On microdystrophin, the FDA is doing what the accelerated-approval standard has always required: asking that a biomarker be shown to predict real benefit before it is trusted.
Much of the commentary around REGENXBIO’s RGX-202 has framed the FDA’s stance as regulatory inconsistency, arguing that the agency is holding one sponsor to a standard it waived for another. But the question before the FDA is simpler: whether microdystrophin expression is actually strong enough to predict clinical benefit.
The charge of inconsistency does not survive contact with the record. And it is worth noticing that its loudest version comes from a sponsor whose commercial position gives it every reason to want the bar set lower.
Nor is REGENXBIO alone in having a commercial interest in how low that bar is set. Solid Biosciences is pursuing the same accelerated-approval opening for SGT-003 while emphasizing high microdystrophin expression and other biomarker results from its early-stage trial. Solid is still gathering the controlled functional evidence that can show whether those laboratory findings actually translate into meaningful improvement for boys with Duchenne. That is precisely why the FDA cannot allow impressive protein numbers themselves to become the standard of proof.
The problem is larger than any one company. If microdystrophin expression can secure accelerated approval without convincing evidence that it predicts how patients actually function, every sponsor developing a similar therapy has an incentive to race toward the biomarker rather than wait for the harder clinical answer.
The standard for surrogate endpoints is not something the agency devised just for this case. Congress wrote it into the Federal Food, Drug, and Cosmetic Act and last refined it in the FDA Safety and Innovation Act of 2012, and that governing statute requires the surrogate to be “reasonably likely to predict clinical benefit.” For a Duchenne gene therapy, that surrogate is microdystrophin expression, essentially the amount of engineered protein the therapy manages to produce in muscle.
That is what REGENXBIO set out to establish, and on May 14, 2026, it reported the topline results meant to do so. In the pivotal portion of their Affinity Duchenne trial, the therapy cleared its primary endpoint decisively. Ninety-three percent of thirty evaluable boys reached at least 10% microdystrophin expression at Week 12, with mean expression of roughly 71% and a p-value below 0.0001.
REGENXBIO says Week 12 expression correlated with functional improvement a year later, making that relationship the foundation of its accelerated-approval case. But the functional analysis rests on just nine boys followed for twelve months and compares them with external natural-history data rather than a randomized control group.
The expression finding is broad. The functional finding is thin. That difference is where the statutory question begins.
The FDA is asking for evidence that expression predicts function rather than simply assuming that it does. Recent experience gives the agency good reason to demand it.
Pfizer’s CIFFREO trial was as rigorous as this field knows how to build: a Phase 3, randomized, double-blind, placebo-controlled study, and the therapy plainly delivered its protein. Mini-dystrophin-positive fibers averaged 34.8% at one year, and roughly 50.6% in the high-dose group.
If expression tracks benefit, as the science of the time reasonably assumed, the boys should have improved. But they didn’t.
The change in the North Star Ambulatory Assessment (NSAA), the standard measure of motor function in ambulant boys with Duchenne, was negligible. NSAA scores came in at 1.46 for treated patients against 1.37 for placebo, a gap of 0.09, and Pfizer shut the program down in 2024.
For anyone deciding whether microdystrophin can stand in for clinical benefit, CIFFREO is close to the whole argument. The protein was there in abundance, and the boys were no better for it. A biomarker that can run that far ahead of the patient is not yet a promise of anything, and it cannot be treated as one until its link to real-world function is shown rather than presumed.
This explains why the FDA’s posture should be demanding.
The agency’s framework has long held that a surrogate must be backed by extensive evidence before it can stand in for a clinical outcome. On the evidence to date, RGX-202 does not clearly meet that standard.
What raises the stakes here, compared with a conventional approval, is that the principal risk cannot be undone.
In practice, a boy who receives an inadequately validated therapy now may lose the option of a better one later, because that first infusion tends to close the door on re-dosing.
Where the consequences of a decision are effectively permanent, wanting the evidence behind it to be correspondingly solid is reasonable rather than overcautious.
The charge of inconsistency deserves a straight answer. Consistency does not mean reaching the same regulatory decision regardless of new evidence. It means applying the same principle to the evidence available at the time.
That principle remains unchanged. A surrogate must reliably predict clinical benefit before it can substitute for clinical benefit. Pfizer gave the FDA important evidence that microdystrophin expression alone may not do that.
The science moved. The standard did not.
Editor’s Note: Do you enjoy PJ Media’s conservative reporting exposing the radical left and woke media? Support our work so we can keep bringing you the truth.
Join PJ Media VIP and use promo code FIGHT to receive 60% off your membership.
