A drug slowed Alzheimer's by hitting the protein everyone ignored — and still "failed" its own test

For thirty years, almost every Alzheimer’s drug has chased the same target: amyloid, the sticky protein that clumps into plaques in the brain. This week, at the Alzheimer’s Association International Conference in London, a drug that goes after the other protein — tau — showed it can work [1].

The result

The drug is called diranersen, made by Biogen with its partner Ionis [1]. It is an antisense oligonucleotide — a short strip of genetic material that blocks a cell’s instructions to build tau, so the brain makes less of it [8]. Tau is the second protein that deforms and tangles inside neurons in Alzheimer’s; amyloid is the first [8].

In a mid-stage (phase 2) trial called Celia, diranersen lowered tau in the brain and slowed patients’ cognitive decline [8]. It is the first tau-targeting drug ever to do both [4]. And it slowed decline at rates roughly on par with the approved anti-amyloid drugs — Biogen’s own Leqembi and Eli Lilly’s Kisunla — but with two advantages: it is dosed just twice a year, and it carries none of the dangerous brain swelling and bleeding those drugs can cause [4].

“It is clear that modulating tau has a positive benefit on patients,” said Holly Kordasiewicz of Ionis [8].

Why it still counts as a miss

Here is the strange part. The trial failed its primary endpoint — the single result it was designed to prove [4].

The design assumed a simple rule: more drug, more effect. But the strongest slowing of decline came in the 60 patients on the lowest dose, 60 milligrams infused into the spinal fluid every six months [8]. Because the benefit did not climb with the dose, the pre-set measure of success — built around that climb — recorded a failure [4]. The drug also did not beat placebo on a separate measure of daily functioning, a result one Biogen scientist called “puzzling” [8].

Investors focused on the miss. Biogen’s stock slid on the day [4]. The company and its outside experts read the data the other way: a proof of concept solid enough to race into a large phase 3 trial, the final test before any bid for approval [4]. The data are the first to “show a tau-targeting drug producing both a robust biomarker effect and a signal of clinical benefit,” said Laura Nisenbaum of the Alzheimer’s Drug Discovery Foundation [4].

The caveats are real. This is one mid-stage trial. The best result rested on 60 people. The lack of a dose response raises hard questions for how the phase 3 is built [4]. Nothing here is a cure, and nothing is approved. But for the first time, the neglected protein has a drug that seems to touch it — and the hope now is to combine amyloid and tau drugs, hitting both culprits at once [4].

The graveyard it climbed out of

The tau result lands against a backdrop of expensive failure. Also this week, GSK formally walked away from a neurodegeneration partnership with the biotech Alector that could have been worth $2.2 billion, writing off a $700 million upfront payment from 2021 [28]. Both drugs in the deal flopped — one failed a late-stage trial in a form of dementia last year, the other was halted in April after being judged unlikely to work in Alzheimer’s [28]. They chased a third mechanism entirely, boosting a protein that helps regulate the brain’s immune system [28].

That is the ordinary weather of this field: most bets fail. It is part of why a drug that merely matched the existing medicines, more safely, could still send a stock down. The bar in Alzheimer’s is set by decades of disappointment.

Two quieter notes from the same conference. A large study replicated earlier findings that lifestyle changes — exercise, diet, mental activity — can reduce the risk of cognitive decline in older adults [25]. And a separate study found that a speed-based brain-training game cleared amyloid from the brains of male participants; it had no such effect in women, and the finding needs testing in larger, more diverse groups before anyone leans on it [12].

Under the radar: a vaccine built in eight weeks

Away from the conference, a faster kind of race is running. An Ebola outbreak has been spreading through the Democratic Republic of Congo and Uganda since mid-May, caused by the Bundibugyo strain of the virus [20]. As of July 9 there were 1,792 confirmed cases and 625 deaths, and the outbreak is still growing [42]. Ebola’s strains are “sisters rather than twins” — similar but different enough that each needs its own vaccine and treatment — and for this one there is no approved vaccine and no approved drug [20][42].

So scientists are building both, fast. UK regulators have cleared the first human trial of a new vaccine, ChAdOx1 BDBV, developed at the University of Oxford in about eight weeks using the same platform as the AstraZeneca COVID vaccine [20][33]. The early trial will test safety and immune response in 50 healthy adults [33]. Three other vaccines, including an mRNA candidate from Moderna, are in development for the same strain [20].

In DRC itself, a treatment trial has begun enrolling patients, randomly assigning them to remdesivir (an antiviral from Gilead), a monoclonal antibody called MBP134 that neutralises the virus, both together, or standard supportive care [42]. Both drugs worked against Bundibugyo in animals; whether they lower deaths in people is the open question [42]. As a benchmark, antibody drugs cut the death rate of the deadlier Zaire strain from about 50% to 35% [42]. A second trial, testing a drug to protect people exposed to the virus, still needs $18 million to run, with only $6 million committed [42].

“What limits an outbreak is our capacity to provide care,” said Yap Boum of Africa CDC [42]. The science is arriving in weeks. Getting it to people in a conflict zone is the slower part.