The Drug Hit Its Target. The Patients Got Worse. What That Quells for Neuroscience.

In 2004, researchers pinned a single gene to a rare form of Parkinson's disease. In 2018, a group of scientists made the case that blocking the protein product of that gene could help every Parkinson's patient, not just the genetic outliers. On Thursday, Biogen and Denali Therapeutics reported the results of the largest test of that theory ever run: 648 patients with early-stage Parkinson's, treated for up to three years, and the drug did not slow the disease, according to a GlobeNewsWire press release.

The drug was BIIB122, a small molecule designed to inhibit LRRK2 — a protein kinase whose gene mutations are the most common known genetic cause of Parkinson's disease, accounting for 4-5% of familial and 1-2% of sporadic cases. It hit its target. More precisely, it demolished it. The study showed greater than 90 percent inhibition of LRRK2 in the blood and roughly 30 percent reduction of a downstream biomarker in cerebrospinal fluid. By every pharmacologic metric, the target was engaged. The patients, by every clinical measure, declined identically to those on placebo, MarketWatch reported.

The companies discontinued development for idiopathic Parkinson's disease, the broad form of the condition that accounts for the overwhelming majority of cases. Denali is continuing a separate study in patients who carry pathogenic LRRK2 mutations, but the broad indication is dead, according to the GlobeNewsWire release.

One open question the data cannot yet answer: whether the gap between 90 percent peripheral inhibition and 30 percent CSF reduction reflects a true mechanism failure — the right target, the wrong disease — or a formulation problem, with insufficient drug reaching the brain to properly test the hypothesis. BEACON, using a different Denali formulation, is designed to help close that gap. Other companies with LRRK2 programs, including Neuron23, now face the same investor re-evaluation of the genetic-to-idiopathic expansion strategy that Thursday's results have forced on Biogen and Denali.

This is the part of drug development that doesn't make it into press releases. The story isn't that the drug failed. The story is how completely it succeeded at the molecular level while failing at every level that matters to a patient.

Biology keeps doing this. Notoriously, not uniquely. The amyloid hypothesis in Alzheimer's consumed three decades and billions of dollars before lecanemab demonstrated marginal slowing of decline in 2023, an outcome so narrow it is still debated whether it matters clinically. BACE inhibitors erased themselves from the pipeline after showing beautiful target engagement and zero cognitive benefit. Gene therapy in the 1990s was going to cure everything; it cured almost nothing, and the field spent a decade rebuilding.

What each of these episodes shares is the same underlying error: mistaking biological plausibility for clinical proof. LRRK2 inhibition made perfect sense. The mutation caused disease in a small number of families. The protein was clearly involved in cellular cleanup mechanisms known to be impaired in Parkinson's. Blocking it reduced the relevant proteins in the blood. The chain of logic was unbroken.

The chain of evidence was not.

A target being the right target for a rare genetic subgroup does not make it the right target for the broader, far more heterogeneous condition that shares the same clinical name. Parkinson's disease affects roughly one million Americans and more than ten million people worldwide, according to the companies' release. Only a small fraction carry LRRK2 mutations. Whether the mechanism that kills those neurons in a mutation carrier is the same mechanism that kills them in a person with no known genetic driver is a question that 648 patients just answered, at enormous cost in time and hope.

Denali's chief medical officer called the LUMA study a robust test of the hypothesis. He meant that as a comfort to participants. It is also a verdict.

The BEACON trial continues, in a narrower population where the genetic logic is tighter. Data is expected in the first half of 2027, according to the release. That study will answer a more honest question: does LRRK2 inhibition work when the patient actually has too much LRRK2 activity? The answer to that question may be yes.

But for the millions of people with Parkinson's who carry no such mutation, Thursday's results close a door that has been open, in various forms, for two decades. The hope that blocking a protein would slow a disease was reasonable. It was also, so far as anyone can tell today, wrong.

That is how science works. It is also, reliably, how it hurts.