Intrathecal Delivery of Adipose-Derived Mesenchymal Stem Cells in Traumatic Spinal Cord Injury: Phase I Trial

Major Points and Findings:

CELLTOP was the Mayo Clinic phase I trial in which ten people with a severe traumatic spinal cord injury received one intrathecal injection of their own fat-derived mesenchymal stem cells and were then followed for two years. This paper is its full report, which is carefully documented and widely quoted. The study was small and open-label with no control group. It shows what the procedure involves and how it was tolerated, but it cannot show whether the cells caused the improvements that were recorded.

The cells were autologous adipose MSCs, which differ from the umbilical cord cells used in our own protocols. The route, lumbar puncture into the spinal fluid, is the same.

Aim:

To find out whether harvesting, expanding and intrathecally injecting 100 million autologous adipose-derived MSCs is safe in people with traumatic spinal cord injury. Sensory and motor scores, MRI, spinal fluid markers and somatosensory evoked potentials (SSEPs) were secondary outcomes.

Methods:

This was a single-arm, open-label phase I trial at Mayo Clinic, Rochester (ClinicalTrials.gov NCT03308565), running from 2017 to October 2021.

Eligible patients were adults aged 18 or over with a traumatic, blunt, non-penetrating, non-degenerative injury in the 12 months before enrolment, graded AIS A or B at the time of injury. People whose grade had already improved since the injury could still take part. Fourteen were screened and ten enrolled.

Participants were eight men and two women, mean age 34.6 years (range 18 to 65). Six had cervical and four had thoracic injuries. At the time of injury eight were AIS A and two were AIS B. By the day of injection five were AIS A, two were B and three were C, so several had already improved on their own. The average time from injury to injection was 12 months (range 7 to 22). One patient waited 22 months because the first fat sample failed to grow and a second biopsy was needed.

Procedure:

• About 15 mL of fat was removed through a 1 to 2 inch incision in the abdomen or thigh under local anaesthetic.
• Cells were grown in medium with 5% platelet lysate for 4 to 6 weeks, then frozen for sterility, karyotype and flow cytometry release tests, and thawed up to four days before use.
• On the day, 100 million cells in 10 mL of lactated Ringer’s solution were injected through a lumbar puncture between L2 and L5 under fluoroscopy, over 2 to 3 minutes, followed by a 1 mL flush.
• Patients were then turned side to back to side every 15 minutes for two hours to spread the cells through the spinal fluid. They were admitted the day before and stayed 24 hours afterwards.

Follow-up visits took place on day 2, day 3 and at weeks 1, 2, 6, 12, 24, 48 and 96. Spinal fluid was sampled before the injection and two weeks later.

Results:

Safety: No serious adverse events occurred in two years. There were 44 adverse events in total, 17 of them judged possibly related to the cells.

• Headache in 8 of 10 patients in the first two weeks.
• Musculoskeletal pain in 9 of 10, comprising back pain (2), hip pain (3), tailbone pain or soreness (3) and leg pain (1).
• Other early events included fever (2 patients), blurry vision, dizziness and one episode of autonomic dysreflexia. Six patients needed only over-the-counter medication.
• Late events were mostly ordinary complications of spinal cord injury: skin breakdown, spasms, depression (2 patients) and one hip fracture after a fall, judged unrelated.
• Blood tests were normal. Two weeks after injection the median spinal fluid cell count had risen from 2 to 9.5 cells per microlitre, and protein was unchanged.

MRI: The injury site itself looked unchanged at one year in all ten patients. However, 8 of 10 developed thickening or clumping of the cauda equina nerve roots in the lower spine, some with nodular enhancement. None had matching symptoms. The authors compare it to the reactive arachnoiditis seen in another trial of intrathecal MSCs and note that case reports have linked similar changes to nerve root compression. They regard the question as unresolved.

Neurological change: Seven of ten patients improved by at least one AIS grade between injection and week 96.

• Of the five who were AIS A at injection, two stayed AIS A and three improved. According to the patient table, one moved to B and two moved to C, driven by return of deep anal pressure, voluntary anal contraction or some motor function below the injury.
• Of the five who were AIS B or C at injection, four improved (two from B to C and two from C to D) and one stayed at C.
• SSEPs improved in three patients, were unchanged in two, were normal throughout in one and could not be interpreted in four.
• VEGF in spinal fluid rose in seven of the nine patients with paired samples. In six patients it went from undetectable to detectable.

The abstract gives 8 patients for both headache and musculoskeletal pain. The full text gives 8 and 9. The results text says two AIS A patients moved to B and one to C, while the patient table and the discussion give one to B and two to C. This summary follows the patient table. A published correction changed only the injury-to-injection interval.

Conclusions:

The trial met its safety endpoint. Manufacturing worked for every patient and a single 100 million cell intrathecal dose was well tolerated, with frequent but minor side effects. The authors write that the improvements “are to be interpreted with caution”. For comparison they cite natural history data in which only about 5% of AIS A patients improve late, against three of five here, but they state that without controls the change cannot be attributed to the cells. They add that these patients had rehabilitation at a highly specialised centre, which most people do not receive, and they call for larger randomised trials.

Background Information:

All patients were injured less than a year before enrolment and nine of the ten were injected 7 to 14 months after injury, a period when some natural recovery still occurs. Four of the ten had already changed AIS grade before receiving any cells. The results say little about people many years from their injury.

The trial deliberately recruited the most severe injuries (AIS A or B at onset). The two patients who stayed AIS A had complete injuries at T11-T12 and C4.

On mechanism, the authors do not claim the cells rebuild the cord. They point to evidence that MSCs survive only briefly and act through released factors, and they present the VEGF rise as support for that idea.