Pipeline

The First NPC Drug to Prove It Saves Lives Failed Its Own Randomized Trial

The FDA approved Lembaz on the strength of a 36-patient survival comparison against a matched external control, months after refusing two gene therapy programs built on the same kind of evidence. The difference is the endpoint.

A parent's hand holding a young child's hand on the rail of a pediatric hospital bed in soft daylight.

The first drug ever shown to keep patients with infantile-onset Niemann-Pick disease type C alive longer won U.S. approval on Nov. 17 without ever winning its own randomized trial. The FDA approved Lembaz (adrabetadex) to reduce the risk of death in adults and children weighing at least 7 kg whose neurological symptoms began before age 6.

The evidence the agency accepted

The approval does not rest on a positive controlled trial. The FDA examined survival in 36 patients treated with Lembaz across three clinical studies and an expanded access program, all with symptom onset before age 6. Treated patients lived longer than a matched external control group, with a 55% reduction in the risk of death. Median survival was 8.5 years versus 4.9 years, and the benefit held whether or not patients also took miglustat. Beren states the FDA determined that this comparison served as the adequate and well-controlled study supporting approval.

NPC is a rare inherited lysosomal storage disorder affecting an estimated 1 in 150,000 people. Cells cannot properly process cholesterol and other fats, which accumulate in the brain and drive progressive loss of movement, coordination and the ability to swallow. Two therapies were already approved, Miplyffa (arimoclomol) and Aqneursa (levacetylleucine), but neither has been shown to reduce the risk of death, and both carry age and weight limits that shut out the youngest patients.

Why the randomized trial failed

The molecule's pivotal test, NCT02534844, was a Phase 2b/3 randomized, double-blind, sham-controlled study of the same drug in 56 patients with neurologic NPC1. It began in 2015 and completed its primary phase in March 2018. The co-primary endpoints were change from baseline to Week 52 on a 4-item composite of the NPC Clinical Severity Scale, and the Clinician Global Impression of Change at Week 52. Posted results show mean change on the composite of -0.1 with sham and 0.0 with adrabetadex. On those endpoints, the drug did not separate from sham.

The program traces to a 2009 finding that cyclodextrin delayed disease onset in mice. Sucampo bought Vtesse in 2017 and Mallinckrodt bought Sucampo later that year; the Phase 2/3 trial then failed in 2018. Mandos, now a Beren subsidiary, picked up the asset in 2021 out of Mallinckrodt's bankruptcy. A February 2026 priority review set a decision date of Aug. 17, which slipped to Nov. 17 after a major amendment.

The line the FDA drew

Earlier in 2026 the FDA refused two other programs that leaned on external controls. On Feb. 7 it issued a complete response letter to Regenxbio for a gene therapy in mucopolysaccharidosis type II built on a single-arm study of 13 children with an external natural history control and a novel CSF biomarker endpoint, writing in the letter that it could not conclude the study was adequate and well-controlled. uniQure disclosed that at a Type A meeting the FDA said it could not agree that its Phase I/II data, compared with an external control, provided the primary evidence of effectiveness for AMT-130 in Huntington's disease, and recommended a randomized, sham surgery-controlled study instead.

The pattern turns on endpoints more than on controls. Death is unambiguous and cannot be argued with; a novel CSF biomarker and a functional rating scale can. A 55% mortality reduction against a matched external control carries a signal that no rating-scale delta from a failed sham-controlled trial can manufacture. That is a reading of three decisions rather than an FDA statement of policy. What it suggests is that an external control argument holds when the outcome is objective, the effect is large, and the comparison population is matched rather than cherry-picked.

What survival does not settle

The prescribing information carries a boxed warning for hearing impairment, including permanent hearing loss. Hearing must be evaluated before treatment and regularly during it, with dose adjustments based on results. Ataxia and fatigue can occur in the first days after a dose, and severe cases requiring hospitalization have occurred. In the trial, hearing impairment, ataxia, fatigue, vomiting, back pain and headache were each reported in 25% or more of patients. The drug is injected into the spinal canal by a trained provider every two weeks.

Those burdens land on families already caring for children with a disease that Beren says kills at a mean age of about 5.6 years when onset comes before age 2. Roughly half of U.S. patients, about 475 children, have the infantile-onset form. Beren says Lembaz will reach commercial sites nationwide in early December, a real shift from the age and weight limits that kept the earlier drugs out of reach.

By Beren's own analysis, five-year overall survival was 84% in the treated population versus 42% in controls. A survival curve does not answer what families will still ask. How much hearing does a child lose over years of biweekly injections? The approval validates a longer life, not a lighter burden. Beren's investigators put the case for earlier treatment plainly: function that is lost is not regained. For the next rare disease program weighing an external control, the lesson is in what it chooses to measure.