Ultragenyx UX111: A $2.65B Gene Therapy Bet on Sanfilippo

Sanfilippo syndrome Type A is the kind of disease that ends childhood. Children born with biallelic SGSH mutations cannot break down heparan sulfate, a glycosaminoglycan that accumulates in the brain and drives relentless neurodegeneration. The median life expectancy is 15 years. There is no approved treatment. On September 19, 2026, the FDA will decide whether Ultragenyx’s UX111 becomes the first.

UX111 (rebisufligene etisparvovec) is a one-time IV infusion of an AAV9 vector carrying a functional SGSH gene. The PDUFA date is the regulatory deadline every gene therapy investor circles, and this one carries more weight than most: it is a resubmitted BLA on the accelerated approval pathway, built on a surrogate biomarker (cerebrospinal fluid heparan sulfate, or CSF-HS) that the FDA already accepted as reasonably likely to predict clinical benefit. That acceptance is the fulcrum of the entire investment thesis.

The Trial Data: A Biomarker Story With Functional Backup

The registrational data comes from the Transpher A study (NCT02716246), which enrolled 28 patients across three dose cohorts at five sites in three countries. The high-dose cohort (3x10^13 vg/kg) contains 22 patients, with 17 in the modified intent-to-treat group. A separate study (NCT04088734) treated five additional patients at the same dose.

The biomarker results are unambiguous. CSF-HS dropped within the first month of treatment. As of a September 2025 data cutoff, the median reduction in CSF-HS exposure was 63.98% (p<0.001). The majority of children, 88.2% of younger patients and 81.5% of the overall efficacy set, achieved a 50% or greater reduction.

The functional data, while not the approval basis, is what makes this more than a biomarker bet. In children under two years old or with earlier-stage disease at treatment (n=17), the Bayley-III cognitive raw score showed a +23.2 point treatment effect versus natural history over 24 to 60 months of age (p<0.0001). Receptive communication improved 8.1 points (p=0.0076). Expressive communication improved 11.1 points (p=0.0008). Fine motor improved 9.0 points (p=0.0026). Eight children reached a 36-month cognitive developmental age that enabled higher-level testing. None of the natural history patients reached that milestone.

These data hold up over time. The February 2026 Ultragenyx presentation reported up to 8.5 years of follow-up. In older patients with more advanced disease at treatment (n=10), all retained some form of communication at a median age of 9.7 years, versus a natural history median communication loss at 7.6 years. Nine of ten retained independent ambulation at a median age of 9.05 years, versus a natural history median loss at 11.3 years. This is not a cure. It is meaningful disease modification in a uniformly fatal condition.

The safety profile is favorable for a systemic AAV9 gene therapy. The most frequent treatment-emergent adverse events were liver enzyme elevations. Treatment-related adverse events were mostly mild or moderate and resolved spontaneously. No patient has dropped out of the study for safety reasons across a median follow-up of 4.8 years. The full trial details are available on ClinicalTrials.gov.

The Valuation Problem

Here is where I get skeptical. Ultragenyx closed at $26.90 on July 24, 2026, with a market cap of roughly $2.65 billion. That is a 19% gain year-to-date, up from $22.63 on January 20. The stock ran into the low $24s around the February 3 BLA resubmission and has ground higher since.

The company is not pre-revenue. Ultragenyx reported $673 million in 2025 total revenue: Crysvita brought in $481 million, Dojolvi $96 million, Mepsevii $37 million. The 2026 guidance is $730 to $760 million, representing 8 to 13% growth. That is a real, commercial rare disease franchise.

But UX111 is being valued as if it is already approved and launching successfully. Here is the comp that makes the valuation hard to defend.

Sarepta Therapeutics, the gene therapy commercial leader, closed at $15.68 with a market cap of $1.66 billion. Sarepta reported $2.198 billion in 2025 total revenue, with Elevidys alone generating $898.7 million. Sarepta has multiple approved DMD products, a commercial infrastructure for gene therapy delivery, and a Roche partnership for ex-US Elevidys distribution.

Ultragenyx is worth $2.65 billion with $673 million in revenue and one gene therapy BLA under review. Sarepta is worth $1.66 billion with $2.2 billion in revenue, an approved gene therapy on the market, and a deep pipeline. RARE is trading at 3.9x trailing revenue. SRPT is trading at 0.75x trailing revenue. Either Sarepta is dramatically undervalued, or Ultragenyx is pricing in a UX111 approval that has not happened yet, plus a launch that has not started, plus peak sales that may take five years to reach.

The addressable market is the constraint. Ultragenyx estimates 3,000 to 5,000 patients in commercially accessible geographies. This is an ultra-rare disease. Even at a $2 million price tag, which is the going rate for AAV gene therapies in fatal pediatric conditions, peak revenue is a function of how many patients can be diagnosed, referred, treated, and reimbursed. If you assume 1,000 patients treated per year at $2 million, that is $2 billion in peak revenue. At a 4x peak-sales multiple, discounted back, you get something close to the current market cap. But that assumes the launch runs smoothly, manufacturing scales, and payers do not push back. The Dynexan ziresanin BLA analysis we published covers the same risk in a larger indication: even in DMD, where Sarepta proved the model, gene therapy launches are slow and reimbursement is contentious.

The Manufacturing and Competitive Picture

Ultragenyx is not outsourcing this launch. Manufacturing will be entirely within the US, split between Andelyn Biosciences in Ohio and an Ultragenyx facility in Massachusetts. That is a deliberate choice. The gene therapy field has been burned by CDMO capacity constraints, and Ultragenyx is paying a premium for vertical integration. It also means higher fixed costs before UX111 revenue materializes.

The competitive picture is thin, which is both a strength and a risk. There is no approved therapy for MPS IIIA. Esteve is developing EGT-101, a competing gene therapy, but it is years behind. The real competitor is the natural history of the disease itself: children deteriorate, families wait, and the diagnostic delay means many patients are too advanced at treatment to benefit. Ultragenyx’s data shows the treatment effect is concentrated in younger, earlier-stage patients. That is the commercial bottleneck. If newborn screening for MPS IIIA is not widespread, the addressable population shrinks to the patients who happen to be diagnosed early through symptomatic workup.

This is the same structural problem that has limited every ultra-rare gene therapy launch. Vertex priced Casgevy at $2.2 million per patient for sickle cell, the most expensive drug in history, and treated fewer than 100 patients in the first year of launch. The CRISPR Therapeutics CAR-T PDUFA analysis covers the same dynamic: ultra-rare gene therapies do not scale like small molecules, and the market cap has to account for that.

What Could Go Wrong

The accelerated approval pathway is a double-edged sword. The FDA accepted CSF-HS as a surrogate endpoint, which is why this BLA is approvable on biomarker data without a completed confirmatory trial. But accelerated approval comes with a post-marketing requirement. Ultragenyx will need to generate confirmatory clinical data after approval to maintain the label. If the confirmatory trial shows a smaller functional benefit than the natural history comparison suggests, the FDA can withdraw approval. This is what happened with Sarepta’s original eteplirsen, and it is the existential risk of every accelerated approval in rare disease.

There is also a manufacturing risk specific to AAV9. Systemic AAV9 gene therapies have been associated with serious adverse events in other programs, including thrombotic microangiopathy and complement activation. UX111’s safety profile is clean so far, but the patient population is small. The Transpher A study treated 28 patients. A post-marketing population of 500 patients will surface adverse events that a 28-patient trial cannot detect. The FDA’s Risk Evaluation and Mitigation Strategy requirements, if any, will determine how quickly this launches.

The specific risk I am watching: diagnostic infrastructure. If newborn screening for MPS IIIA is not adopted, Ultragenyx will be treating symptomatic patients who are already past the window of maximum benefit. The trial data is clearest in children under two. The commercial reality is that most patients are diagnosed between two and four, after developmental delay is already apparent. That gap is the difference between the +23.2 point cognitive benefit and a much smaller effect in real-world practice.

The Verdict

At $2.65 billion, Ultragenyx is pricing in UX111 approval plus a successful launch plus peak sales of roughly $1.5 to $2 billion. The biomarker data supports approval. The functional data supports clinical benefit. The valuation does not support a margin of safety.

I would not buy RARE here. The catalyst is real and the data is strong, but the stock has already moved 19% year-to-date on the approval expectation. The comp is damning: Sarepta, with $2.2 billion in revenue and a proven gene therapy platform, trades at 40% of Ultragenyx’s market cap. Either SRPT is a generational buy, or RARE is 30 to 40% overvalued at current levels.

The trade I would consider: buy the September 19 PDUFA outcome, not the stock. If UX111 is approved, the stock may sell off on the news because the approval is already priced in. If UX111 receives a Complete Response Letter, the stock drops 40 to 50%. The asymmetry favors waiting. Ultragenyx is a quality company with a real franchise and a gene therapy that children with Sanfilippo desperately need. At $26.90, it is not a quality stock. The PDUFA date primer explains why these binary events are for trading, not for buying and holding through. Ultragenyx is no exception.