Neural stem cell transplantation in progressive multiple sclerosis: STEMS study long-term outcomes

STEMS phase I trial design and population

STEMS was a prospective, non-randomized, single-centre, open-label phase I clinical trial (NCT03269071, EudraCT2016-002020-86) evaluating safety of hfNPC transplantation in PMS [8]. Patients were enrolled according to inclusion and exclusion criteria, as follows: (1) age 18–55 years; (2) diagnosis of MS according to revised 2010 McDonald criteria [11], with a progressive course according to 2013 Lublin definition [1]; (3) failure or intolerance to all approved therapies; (4) Expanded Disability Status Scale (EDSS) score [12] ≥ 6.5 and disease duration 2–20 years; (5) evidence of progression of the disease, defined by an increase of ≥ 0.5 EDSS points in the last 12 months; (6) presence of oligoclonal bands in the CSF required for primary PMS; (7) no treatment with any immunosuppressive therapy within 3 months before screening or with interferon-beta, glatiramer acetate or corticosteroids within 30 days before screening; (8) no relapses within 30 days before screening.

Briefly, STEMS-patients were consecutively enrolled in 4 treatment cohorts (TCs). Given the first-in-human, phase I nature of the study, a standard 3 + 3 design, was used. Each TC received a single intrathecal administration of one of the four increasing treatment doses of hfNPCs (TC-A: 0.7 × 10⁶, TC-B: 1.4 × 10⁶, TC-C: 2.8 × 10⁶ and TC-D: 5.7 × 10⁶ ± 10% viable cells per kilogram of body weight), derived from a single 10–12 weeks post-conception human fetus (elective abortion), expanded under good manufacturing practice conditions, and administered through lumbar puncture. To prevent rejection of transplanted hfNPCs, they received immune suppression with tacrolimus from day 7 before transplantation until week 96 after. All patients in the trial discontinued tacrolimus at the end of the 2-year follow-up, as per protocol, or according to the timelines reported in the phase I trial for specific cases. Primary outcome was to assess feasibility, safety and tolerability of the procedure over a period of 96 weeks. Secondary exploratory outcomes included longitudinal variations from baseline (pre-hfNPC administration) of clinical (EDSS), neuropsychological (Symbol Digit Modalities Test [SDMT]), optical coherence tomography (OCT), plasma biomarkers and MRI variables over a period of 96 weeks of follow-up (2Y-FU). The results of this phase I clinical trial have already been described [8].

Long term follow-up design and population

This retrospective observational study collected long-term follow-up (LT-FU) data of the patients enrolled in the phase I study, who underwent continuous monitoring according to clinical practice. All clinical assessments (including neurological examinations, SDMT, 9HPT, and adverse events) were collected retrospectively from the medical records, as documented during routine clinical visits. STEMS-patients with an available neurological visit and MRI performed according to clinical practice at least 4.5 years after hfNPC transplantation were assessed. OCT scans were evaluated when available. Available serum samples collected and stored in the INSPE Biobank during routine clinical care and obtained at least 4.5 years after the hfNPC transplantation have been analyzed for neurofilament light chain sNfL, tau, glial fibrillary acidic protein (GFAP), ubiquitin C-terminal hydrolase-L1 (UCH-L1) content. Retrospective data have been collected from the end of the phase I study to July 2024. Although a baseline MRI scan prior to hfNPC transplantation was available for all STEMS patients, a standardized pre-transplant MRI assessment was lacking, precluding direct estimation of individual atrophy rates before transplantation. For this reason, demographic, clinical and MRI data from three longitudinal cohorts – healthy controls (HC), early MS, and PMS - were retrospectively collected as a reference for atrophy rates at LT-FU.

Clinical evaluation

At LT-FU visit (last available neurological visit according clinical practice at least 4.5 years after transplantation), STEMS-patients underwent a comprehensive neurological evaluation with collection of demographics (age and sex) and clinical variables, including clinical relapses, possible disease modifying treatment initiation and EDSS score [12]. AEs were collected and their rate and nature have been classified according to the Common Terminology Criteria for Adverse Events version 5.0 and attributed to the investigational medicinal product accordingly to National Cancer Institute guidelines of AE reporting requirements. As a measure of cognitive impairment, the oral SDMT has been collected to quantify the speed of information processing, by asking the patient to associate symbols and numbers as quickly as possible, an essential cognitive function early affected in MS [13, 14]. The patient’s score is estimated as the number of correct associations produced in 90 s. The 9-HPT evaluates hand function and coordination. To conduct this timed test, the patients are seated with a peg board in front of them. The board has nine holes on one side and a shallow round dish on the other side, and the patients are asked to place the pegs into each hole on the peg board, one at a time, and remove them again one at a time, while the test speed is measured. The patients were tested twice for each arm, the dominant arm first. An average time was calculated separately for the dominant and non-dominant hands.

OCT

OCT was performed using spectral domain Spectralis OCT (Heidelberg Engineering, software version 5.8). Standard peripapillary retinal nerve fiber layer (pRNFL) was measured on a standard 3.5-Ø-mm circle scan protocol centered on the optic disc by the scanning technician. Global pRNFL thickness was automatically measured through the pre-installed software provided by the manufacturer. As previously described [8], for subjects with a history of unilateral optic neuritis (ON), we studied pRNFL of unaffected eye, while for subjects with an history of bilateral or no ON, we considered pRNFL mean of both eyes.

Reference cohorts for the evaluation of atrophy

Since standardized MRI scans of STEMS-patients before the transplantation were lacking, we retrospectively collected demographic, clinical and MRI data from three longitudinal cohorts as a reference for atrophy rates at LT-FU (Table 3). The MRI data related to the year prior to NPCs transplantation summarized in Fig. 2a were collected retrospectively and refer exclusively to the presence of new T2 lesions or gadolinium-enhancing lesions. Since these MRIs were performed by using scanners and acquisition protocols different from those used in the study, they do not allow for a longitudinal analysis of brain volumes. From the local database at the Neuroimaging Research Unit of IRCCS San Raffaele Scientific Institute in Milan, we selected:

(1) 55 patients studied at time of their clinically isolated syndrome suggestive of MS (early MS). Inclusion and exclusion criteria were: age between 18 and 60 years, a typical clinical presentation of relapsing-onset MS, baseline neurological evaluation and brain MRI performed within 5 months of the clinical onset. Patients underwent clinical and brain MRI evaluation after 2-years of follow-up; (2) 19 age- and sex-matched PMS patients (3 PPMS and 16 SPMS) with a neurological evaluation and brain MRI performed both at baseline and after at least one year of follow-up. The 58% of PMS patients were under treatment with disease modifying therapies; (3) 23 age-matched HC with the following inclusion criteria: (a) age ≥ 45 years at baseline; (b) at least 2 longitudinal MRI acquisitions on the same scanner, with a minimum follow-up of 1.5 years; (c) no history of major neurological or psychiatric disorder.

MRI acquisition and analysis

All brain MRI scans of STEMS-patients and early MS were acquired on a 1.5-Tesla MR scanner (Achieva dStream, Philips Medical Systems), with the following sequences, respectively for the two cohorts: 3D or 2D fluid-attenuated inversion recovery (FLAIR): repetition time (TR) = 4800 or 8000 ms, echo time (TE) = 277 or 90 ms, inversion time (TI) = 1660 or 2500 ms, echo train length (ETL) = 171 or 9, flip angle (FA) = 90°, field of view (FOV) = 230 × 230 × 184 mm³ or 250 × 250 mm², 160 sagittal or 44 axial slices, voxel size = 1.15 × 1.15 × 1.15 mm³ or 0.98 × 0.98 mm² with slice thickness = 3 mm and gap = 1 mm, matrix = 200 × 200 × 160 or 256 × 256, number of signal averages (NSA) = 2 or 1; 2D T2-weighted turbo spin echo (TSE): TR = 4446 or 3900 ms, TE = 100 or 90 ms, ETL = 15 or 6, FA = 90°, FOV = 230 × 192 mm² or 250 × 250 mm², 44 axial slices; slice thickness = 5–3 mm, gap = 1 mm, pixel size = 0.6 × 0.9 mm² or 0.98 × 0.98 mm², matrix = 384 × 224 or 256 × 256, NSA = 3 or 1; 3D T1-weighted turbo field echo (TFE) or magnetization-prepared rapid gradient echo (MPRAGE): TR = 8 ms, TE = 3.7 ms, TI = 1000 ms, ETL = 171, FA = 8°, FOV = 256 × 256 × 192 mm³, 160 sagittal slices, voxel size = 1.0 × 1.0 × 1.2 mm³, matrix = 256 × 256 × 160, NSA = 1; Post-contrast 2D T1-weighted TSE: TR = 580 or 600 ms, TE = 12 or 15 ms, FA = 69°, FOV = 230 × 183 mm² or 250 × 250 mm², 22 or 44 axial slices, slice thickness = 5–3 mm, gap = 1 mm, pixel size = 0.85 × 1.12 mm² or 0.98 × 0.98 mm²; matrix = 272 × 164 or 256 × 256, NSA = 2, 5 min after the injection of gadobutrol 0.1 ml/kg (for STEMS-patients, see [8] for further details on the MRI protocol). STEMS patients underwent MRI acquisition at baseline (pre-hfNPC administration) and at 96 weeks of follow-up (2Y-FU). For LT-FU, the last available MRI scan acquired according to clinical practice on the same MR scanner (Achieva dStream, Philips Medical Systems) and performed at least 4.5 years after hfNPC transplantation was assessed.

The brain MRI scans of HC and PMS were acquired on a Intera and Achieva (Philips Medical Systems, Best, The Netherlands) or a Signa HDxt (GE Healthcare, Milwaukee, USA) 3.0-Tesla scanners. 3D T1-weighted and T2-weighted scans were acquired using the following protocol (both at baseline and at follow-up): 3D T1-weighted TFE or inversion recovery fast spoiled gradient echo (IR–FSPGR): TR = 7 ms, TE = 2.85 to 3.2 ms, TI = 650 to 900 ms, FA = 8°, FOV = 256 × 240 or 256 × 256, voxel size = 1.0 × 1.0 × 1.0 mm³ or 1.0 × 1.0 × 1.2 mm³; 2D dual-echo or FLAIR T2-weighted sequence: TR = ranging from 2900 to 9002 ms, TE = 16 to 120 ms, FA = 90°, FOV = 243 × 243 mm² or 256 × 256 mm²; slice thickness = 3 mm, pixel size = 1 × 1 mm².

At follow-up, subjects were carefully repositioned, with slices running parallel to a line that joins the most infero-anterior and infero-posterior parts of the corpus callosum, covering the entire brain from the skull base to the vertex.

For MRI analyses, all sagittal acquisitions were reoriented in axial plane and the exceeding portion of the neck was removed.

Focal lesion assessment. In STEMS-, early MS and PMS patients, focal brain T2-hyperintense white matter (WM) lesions were identified using a fully automated deep-learning approach on co-registered 3D FLAIR and 3D T1-weighted input images (nicMSlesions, relase 0.2, Python version 2.7) [15]. Lesion segmentation masks were carefully reviewed by an experienced observer, blinded to patients’ identity, and eventually edited to remove false-positive lesions. For HC, the presence of WM lesions was visually checked and eventually segmented by the same experienced observer. T2-hyperintense lesion volume was then obtained from lesion segmentation masks. For STEMS-, early MS and PMS patients, the number of new/enlarging brain T2-hyperintense lesions were counted in respect to the previous visit. The percentage of T2-hyperintense lesion volume change was also obtained. At all timepoints, Gd-enhancing lesions were manually counted on post-contrast T1-weighted scans.

Brain atrophy analysis

Whole-brain, gray matter (GM) and WM normalized volumes were measured on 3D T1-weighted images using the SIENAX version 2.6, FSL 5.0.1 software [16], after applying lesion filling technique to replace intensity values within lesions so that they resemble normal tissue [17]. The percentage brain volume changes (PBVCs) were estimated at follow-up timepoint vs. the previous visit using the FSL SIENA software [16]. The percentage of GM and WM volume changes (PGMVC and PWMVC) were calculated using the segmentation tool and pairwise longitudinal registration in the Statistical Parametric Mapping (SPM) toolbox version12, MATLAB version 8.0, to segment the tissues and quantify volume changes from images of the Jacobian determinants derived from the non-linear registration between two visits [18, 19]. All percentage of longitudinal brain tissues volume changes were annualized, dividing by the follow-up duration.

For a better visual assessment of brain atrophy rates over time after transplantation, MRI PBVC, PGMVC and PWMVC quantifications at week 48 of the STEMS protocol were also retrieved.

Plasma biomarker analysis

Available samples collected and stored in the INSPE Biobank at least 4.5 years after the hfNPC transplantation have been analyzed for sNfL, tau, GFAP, UCH-L1 content using Simoa Human Neurology 4-Plex A (N4PA, 102153) digital immunoassay on an HD-1 single-molecule array (Simoa) instrument according to instructions from the manufacturer (Quanterix). The measurements were performed by a technician blinded to clinical data, in one round of experiment using one batch of reagents. Plasma samples were diluted four times. This is a two-steps immunoassay, in which target antibody coated paramagnetic beads are combined with sample and biotinylated detector antibody in the same incubation. Target molecules present in the sample are captured by the antibody-coated beads and bind with the biotinylated antibody detector simultaneously. After a wash, a conjugate of streptavidin-ß-galactosidase (SBG) is mixed with the beads. SBG binds to the biotinylated detector antibodies, resulting in enzyme labelling of the captured target. After a final wash, the beads are resuspended in a resorufin ß-d-galactopyranoside (RGP) substrate solution and transferred to the Simoa Disc. Individual beads are then sealed within microwells in the array. If the target has been captured and labeled on the bead, ß-galactosidase hydrolyzes the RGP substrate in the microwell into a fluorescent product that provides the signal for measurement. A single-labeled target molecule results in sufficient fluorescent signal in 30 s to be detected and counted by the Simoa optical system.

Statistical analysis

Statistical analyses and graphs were produced using R software (version 4.3.3) and with GraphPad Prism (version 8.0, GraphPad Software). All variables are summarized as median values or changes (with inter-quartile ranges) or absolute numbers (and percentages), according to the type of data.

Clinical, biological and neurophysiological variables of STEMS-patients were compared over time using a Wilcoxon Signed Ranks Test for paired samples. The yearly rate of changes (slope) after transplantation of EDSS, SDMT and pRNFL measures were estimated within subjects. The subjects’ specific slopes two years after transplantation and between LT-FU and 2Y-FU were compared by a non-parametric exact Wilcoxon test.

Demographic variables were compared among STEMS-patients and reference cohorts using Fisher’s exact test for categorical variables and the Independent-Samples Kruskal-Wallis test with post-hoc comparisons for continuous variables. Changes of all MRI measures were estimated between LT-FU and 2Y-FU for STEMS-patients, and between follow-up and baseline visit for reference cohorts. The within-patient longitudinal changes of MRI variables were compared by a non-parametric exact Wilcoxon Signed Ranks Test for paired samples. The associations of longitudinal variations of clinical and MRI measures with the number of injected cells at baseline were obtained using multivariate regression analysis, adjusting for baseline scores. Longitudinal changes of clinical and MRI variables among STEMS-patients and reference cohorts were compared using multivariate ANCOVA models corrected for age, sex, normalized brain tissue volumes, EDSS and MR scanner (as appropriate). To account for the potential imbalance in clinical phenotype distribution between STEMS and PMS groups, a sensitivity analysis was performed including clinical phenotype (SPMS vs. PPMS) as an additional covariate in the statistical models.

Longitudinal comparisons of clinical and MRI variables between LT-FU and 2Y-FU were performed only on the STEMS-patients with available LT-FU data (11/12). P values less than 0.05 were considered statistically significant. Owing to the exploratory nature of the analyses, no corrections for multiple testing were performed, and P values must be considered as descriptive.

Patient characteristics and safety outcome

Detailed characterization of the trial population is reported in Table 1.

Complete LT-FU data were available for 11/12 (91.7%) patients with a median follow-up duration of 5.7 years after the hfNPCs transplantation (interquartile range [IQR] 4.8-6.0 years). The twelfth patient discontinued follow-up visits at our center due to disability-related logistical constraints. No AEs related to the procedure were observed. The most relevant AEs observed are detailed in Table 2.

No MS relapses occurred during LT-FU. 4/11 (36%) patients started a disease-modifying treatment (DMT) after discontinuation of tacrolimus according to STEMS protocol; of these, 1 started teriflunomide due to minimal radiologic disease activity (a small brain lesion without gadolinium -gd- enhancement) during the LT-FU, and 3 started ocrelizumab. Reasons for starting ocrelizumab were clinico-radiological activity during STEMS phase I protocol (n = 1) or radiological activity during STEMS LT-FU (n = 2).

Clinical, neuropsychological and neurophysiological outcomes

EDSS in STEMS-patients significantly worsened at LT-FU (median EDSS [IQR]: 8.5 [8.0–9.0]) compared to baseline (median EDSS [IQR]: 7.25 [6.5-8.0]) and 2-year FU (median EDSS [IQR]: 7.75 [7.0–8.0]. P < 0.01, for both comparisons). The mean rate of EDSS change (slope) in the LT-FU (+ 0.24 EDSS points per year [range, 0–0.63]) did not significantly differ from the slope in the 4 years before (+ 0.24 EDSS points per year [range, 0–0.70]; P = 0.56) and 2 years after transplantation (+ 0.13 EDSS points per year [range, 0–0.80]; P = 0.49) (Fig. 1a). Out of 11 patients, only 4 and 3 could perform the nine-hole peg test with their dominant (DH) and non-dominant hand (NDH), respectively, suggesting an overall worsening in manual dexterity. Among these, 3 for DH and 1 for NDH experienced a score decline exceeding 20%. The LT-FU SDMT score (median SDMT 42, IQR 32.5–46.5) significantly worsened compared to 2-year FU (median SDMT 49, IQR 42.5–50; P < 0.01), whereas it remained stable compared to baseline visit (median SDMT 38, IQR 30.15-49; P = 0.92; Fig. 1b).

Long-term OCT assessment was available for 8/11 patients. pRNFL thickness values at LT-FU (mean ± SD pRNFL thickness 92.9 ± 9 μm) showed no significant differences compared to baseline (mean ± SD pRNFL thickness 91.7 ± 11.5 μm, P = 0.05) and 2Y-FU (mean ± SD pRNFL thickness 90.7 ± 13.3 μm, P = 0.30; Fig. 1c). No significant correlation was found between number of injected hfNPCs and slopes of EDSS (Spearman R² = 0.19, P = 0.12), SDMT (Spearman R² = 0.26, P = 0.09) and pRNFL (Spearman R² = 0.19, P = 0.65) between LT-FU and 2Y-FU.

MRI assessments and reference populations

At LT-FU, 5 (45%) STEMS-patients showed new/enlarged T2-hyperintense brain lesions (Fig. 2, patient-by-patient representation of MRI inflammatory activity) with a median annualized percentage change in T2-hyperintense lesion volume (T2LV) of 8.94 (IQR − 1.56–14.88) %. Three of these 5 patients were not active at previous MRI timepoints. The proportion of STEMS-patients showing MRI disease activity and the annualized variation of total T2LV was not significantly different between LT-FU vs. 2Y-FU (P = 0.83 and P = 0.12, respectively; Table 3).

No patient showed Gd-enhancing lesions during the LT-FU. Between 2Y-FU and LT-FU, STEMS-patients showed a statistically significant reduction in the annualized rate of gray matter atrophy (PGMVC, P = 0.04, Fig. 3b), while no statistically significant differences were found for WM atrophy (PWMVC, P = 0.1, Fig. 3c) and brain atrophy (PBVC, P = 0.6, Fig. 3a). A patient-level representation of PBVC, PGMVC, and PWMVC at 2Y-FU and LT-FU are reported in Additional file 1. After hfNPC transplantation, STEMS patients exhibited a trend toward lower annualized rates of brain (P = 0.6), GM (P = 0.04), and WM atrophy (P = 0.1) during the post-transplantation follow-up period (Figs. 3d and 4). Considering that 4 out 11 patients started a DMT during the LT-FU, we performed a descriptive sub-analysis evaluating the differences between PBVC, PGMVC and PWMVC between treated and untreated STEMS patients, to explore whether the lower rate of atrophy observed in the LT-FU could be due to the introduction of DMTs. Our analysis showed no significant difference of PBVC between treated and untreated patients. On the contrary, patients who did not receive DMTs in the LT-FU seem to have a significant lower rate of GM and WM atrophy compared to those who underwent therapy (LT-FU median [IQR] PGMVC in treated − 0.55 [-0.88; -0.49] vs. untreated patients − 0.19 [-0.47; -0.09], p = 0.04. LT-FU median [IQR] PWMVC in treated − 0.93 [-1.34; -0.71] vs. untreated patients − 0.26 [-0.49; -0.01], p < 0.01).

No significant correlation was found between annualized PBVC (adjusted R² = 0.33, P = 0.3), PGMVC (adjusted R² = 0.09, P = 0.8) and PWMVC (adjusted R² = 0.24, P = 0.5) at LT-FU and number of injected hfNPCs.

Since trial design did not include a standardized longitudinal MRI assessment prior to hfNPC transplantation to estimate individual pre-transplant atrophy trajectories, nor a control group, causal relationship between changes in atrophy rates and the procedure cannot be established. To better contextualize long-term atrophy rates observed after transplantation in our patients, we compared them with those of longitudinal reference cohorts of early MS, PMS and HC. Demographic and MRI findings of STEMS-patients and reference cohorts (early MS, PMS and HC) are reported in Table 3. At 2Y-FU, STEMS-patients showed significantly higher annualized atrophy rates in whole brain (P < 0.01 vs. early MS, P < 0.01 vs. HC), GM (P < 0.05 vs. early MS, P < 0.01 vs. HC) and WM (P < 0.05 vs. early MS, P < 0.01 vs. HC) compared to the early MS and HC reference cohorts (indicated by lower PBVC, PGMVC and PWMVC distributions, Fig. 3a-c). However, at the LT-FU, they reached annualized GM and WM atrophy rates that were not significantly different from those observed in early MS patients (P = 0.7 for GM, P = 0.9 for WM) and age-matched HC (P = 0.8 for GM, P = 0.08 for WM, Fig. 3a-c). When comparing atrophy rates between STEMS-patients and a reference cohort of PMS patients - who had comparable age and disease duration but lower disability as assessed by EDSS (Table 3) - no significant differences were found at 2Y-FU (P = 0.7 for GM, P = 0.1 for WM). Interestingly, at LT-FU there was a trend toward a lower GM and WM atrophy rates in STEMS patients compared to PMS (P = 0.06 for GM, P = 0.08 for WM). No differences in whole brain volume were observed between STEMS-patients and PMS at either timepoints (P = 0.6 for 2Y-FU, P = 0.7 for LT-FU. Figure 3a-c). The inclusion of phenotype in the sensitivity analysis did not materially modify the results, and the level of statistical significance remained unchanged (GM at LT-FU: P = 0.07; WM at LT-FU: P = 0.09; whole brain volume: P = 0.8 and P = 0.9 at 2Y-FU and LT-FU, respectively).

Plasma biomarkers

Plasma biomarkers of neurodegeneration and inflammation were evaluated longitudinally in 8 out of 12 patients. Serum levels of GFAP decreased at LT-FU (mean ± SD: 151 ± 64.5 pg/ml) compared to baseline (mean ± SD: 199.9 ± 89 pg/ml, P = 0.05) and 2-year FU (mean ± SD: 253.2± 131.2 pg/ml; P = 0.01). For NfL, tau and UCH-L1, no statistically significant differences were observed between LT-FU (mean ± SD Nfl: 144.3 ± 357.4 pg/ml; tau: 5.5 ± 1 pg/ml; mean ± SD UCH-L1: 30.6 ± 28.9 pg/ml) and baseline (mean ± SD NfL 18.9 ± 9.8 pg/ml, P = 0.64; tau: 13.6 ± 15.5 pg/ml, P = 0.46; mean ± SD UCH-L1: 61.1 ± 84.6 pg/ml, P = 0.11), nor between LT-FU and 2Y-FU (mean ± SD NfL: 24.4 ± 12.2 pg/ml, P = 0.84; tau: 25.2 ± 49.4 pg/ml, P = 0.99; mean ± SD UCH-L1: 54.17 ± 55.33 pg/ml, P = 0.11).

Ethics approval and consent to participate

The Neural Stem Cell Transplantation in Multiple Sclerosis Patient: a Phase I Study was approved by IRCCS Ospedale San Raffaele ethics committee and authorized by the Italian Medicines Agency (AIFA/SC/P/28404) on 20 march 2017. All participants provided written consent to participate. Long-term follow up (LT-FU) data have been collected according to EMA Guideline on safety and efficacy follow-up and risk management of Advanced Therapy Medicinal Products (EMEA/149995/2008 rev.1).

Consent for publication

Not applicable.

Competing interests

Prof. G. Martino is Editor of Stem Cell Research & Therapy. The authors declare that they have no other potential conflicts of interest. This is a non-profit study, and none of the authors have any commercial relationships with companies whose products are used in this study or could be affected by its outcomes.