The FDA Approved a Pathway for Huntington's Gene Therapy. Then Reversed It. Then Reversed the Reversal. 9,000 Patient-Years of Function Were Lost in Between.
AMT-130 slowed Huntington's disease progression by 75% in 29 patients. An 18-month regulatory zigzag cost an estimated 10,000 early-manifest patients 0.9 points each on a 13-point functional scale that only moves in one direction.
On June 17, uniQure's stock price rose 78% in a single session after the Amsterdam-based biotech announced that the U.S. Food and Drug Administration had reversed itself for the second time in 18 months, agreeing once again that existing clinical trial data from 29 patients could support a marketing application for AMT-130, a one-time gene therapy for Huntington's disease. No new Phase 3 trial required, with filing expected in the third quarter of 2026.
Markets celebrated, but they should not have needed to. A reversal this welcome implies a preceding error this costly, and the approximately 30,000 Americans living with Huntington's disease, a neurodegenerative condition for which no approved treatment slows progression, bore that cost in a currency they cannot earn back: functional capacity measured on a scale that only ticks downward. We ran the numbers on what the regulatory zigzag took from them.
A Drug That Actually Works
AMT-130 is a single-administration gene therapy delivered directly into the striatum via MRI-guided stereotactic neurosurgery, using an AAV5 viral vector to deliver a microRNA that silences the mutant huntingtin gene at the root molecular cause of Huntington's disease. In the completed Phase I/II trial, 29 patients received treatment across two dose levels in U.S. and European studies, and 12 patients per dose group reached the 36-month evaluation point.
High-dose patients showed a 75% reduction in disease progression compared to a propensity-matched external control drawn from the Enroll-HD natural history dataset of 940 patients, measured on the composite Unified Huntington's Disease Rating Scale (cUHDRS). Both primary and key secondary endpoints were met, including statistically significant results on Total Functional Capacity at 36 months. Cerebrospinal fluid neurofilament light chain, a biomarker of active neurodegeneration, dropped 8.2% from baseline in treated patients while untreated patients in natural history studies showed a 26% increase over the same period, a directional divergence of 34 percentage points that external confounding cannot plausibly explain.
"These are the most convincing data in the field to date," said Sarah Tabrizi, professor of clinical neurology and director of the University College London Huntington's Disease Centre, in a statement accompanying the data.
For context, every previous Huntington's therapeutic candidate has failed at scale. Roche's tominersen was terminated in 2021 after worsening outcomes in a 791-patient Phase 3 trial, and no other investigational compound has demonstrated statistically significant slowing of disease progression in a clinical setting. Twenty-nine patients, one injection each, 75% slower decline over three years stands as the first clinical evidence that this disease's trajectory can be meaningfully altered.
A Timeline of Reversals
Regulatory history for AMT-130 deserves to be laid out precisely, because the sequence itself constitutes the story.
May 2024: FDA grants AMT-130 Regenerative Medicine Advanced Therapy (RMAT) designation, a status reserved for therapies that show "preliminary clinical evidence" of the potential to address unmet medical needs for serious conditions.
December 2024: After reviewing interim and emerging data, the FDA and uniQure formally align on an accelerated approval pathway, with the agency concurring that Phase I/II data combined with the Enroll-HD natural history comparator can serve as the primary basis for a Biologics License Application requiring no additional presubmission study.
Late 2025: Reversal arrives when agency staff declare available data insufficient, with one senior regulator describing AMT-130 as a "failed product," according to Reuters. A new demand emerges: run a randomized, sham-controlled Phase 3 trial before any filing can proceed.
Early 2026: Full scope of that demand becomes clear when the required Phase 3 design is outlined, calling for patients randomized to receive either AMT-130 or a sham surgical procedure involving general anesthesia, burr holes drilled through the skull, placement in a stereotactic frame under MRI guidance, and closure without delivery of any therapy.
June 17, 2026: Roughly one month after the departure of FDA Commissioner Marty Makary and vaccines chief Vinay Prasad, the agency reverses course again, accepting three-year Phase I/II data under accelerated approval, with the confirmatory study permitted to use patients on standard treatment as comparators instead of a sham procedure.
What 18 Months Cost: A Calculation Nobody Ran
Huntington's disease progression is mercilessly quantifiable through the Total Functional Capacity scale, which runs from 13 (normal function) to 0 (total dependence) and moves in one direction only. According to a prospective longitudinal study published in JAMA Neurology tracking 334 patients across 44 research sites, early-manifest patients lose an average of 0.6 TFC points per year (95% CI, 0.5-0.7), with decline that is monotonic and linear in early disease, neither plateauing nor reversing at any point before the scale bottoms out.
Each TFC point represents a concrete functional milestone: the difference between a score of 11 and 10 might mean losing the ability to manage personal finances independently, between 8 and 7 the ability to perform household chores, and between 5 and 4 the ability to dress without assistance. Not abstract clinical endpoints, but a parent who can no longer button their child's shirt.
Had uniQure filed its application on the originally agreed-upon timeline in early 2025 and the FDA reviewed it under the standard 8-month priority review clock, accelerated approval could have arrived by late 2025 or early 2026. Instead, the 18-month delay pushed filing to Q3 2026, with approval now unlikely before Q1 or Q2 2027.
| Variable | Value | Source |
|---|---|---|
| Americans with Huntington's disease | ~30,000 | Huntington's Disease Society of America |
| Early-manifest patients (AMT-130 eligible estimate) | ~10,000 | Conservative estimate from Enroll-HD enrollment data |
| Annual TFC decline rate | 0.6 points/year | Ross et al., JAMA Neurology (n=334) |
| Net regulatory delay | ~18 months | Dec 2024 alignment vs. Q3 2026 refiling |
| TFC loss per patient during delay | ~0.9 points | 0.6 × 1.5 years |
| TFC scale total range | 13 points | UHDRS standard |
| Functional capacity lost per patient | ~7% of total scale | 0.9 / 13 |
| Population-level TFC loss | ~9,000 points | 10,000 patients × 0.9 points |
Nine thousand TFC points of irreversible functional decline across the eligible U.S. population, accumulated during a period when the agency was deciding whether to accept data it had already agreed to accept. Some of those patients crossed clinical thresholds that cannot be uncrossed, progressing to disease stages where their striatal volume, the target tissue for AMT-130 delivery, has atrophied to a point where the therapy may be less effective or the neurosurgical procedure more difficult to execute precisely. Disease kept running while paperwork went in circles.
Why Sham Brain Surgery Was Ever on the Table
Mid-course demand for a sham-controlled Phase 3 deserves separate scrutiny because it reveals a deeper tension in how regulators evaluate therapies delivered by brain surgery, a category that did not exist when current trial design frameworks were established for conventional pharmaceuticals.
Stereotactic neurosurgery for AMT-130 involves general anesthesia, placement in a rigid stereotactic head frame, MRI-guided targeting, and bilateral administration via convection-enhanced delivery into the caudate and putamen through small burr holes, carrying published complication rates between 2% and 5% for hemorrhage, infection, seizure, and transient neurological deficits. Demanding a sham arm means proposing to drill into the skulls of 25 to 50 patients, subject them to all the surgical risks, and give them nothing in return, in a disease for which no alternative disease-modifying treatment exists and the natural trajectory is total functional collapse over 15 to 20 years.
Patient advocacy organizations, clinical ethicists, and several investigators involved in AMT-130 research objected publicly before the agency ultimately withdrew the demand as part of its June reversal, permitting the confirmatory study to use standard-of-care comparators instead. But the fact that sham brain surgery was ever on the table for a rare neurodegenerative disease with 36 months of statistically significant efficacy data illuminates how poorly regulatory frameworks for surgical gene therapies map onto conventional drug development paradigms built around pills and injections where placebo risk is trivial.
Bigger Than One Drug
uniQure is not an isolated case, because gene therapy regulation has become a landscape of shifting goalposts where consequences compound on a population that cannot absorb delay.
More than 30 FDA-approved gene therapy products now exist, with price tags reflecting the impossible economics of rare disease: Zolgensma at $2.125 million per dose for spinal muscular atrophy, Casgevy at $2.2 million for sickle cell disease, Hemgenix at $3.5 million for hemophilia B. Bluebird bio, which brought Lyfgenia to market at $3.1 million for sickle cell disease, nearly went bankrupt despite approval because the commercial infrastructure for million-dollar one-time therapies barely exists, and every month of regulatory delay adds $30 to $80 million in operational burn that must be amortized across a patient population that often numbers in the low thousands.
When goalposts move, capital evaporates, investors take note, and the next rare disease gene therapy startup faces a harder fundraising environment because uniQure's story became a cautionary tale about regulatory unpredictability rather than clinical science. FDA itself appears to recognize this problem, having introduced Rare Disease Evidence Principles in September 2025, a Plausible Mechanism Framework for individualized therapies in February 2026, and new draft guidance on leveraging prior knowledge in June 2026. Genuine, substantive reforms that represent exactly the kind of regulatory modernization gene therapy requires, but they arrived during the same 18 months that the agency was jerking uniQure between approval and rejection of the exact same dataset, a contradiction that no new guidance document can paper over.
Limitations
Our population-level TFC calculation makes several assumptions that deserve explicit challenge, and the strongest of these concerns the gap between theoretical eligibility and real-world access. Not all 10,000 estimated early-manifest patients would qualify for or receive AMT-130 immediately upon approval because real-world rollout of surgical gene therapies is constrained by the number of qualified neurosurgical centers, manufacturing capacity, and insurance coverage battles that routinely delay treatment by months or years. Actual patients who would have been treated during the 18-month delay window likely number in the hundreds, not thousands, making the 9,000-point figure a measure of population exposure to irreversible decline rather than a count of directly prevented treatments.
Our 0.6 TFC points/year decline rate is a population mean from Ross et al., and individual decline rates vary substantially depending on CAG repeat length, age of onset, and other genetic and environmental modifiers. Counterfactual timing assumes the FDA would have completed its review within the standard priority review clock, but PDUFA timelines are targets rather than guarantees, and biologics reviews sometimes run long. And because AMT-130 slows but does not halt progression, the 0.9 TFC points lost are not all recoverable even with eventual treatment; our calculation represents irreversible background loss during a delay, not a benefit that would have been fully captured had approval arrived on time.
Strongest Counterargument
Defenders of the FDA's caution will argue, correctly, that accelerated approval based on 29 treated patients carries genuine risk because the history of medicine includes therapies that looked promising in small trials and caused harm at scale. Aducanumab, approved for Alzheimer's under accelerated approval in 2021 with contested data, became a case study in premature regulatory action when confirmatory evidence remained ambiguous and Biogen ultimately withdrew the drug from market. An agency that insists on stronger evidence before exposing a wider population to a therapy delivered via brain surgery is operating within a defensible interpretation of its statutory mandate to protect public health.
Stronger still is the specific evidentiary concern that AMT-130's primary comparison was against an external natural history cohort rather than a randomized concurrent control, because external comparators introduce confounding variables that randomization eliminates, propensity matching cannot account for unmeasured confounders, and a regulator who insists on demonstrating efficacy against that methodological gold standard is not being irrational but applying the evidentiary discipline that has prevented countless iatrogenic disasters across the history of pharmaceutical regulation.
Rebuttal runs equally specific to the disease in question. Huntington's is genetically determined by CAG trinucleotide repeat expansion in the HTT gene, and its natural history is among the most thoroughly characterized of any neurodegenerative condition, with Enroll-HD tracking tens of thousands of participants across decades to establish the most robust rare-disease natural history dataset in existence. External control here was not a convenience sample but a rigorously propensity-matched cohort of 940 patients drawn from that gold-standard registry, and biomarker data closes the confounding gap decisively: cerebrospinal fluid NfL dropped 8.2% in treated patients against a background trajectory of 26% increase in matched controls, a 34-percentage-point directional divergence in a well-validated neurodegeneration biomarker that no plausible unmeasured confounder can explain. When evidence this clean meets a disease this well-characterized and this uniformly fatal, the risk calculus must shift, because holding approval hostage to a sham-controlled surgical trial does not just delay treatment but delays it while proposing to expose placebo patients to brain surgery for a disease with no alternative recourse.
What To Do About It
If you carry the Huntington's gene or have a family member who does, monitor uniQure's Q3 2026 filing and identify the neurosurgical centers likely to be authorized for AMT-130 delivery, positioning yourself to seek treatment as soon as approval arrives rather than entering a queue after the fact. Huntington's Disease Society of America maintains a directory of HDSA Centers of Excellence with neurosurgery capabilities, and early contact with these centers will matter because surgical gene therapy capacity, not drug supply, will be the binding constraint on initial access. If you are a biotech investor evaluating gene therapy companies, this case is a $400 million object lesson in how regulatory inconsistency rather than clinical failure can destroy value and timeline, so build that risk into your models explicitly with scenario analysis that prices the probability of mid-process pathway changes rather than treating FDA alignment as a fixed input. If you are a legislator or FDA official, this case demonstrates that writing new frameworks such as Rare Disease Evidence Principles while simultaneously reversing agreements made under old ones is worse than having no framework at all, because it teaches companies that no agreement with the agency is binding across a change in leadership, and the only durable fix is legislative codification of pathway commitments that survive personnel turnover at the agency that made them.