Remyelination in multiple sclerosis is biologically achievable, according to a review of a decade of proof-of-concept trials — but no remyelinating treatment has yet shown a clinically meaningful benefit for patients. That is the assessment of Nick Cunniffe, MD, PhD, assistant professor and consultant neurologist at the University of Cambridge, speaking at the 10th Joint ACTRIMS-ECTRIMS Meeting (MSToronto2026), held October 21-23 in Toronto, and reported by NeurologyLive on November 14, 2026.
The distinction matters. Several trials have produced signals on biological measures of myelin repair. None has demonstrated that those signals change disability outcomes. Cunniffe's own framing is candid: the field should be encouraged, he said, but "we have to stay grounded because we haven't yet seen a real clinical impact from these treatments." For a field that has run proof-of-concept studies for roughly ten years, that gap between biology and benefit is the central unresolved question.
What did the trials actually show?
The positive reading is real. Across several clinical trials, researchers have demonstrated that myelin repair in people with MS is achievable, per Cunniffe's talk. The proof-of-concept trials — including two he conducted in Cambridge — have tested whether drugs can produce a biological effect over short-duration studies. For related coverage, see Epigenetics Platforms Explained: Reading the Marks That Sit Above the Genome.
Those effects were measured two ways: tests of nerve function and tests of nerve structure. On the function side, researchers tracked the latency of the visual evoked potential, or VEP — the speed of electrical signaling along the optic nerve. As myelin repair proceeds, that latency shortens. On the structure side, trials applied myelin-sensitive MRI sequences. Both approaches, Cunniffe said, have been used "with varying degrees of success over the last 10 years."
The skeptical reading follows directly from that phrase. Varying degrees of success is not consistent success. And the measures that did work are surrogate measures — proxies for repair, not proof that patients function better. Cunniffe's summary of the field's position is that translating those early signals "into long-term benefit is the next step we need to make."
Why full-field VEP latency leads the endpoint pack
Among the measures tested, one has separated from the rest. Full-field VEP latency has emerged as a particularly sensitive, reproducible, short-duration endpoint for detecting remyelination effects, compared with other functional and imaging measures, according to the meeting coverage. Cunniffe called visual outcome measures "highly sensitive to detecting a remyelination effect over a short-duration trial," and said full-field VEP latency "has shown statistically significant effects across clinical trials to date."
That track record is the strongest technical asset the field has. It is also worth reading with some caution. A statistically significant change in nerve conduction speed over weeks is a long way from a slower progression of disability over years. The coverage itself flags the timing mismatch: some measures of myelin repair may show changes over relatively short periods, while the potential clinical benefits of neuroprotection may take years to emerge. Trials built to detect the first may simply be too short to detect the second.
Who should be enrolled — and who probably should not
The second design lesson concerns participant selection. A remyelinating treatment needs abundant axons — nerve fibers — that are intact and ready to be repaired, Cunniffe explained. In more advanced MS, many of those fibers have already degenerated because they were left without myelin for too long.
His conclusion is that enriching trials with earlier MS populations may maximize measurable repair. Progressive disease cohorts, by contrast, may dilute treatment effects because of longstanding neuroaxonal loss. He was careful not to overstate the point: "That's not to say our drugs won't work in people with progressive or more advanced MS," he said, "but they probably won't be as effective or as easy to measure."
There is a trade-off embedded here that the field will have to confront. Trials enriched with early-stage patients are more likely to show a measurable biological signal — but early-stage patients already have the least disability to recover. A treatment that works best in people with the least to lose faces a harder path to demonstrating clinical value than the surrogate data alone suggests. We covered a connected angle in Who Protects Your Genomic Data? GINA, NIH Policy and the Gaps.
What comes next
The next phase of development, per the meeting coverage, prioritizes three things: mechanism-targeted remyelinating drugs, enhanced myelin-sensitive MRI for lesion stratification, and discovery of electrophysiologic and blood biomarkers that reflect central nervous system repair.
Better biomarkers would address the field's core measurement problem. Stratifying lesions by myelin-sensitive MRI would help identify which lesions in a given patient still contain repairable tissue. Neither advance, on its own, closes the gap between biological signal and clinical benefit — that still depends on longer follow-up to detect any disability impact, which the coverage identifies as likely necessary.
The broader context frames the stakes. Disease-modifying therapies have substantially improved the management of MS by preventing inflammatory disease activity, per Cunniffe's talk. But preventing inflammation and addressing long-term disability progression remain distinct challenges. Remyelination research aims at the second problem — restoring damaged nerve fibers to help preserve neurological function — and it remains, after a decade of work, an unproven one.
What the evidence has established is feasibility: the biology works in humans. What remains unknown is whether any of it matters for how patients live. The honest reading of ten years of trials is that the field has solved the easier half of its problem and has not yet started on the harder half.
