UniQure gene therapy slows Huntington’s less at four years than at three

“As pioneers in this space, we are blazing a new trail in HD, learning and deepening our understanding with every data point we generate,” the chief executive said.

The market did not wait. UniQure shares opened down sharply Tuesday morning, falling as much as 38 percent to $24.11 an hour after the open. Other desks put the slide near 40 percent. Premarket prints had already run as weak as the mid-50s percent range, with one report showing the stock off 45 percent at $22.00 and another noting a 55 percent premarket drop.

Investors had come looking for durability proof on AMT-130. Earlier data had drawn the crowd; the longer follow-up was the test they wanted answered. At 48 months, data cutoff June 30, 2026, twelve high-dose patients showed 44 percent slowing on the composite Unified Huntington’s Disease Rating Scale, or cUHDRS, versus updated matched ENROLL-HD external controls. The difference was not statistically significant (p=0.144). Three years earlier the same measure had shown 75 percent slowing. The primary composite had slowed less at four years than at three, and the gap no longer cleared statistical significance.

Huntington’s is a fatal inherited neurodegenerative disease of progressive psychiatric, cognitive, and motor decline. No approved therapy slows the underlying course. Existing drugs ease symptoms only. AMT-130 is uniQure’s one-time gene therapy, delivered directly to the brain by an adeno-associated virus vector to silence mutant huntingtin and limit the damage it does to neurons. The Phase I/II studies’ first two cohorts treated 29 patients in all—17 at the high dose, 12 at the low dose. In September 2025 an analysis of 12 high-dose patients at three years reported that 75 percent slowing on cUHDRS, a combined measure of movement, thinking, and daily function. That figure lit the stock and underwrote the regulatory push that followed.

On September 2 the company filed its biologics license application with the FDA and a parallel submission with the UK’s MHRA, requesting priority review. Those filings rested on the three-year package and did not include the longer look now being debated. Years of work, a treated cohort still counted in the dozens, and a design that matches patients against external controls drawn from the ENROLL-HD natural-history database now meet the four-year numbers.

On Total Functional Capacity, which tracks the ability to work, manage finances, and handle daily self-care, the slowing at 48 months was 61 percent (nominal p=0.008). The updated 36-month analysis in fifteen high-dose patients—the regulatory-anchor timepoint—showed 80 percent slowing on cUHDRS (nominal p=0.005) and 67 percent on TFC (nominal p=0.011). “At 48 months, we continue to see evidence of meaningful disease slowing,” Kapusta said on the call. Stifel analysts wrote that the four-year package “continues to clearly support a drug effect, though there is regression in efficacy on cUHDRS . . . whereas TFC data (61% slowing) hold up.” Chief Medical Officer Walid Abi-Saab framed TFC as the primary outcome planned for the confirmatory study and the measure most reflective of quality of life. The split left open how much of the thinner margin belonged to the therapy and how much to the controls against which it was measured.

Updated ENROLL-HD matched controls at 48 months had 53 percent missing data. UniQure said faster-progressing patients were more likely to discontinue, biasing the remaining controls toward healthier survivors and understating the treatment effect—what Walid Abi-Saab called an “underestimation of disease progression.” Kapusta’s caveat was direct: the missing control data “makes these statistical comparisons and percent slowing more challenging.” A post-hoc analysis using prior external control data showed 54 percent slowing on cUHDRS and 68 percent on TFC at the same horizon. Treated patients’ TFC score fell 0.37 points on average versus 0.94 points in controls. High-dose patients declined less than low-dose patients at 48 months, consistent with a dose-dependent effect.

Victor Sung, a neurologist at the University of Alabama at Birmingham who treats several of the participants, put the clinic picture plainly. “Overall, my patients are not progressing very much at all, which is not the typical thing.” He added, “Huntington’s disease does not slow on its own; the biology is one of inevitable progressive decline.” Leerink recognized the caveats and still wrote that slowing of any kind, especially on TFC that has remained stable, is absolutely unheard of.

Five high-dose participants, 17 percent, had treatment-related serious adverse events of central nervous system inflammation. All resolved. AMT-130 was otherwise generally well tolerated, and procedure-linked events resolved too. One low-dose patient died by suicide about five years after treatment, assessed as unrelated. Suicide risk is elevated in Huntington’s.

The regulatory path had already swung hard. In November of the prior year the FDA rejected the Phase I/II data as adequate primary support for a biologics license application. In March it strongly recommended a sham-surgery controlled trial, an ethical flashpoint for a delivery that takes roughly ten hours in the brain. Then in June 2026—after Marty Makary and Vinay Prasad had departed—the agency reversed course and agreed that three-year data from twelve high-dose patients could support accelerated approval. The BLA filed on September 2 rests on that 36-month package, not the four-year cut released Tuesday.

“We strongly believe in our data, and we strongly believe that what we've demonstrated in four years, quite frankly, is unprecedented...it really doesn't matter who's leading CBER, who's the commissioner of the FDA, who's on the review team,” Kapusta said. Sung put the stakes plainly: “Seeing that treatment difference maintained at four years is meaningful for people living with this relentlessly progressive degenerative disease.”

UniQure expects acceptance of the biologics license application in the fourth quarter. Kapusta said the confirmatory study will enroll about 200 patients, with screening aimed to start before year end. The trial will not include sham surgery. Paul Matteis of Stifel wrote that the four-year results are still impressive generally even if they represent some regression, while the fresh three-year results look robust and are “arguably most important, since this data cut is the crux” of the company’s alignment with the FDA. The big picture, he added, continues to support accelerated approval. Joseph P. Schwartz of Leerink judged that the full package lands nearer neutral than the market reaction implied. “At the end of the day, we think these data remain approvable,” he wrote. Making patients wait for a better-conducted Phase 3 trial or other options just seems inappropriate, he added, when rare disease communities are willing to take on a greater degree of risk in exchange for time and an option.

The dozen high-dose patients who have already reached four years keep walking the same scales the agency will re-read.