Beneficial Effects of Autologous Mesenchymal Stem Cell Transplantation in Active Progressive Multiple Sclerosis

Panayiota Petrou; Ibrahim Kassis; Netta Levin; Friedemann Paul; Yael Backner; Tal Benoliel; Frederike Cosima Oertel; Michael Scheel; Michelle Hallimi; Nour Yaghmour; Tamir Ben Hur; Ariel Ginzberg; Yarden Levy; Oded Abramsky; Dimitrios Karussis


Brain. 2020;143(12):3574-3588. 

In This Article

Materials and Methods

Study Design

The study (NIH registration: NCT02166021) was initiated in February 2015 and completed in June 2018. It was approved by the local ethics committee and Ministry of Health, and monitored by an external contract research organization (CRO) (BRD, Israel) and an external safety committee. The inclusion criteria included: age ≤65 years; diagnosis of either active (with evidence of a relapse or MRI activity) or worsening [with deterioration in Expanded Disability Status Scale (EDSS)] progressive multiple sclerosis (according to the 2013 revised criteria by Lublin et al., 2014b); treatment failure with at least one line of multiple sclerosis therapy (as evidenced by either ≥2 relapses or by deterioration in EDSS and new MRI activity); and EDSS score between 3.0 and 6.5. The study included two phases (cycles of treatment): in the first phase, three groups of patients were formulated after randomization: one group (group 1) was assigned to receive an intrathecal injection of 1 × 106 MSCs (MSC-IT)/kg of body weight, and an intravenous sham injection of normal saline. The second group (group 2) was treated with an intrathecal sham injection (normal saline) and an intravenous injection of 1 × 106 MSCs (MSC-IV)/kg of body weight, and the third group (group 3: sham treatment) was treated only with normal saline (intravenously and intrathecally). In the second phase after 6 months, the treatment groups were crossed over and each one subdivided into two subgroups of eight patients (Figure 1). The patients in group 3 who were in the placebo arm in the first phase, were treated after the 6 months with either MSC-IT (subgroup 3A, n = 8) or MSC-IV (subgroup 3B, n = 8). Half of the patients of groups 1 and 2, were retreated with the same treatment as in the first cycle (subgroup 1A with MSC-IT, n = 8 and subgroup 2A with MSC-IV, n = 8). The other halves of groups 1 and 2 (i.e. subgroups 1B and 2B, n = 8 each), were treated in this second cycle with placebo (Figure 1).

Figure 1.

Study design and flow chart. fMRI = functional MRI; MS = multiple sclerosis; MSFC = Multiple Sclerosis Functional Composite; OCT = optical coherence tomography; VEP = visual evoked potentials.

The patients were followed-up in the outpatient multiple sclerosis clinic by the assigned treating physician (for adverse events) and by the examining physicians for the neurological testing. All physicians, clinic personnel, and patients were blinded to the treatment assignment. Patients underwent neurological examination (EDSS/Functional Systems scoring), the 9-hole peg test, timed 25-foot walking test, MRI (including resting functional MRI), visual evoked potentials (VEP), optical coherence tomography (OCT), and cognitive, immunological, and visual dynamic tests at predetermined time points, as displayed in the study flow chart (Figure 1). In case of a relapse, the patient was treated with intravenous methylprednisolone (Solu-Medrol®) 1000 mg for 3–5 days and the scheduled neurological examination and scoring were postponed to 1 week after the end of steroid treatment. The MRI and OCT scans were transferred anonymously to a referral collaborating centre (Berlin University) and evaluated in a blinded way. The results were sent to the CRO.

MSC Preparation and Administration

MSCs were obtained from the bone marrow of each patient and prepared using a previously described protocol with slight modification (Karussis et al., 2010) (details of MSC preparation and culture are presented in the Supplementary material).

The cultured cells were diluted with normal saline and transferred to one of the two syringes that were prepared for each patient (according to the treatment group assignment by the CRO). One syringe contained MSCs (1 × 106/kg of body weight) resuspended in 3 ml of normal saline and one syringe contained only normal saline. To ensure blinding, the treating physician received from the Laboratory Unit (according to the randomization number), two sealed syringes covered with black adhesive, for every patient, at each treatment cycle, The full 3 ml of the content in the sealed syringe were injected to the cerebrospinal fluid via lumbar puncture at the level of L4–5, using a 20-gauge needle and three-way cannula. The 3 ml of the second syringe were injected into an aluminum-covered 500-ml sac with normal saline and infused into the patient over 30 min using a 20-gauge vein catheter. A volume of 3 ml of CSF was removed for future testing. During the two phases of the trial, each of the 48 patients received one intravenous and one intrathecal injection (of which, one or both was placebo), at each of the two treatment cycles.

Primary and Secondary End Points

The two predetermined primary end points of the trial were: (i) the safety of the MSC-IV and -IT treatments (incidence of adverse events versus those in the sham-treated group); and (ii) the differences among the three groups in EDSS score changes and the proportion of patients with treatment failure, as evidenced by an increase in EDSS score or deterioration in any of the functional systems, at 6 and 12 months. Secondary end points included the differences between the sham-treated and the MSC-IT or MSC-IV treated groups in: (i) the number of relapses and the relapse rate; (ii) the number of MRI gadolinium-enhancing lesions; (iii) the annualized rate of change in the T2 lesion load on MRI, total normalized brain volume (per cent brain volume change), and functional MRI-network connectivity strength versus the rates during the run-in period; (iv) the timed 25-foot walking and 9-hole peg tests; (v) cognitive functions; and (vi) the retinal nerve fibre layer thickness, and macular thickness and volume, as evaluated via OCT.

Lesion Load on MRI and Brain Volumetric Changes

For conventional 3 T MRI, raw data were sent to the NeuroCure Clinical Research Center, Charité – Universitätsmedizin Berlin, Germany and evaluated in a blinded manner. The methods of evaluation are described in the Supplementary material.

Functional MRI, Visual Evoked Potential, Optical Coherence Tomography and Cognitive Tests

Resting-state blood oxygenation level-dependent functional MRI, VEP, OCT, and a battery of cognitive tests that are sensitive to multiple sclerosis were performed [including the Paced Auditory Serial Addition Test (PASAT); Brief Visuospatial Memory Test-Revised (BVMT-R); Symbol Digit Modalities Test (SDMT); Owatonna Cognitive Behavioral Test (OWAT); KAVE-naming and fluency test (KAVE); Rey Auditory Verbal Learning Test (RAVLT); and Trail Making Test (TMT)], by standard techniques described in the Supplementary material.

Statistical Analysis

The sizes of the experimental groups were calculated based on an assumption of efficacy of at least 50%, in reduction of the number of patients with treatment failure (evidenced by an EDSS step change) versus sham treatment to provide 80% power, assuming a standard deviation of differences of <75%, with a 0.050 two-sided significance level. We based our calculations both on published cohorts showing a mean annual change of 0.5 in EDSS and on our Centre cohort, in which there was a higher annual EDSS change of 0.7 (due to the selection criteria of only active or worsening patients who stopped all immunotherapies). The calculations were done for 6-month periods. Various models were evaluated that included the assumptions of the two cohorts and different scenarios of efficacy ranging from 50% to 72% and SDs of 50–75%. The locked database was transferred from the CRO (BRD, Israel) to an external company with expertise in medical statistical analysis (MediStat, Israel). All measured variables and derived parameters were assessed individually and tabulated using descriptive statistics. For categorical variables, summary tables were provided, which included the sample size and absolute and relative frequencies by study group. For continuous variables, summary tables presenting the sample size, arithmetic mean, standard deviation, median, minimum, and maximum by study group were formulated. Within-group changes from baseline or from the run-in period were analysed using paired t-tests (for the continuous values) and Wilcoxon signed-rank test (for EDSS).

For most of the statistical comparisons of the efficacy parameters, the two treatment cycles for each mode of treatment (i.e. sham, MSC-IV and MSC-IT) were pooled together: the 16 placebo patients from the first cycle were pooled with the two placebo subgroups of the second cycle (eight patients from group 1B and eight from the group 2B). Similarly, the 16 MSC-IT treated patients from the first cycle (group 1), were pooled with eight of each MSC-IT subgroup (1A and 3A) from the second cycle and the 16 MSC-IV treated patients from the first cycle (group 2), were pooled with eight of each MSC-IV subgroup (2A and 3B). In this way, three pooled groups of 6-months exposure to treatment with MSC-IT, MSC-IV or placebo were formulated (n = 32 each) for statistical comparisons. The two-sample non-parametric Wilcoxon-Mann-Whitney rank sum test was used to analyse the differences in quantitative parameters between the MSC-IT or MSC-IV and sham-treated groups. Noticeably, the 'pooling' of the two treatment periods, may have influenced somehow the statistical analysis of the treatment groups as independent ones, due to a possible 'carry over effect' in some of the patients from the first period to the second and therefore, the interpretation of this statistical analysis should be taken with caution. The chi-squared test was used to analyse the differences in binary parameters between the MSC-IT or MSC-IV and sham-treated groups. For functional MRI, a single functional z-score reflecting network connectivity was calculated per scan for each subject. For cognitive functions, values were expressed as z-scores, which were calculated using normative data from the literature, adjusted for age, sex, and educational level. All tests were two-tailed, and results with P-value ≤ 0.05 were considered statistically significant. The data were analysed using the SAS® version 9.3 software (SAS Institute, Cary, NC, USA).

Data Availability

The authors confirm that the core of the data supporting the findings of this study are available within the article and its Supplementary material.

The full raw data are available on request from the corresponding author or the study coordinator (Ariel Ginzberg, or the Director of the external CRO, Dr Moshe Neuman, The data are not publicly available because they contain information that could compromise the privacy of research participants.