Cell-Based Therapies for Traumatic Optic Neuropathy: Recent Advances, Challenges, and Perspectives

Major Points and Findings:

This narrative review from Shanghai covers the cell-based approaches tried for traumatic optic neuropathy (TON), the optic nerve damage that follows a blow to the head or face. It draws on 189 articles. Almost all of the evidence is preclinical and comes from experiments in rats and mice, usually after a deliberate crush of the optic nerve. Only two small phase I studies in people exist.

The parts most relevant to families are the two human trials (one used umbilical cord MSCs and found no difference in vision compared with surgery alone), the animal results for MSCs from umbilical cord, Wharton’s jelly and placenta, and the list of what is still unknown. One discrepancy should be mentioned. The abstract says the two trials “confirmed the safety and potential benefits” with “reported improvements in visual acuity”. In the body of the paper, only one of the two reported better vision, in four patients without a control group.

Aim:

To review the mechanisms and animal models of TON, summarise experimental and clinical work on cell-based therapies, and set out the obstacles to clinical use.

Methods:

One author searched PubMed and Web of Science between May 2024 and January 2025 for English-language articles from 2000 to 2025. ClinicalTrials.gov and the Chinese registry ChiCTR were searched for trials (accessed 21 January 2025). The cell therapy section rests on 11 articles on ganglion cell replacement, 13 on MSCs, 5 on neural stem/progenitor cells and 11 on cell-derived products. There was no formal quality scoring or pooling.

Results:

TON occurs in 0.5 to 5% of closed head injuries and is one-sided in more than 96% of cases. The nerve segment inside the bony optic canal is the site of injury in 71.4%. On OCT the ganglion cell layer starts to thin within 2 weeks and thinning levels off at about 20 weeks, which the authors take as the period in which treatment should be given. There is no agreed standard treatment. Steroids, decompression surgery and observation all have conflicting results.

Ganglion cell replacement (animal studies): Ganglion cell-like cells can be made from embryonic or induced pluripotent stem cells in 7 to 60 days. Injected into rodent eyes (10,000 to 500,000 cells), they survived 3 weeks to 5 months and some settled in the ganglion cell layer. Survival was lower in crushed-nerve eyes than in healthy ones, and no study showed reconnection to the brain. Named risks are tumours from leftover stem cells and immune rejection.

Mesenchymal stem cells (animal studies): Twelve studies are tabulated, with MSCs from bone marrow, fat, periodontal ligament, placenta, cord blood and Wharton’s jelly.

• Routes and doses per eye were intravitreal 20,000 to 500,000 cells, subtenon (beside the eye) 1 to 2 million and peribulbar 10 million. The one intravenous study used 1 million placenta-derived MSCs and reported better axon survival at 28 days.
• The consistent finding is protection, seen as more ganglion cells surviving and more axon regrowth, for 14 to 240 days. The longest effect (240 days) followed the highest intravitreal dose of rat bone marrow MSCs. In that study regenerated axons reached the brain, yet the animals’ visual behaviour did not improve.
• Wharton’s jelly MSCs gave long-term ganglion cell survival and partial recovery of synaptic function at 120 days. One study, however, found that they drew in many activated immune cells that disrupted the layered structure of the retina, apparently temporarily.
• Umbilical cord blood MSCs grown as a flat layer improved flash visual evoked potentials. The same cells grown as spheroids did not. Hypoxia preconditioning made placenta-derived MSCs more effective.
• Most injected cells stay in the vitreous for 2 to 18 weeks and few enter the retina. MSCs did not need to become nerve cells to have an effect, which is attributed to secreted factors (BDNF, CNTF, VEGF) and calming of inflammation.

Five studies used neural stem/progenitor cells in animals. Given intravenously, cells derived from human embryonic stem cells roughly doubled ganglion cell survival but did not improve a visual behaviour test.

Extracellular vesicles (animal studies): Eleven studies were found, with observation for 10 to 60 days. Vesicles from umbilical cord, bone marrow and placenta MSCs, Schwann cells and others improved ganglion cell survival. Two umbilical cord MSC vesicle studies disagreed. In the first, three doses of 1 × 10⁹ particles protected cells but did not regrow axons. In the other, a single dose of 3 × 10⁹ improved survival, axon regrowth, nerve fibre layer thickness and pattern ERG at 21 days. Vesicles cannot divide and provoke little immune response, but they are cleared from the vitreous quickly and a higher dose is not always better. The right dose, timing and source are all unknown.

The two human trials:

• China, ChiCTR-TRC-14005093 (Li and colleagues, 2021): a phase I, open-label study of 20 patients, run from September 2014 to July 2016. Ten had umbilical cord MSCs on a gelatin sponge placed directly on the injured nerve during optic canal decompression surgery, and the rest had surgery alone. Over 6 months there were no systemic or eye complications and no adverse events linked to the transplant. Recovery continued for about 1 month in the MSC group against about 1 week in controls, but there was no significant difference in vision improvement between the groups. The review suggests the small sample and very poor starting vision as possible reasons.
• South Korea, NCT05147701 (Sung and colleagues, 2020): a phase I, open-label study. Five patients enrolled and one dropped out because of coexisting damage at the fovea. Four received placenta-derived MSCs by subtenon injection. Over 12 months there was no abnormal cell growth, tumour, severe inflammation or other serious event, and visual acuity improved in all four. There was no control group.

Conclusions:

The authors regard cell-based therapy for TON as promising and not yet ready for clinical use. They call for longer observation (most animal studies stop at about 30 days), testing in primates before humans, and standardised doses and routes. Regrown axons in animals are mostly unmyelinated, which limits function. They state that cell therapy “cannot eliminate the risks of inflammation following cell administration” and that the need for immunosuppression remains open.

The limitations they list are the restriction to English-language papers, a focus on 2018 to January 2025, and a thin evidence base for neural progenitor cells.

Background Information:

The crush model has limits as a guide to clinical practice. In the standard rodent model about 70% of ganglion cells degenerate between days 3 and 8, with roughly 20% surviving at day 14 and 10% at day 28. Most studies treat immediately after injury, whereas a patient may seek treatment months or years after trauma. The review gives no evidence of benefit in long-established traumatic optic atrophy.

The review names intravitreal, subtenon, intravenous and peribulbar delivery as the routes used in TON research, and describes subtenon injection as less invasive and safer for repeated treatment than intravitreal injection. Intrathecal delivery was not tested in any TON study it covers.