Three people is a small study.
Five years is not a small follow-up.
That is the tension inside the new Stem Cell Reports paper on NSI-566 in chronic cervical spinal cord injury. The study is not built to answer the final functional-recovery question. It does something earlier and less glamorous, but still necessary: it follows three chronic AIS-A cervical SCI patients for 54 to 60 months after direct implantation of a human spinal cord-derived neural stem cell line into the injured cervical cord.
For CNS cell therapy, that matters. The risk is not only whether a cellular product has biological activity. The procedure itself is part of the evidence. This was not a pill, an infusion, or a nasal spray. The intervention required open spinal cord exposure, intraparenchymal injections, perioperative management, and temporary immunosuppression. If a field wants to ask whether such a product can help, it first has to show that the intervention can be delivered and followed without producing unacceptable harm.
This paper belongs at that step.
The full text gives a much sharper read than the abstract alone. Six people were screened; three were enrolled. All three were white men with traumatic chronic cervical SCI, neurological levels C5-C7, AIS-A motor and sensory complete injuries, and 12 to 24 months since injury. The mean age was 38.7 years, and the mean follow-up period was 58 months.
The product was NSI-566, described as a human spinal cord-derived neural stem cell line originally derived from fetal spinal cord tissue and expanded as a single line. For administration, the cells were prepared as a live suspension at a final concentration of 2 x 10^6 cells/mL. Each subject received six intraspinal injections, with 2 x 10^5 cells per injection in 10 uL, placed bilaterally into parenchyma adjacent to the injury site and near medial white matter tracts roughly one segment below the injury epicenter.
That detail is not procedural clutter. It is the product story.
Cell therapy in the cord is a system: cell source, manufacturing, shipping, viability inspection, surgical access, injection hardware, targeting, immunosuppression, imaging, neurological exams, rehabilitation, and long-term surveillance. If one piece fails, the clinical interpretation changes.
Here, the first public reading is safety. The paper reports that all three subjects had successful surgery, no intraoperative complications, stable neurological status before discharge, and no procedure-related adverse events prior to discharge. The primary safety endpoint, adverse events and clinically significant laboratory abnormalities up to six months, was met. Serial MRI did not show evidence of post-injection inflammatory change, swelling, syrinx formation, tumorigenicity, enlarging mass, injection-related spinal cord damage, or inflammatory lesions within the spinal cord.
That is the important signal: a safety signal, and a meaningful one at this stage.
The adverse-event table still deserves a careful read. Across the three subjects, reported events included gastrointestinal symptoms, urinary tract infections, pain requiring a longer hospital stay for pain-control transition, dysreflexia, pneumonia, seizure, and one death. The death occurred in subject 0203 between the 54- and 60-month visits after cardiac arrest during vigorous exercise. The subject had a documented do-not-resuscitate order, and the family declined autopsy. The paper states that independent medical review found no evidence implicating the surgical procedure, stem cell product, or immunosuppressive therapy.
That sentence should not be softened into "nothing happened." Something did happen: a participant died. The disciplined reading is narrower. In this very small cohort, the investigators did not identify the death or other serious events as attributable to NSI-566, surgery, or immunosuppression. That is enough to preserve the safety conclusion. It is not enough to make broad safety claims.
The secondary signals are where the paper becomes easy to overread.
Subject 0202 had no overall ISNCSCI change at 60 months, but the motor level improved from C7 to C8 bilaterally beginning at 42 months and remained through follow-up; sensory level decreased from C8 to C7 bilaterally. EMG did not show new spontaneous activity below the injury level, but BMCA showed new activation during left wrist extension and flexion tasks at 36 months, sustained to study completion.
Subject 0203 moved in the other direction clinically, with a one-level ISNCSCI decline from C6 to C5 at 54 months. At the same time, EMG and BMCA showed new or stronger right upper-extremity activity, including deltoid motor unit action potentials and improved right wrist-extension motor activation.
Subject 0206 had no overall ISNCSCI change and no motor-score change at 60 months, but sensory level improved on the right from C6 to C7 at 54 months and stayed there. EMG and BMCA showed increased or better localized motor activity in some upper-extremity recordings.
This is exactly why the piece should be read as earned hope, not efficacy.
There are signs worth preserving. A cervical cord intervention that is followed for five years, with no obvious graft-related mass effect or tumor signal and some neurophysiological changes, is not trivial. But the study was not controlled, was not powered for inference, and analyzed outcomes descriptively. The authors themselves emphasize that modality discrepancies, small sample size, lack of controls, post-surgical physical therapy, and interrater variability limit interpretation. All three patients also had two to three mobility sessions per week for 6 to 12 weeks after surgery, which makes functional attribution harder.
So the fair public sentence is not "NSI-566 improved function."
It is closer to this: in a three-patient cervical SCI Phase 1 cohort, direct NSI-566 implantation was feasible and followed for five years without product-, surgery-, or immunosuppression-attributed serious safety problems, while exploratory clinical and neurophysiological observations justify more rigorous testing.
That sentence is less dramatic than "stem cells restore movement." It is also the sentence that can survive an expert reader.
The cervical location is what gives the paper its weight. Cervical injuries are unforgiving, but they are also where modest changes in upper-limb function can matter deeply for independence. The paper notes that higher cervical targeting may increase both risk and potential benefit. That is the development logic: if safety holds, the next question becomes whether a controlled trial can separate true recovery from measurement noise, rehabilitation effects, spontaneous late changes, and site-specific execution.
The paper also keeps a manufacturing and ethics issue visible. NSI-566 comes from a single fetal spinal cord tissue source. The authors state that the tissue was obtained under regulatory and ethical oversight, but the origin still matters for scalability, public acceptance, and future product strategy. CNS replacement or repair programs do not escape those questions just because the biology is interesting.
The next threshold is not mysterious. Larger cohorts. Controlled comparison. Cleaner dose logic. Standardized rehabilitation context. Prespecified handling of discordant EMG, BMCA, and ISNCSCI findings. Longer imaging and immune surveillance. Replication outside one expert environment.
If the biological signal is real, it has to survive that harder setting.
And if it does not, the safety data still have value. CNS regenerative medicine needs negative space as much as positive signal: which delivery routes are tolerable, which monitoring plans catch the right risks, which secondary readouts are too noisy, and which product-procedure combinations can actually be executed.
That is why this paper is useful. It is not a destination. It is a checked bridge on the way to the real efficacy question.
Source Boundary
- Primary full text: Stem Cell Reports / Cell Press, "Long-term safety and clinical outcomes from a single-site phase 1 study of neural stem cell transplantation for chronic cervical spinal cord injury" - https://www.cell.com/stem-cell-reports/fulltext/S2213-6711(26)00137-2
- PubMed record: https://pubmed.ncbi.nlm.nih.gov/42167221/
- Evidence level: single-site, open-label Phase 1 human safety follow-up; N=3.
- Do not imply: proven functional recovery, generalizable efficacy, clinical availability, or broad applicability to spinal cord injury patients.