The Dystrophin Standard: How a 1% Biomarker Signal Became a $967 Million Accelerated Approval Precedent

Pull up the FDA’s accelerated approval files for eteplirsen, golodirsen, viltolarsen, and casimersen, and look for the dystrophin numbers. The increases that justified approval, the surrogate endpoint the agency staked four regulatory decisions on, average roughly one percentage point above baseline. Not ten percent. Not the threshold that a reasonable clinical pharmacologist might connect to functional muscle performance. One percentage point, measured against healthy controls, in a disease where patients are losing the ability to walk. Sarepta Therapeutics and its competitors collected nearly $967 million in net product revenue from these four drugs in 2024 alone. The question the Lancet Neurology commentary on exon skipping forces into the open is not whether antisense oligonucleotides represent scientific progress. They do. The question is whether the FDA’s surrogate endpoint standard has drifted so far from clinical meaningfulness that the agency is now approving drugs faster than it can prove they work.

This is not a hypothetical concern about regulatory philosophy. It is an operational accountability problem with a measurable cost, and someone needs to say so plainly.

A Surrogate That Doesn’t Speak For Itself

Accelerated approval exists because rare, life-threatening diseases cannot always wait for full clinical endpoint confirmation. The statute is sensible. The FDA’s application of it in DMD has been something else. Under 21 CFR § 601.41, accelerated approval for biologics rests on a surrogate endpoint that is “reasonably likely to predict clinical benefit.” The agency interpreted dystrophin protein expression, measured as a percentage of levels found in healthy controls, as that surrogate. The PROMOVI trial, an open-label study evaluating eteplirsen in 78 ambulatory DMD patients aged 7 to 16 over 96 weeks, reported a 7-fold increase in dystrophin protein, but the absolute level remained so low that functional correlations were difficult to establish in any controlled setting. When FDA reviewers approved eteplirsen in 2016, the agency’s own statistical reviewer dissented, arguing that the evidence was insufficient. That dissent did not stop the approval.

What followed was precedent-setting in the most dangerous regulatory sense. Once eteplirsen cleared, each subsequent antisense oligonucleotide (golodirsen in 2019, viltolarsen in 2020, casimersen in 2021) inherited the same surrogate endpoint framework, with the same approximate magnitude of effect. The FDA had, in effect, defined the minimum acceptable signal not by clinical science but by the first approval it chose not to block.

This is the pattern that accelerated approval critics warned about, and the DMD case makes it concrete. According to data compiled by the Center for Research on Environmental Chemicals, between 2015 and 2024, 24 out of 165 drugs that received accelerated approval were ultimately withdrawn from the market. Of those withdrawals, 62.5 percent (15 of 24) were pulled because confirmatory studies failed to verify clinical benefit, with a median time from approval to withdrawal of 1,243 days. The DMD antisense oligonucleotides have not been withdrawn. But the evidentiary trail behind them raises a structural question that the Lancet Neurology commentary articulates precisely: when increases are small, measured at roughly one percentage point, and the connection between that biomarker increment and functional patient outcomes remains unconfirmed at scale, what exactly did the FDA approve?

The counterpoint the sponsors would offer is that these patients have no alternatives, that any dystrophin is better than none, and that the regulatory pathway was designed for exactly this situation. That argument has merit on humanitarian grounds. It falls apart as a scientific defense when the same surrogate gets recycled across four separate approvals without a single completed, placebo-controlled confirmatory trial demonstrating that the biomarker increment translates to delayed disease progression, preserved ambulation, or improved respiratory function. Compassion and evidentiary rigor are not mutually exclusive, and treating them as if they are has real costs for the patients these drugs are supposed to help.

The Confirmatory Trial Gap Nobody Closes

Accelerated approval is a conditional grant. The condition is a post-market confirmatory trial that verifies the surrogate endpoint predicts what it is supposed to predict. In oncology, FDA issued a draft guidance in March 2023 specifically to tighten confirmatory trial requirements, signaling that the agency recognized the pathway was being used to park products in an approval limbo where commercial launch precedes clinical proof. The DMD antisense oligonucleotide situation predates that guidance and, critically, sits outside oncology, where the FDA’s enforcement attention has been most concentrated.

ALS offers the most instructive parallel. Tofersen (Qalsody) received accelerated approval from FDA on April 25, 2023, for SOD1-ALS based on a reduction in plasma neurofilament light (NfL) concentration, a surrogate with a more established mechanistic link to neurodegeneration. The NfL reduction reached 50%, a signal an order of magnitude larger than anything the DMD exon-skipping drugs have demonstrated on dystrophin. Even there, the clinical benefit confirmation process faces timelines measured in years. The DMD case involves smaller biomarker signals in a disease with longer natural history, which means confirmatory trials must run longer, enroll patients in a rare population, and detect functional differences against a disease trajectory that varies enormously between individuals. The structural incentive for sponsors is plain: launch commercially under accelerated approval, enroll a confirmatory trial slowly, and generate revenue while the evidentiary clock runs.

Sarepta’s 2024 annual revenue of $967.2 million from its RNA-based PMOs, including eteplirsen, golodirsen, and casimersen, represents a commercial outcome that dwarfs the cost of any confirmatory trial. When the financial upside is that asymmetric, the regulatory pressure to complete confirmatory trials quickly has to come from FDA, not from sponsor goodwill. And FDA’s enforcement bandwidth is not infinite, particularly under the resource constraints the agency has faced since 2025.

What Sponsors and Sites Must Do Before FDA Acts

The operational accountability failure here runs in two directions. Sponsors who built development programs on a one-percentage-point surrogate exploited a permissive precedent without building the confirmatory evidence base that would actually justify long-term market position. Sites that enrolled patients in open-label studies measuring dystrophin expression without randomized controls contributed to a regulatory record that looks like a confirmatory trial program but functions as a perpetual biomarker monitoring exercise.

The structural read is this: FDA’s accelerated approval pathway, when applied to rare diseases with small patient populations and weak biomarker-to-function correlations, creates a category of conditional approvals that may never get confirmed because the confirmatory trials are too hard to run at the speed the pathway assumes. The agency has not acknowledged this structural tension explicitly for DMD. It should. The 2023 oncology confirmatory trial guidance is a template that deserves a rare disease analogue, with specific timelines, pre-specified functional endpoints, and a public registry of confirmatory trial status tied to each accelerated approval.

For sponsors with DMD programs or analogous rare disease antisense platforms, the watchdog directive is blunt: design your confirmatory trial before your accelerated approval filing, not after. Pre-specify functional endpoints in your confirmatory protocol at the IND stage, document the mechanistic rationale connecting your surrogate to those endpoints in your Phase 2 briefing document, and negotiate a binding confirmatory trial timeline with FDA as a condition of your accelerated approval letter. “Reasonably likely to predict clinical benefit” is not a permanent safe harbor. The 62.5 percent withdrawal rate among failed confirmatory programs proves that accelerated approval is a bet, not a guarantee, and the patients enrolled under that bet deserve to know the odds are being tracked rigorously.

For FDA-watchers, the signal to monitor is the agency’s next rare disease accelerated approval withdrawal decision, particularly whether the public record includes the underlying BIMO inspection findings and the confirmatory trial protocol deviations that preceded it. Fifteen withdrawals for failed clinical benefit confirmation between 2015 and 2024 produced almost no public accountability documentation at the operational level. That gap between the withdrawal announcement and the operational record of what went wrong is where the next DMD-type precedent is already forming.

The Lancet Neurology commentary frames the DMD exon-skipping story as progress. On the science, it is. On the regulatory precedent, four approvals resting on a one-percentage-point biomarker increment, with $967 million in annual revenue generating commercial facts before confirmatory facts, the next Advisory Committee review of an antisense oligonucleotide program will tell us whether FDA has learned to ask a harder question before the approval letter goes out.

References

1. Lancet Neurology, “Progress in exon skipping for Duchenne muscular dystrophy”
2. Johns Hopkins, “Open-label evaluation of eteplirsen in patients with Duchenne muscular dystrophy” (PROMOVI trial data)
3. Center for Research on Environmental Chemicals, “Accelerated Approval Pathway Summary: Withdrawal rates and confirmatory trial outcomes 2015-2024”
4. Sarepta Therapeutics, 2024 Annual Revenue Filing (RNA-based PMO net product revenue $967.2M)
5. FDA, “FDA Issues Draft Guidance Aimed at Improving Oncology Clinical Trials for Accelerated Approval” (March 2023)
6. ALS Northwest, Tofersen (Qalsody) accelerated approval summary, SOD1-ALS, April 25, 2023, 50% NfL reduction