A one-time gene therapy for Huntington's is about to face its hardest test

A one-time gene therapy for Huntington's disease, a fatal inherited brain disorder with no approved treatment that slows it, is about to face its hardest test. Before the end of September, UniQure is expected to report four-year follow-up data from a 12-patient high-dose cohort of AMT-130, the company's AAV-based candidate delivered by a single brain injection. The readout will test whether the 75% slowing figure that electrified the field a year ago can survive a longer follow-up and a more skeptical read.

The 75% number is already public. Roughly twelve months ago, UniQure reported three-year results showing that high-dose AMT-130 patients progressed 75% slower on a standard clinical measure than matched participants drawn from an external natural-history study. The figure was the strongest disease-modifying signal the Huntington's field has reported, and it moved the stock on a small dataset that has no randomized placebo arm. The four-year update is the first chance to see whether the effect holds, plateaus, or quietly narrows.

That makes the comparator, not the molecule, the real story. AMT-130 is built on a familiar gene-therapy template: a harmless adeno-associated virus carries a small interfering RNA payload designed to lower the production of mutant huntingtin protein in the striatum, the brain region that degenerates first in Huntington's. The one-and-done dosing is delivered by a stereotactic brain injection, a neurosurgical procedure that threads a thin cannula to the target nuclei. The biology is the easy part to explain. The harder part is the benchmark.

A randomized placebo arm, the gold standard in neurodegenerative trials, was not part of the early AMT-130 program. Instead, treated patients are compared against a large external natural-history cohort, including TRACK-HD and ENROLL-HD, the long-running observational studies that track how the disease behaves in untreated carriers. The natural-history approach is the right call for a one-shot surgery with no sham-procedure placebo, but it is also the result's soft underbelly. Selection effects, including who agrees to brain surgery, who enrolls in a natural-history registry, and how missing data are handled, can move a 75% effect meaningfully. The 12-patient high-dose arm makes any single noisy trajectory count.

Three watch items will decide how the September readout lands. First, durability: do the year-four scores stay within the same slowed-trajectory band, or do they drift back toward the natural-history slope? Second, safety: a one-time brain injection that lowers a protein patients live with for decades is asking a long safety question, and any new serious adverse event would dominate the conversation. Third, biomarkers: neurofilament light chain, a blood marker of neuronal damage, has become a tie-breaker in prior Huntington's readouts. An NfL signal in the same direction as the clinical slowing would harden the case. An unfavorable one would reopen it.

If the year-four numbers look like the year-three numbers, the field will treat AMT-130 as the leading one-and-done candidate for Huntington's and start arguing about the regulatory path for a small, deeply characterized dataset. If they don't, the 75% headline will be remembered as a reminder of how far a striking early signal can travel without a randomized control underneath it. Either way, the question UniQure is actually asking the data is the same one regulators and patients have been asking for a decade: can a single injection into the brain durably slow a disease that no current drug can touch? The September readout will not end that argument, but it will move it from preview to evidence.