STOK Zorevunersen: 4-Year Durability in Dravet Syndrome

Stoke Therapeutics (STOK) and Biogen reported that zorevunersen, their antisense drug for Dravet syndrome, sustained cognition and behavior improvements through four years of treatment, alongside durable seizure reductions and a clean safety profile. The data, from the Phase 1/2a open-label extension (OLE), will be presented at the 16th European Epilepsy Congress in Athens this week.

The new element is a set of exploratory sub-analyses showing effects on the most severe seizure types, the leading risk factor for SUDEP (sudden unexpected death in epilepsy). Up to 20% of children with Dravet syndrome die before adulthood, and no drug yet treats the disease’s underlying cause, a mutated SCN1A gene that leaves the brain short of the NaV1.1 sodium channel. Zorevunersen is an antisense oligonucleotide that nudges the healthy copy of SCN1A to make more of that protein, rather than simply suppressing seizures the way Epidiolex and Fintepla do.

The biology is specific: most Dravet cases come from a loss-of-function mutation in one copy of SCN1A, leaving the brain with half the normal NaV1.1 channel, a protein that calms overexcited neurons. By boosting the healthy allele, zorevunersen aims to restore that channel instead of just damping seizures. Dravet affects roughly 16,000 people in the US and about 38,000 across the US, UK, EU-4, and Japan, a small but high-need population where a disease-modifying drug would command premium pricing.

STOK closed at $30.89 on September 1, valuing the company near $1.9 billion.

The honest read is that this is long-term durability data, not the registrational result. It de-risks the question that hangs over central nervous system antisense programs: whether the benefit fades. The SUDEP-linked sub-analysis matters because severe seizures, not overall seizure counts, are what kill these children. But it is exploratory, and the companies did not quantify the effect.

The binary event remains the Phase 3 EMPEROR Dravet study, a randomized sham-controlled trial that is fully enrolled, with a readout expected in Q3 2027. The 162-patient study runs a 52-week treatment period followed by an open-label extension, with key secondary endpoints tracking major motor seizure frequency and cognition and behavior on the Vineland-3 scales. Stoke retains US, Canada, and Mexico rights; Biogen holds the rest of the world. The FDA has granted zorevunersen Breakthrough Therapy, orphan drug, and rare pediatric disease designations.

What to watch: the full four-year dataset at EEC and any update on EMPEROR’s data-cut date. A miss there would cut the $1.9 billion valuation sharply, since zorevunersen is Stoke’s lead asset.

What It Means for the Trade

Four-year open-label durability data does not move a $1.9 billion valuation — the Phase 3 EMPEROR readout in Q3 2027 does. But this dataset does something specific and valuable: it attacks the main historical objection to CNS antisense therapy, which is that the benefit fades or the tolerability craters with chronic exposure. Sustained cognition and behavior improvement at year four, on top of durable seizure reduction and a clean safety profile, is the profile that makes the EMPEROR bet credible. In a space where most neurology programs die on durability, zorevunersen keeps answering the question before it is asked.

The positioning nuance is who this drug is for. Epidiolex and Fintepla suppress seizures; zorevunersen is designed to restore the missing NaV1.1 channel — disease modification, not symptom control. The Vineland-3 cognition and behavior endpoints in EMPEROR are the commercial differentiator, because developmental improvement is what parents and payers will pay premium prices for in a disease that kills up to 20% of affected children before adulthood. A drug that only matches Epidiolex on seizures has no value proposition; a drug that changes trajectory does.

Mechanism and Market Context

The biology is among the cleanest in rare neurology: most Dravet cases trace to a loss-of-function mutation in one SCN1A copy, leaving roughly half the normal NaV1.1 sodium channel in inhibitory interneurons. Zorevunersen is an antisense oligonucleotide that nudges the intact allele to upregulate NaV1.1 production — restoring the deficiency instead of masking downstream seizures. That precision explains why the effect generalizes to cognition and behavior, not just seizure counts.

The epidemiology defines a premium orphan market: roughly 16,000 US patients, about 38,000 across the US, UK, EU-4, and Japan. The structure splits geography — Stoke keeps North America, Biogen holds the rest of the world — so the EMPEROR readout is a proportionally bigger event for STOK than for BIIB. The Breakthrough Therapy, orphan drug, and rare pediatric disease designations stack the regulatory deck.

Risks and What Could Go Wrong

1. Open-label inflation. Four years of uncontrolled data with no placebo comparison — patients, families, and investigators all know who is treated. The improvements are encouraging but exploratory; randomized-controlled EMPEROR is the only result that counts.
2. The sham-controlled bar. EMPEROR runs a sham-controlled design in a pediatric seizure population, which is methodologically rigorous but produces lower apparent effect sizes than open-label extension data suggests — expect the market to misprice this gap when the readout approaches.
3. Endpoint spread. The primary endpoint and the major motor seizure secondary must both move; cognition and behavior on Vineland-3 are harder instruments and noisier than seizure diaries.
4. Competitive timing. Other SCN1A-directed approaches are advancing across the industry; an EMPEROR miss would hand the disease-modifying window to whoever reads out next.

Hold through the data-cut updates, but recognize the real trade is sized for Q3 2027. Everything until then is evidence accumulation, not catalyst.