Assessment of Mesenchymal Stem Cells for the Treatment of Spinal Cord Injury: A Systematic Review and Network Meta-Analysis

Major Points and Findings:

This network meta-analysis pools 18 randomised controlled trials (950 patients) of mesenchymal stem cell (MSC) transplantation for spinal cord injury and tries to rank cell sources and delivery routes against each other. The main finding is that umbilical cord MSCs ranked above bone marrow MSCs, and that intrathecal injection ranked above intravenous and intralesional delivery.

That conclusion favours the cell type and route this clinic uses, so it should be read with particular care. The underlying trials are mostly small and unblinded, with an unclear risk of bias. The authors’ own funnel plots point to possible publication bias. The review contains no analysis of side effects, and the combination “umbilical cord cells by the intrathecal route” was never analysed as a single strategy. The review is a useful map of the randomised evidence up to early 2024. As proof of what works best it is weak.

Aim:

To evaluate the effect of MSC transplantation on motor function, sensory function and activities of daily living after spinal cord injury, and to identify the best cell source and transplantation route using network meta-analysis.

Methods:

The review followed PRISMA and was registered in PROSPERO (CRD42023466102).

Search: The authors searched four English-language databases (PubMed, Cochrane Library, Web of Science, Embase) and four Chinese ones (CNKI, VIP, Wanfang, SinoMed), from inception to 5 March 2024.

Eligible studies were randomised controlled trials in patients, where the treatment group received MSCs on top of whatever the control group received. They had to report ASIA motor or sensory scores, ASIA grade improvement, or a daily living scale (Barthel Index, Modified Barthel Index, FIM or SCIM-III). Conference papers, non-randomised studies, studies without extractable data and studies with a “low quality score” were excluded.

Two reviewers screened and assessed bias with the Cochrane tool. Where a trial reported several time points, only the last was used. Change scores were calculated from baseline and final means, with the standard deviation of change estimated using an assumed correlation of 0.5. Standard pairwise meta-analysis used standardised mean differences (SMD) and odds ratios. The network analysis was run in Stata and interventions were ranked by SUCRA, a 0 to 100% score where higher means more likely to be best.

Results:

From 5,605 records, 18 trials qualified. Of these, 8 were published in English and 10 in Chinese, dated 2007 to 2022. There were 531 patients in MSC groups and 419 controls. Trial size ranged from 13 to 100 participants.

• Cell source: autologous bone marrow MSCs in 12 trials, allogeneic umbilical cord MSCs in 6.
• Route: intrathecal in 7 trials, intralesional (into the injured area of the cord) in 7, intravenous in 2. The rest mixed intrathecal and intravenous dosing or did not report the route.
• Doses varied by several orders of magnitude between trials.
• Timing: most trials enrolled chronic injuries, with mean times since injury up to about four years. Several treated people within hours or days of injury, including the largest umbilical cord trial.
• Follow-up at final assessment ranged from 7 days (two trials) to 18 months.

Quality of the trials: Only 7 of 18 described how randomisation was done. One used concealed allocation and one blinded participants and staff. Six blinded the outcome assessors.

Pooled effects of MSCs against control:

• ASIA motor score (16 trials): no significant difference at 0 to 3 months (SMD 0.27, 95% CI -0.13 to 0.67). Significant differences at 4 to 6 months (SMD 0.56), 7 to 12 months (SMD 2.00, 95% CI 1.64 to 2.37) and 13 to 18 months (SMD 0.78). Heterogeneity was high overall (I² 78.6%).
• ASIA sensory score (12 trials): SMD 0.35 at 0 to 3 months, 0.79 at 4 to 6 months and 1.81 at 7 to 12 months.
• Improvement of at least one ASIA grade (8 trials): odds ratio 11.19 (95% CI 4.70 to 26.64), calculated after one trial was removed to eliminate heterogeneity.
• Activities of daily living (10 trials): SMD 0.57, 0.86, 1.71 and 1.28 across the four time bands.

Rankings:

• By cell source: umbilical cord MSCs ranked first for motor score (SUCRA 82.2%, against 67.7% for bone marrow MSCs and 0.1% for control), sensory score (81.4% against 68.5%) and daily living (79.4% against 70.5%).
• By route: intrathecal ranked first for motor score (93.1%, against 54.5% intravenous, 49.6% intralesional, 2.8% control), sensory score (97.1%) and daily living (98.5%).
• For both rankings, the funnel plots were not symmetrical, which the authors read as possible publication bias and small-study effects.

The review contains no pooled or tabulated adverse event data, and safety was not an outcome. It also has no subgroup analysis by completeness of injury (AIS grade) or by acute against chronic injury, although the included trials differ widely on both.

Conclusions:

The authors conclude that intrathecal transplantation of umbilical cord MSCs “may be the best transplantation strategy” for functional impairment after spinal cord injury, and ask for further high-quality studies. Their stated limitations include small samples and the fact that most trials did not use blinding or allocation concealment. They are also uncertain whether the cells used in some trials met the International Society for Cellular Therapy minimum criteria for MSCs, and they accept that a ranking cannot explain why one approach would be better. They call for standardised, clinical-grade cell products and large randomised trials to establish the best route, dose and timing.

Background Information:

The ranking needs careful interpretation. Cell source and route were ranked in two separate networks. The paper never tests “umbilical cord cells given intrathecally” as one strategy. In the table of included trials, only one used umbilical cord MSCs by the intrathecal route alone (100 patients treated about six and a half hours after injury, AIS B to D), and one small trial of 24 patients combined one intravenous with three intrathecal doses. The other umbilical cord trials used intralesional or intravenous delivery. The recommendation is therefore an inference from two rankings, resting on very few directly relevant trials. Neither of those two can be identified from the review as a trial in chronic complete injury.

The effect sizes also call for caution. An SMD of 2.00 is very large by the standards of rehabilitation research. With high heterogeneity, and with blinding of participants reported in one trial and of assessors in six, effects of this size should be treated as upper estimates.

The search closed in March 2024 and the newest included trial is from 2022. A double-blind, placebo-controlled trial of intrathecal Wharton’s jelly MSCs in chronic incomplete cervical injury, published in 2026, found no difference from placebo at 12 months. That trial is not part of this analysis.