Efficacy of Mesenchymal and Embryonic Stem Cell Therapy for the Treatment of Spinal Cord Injury: A Systematic Review and Meta-Analysis of Human Studies

Major Points and Findings:

A team of orthopaedic researchers at the University at Buffalo collected every English-language human study they could find of mesenchymal, embryonic or induced pluripotent stem cells for traumatic spinal cord injury. They found 30 studies with 656 patients. Across them, 43.3% of treated patients improved by at least one ASIA grade, 49.4% gained motor function and 73.6% gained sensation. No study reported a death, tumour or lasting neurological harm.

These figures are often quoted to families and need context. Most of the 30 studies had no control group and only six were randomised. The authors write that this “may contribute to an overestimation” of the benefit. The percentages describe what happened to people after treatment. How much of the change was due to the cells cannot be determined from them.

Aim:

To review the effect of embryonic, induced pluripotent and mesenchymal stem cells in people with traumatic spinal cord injury, measured by ASIA grade, ASIA motor and sensory scores, bladder and bowel function, pain and adverse events.

Methods:

The authors conducted a PRISMA systematic review, registered in PROSPERO (CRD42024568194). PubMed and Embase were searched up to August 2024, English full-text papers only.

Any study with original human data (randomised trials, cohort studies, case-control studies, case series) was included if it concerned acute or chronic spinal cord injury treated with embryonic, induced pluripotent or mesenchymal stem cells.

Quality was rated with Cochrane Risk of Bias 2 for randomised trials and the Newcastle-Ottawa Scale for the rest.

Because most studies had a single arm, a conventional meta-analysis of treatment against control was not possible. The authors instead pooled the proportion of patients who improved, using a random effects model. The analysis was not adjusted for study quality. A planned comparison of mesenchymal against embryonic cells was dropped because only three embryonic studies existed, all from one author.

Results:

The search returned 788 records and 30 studies were kept, published 2009 to 2024, with 4 to 226 patients each.

• Mean age was 34.2 years and 79.2% were male.
• Mean time from injury to treatment was 51.6 months, with a standard deviation of 55.7 months, so the range runs from acute to very long-standing injuries.
• Mean follow-up was only 7.9 months.
• Injury level: 39.5% cervical, 32.6% thoracic, 23.9% thoracolumbar, 3.8% lumbar.
• Cells: mesenchymal in all but three studies, from autologous bone marrow (most), autologous adipose tissue, or allogeneic umbilical cord or Wharton’s jelly (five studies). No study used induced pluripotent cells.
• The three embryonic cell studies, all by one author, supplied 246 of the 656 patients, because one of them was a retrospective cohort of 226.
• Route: intrathecal (“intradural”) in 18 studies, intralesional in 4, intravenous in 1, CT-guided percutaneous in 1 and a combination of routes in 7.

ASIA grade: Twenty-two studies (578 patients) reported grades before and after. At baseline 70.6% were ASIA A (complete), 14.2% B, 13.3% C and 1.9% D.

• Pooled proportion improving by at least one grade: 43.3% (95% CI 38.4% to 48.3%).
• Of 18 studies reporting the average change, 13 reported at least one grade, 2 reported less than one and 3 reported none. No study reported a worse grade after treatment.
• Results varied widely between studies. Three studies saw no grade improvement in any patient (0 of 9, 0 of 9 and 0 of 24), one of them a randomised Wharton’s jelly trial in chronic complete injury. Others reported improvement in 10 of 11 or 12 of 13 patients, and those were studies in acute or subacute injury.
• Only four studies gave grade data for controls, covering 68 patients, 67 of them ASIA A. 9.6% of those controls improved (95% CI 3.0% to 18.5%), significantly fewer than in the treated groups (p < .001).

Motor improvement was reported in 49.4% (16 studies, 209 patients) and sensory improvement in 73.6% (12 studies, 129 patients).

Bladder, bowel and pain: These outcomes could not be pooled because every study measured them differently.

• Bladder: 17 of 19 studies reported some improvement, ranging from better sensation to larger capacity or lower residual volume. One study reported complete recovery of bladder function in 2 of 24 patients and another in 5 of 12.
• Bowel: 14 of 15 studies reported improvement. The randomised Wharton’s jelly trial found no significant change in bladder or bowel function.
• Pain: seven studies used a visual analogue scale. Pain improved in 3 of 5, 9 of 16 and 7 of 8 patients in the studies that counted responders.

Adverse events: Eleven studies reported none and 19 reported minor or moderate events. These were headache, nausea and vomiting, dizziness, fever, muscle pain, radiating nerve pain, tingling and urinary tract infection, resolving within 24 hours to two weeks. One patient had a cerebrospinal fluid leak after intralesional injection, managed without surgery. One patient had yellow-tinged spinal fluid at a second dose with normal laboratory tests. No deaths, surgical site or intracranial infections, tumours, immune reactions or permanent neurological worsening were reported.

Study quality: Five studies were rated low risk of bias, 24 “some concerns” and one high risk. Six (20%) used randomisation and eight (26.7%) had any control group.

The abstract says no serious adverse events were reported. The results section describes the cerebrospinal fluid leak as the only serious event. This summary gives the figures as printed.

Conclusions:

The authors conclude that stem cell transplantation “appears to offer moderate improvements” in ASIA grade, motor, sensory, bladder and bowel function with mild, short-lived side effects, and that blinded randomised trials are essential. As limitations they list the predominance of non-randomised studies of moderate quality and an ASIA scale that depends on the examiner’s judgement and can be misclassified. They also cite wide variation in cell dose, timing, follow-up, injury level, severity and route, which prevented any deeper analysis.

Background Information:

The comparison with controls is indirect. The 43.3% and the 9.6% do not come from the same trials. The control figure rests on four studies whose patients were almost all ASIA A, while the treated figure mixes complete and incomplete, acute and chronic injuries from 22 studies. People with incomplete or recent injuries are more likely to change grade whatever is done.

The review gives no pooled result for complete against incomplete injury, for chronic against subacute injury, or for one cell source or route against another. The authors list these as questions for future trials. The 43.3% figure therefore cannot be applied to a specific patient, for example someone with a complete injury several years old, without going beyond what the review shows.

Follow-up was short. With an average of under eight months, the review cannot say whether gains last or whether late complications occur.