 95% confidence interval (CI) was used to measure the effect sizes that were classified as small (d 0 to 0.20), medium (d 0.20 to 0.50), and large (d> 0.50) using R statistical package (R Core Team, 2013).