What cleared with Fayuvi: Sanfilippo A’s first gene-therapy approval

In Short. On September 17, 2026, FDA approved Fayuvi (rebisufligene etisparvovec-hopf), also known as UX111, for neurologic manifestations of pediatric mucopolysaccharidosis type IIIA in children with preserved neurodevelopmental function. Ultragenyx’s Form 8-K calls that decision a standard full approval and notes a Priority Review Voucher. A first-in-disease gene-therapy approval is an evidence-package claim: disease, product, approval type, and which public materials support cognition versus historical controls, CSF heparan sulfate as supportive biomarker language, and a chemistry-manufacturing complete-response path that had to clear before the calendar could finish.

---

Mucopolysaccharidosis type IIIA, Sanfilippo syndrome type A, is a rare inherited lysosomal storage disease of childhood. Children lack enough sulfamidase, the enzyme made by the SGSH gene, so heparan sulfate builds up and the brain’s developmental trajectory turns from early progress into plateau and loss. Until this week, U.S. care was supportive. FDA’s September 17 press announcement states that there had been no approved therapy designed to change the underlying course of the disease.

Fayuvi is a one-time intravenous AAV9 gene therapy that delivers a working SGSH copy so cells can make sulfamidase and break down heparan sulfate. Ultragenyx markets it; the product was developed earlier as ABO-102 / UX111. The company’s September 17, 2026 Form 8-K (accession 0001515673-26-000006) says FDA granted standard full approval of FAYUVI for pediatric patients with MPS IIIA, calls it the first FDA-approved Sanfilippo A treatment, and records a Priority Review Voucher. That filing is the clean primary for approval type on the public company side.

Indication language on the public science face is narrower than a ticker headline. Company materials and FDA’s announcement frame use for neurologic manifestations of MPS IIIA in pediatric patients with preserved neurodevelopmental function. That second clause is part of the package. The therapy is aimed at children who still have neurodevelopmental function to preserve or improve, not at every stage of a relentless storage disease. The full FDA approval letter PDF (STN BL 125845/0 on fda.gov/media/194921) returned HTTP 401 from this environment, so letter-paragraph quotation stays limited; openFDA also returned no Fayuvi label row yet. Indication wording above follows the company’s post-approval materials and FDA’s public announcement, and waits on a readable letter or SPL label for finer legal text.

What FDA says publicly about effectiveness is a cognitive comparison to history, not a press-release scoreboard. The Agency’s September 17 announcement describes an open-label, single-arm, multicenter study in pediatric MPS IIIA that measured mean changes in cognitive scores in patients between ages 2 and 5. Fayuvi-treated patients maintained or improved cognitive function compared with an untreated historical control cohort, which FDA calls a meaningful divergence from the expected natural course of plateau and decline in that window. Megha Kaushal, acting deputy director of the Office of Therapeutic Products, framed the result as evidence that systemic AAV9 delivery can reach the central nervous system at therapeutically relevant levels in these children. That is the primary public efficacy verb available without opening the closed review memo: cognition versus natural history in a defined age band.

Secondary coverage, including Fierce Pharma as relayed in the September 18 Editor’s Briefing, quotes finer Bayley subgroup arithmetic (including a reported 23.2-point cognitive treatment effect in a younger or earlier-stage cohort of 17, and retention claims in an older cohort of 10, plus a median CSF heparan sulfate reduction near 64%). Those numbers are useful as reporting to chase. They are not treated here as primary until they appear in a label, approval letter, or FDA review document that this draft could open.

Biomarker language has its own lane. On January 30, 2026, Ultragenyx’s Form 8-K said the company had resubmitted the BLA seeking accelerated approval, with longer-term neurologic-benefit measures intended as an intermediate clinical endpoint and CSF heparan sulfate plus other biomarkers as further support, “as agreed with the FDA during the last clinical review.” Live ClinicalTrials.gov still lists NCT02716246, a Phase 2/3 UX111 study in MPS IIIA, with a primary of CSF heparan sulfate (disaccharide) exposure through Month 24. The approval letter’s public indexes, as summarized in FDA media metadata, associate the review with NCT02716246, NCT04360265 (long-term follow-up), and NCT04088734 (terminated middle/advanced-phase ABO-102 study). Registry primaries explain what the program measured. They do not, by themselves, equal the Agency’s final evidentiary weighting inside a full approval. The teachable split is simple. Company resubmission language put neurologic benefit at the center of an accelerated ask and called CSF heparan sulfate supportive. FDA’s public approval announcement leads with cognitive maintenance or improvement versus historical controls. Ultragenyx’s approval 8-K then names the outcome standard full approval. Readers who collapse those three sentences into “the biomarker approved the drug” lose the package.

Manufacturing history is the third pile, and it is documented on the CRL path. On July 11, 2025, Ultragenyx reported a Complete Response Letter. The July 11 Form 8-K says FDA asked for additional CMC information and improvements tied to observations from manufacturing facility inspections. The company characterized those observations as related to facilities and processes rather than product quality. The same filing says clinical review had been ongoing, that FDA had acknowledged neurodevelopmental outcome data as robust and biomarker data as additional supportive evidence, that the CRL did not note review issues related to the clinical data package or clinical inspections, and that updated clinical data from current patients should be included in resubmission. The January 30, 2026 resubmission 8-K says the new BLA answered those CMC observations and added longer-term clinical data the Agency had requested. April 2, 2026’s acceptance 8-K set the September 19, 2026 PDUFA date that the weekday calendar piece tracked. Approval landed on September 17, ahead of that posted date. The durable lesson is sequencing. A clinical package can look robust to the Agency while a CRL still stops the clock on facilities; clearing CMC is part of what “cleared” means for a gene-therapy BLA even when the public celebration is about cognition and biomarkers.

Safety belongs in the same primary announcement. FDA lists common adverse reactions (more than 5%) as increases in AST, nausea and vomiting, fever, decreased appetite, decreased white blood cell and platelet counts, and increased amylase. Important warnings include thrombotic microangiopathy and the class concern that inserted genetic material could integrate and, over the long term, contribute to tumor risk. Patients receive corticosteroids from the day before infusion through at least eight weeks afterward, and the product is given in a setting equipped for infusion reactions. Those facts are part of the cleared package, not footnotes under a first-in-disease headline.

What still waits is honesty about closed paper. The approval letter PDF and the SPL label were not readable here (FDA media 401; openFDA empty), though public indexes and secondary mirrors give the indication’s neurologic-manifestations / preserved-neurodevelopmental-function frame and the associated NCT numbers. Confirmatory-trial language, if any appears on the letter for a full approval, is therefore unverified in this draft. Exact Bayley arithmetic in secondary press stays secondary. Plant-level inspection findings beyond the company’s CRL summary are not invented. Anyone reading the next first-in-disease gene-therapy approval can ask the same questions. What is the approval type. What clinical boundary sits in the indication. How does cognition-versus-history separate from biomarker support. Where does the CMC path sit on the timeline. A calendar sticker or a nested NCT cannot do that work for the package.

---