Intrathecal Mesenchymal Stem Cells in Progressive Multiple Sclerosis

Abstract

Background and Objectives

There is an unmet need for neuroregenerative therapies in multiple sclerosis (MS). Mesenchymal stem cells (MSCs) have shown immunomodulatory and regenerative effects in preclinical models and early clinical studies. Intrathecal administration may enhance therapeutic potential by direct delivery to the CNS. However, randomized, placebo-controlled trials are needed to establish safety and efficacy. The objective of this trial was to assess whether a single intrathecal administration of autologous MSCs could provide evidence of a neuroregenerative effect in progressive MS.

Methods

In this randomized, double-blind, placebo-controlled phase I/II trial (NCT04749667), patients with progressive MS enrolled at 4 Norwegian tertiary hospitals received a single intrathecal injection of autologous bone marrow–derived MSCs (1 × 10⁶ cells/kg) in a crossover design. The primary end point was the change in latency of combined evoked potentials at 6 months. Secondary end points included safety, brain MRI measures, functional and ophthalmologic assessments, and serum biomarkers at 6 and 12 months. Exploratory analyses comprised proteomic profiling of CSF. Outcomes were analyzed using baseline-adjusted regression models.

Results

A total of 18 patients were included (mean age 46.7 years; 55.6% female). No significant between-group difference was observed for the primary end point (β = −0.31, 95% CI –1.84 to 1.22, p = 0.668). At 6 months, patients in the MSC group showed reduced cerebral atrophy on MRI (β = 9.37, 95% CI 0.29 to 18.45, p = 0.044) and lower serum glial fibrillary acidic protein levels (β = −16.3 pg/mL, 95% CI –33.0 to 0.3, p = 0.054), but neither were sustained at 12 months. Exploratory CSF proteomics revealed reductions in multiple inflammation-related proteins at 6 months. One serious adverse event was deemed probably related to MSC treatment. Common adverse events included fever (n = 9) and low back pain (n = 10) after MSC administration, and spinal MRI abnormalities with fluid loculations and nerve root clumping (n = 7) at 6 months. One patient developed chronic coccygeal pain attributed to arachnoiditis.

Discussion

A neuroregenerative effect was not detected, although interpretation may be limited by the small sample size. Adverse events suggest an acute localized inflammatory reaction after MSC administration. Our findings suggest that intrathecal administration of MSCs in progressive MS should be approached with caution in future studies.

Classification of Evidence

This study provides Class III evidence that, in patients with progressive multiple sclerosis, treatment with a single intrathecal administration of autologous mesenchymal stem cells does not provide neuroregenerative effect, as assessed by a composite evoked potential score.

Trial registration

ClinicalTrials.gov (NCT04749667); registered February 8, 2021; first patient enrolled August 9, 2021.

Introduction

Multiple sclerosis (MS) is a chronic, immune-mediated disorder of the CNS, characterized by inflammation, multifocal demyelination, and neurodegeneration. Globally, more than 2 million individuals are affected, and its prevalence has increased in recent decades, making MS one of the leading nontraumatic causes of disability among young adults in Western countries.¹

A substantial proportion of patients with progressive MS experience ongoing disability accumulation despite optimized therapy. Currently, no treatment has been shown to halt progression, and no intervention is available to promote remyelination or repair axonal injury. The absence of regenerative therapies represents a major unmet need in MS care and highlights a broader challenge in neurology.

Mesenchymal stem cells (MSCs) have attracted considerable interest as a potential immunomodulatory and regenerative therapy for MS. Preclinical studies demonstrate that MSCs migrate to sites of CNS injury and secrete trophic factors that modulate microglial activation, support oligodendrocyte precursor differentiation, promote remyelination, and stimulate endogenous repair pathways.^2-7

Clinical studies of MSC transplantation in MS and other neurologic conditions have reported favorable safety profiles and suggested potential regenerative effects, including improvements in visual function and evoked potentials.^8-16 However, a recent randomized trial of intravenous MSCs in MS showed no benefit compared with placebo.¹⁷ The intrathecal route may offer advantages by bypassing the blood-brain barrier and delivering higher concentrations directly to sites of pathology.¹⁸ Nevertheless, most studies of intrathecal MSCs have lacked placebo controls, making them prone to bias. The primary objective of this trial was, therefore, to determine whether a single intrathecal administration of autologous MSCs could provide evidence of neuroregenerative effects in patients with progressive MS, assessed by a composite evoked potential score. Accordingly, the primary research question addressed was whether a single intrathecal administration of autologous mesenchymal stem cells induces neuroregenerative effects in patients with progressive multiple sclerosis, as measured by a composite evoked potential score.

Methods

Study Design

This study was a randomized, controlled, crossover, phase I/II trial designed to assess the neuroregenerative efficacy of treatment with autologous bone marrow–derived MSCs compared with placebo in patients with progressive MS. Eligible patients had primary or secondary MS according to the revised McDonald criteria, were between 18 and 55 years, had an Expanded Disability Status Scale (EDSS) score of 4–7, and had a disease duration of 2–18 years.¹⁹ Patients were required to have no relapses and no new or enlarging MRI lesions for at least 24 months before enrollment and to remain untreated with DMTs during the study to avoid potential interference with study end points. Other exclusion criteria included active or chronic infection, history of malignancy, polyneuropathy, severe comorbid illness, or other conditions that could affect cognition or compliance. Full eligibility criteria are detailed in the clinical trial protocol (eSAP 1).

Patients were randomized to receive either autologous MSCs (Arm A) or placebo (Arm B) at baseline, with crossover treatment administered at 6 months (Figure 1). A crossover design was chosen to ensure that all participants received MSC treatment while maintaining blinding. Patients were randomized in blocks of 4 using a computer-generated sequence prepared by a statistician not otherwise involved in the trial. Allocation was concealed, and only the study nurses involved in bone marrow aspiration and lumbar puncture procedures, as well as the principal investigator (C.E.K.), had access to the allocation list. To maintain blinding, all patients underwent bone marrow aspiration twice under identical conditions. During intrathecal administration, the study drug (MSCs or saline) was prepared in a separate room, concealed from view, and injected behind the patient's back during lumbar puncture, ensuring full concealment throughout administration. A safety visit was conducted at 18 months as the end-of-study assessment. A sample size of 18 patients was determined based on feasibility and precedent from similar early-phase trials designed to provide proof-of-concept and assess safety.^10-12,15 Patients were recruited from Haukeland University Hospital (HUS), Akershus University Hospital, St. Olavs Hospital, and University Hospital of North Norway. Outcome assessments and procedures were performed at HUS, whereas safety visits were performed at all centers.

Figure 1. Randomized Crossover Trial Design.

Patients with progressive MS were randomized to 1 of 2 arms in this double-blind, placebo-controlled, crossover phase I/II trial. Arm A received autologous intrathecal MSCs at baseline (week 0) and placebo (saline) at 6 months. Arm B received placebo at baseline and MSCs at 6 months. All patients underwent bone marrow aspiration twice to ensure blinding. Clinical, neurophysiologic, and MRI assessments were performed at baseline, 6, and 12 months, with a final safety visit at 18 months. CSF was collected before each injection for biomarker and proteomic analyses (Figure created with Biorender). MSC = mesenchymal stem cell.

MSC Manufacturing and Administration

MSC manufacturing, quality control, and transportation are described in the Supplement (eMethods 1). In brief, autologous MSCs were generated from bone marrow aspirates and manufactured under Good Manufacturing Practice conditions at a certified cell manufacturing facility (Ulm, Germany). Bone marrow cells were directly seeded and culture-expanded for approximately 21 days in alpha-MEM supplemented with human platelet lysate and heparin, using a two-step expansion protocol. Cells were harvested, washed, and resuspended in 0.9% saline at a concentration of 20 × 10⁶ MSCs/mL for intrathecal administration. Release criteria included viability, sterility, endotoxin testing, mycoplasma testing, and immunophenotypic characterization by flow cytometry. Each patient received an intrathecal single dose of 1 × 10⁶ MSCs/kg body weight (maximum of 100 × 10⁶ MSCs) through lumbar puncture. The dose was chosen based on previous open-label studies reporting acceptable safety at comparable dose levels.^{10-12,14-16,20,21} Placebo consisted of an identical volume of 0.9% saline solution. Before each intrathecal injection, approximately 12 mL of CSF was collected for analysis and biobanking.

End Points

The primary end point was the change in combined evoked potentials (CEPs) from baseline to 6 months, calculated as the sum of z-transformed latencies for visual (VEP), motor (MEP), and somatosensory (SEP) evoked potentials bilaterally. Secondary end points included individual evoked potential latencies, EDSS scores, MRI-based measures, visual function, retinal thickness assessed by optical coherence tomography, motor and cognitive performance, patient-reported outcomes, and serum biomarkers including neurofilament light chain (NfL) and glial fibrillary acidic protein (GFAP). Further end point details are provided in the Supplement (eMethods 2). Primary outcomes were assessed at 6 months after baseline, while secondary outcomes were evaluated at both 6 and 12 months. Safety assessments were conducted at all study visits, including the end-of-study visit at 18 months. Adverse events (AEs) and serious adverse events (SAEs) were defined according to standard clinical trial guidelines and were systematically recorded using structured case report forms and clinician interviews. No important changes to the trial design, outcomes, or analyses were made after the study commenced.

CSF Proteomics

CSF samples from baseline and 6 months (before crossover) were analyzed using tandem mass tag (TMT)–based quantitative proteomics. Sample preparation, liquid chromatography-tandem mass spectrometry (LC-MS/MS), and data analysis were performed as previously described, with protocol modifications.²² Samples were processed in 3 TMT-16plex sets, each including pooled controls for normalization. Peptides were analyzed using an Orbitrap Eclipse high-resolution mass spectrometer and processed with Proteome Discoverer v2.5 (Thermo Fisher Scientific) and Perseus.²³ Differential protein expression, protein-protein interaction networks, and functional enrichment analyses were conducted using the STRING database and Gene Ontology.^24,25 A detailed description of sample handling, data processing, and statistical methods is provided in the Supplement (eMethods 3).

Statistical Analysis

Efficacy analyses were, according to the statistical analysis plan, conducted on the per-protocol population, while safety analyses included all participants who received at least 1 dose of the investigational product. For the 6-month analyses, patients were not excluded from the per-protocol population if protocol deviations occurred after the 6-month visit. No interim analyses or formal stopping guidelines were specified. Classic ANCOVA (Y = β0+βt⋅treatment+γ⋅X+ϵ) was used to assess differences in outcomes after 6 months (Arm A vs Arm B), with baseline values as covariates, where β_t represents the adjusted mean difference between treatment groups. The primary outcome was assessed at 6 months, before the crossover, to ensure a clean between-group comparison unaffected by possible carry-over effect.

At 12 months, ANCOVA adapted for the crossover trial design ((Yti − Yci) = βt + γ (Xti − Xci)) was used to assess treatment effect and included order of treatments/randomization as a covariate to adjust for potential carry-over effects. In this study, Yti and Yci denote each participant's post-treatment outcomes under the treatment and control periods, respectively; Xti and Xci denote the corresponding baseline values; γ estimates the association between baseline and post-treatment values; and βt represents the adjusted treatment difference (i.e., treatment effect). This model has been validated as taking optimal advantage of baseline measurements to improve precision, while minimizing bias due to chance baseline imbalances, particularly in small samples.²⁶ Missing data for evoked potentials were imputed using the longest conduction time recorded within the same modality, in accordance with previous methodological work identifying this approach as the most robust option for retaining sensitivity in the presence of absent cortical responses.²⁷ No imputation was applied to other outcomes. To comply with the SAP, which specifies the use of cross-sectional values in ANCOVA models, normalized brain volume and T1/T2 lesion volume at months 6 and 12 were derived from baseline values combined with longitudinal change measures (percentage brain volume change and total lesion volume change). This allowed us to retain the higher sensitivity of longitudinal analyses while generating time point–specific values compatible with the planned statistical models. For descriptive analyses of baseline characteristics, the Student t test, Mann-Whitney U test, and χ² test were used as appropriate. Safety outcomes, including AEs and SAEs, were summarized descriptively. A representative from the Norwegian MS Society participated as a patient representative throughout the design, conduct, and reporting of the study.

Standard Protocol Approvals, Registrations, and Patient Consents

The study was approved by the Regional Committee for Medical and Health Research Ethics (nr. 159326) and the Norwegian Medicines Agency and was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice (GCP) guidelines. All participants provided written informed consent before enrollment. Oversight was provided by external monitors from the Western Norway Health Trust Research & Development to assure alignment with GCP, and safety was regularly evaluated by an independent Data Monitoring Committee (DMC). The trial was registered at ClinicalTrials.gov (NCT04749667) and European Union Clinical Trials Register (EudraCT no. 2020-002373-95) before participant enrollment.

Data Availability

Deidentified participant data and related documents are available from the corresponding author on reasonable request after publication, for up to 5 years, to qualified researchers under a data-sharing agreement. The protocol and SAP are provided in the supplemental material (eSAP 1).

Results

Between August 9, 2021, and July 7, 2023, a total of 18 patients were enrolled according to the protocol and randomized (9 per group). All participants received the assigned intervention (MSC or placebo) according to the protocol, with no deviations. Follow-up assessments for efficacy and safety were completed by March 2025. All patients completed study visits as scheduled, except for 1 patient whose 12-month follow-up was delayed by 4 months because of treatment for colorectal malignancy, which was considered unrelated to the study treatment. In line with the SAP, data from this visit were excluded from the per-protocol analyses at 12 months but included in the safety analyses. All participants continued their usual symptomatic and supportive care during the trial. Overall, missing data accounted for 7.1% of all planned assessments across parameters (eTable 1).

Baseline Characteristics

Baseline demographic and clinical characteristics were similar between groups, with no statistically significant differences (Table 1). The mean age was 46.0 years (SD 4.3) in Arm A and 47.3 years (SD 4.2) in Arm B. Both groups had 55.6% female participants. Most had secondary progressive MS (Arm A: 66.7%, Arm B: 44.4%). The median EDSS score was 6.0 (interquartile range [IQR] 4.5–6.0) in Arm A and 6.5 (IQR 6.0–6.5) in Arm B. MRI lesion load and brain volume were comparable, although Arm B had more T2 lesions on average.

Table 1.

Demographic Baseline Data of Patients Randomized to Group A (MSCs at Baseline and Placebo at 6 Months) and Group B (Placebo at Baseline and MSCs at 6 Months)

	Arm A (n = 9)	Arm B (n = 9)
Age (SD)	46 (4.3)	47.3 (4.2)
Female sex (%)	5 (55.6)	5 (55.6)
MS type
Secondary progressive (%)	6 (66.7)	4 (44.4)
Primary progressive (%)	3 (33.3)	5 (55.6)
Years since first clinical event (SD)	14.6 (5.5)	14.3 (8.3)
Years since MS diagnosis (SD)	11.8 (3.8)	8.3 (4.1)
Baseline EDSS score (IQR)	6 (4.5–6)	6.5 (6–6.5)
Use of fampridine (%)	6 (66.7)	7 (77.8)
Number of MRI T2 lesions (SD)	14.6 (8.4)	22.1 (4.1)
Normalized brain volume, ml (SD)	1,452.2 (38.7)	1,486.2 (48.5)
Number of previous DMTs (IQR)	1 (1–4)	2 (1–3)

Abbreviations: DMT = disease-modifying therapy; EDSS = Expanded Disability Status Scale; IQR = interquartile range; MS = multiple sclerosis; MSC = mesenchymal stem cell.

Evoked Potentials

No statistical significant differences were observed in the primary end point of CEP latency z-scores between patients receiving MSCs and those receiving placebo at 6 months (β = −0.31, 95% CI –1.84 to 1.22, p = 0.668) or 12 months (β = −0.32, 95% CI –1.26 to 0.62, p = 0.476) (Table 2). Removing the covariate for treatment order from the ANCOVA model had no significant impact on the estimated treatment effect, and the intercept (−0.33, 95% CI –0.9 to 0.3, p = 0.249) remained nonsignificant at 12 months. There were no significant changes at 6 or 12 months in individual VEP, SEP, or MEP latencies or amplitudes.

Table 2.

Outcomes at 6 and 12 Months in Patients Randomized to Group A (MSCs at Baseline and Placebo at 6 Months) and Group B (Placebo at Baseline and MSCs at 6 Months)

Arm (SD)	Baseline	6 mo	12 mo	Β coefficient 6 mo (95% CI)	p Value	Intercept coefficient 12 mo (95% CI)	p Value
Evoked potentials
Latencies, z-scores combined evoked potentials
A	−1.02 (2.93)	−0.60 (2.97)	−0.30 (2.89)	−0.31 (−1.84 to 1.22)	0.668a	−0.32 (−1.26 to 0.62)	0.476
B	0.31 (2.36)	1.02 (2.98)	0.51 (3.46)
Amplitudes, z-scores combined evoked potentials
A	−0.07 (1.61)	0.29 (2.15)	−0.31 (1.65)	0.57 (−1.92 to 3.05)	0.635	−0.54 (−1.69 to 0.61)	0.332
B	0.37 (3.45)	−0.02 (3.41)	0.07 (2.84)
Visual evoked potentials
Latencies (ms)
A	113.9(12.1)	124.9 (29.7)	134.5 (39.5)	6.14 (−14.73−27.00)	0.540	−0.67 (−15.82 to 14.48)	0.926
B	117.0 (13.5)	125.1 (35.2)	124.2 (35.3)
Somatosensory evoked potentials
Latencies (ms)
A	55.5 (22.4)	53.5 (26.3)	56.8 (23.9)	−0.23 (−14.52 to 14.06)	0.973	−0.24 (−5.76 to 5.28)	0.928
B	74.8 (28.5)	71.9 (28.2)	72.5 (27.7)
Motor evoked potentials
Latencies (ms)
A	20.0 (10.7)	19.9 (9.9)	19.2 (9.9)	−6.15 (−14.34 to 2.04)	0.130	−3.19 (−7.39 to 1.01)	0.126
B	23.0 (8.7)	27.9 (9.6)	23.0 (9.2)
MRI
T2 lesion volume (mL)
A	7.0 (7.7)	7.0 (7.7)	7.0 (7.7)	−0.01 (−0.03 to 0.00)	0.115	0.33 (−0.81 to 1.47)	0.545
B	5.8 (3.9)	6.5 (3.5)	6.5 (3.5)
Cerebral volume (mL)
A	1,448 (39)	1,447 (43)	1,442 (41)	9.37 (0.29 to 18.45)	0.044	−4.64 (−14.66 to 5.38)	0.330
B	1,493 (32)	1,486 (34)	1,471 (38)
Ophthalmologic assessments
Visual acuity (logMAR)
Right eye
A	0.08 (0.36)	0.07 (0.33)	0.08 (0.32)	−0.01 (−0.06 to 0.04)	0.669	0.01 (−0.04 to 0.05)	0.710
B	−0.04 (0.05)	−0.03 (0.07)	−0.02 (0.07)
Left eye
A	−0.02 (0.10)	−0.03 (0.10)	−0.02 (0.07)	−0.01 (−0.05 to 0.03)	0.600	0.01 (−0.03 to 0.05)	0.662
B	−0.04 (0.05)	−0.03 (0.04)	−0.02 (0.05)
Visual field (decibel)
Right eye
A	4.6 (2.2)	3.3 (0.9)	3.6 (2.1)	0.94 (−1.7 to 2.04)	0.090	−0.07 (−1.51 to 1.38)	0.923
B	2.7 (2.1)	1.7 (1.4)	1.6 (1.7)
Left eye
A	4.0 (2.5)	3.7 (2.2)	3.9 (1.3)	0.98 (−0.47 to 2.44)	0.170	0.17 (−0.26 to 0.61)	0.400
B	3.1(3.1)	2.2 (2.2)	2.8 (2.13)
Contrast sensitivity (AUC)
Right eye				−151.0 (−573.9 to 271.9)	0.440	−125.0 (−314.7 to 64.7)	0.170
A	992.4 (267.7)	980.0 (317.8)	1,074.4 (268.0)
B	1,235.5 (264.9)	1,323.0 (215.5)	1,204.1 (237.5)
Left eye
A	1,004.4 (264.0)	972.7 (281.5)	1,009.2 (260.8)	−22.5 (−374.4 to 329.4)	0.889	−15.6 (−236.9 to 205.8)	0.879
B	1,260.7 (169.3)	1,202.1 (277.6)	1,186.0 (212.7)
Fovea thickness (µm)
Right eye
A	277.4 (27.8)	277.4 (30.6)	271.3 (23.5)	−2.16 (−7.79 to 3.46)	0.425	1.84 (−1.47 to 5.16)	0.253
B	275.6 (21.2)	277.7 (20.4)	279.1 (23.4)
Left eye
A	277.5 (17.5)	278.4 (16.9)	272.9 (23.2)	0.80 (−3.95 to 5.55)	0.723	2.07 (−0.05 to 4.19)	0.055
B	273.2 (21.5)	273.6 (20.7)	275.4 (22.1)
Retinal nerve layer thickness (µm)
Right eye
A	86.7 (8.7)	94.1 (14.2)	91.3 (11.3)	5.60 (−4.31 to 15.52)	0.245	0.47 (−3.65 to 4.60)	0.809
B	89.0 (9.4)	89.7 (8.6)	89.9 (11.0)
Left eye
A	88.4 (10.2)	94.1 (10.4)	93.6 (9.8)	3.66 (−4.42 to 11.75)	0.348	0.44 (−0.71 to 1.61)	0.419
B	90.7 (11.4)	90.8 (11.3)	91.2 (12.1)
Clinical assessments
EDSS score
A	5.6 (1.0)	5.5 (1.2)	5.4 (1.4)	−0.18 (−0.62 to 0.26)	0.395	−0.19 (−0.61 to 0.24)	0.363
B	6.10 (0.9)	6.2 (0.7)	6.0 (1.4)
9-Hole Peg Test score, dominant hand (s)
A	25.9 (7.5)	25.7 (9.8)	26.8 (7.8)	−6.0 (−16.9 to 4.9)	0.257	−0.2 (−1.8 to 1.5)	0.808
B	38.1 (39.8)	30.0 (21.8)	28.7 (20.9)
Timed 25-Foot Walk Test (s)
A	11.8 (8.0)	9.7 (12.8)	12.8 (8.1)	−0.6 (−3.0 to 1.8)	0.590	1.6 (−1.7 to 4.8)	0.302
B	15.6 (18.3)	9.8 (3.1)	11.1 (4.6)
Patient-reported outcomes (PROs)
Symbol Digit Modalities Test (SDMT)
A	43.1 (10.6)	47.0 (11.2)	44.8 (12.0)	2.8 (−1.0 to 6.6)	0.135	2.2 (−1.8 to 6.2)	0.261
B	49.3 (5.9)	50.4 (6.8)	53.0 (9.2)
California Verbal Learning Test II (CVLT-II)
A	56.4 (13.1)	56.7 (13.5)	62.4 (12.4)	−4.8 (−12.9 to 4.7)	0.338	4.8 (−0.4 to 9.9)	0.066
B	55.3 (6.8)	60.0 (7.9)	62.0 (9.0)
Brief Visuospatial Memory Test–Revised (BVMT-R)
A	7.3 (3.2)	7.5 (2.0)	8.6 (2.3)	0.6 (−1.4 to 2.6)	0.546	0.37 (−0.78 to 1.5)	0.502
B	9.0 (2.2)	7.9 (2.9)	8.9 (2.5)
EQ-5D-5L scale
A	41.7 (17.9)	50.3 (19.9)	48.9 (15.0)	3.8 (−13.8 to 21.5)	0.650	2.0 (−3.8 to 8.0)	0.471
B	43.4 (12.3)	47.8 (20.6)	46.9 (24.8)
Fatigue severity scale
A	5.2 (1.2)	5.1 (1.3)	5.0 (1.5)	0.4 (−0.5 to 1.4)	0.317	0.2 (−0.5 to 0.6)	0.944
B	5.6 (1.4)	5.0 (1.6)	5.2 (1.6)
MSIS-29 (Motoric Subscale)
A	66.7 (15.1)	65.4 (15.8)	60.3 (17.9)	0.0 (−7.2 to 7.2)	0.992	−4.1 (−11 to 2.9)	0.229
B	68.8 (9.8)	67.4 (11.3)	65.0 (13.3)
MSIS-29 (Mental Subscale)
A	18.2 (4.7)	19.3 (5.2)	17.8 (6.1)	0.2 (−4.6 to 5.0)	0.940	−0.2 (−3.4 to 3.0)	0.893
B	20.8 (5.4)	20.6 (5.3)	20.6 (5.9)
Biomarkers in serum
Glial fibrillary acidic protein (pg/mL)
A	122.3 (55.6)	115.9 (47.4)	114.9 (38.4)	−16.3 (−33.0 to 0.3)	0.054	15.1 (−10.1 to 40.7)	0.228
B	111.6 (20.9)	123.3 (25.4)	127.0 (26.3)
Neurofilament light chain (pg/mL)
A	7.9 (1.5)	8.9 (2.1)	8.3 (0.9)	−0.4 (−4.5 to 3.7)	0.826	−0.1 (−1.7 to 1.4)	0.852
B	12.3 (6.6)	11.0 (5.2)	9.4 (1.7)

Abbreviations: AUC = area under the curve; BVMT-R = Brief Visuospatial Memory Test–Revised; CVLT-II = California Verbal Learning Test–Second Edition; EDSS = Expanded Disability Status Scale; EQ-5D-5L = EuroQol five-dimension five-level scale; FSS = Fatigue Severity Scale; logMAR = logarithm of the minimum angle of resolution; MS = multiple sclerosis; MSC = mesenchymal stem cell; MSIS-29 = Multiple Sclerosis Impact Scale–29 items; PRO = patient-reported outcome.

Analysis population: evoked potentials (6 mo: n = 18; 12 mo: n = 17); MRI (6 mo: n = 17; 12 mo: n = 16, 1 participant did not undergo MRI at visits 4 and 8 due to claustrophobia); ophthalmologic assessments (6 mo: n = 18; 12 mo: n = 17); clinical tests (6 mo: n = 18; 12 mo: n = 17); and biomarkers (6 mo: n = 18; 12 mo: n = 17).

^aPrimary end point.

Safety and Tolerability

A total of 76 AEs were reported during the MSC treatment period, compared with 54 during placebo (eTable 2). The most frequently reported events after MSC administration were headache and fever, each occurring in 9 patients, vs 6 and 1 patients, respectively, after placebo. Fever typically developed within a few hours of MSC administration and resolved spontaneously within the next day. Low back or gluteal pain was reported in 10 patients after MSC treatment. The pain typically developed shortly after injection and tended to persist from days to a few weeks before improvement, with good effect of standard analgesics.

One SAE was registered as probably related to MSC treatment. The patient was hospitalized because of severe low back pain 1 week after intrathecal injection. Initial lumbar spine MRI was interpreted as unremarkable, but re-analysis at a later time point revealed small cystic CSF loculations, likely representing a local inflammatory reaction. The condition was successfully treated with a course of prednisolone, and the patient recovered fully without sequelae.

Spinal MRI abnormalities were identified in 7 patients in total (including 1 patient diagnosed with arachnoiditis). Three cases were detected during the study period, while 4 additional cases (including the SAE) were identified retrospectively. The findings included CSF loculations and clumping of nerve roots (Figure 2) and were considered to be likely related to MSC administration. In one case, the cystic loculation resulted in radiologically confirmed lumbar spinal stenosis, which was not considered symptomatic at the time of assessment.

Figure 2. MRI Findings Before and After MSC Transplantation.

Sagittal and coronal MR images at the level of the L5 vertebral body, acquired 6 days before (A) and 6 months after (B) MSC transplantation. At 6 months, the cauda equina nerve roots seem clumped and no longer suspended freely within the thecal sac (arrows). This patient reported low back pain for 2 months after MSC injection before complete resolution of symptoms. MSC = mesenchymal stem cell.

One patient was diagnosed with arachnoiditis, presenting with chronic gluteal pain occurring after MSC injection. This adverse event was also considered probably related to MSC treatment. The condition was managed with daily antiphlogistic therapy, resulting in partial symptom relief. However, the condition did not improve during the study period.

No cases of treatment-related infection or malignancy were observed. No patients showed increased cell counts in the CSF, apart from cases with suspected blood contamination. CSF protein was mildly elevated in 1 patient (0.64 g/L) at 6 months, but normal in the others. This patient also had mildly elevated CSF protein levels at baseline (0.65 g/L) before any intervention.

MRI Outcomes

There were no statistical significant between-group differences in changes in T1 or T2 lesion volumes at 6 or 12 months. A statistically significant between-group difference in cerebral volume was observed at 6 months, after adjustment for baseline values (β = 9.37, 95% CI 0.29 to 18.45, p = 0.044), indicating higher volume in the MSC group compared with placebo. However, no significant difference was observed at 12 months (β = −4.64, 95% CI –14.66 to 5.38, p = 0.330). New T2 lesions were observed in 1 participant: one occurred after MSC treatment and one after placebo. Another participant developed a new T1 lesion after MSC treatment. No additional new T2 or T1 lesions were observed in any other participants.

Ophthalmologic Assessments

No significant between-group differences were observed in visual acuity, visual field, color vision, foveal thickness, or RNFL thickness at either 6 or 12 months. All ophthalmologic measures remained stable over time in both treatment groups. A nonsignificant increase in foveal thickness in the left eye at 12 months was noted in the MSC group (β = 2.07, 95% CI –0.05 to 4.19, p = 0.055).

Clinical and Functional Outcomes

No significant between-group differences were observed in EDSS scores at either 6 or 12 months (6 months: β = −0.18, p = 0.395; 12 months: β = −0.19, p = 0.363). Motor performance, as assessed by the Timed 25-Foot Walk and 9-Hole Peg Test, showed no differences between the 2 treatment arms. Similarly, cognitive outcomes (SDMT, CVLT-II, BVMT-R) and patient-reported outcomes (MSIS-29, EQ-5D-5L, FSS) remained stable across both groups and time points, with no statistically significant treatment effects.

Biomarker Analyses

Serum GFAP levels showed a nonsignificant reduction at 6 months in the MSC-treated group (β = −16.3, 95% CI –33.0 to 0.3, p = 0.054), but this was not sustained at 12 months. No significant differences were found for serum NfL.

Proteomic Analyses

In the initial blinded phase, the aim was to see whether we could separate the treated and untreated patients based on the proteomics data set without previous knowledge of group assignment. A total of 1,547 proteins were identified across the samples. After excluding proteins with missing values (222), extreme values/contaminants (20), and proteins with only minor changes (714), unsupervised clustering of the remaining 591 proteins identified a subset of 184 proteins mainly involved in inflammation, differentiating most of the patients into 2 main clusters (Figure 3A). Principal component analysis (PCA) based on these 184 proteins (Figure 3B) revealed separation of the patients into 2 groups. Because inflammation was reduced in one of the groups, as shown in the heat map, this patient group was assumed to represent those receiving treatment at baseline. Based on this assumption, the heat map, and PCA, we were able to correctly classify 13 of the 16 patients as treated or nontreated at baseline in this blinded study (2 patients were excluded because of heavy blood contamination of the CSF).

Figure 3. CSF Proteomic Changes After MSC Treatment.

(A) Heat map showing relative changes in 184 proteins from baseline to 6 months. Rows represent proteins; columns represent individual patients. Blue indicates a relative decrease in protein abundance, and red indicates a relative increase. Patients who received MSCs at baseline (Arm A) are marked green and patients who received saline (Arm B) yellow. (B) Principal component analysis (PCA) based on the same 184 proteins. Patients are colored by actual treatment group at baseline (MSC = red triangles; placebo = blue circles). Separation between groups reflects treatment-related proteomic shifts. (C) Gene Ontology enrichment analysis of differentially expressed proteins, highlighting biological processes affected by MSC treatment. Dot size indicates gene count; color indicates false discovery rate (FDR). Top enriched processes include complement activation, coagulation regulation, and inflammatory response. (D) STRING-based protein-protein interaction network of proteins differentially expressed between MSC-treated and untreated patients. Node color reflects direction of change (blue = lower in MSC group; red = higher). Distinct clusters include complement components and collagens. MSC = mesenchymal stem cell.

In the next phase of analysis, the groups were unblinded and statistical comparisons were conducted to reveal the most significant changes in the patients treated with MSCs at baseline vs those in the placebo group. The proteins with a p value less than 0.05 were used as input to STRING, and the analyses revealed that the most affected protein clusters were associated with complement activation (downregulated in treated), extracellular matrix remodeling (upregulated in treated), neuroprotective and reparative mechanisms (regulated both ways), and key glycolytic enzymes (upregulated in treated) (Figure 3D). Gene Ontology enrichment analysis highlighted biological processes including complement activation (false discovery rate [FDR] = 8.25e-13), regulation of hemostasis (FDR = 5.45e-09), and acute inflammatory response (FDR = 1.53e-09) (Figure 3C). Raw data are given in Supplement (eAppendix 1).

Subjective Patient Reports at the End of Follow-Up

At the 18-month safety visit, neurologists noted patients' subjective assessments of their disease status based on routine clinical interviews (eTable 3). Of the 18 participants, 5 described continued progression in line with their prestudy disease course and 4 were assessed as clinically stable without notable improvement or worsening. Seven patients reported varying degrees of perceived improvement, including reduced fatigue, improved gait or walking distance, increased strength in upper or lower limbs, and better balance. Two participants reported clinical worsening: one developed persistent gluteal pain consistent with post-treatment arachnoiditis (in the Safety section), and another experienced increased numbness in the arm, weakness in the left leg, and worsening gait function. This latter patient was also diagnosed with colorectal malignancy during the follow-up period; the condition was treated and judged unrelated to MSC therapy.

Classification of Evidence

This study provides Class III evidence that, in patients with progressive multiple sclerosis, treatment with a single intrathecal administration of autologous mesenchymal stem cells does not provide neuroregenerative effect, as assessed by a composite evoked potential score.

Discussion

Our study did not demonstrate a clear neuroregenerative effect of intrathecal MSC administration in patients with progressive MS. Neither evoked potentials nor other predefined clinical or radiologic outcomes showed consistent improvement with MSCs compared with placebo. At 6 months, a statistically significant attenuation of cerebral volume loss was observed in the MSC group. Adjusted serum GFAP levels were also lower in the MSC group, although this difference did not reach statistical significance. Neither finding was sustained at the 12-month analysis, making it difficult to interpret the results as an indication of therapeutic effect.

In this study, combined evoked potentials, summarizing visual, motor, and somatosensory evoked potentials, were used as the primary end point to assess potential remyelination after MSC therapy. Evoked potentials were chosen because of their objectivity and relevance to MS pathophysiology, where delayed conduction represents a core feature. Previous trials have shown latency improvements after MSC treatment, suggesting that evoked potentials may capture early signs of functional repair.^9,21

The overall negative results contrast with those of a previous placebo-controlled study investigating intrathecal MSC administration in progressive MS, which reported beneficial effects across both clinical and radiologic outcomes, including signs of remyelination on visual evoked potentials.²¹ One key difference may lie in the patient populations: the aforementioned study included patients with active progressive disease, whereas no patients in our cohort had experienced clinical relapses or radiologic activity in the past 2 years before enrollment. Preclinical studies have shown that MSCs migrate to sites of inflammation and exert regenerative effects, but this mechanism has yet to be demonstrated in humans.² It may be plausible that ongoing inflammatory activity is required to enable MSC homing and therapeutic action, which may partly explain the lack of effect observed in our study.

Our findings also differ partly from those of a recent randomized trial investigating repeated intrathecal injections of mesenchymal stem cell–derived neural progenitors in progressive MS.²⁸ Although the primary end point of clinical improvement was not met, subgroup analysis indicated a potential benefit among patients requiring assistance for ambulation (EDSS 6.0–6.5). Of interest, the abovementioned study reported a much lower frequency of lumbar pain and fever than seen in our trial. It is possible that neural progenitor cells, delivered in multiple smaller doses, are less prone to trigger local inflammatory reactions than unmodified MSCs.

To gain insight into potential molecular effects within the CNS compartment, we performed exploratory CSF proteomics. STRING analysis identified 2 main clusters of differentially expressed proteins. One network involved components of the complement cascade, likely reflecting an anti-inflammatory effect, which may be consistent with their known immunomodulatory properties. The other network contained proteins related to extracellular matrix organization, particularly collagens, as well as a smaller network of upregulated glycolytic key enzymes. In context of downregulated inflammatory proteins, these changes may reflect a shift toward a more supportive and less stressed CNS environment.

However, an alternative explanation may be related to the adverse events. Approximately half of the patients developed fever, and many reported back pain or discomfort shortly after MSC injection. As a result, lumbar imaging was incorporated into the MRI protocol, which revealed spinal abnormalities in 7 patients, including CSF loculations and nerve root clumping, consistent with a localized inflammatory reaction. These findings suggest that, in at least some patients, MSCs triggered a local inflammatory response. This led to one serious adverse event requiring hospital admission and one case of arachnoiditis associated with chronic pain.

In light of these findings, the observed proteomic changes might also represent a counter-regulatory or regenerative response secondary to the initial local inflammation triggered by MSC administration. An argument against this interpretation is that inflammatory symptoms in other patients were generally short-lived, and spinal MRI abnormalities remained stable during the study period without evidence of ongoing tissue injury. Furthermore, no patients showed increased CSF cell counts or protein levels, indicating the absence of ongoing inflammation at the time of sampling.

Previous proteomic studies have demonstrated persistent molecular signatures of low-grade inflammation after the acute inflammatory phase of spinal cord injury.^29,30 Although such injuries were not induced in our study, the temporal pattern of molecular responses may share similarities. Thus, if the proteomic changes were secondary to an initial localized inflammation, one would suspect elevated levels of inflammatory proteins compared with placebo, not reduced as observed in our study.

Several patients reported subjective improvements, including reduced fatigue, increased energy, improved gait function, reduced spasticity, and less urgency. In the absence of corresponding objective changes in clinical outcomes, these self-reported effects may be attributable to expectancy or placebo mechanisms. This underscores the importance of evaluating such interventions in blinded, placebo-controlled trials using objective outcome measures. Nevertheless, it is difficult to entirely exclude the possibility of a biological effect.

In our opinion, the occurrence of adverse events, including a serious adverse event requiring hospital admission and a case of arachnoiditis with chronic pain, suggests that further studies using this mode of MSC administration should be approached with caution. Fever and back pain are among the most frequently documented adverse events after intrathecal administration of unmodified MSCs.^10,13,16 Yet, few studies have included imaging of the lower spinal cord, as this region is typically not assessed in MS outcome evaluations.^10,12,21 There may, however, be growing recognition of local inflammatory responses after intrathecal MSC therapy. One trial in patients with multiple system atrophy (MSA) reported a dose-dependent increase in the incidence of radiologic signs of inflammation after intrathecal MSC administration, although the overall outcomes were considered positive compared with historical controls.³¹ This has also been shown in patients with MS³² and spinal cord injury³³ receiving MSCs intrathecally, as well as in case reports of stem cell tourism.³⁴ Taken together, these findings suggest that acute spinal inflammation may occur more frequently than previously recognized after intrathecal MSC administration and is not unique to our study.

Limitations of our study include the small sample size, which reduces statistical power to detect subtle treatment effects and limits the generalizability of findings. Given the exploratory nature of the study and absence of multiplicity correction for secondary outcomes, p values should be interpreted with caution. In addition, patients only received a single dose of MSCs and follow-up time was relatively short from a progressive MS perspective. However, owing to the occurrence of the adverse events, we do not consider it appropriate to pursue a larger trial with the current administration route. An extension study is ongoing, with follow-up assessments planned at 2 and 3 years to evaluate long-term outcomes. Another limitation is use of CEP as the primary end point in a population of patients with progressive MS. However, there is no established gold standard for assessment of remyelination or neural regeneration in MS, and similar neurophysiologic outcome measures have been applied in other regenerative MS trials.^35,36 Moreover, missing data in neurophysiologic assessments occurred in 13% of cases because of unreadable signals in patients with pronounced pathology (suppl Table S3). This is an unavoidable issue in progressive MS trials because severe demyelination may prevent reliable signal detection. For these outcomes, missing values were imputed using the longest conduction time within the same modality, which is considered the most appropriate approach to avoid loss of sensitivity from excluding the most severely affected patients.²⁷ Finally, there was a 24-hour delay between MSC production and injection because the cells were manufactured in another country. Nevertheless, this approach has been validated and applied for other clinical studies assessing bone regeneration, and prestudy testing confirmed preserved viability and stemness of the MSCs.^37,38 These validation experiments are described in the Supplement (eAppendix 2). The use of freshly expanded MSCs may actually be considered advantageous compared with cryopreserved cells, which have been reported to exhibit altered biological properties in some studies.^39,40 Additional strengths of the study include the double-blind, placebo-controlled, crossover design; full retention of participants; and effective blinding of both patients and outcome assessors throughout the study period.

In conclusion, this randomized, placebo-controlled trial did not demonstrate a neuroregenerative effect of intrathecal MSC administration in progressive MS, although CSF proteomics indicated potentially beneficial immunomodulatory activity. Owing to adverse events likely related to inflammatory reactions at the injection site, our findings warrant caution for future studies evaluating intrathecal MSC administration in progressive MS.

Glossary

AE

adverse event

EDSS

Expanded Disability Status Scale

FDR

false discovery rate

GFAP

glial fibrillary acidic protein

IQR

interquartile range

MEP

motor evoked potential

NfL

neurofilament light chain

PCA

principal component analysis

SAE

serious adverse event

SEP

somatosensory evoked potential

TMT

tandem mass tag

VEP

visual evoked potential

Author Contributions

C.E. Kvistad: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data. T. Kråkenes: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data. T. Holmøy: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data. L.E. Eidem: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data. L. Steffensen: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design. K. Wesnes: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design. M. Rojewski: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design. T. Eichele: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design. S. Wergeland: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design. S. Gavasso: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design. H. Barsnes: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; analysis or interpretation of data. J. Assmus: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design. N. Al-Sharabi: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design. S. Mohamed-Ahmed: drafting/revision of the manuscript for content, including medical writing for content; study concept or design. H. Vrenken: drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data. I. Brouwer: drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data. H. Mutsaerts: drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data. M. Tranfa: drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data. C. Gjerde: drafting/revision of the manuscript for content, including medical writing for content; study concept or design. M. Ytterdal: drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data. E. Rødahl: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data. F.S. Berven: major role in the acquisition of data; study concept or design. H. Schrezenmeier: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design. K. Mustafa: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design. L. Bø: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data.

Study Funding

This study was funded by KLINBEFORSK (The Norwegian Clinical Research Program), Helse Vest, the Norwegian Red Cross, and the Norwegian MS Society. The funders had no role in the design, conduct, analysis, or reporting of the trial.

Disclosure

C.E. Kvistad and T. Kråkenes report no disclosures relevant to the manuscript. T. Holmøy has received speakers' honoraria from Biogen, Roche, Merck, Novartis, Alexion, and Amgen; and honoraria for international advisory boards from Merck, Sandoz, and Alexion. L.E. Eidem, L. Steffensen, K. Wesnes, M. Rojewski, and T. Eichele report no disclosures relevant to the manuscript. S. Wergeland has received honoraria for serving on advisory boards for Biogen and Janssen; has received speaker fees from Biogen, Janssen, and Novartis; and has served as Principal Investigator for projects from EMD Serono and Merck KGaA. S. Gavasso, H. Barsnes, J. Assmus, N. Al-Sharabi, and S. Mohamed-Ahmed report no disclosures relevant to the manuscript. H. Vrenken has received research support from Merck, Novartis, Pfizer, and Teva; consulting fees from Merck; and speaker honoraria from Novartis; all funds were paid to his institution. I. Brouwer, H. Mutsaerts, M. Tranfa, C. Gjerde, M. Ytterdal, E. Rødahl, F.S. Berven, H. Schrezenmeier, K. Mustafa, and L. Bø report no disclosures relevant to the manuscript. During the preparation of this work, the authors used ChatGPT in order to improve language and readability. After using this tool/service, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication. Go to Neurology.org/N for full disclosures.