News · Clinical trials

EVOKE and EVOKE+: Oral Semaglutide Made No Difference to Alzheimer's Progression

A 3,808-patient, two-year programme returned a null result — and null results in this field deserve more coverage than they get.

Event date 2026-06-08 · Published 2026-07-24 · EV-0815

The result

The EVOKE and EVOKE+ trials enrolled 3,808 participants over two years to test whether oral semaglutide slowed Alzheimer's disease progression. It did not.

That is a clean, well-powered negative finding, and it is worth reporting precisely because negative findings attract a fraction of the coverage positive ones do.

Why the hypothesis was reasonable

GLP-1 receptors are expressed in the brain. Observational work had suggested associations between GLP-1 use and lower dementia incidence. Metabolic dysfunction and neurodegeneration share pathways, and insulin signalling in the brain has been a research target for years.

It was a serious hypothesis with a plausible mechanism, tested properly at scale. That is how the process is supposed to work, and the answer is no.

Why this matters beyond Alzheimer's

The GLP-1 class has expanded into indication after indication — sleep apnoea, heart failure, kidney outcomes, osteoarthritis, addiction, and more. Much of that expansion is genuine, backed by trials with hard endpoints. Some of it is extrapolation from mechanism and observational association.

EVOKE is a useful corrective. A plausible mechanism plus observational signal does not reliably predict a trial result, and this class is not a general-purpose intervention for every condition that correlates with metabolic dysfunction.

Anyone marketing these drugs on the breadth of their potential indications should be asked which ones have completed randomised trials with hard endpoints, and which are hypotheses.

What a null result does not mean

It does not mean GLP-1s harm cognition. It does not close the question for other agents in the class, other doses, other populations or earlier disease stages. And it does not undermine the established indications, which rest on their own evidence.

It means this specific drug, at this dose, over two years, in this population, did not slow progression.

The publication-bias point

Negative results are systematically under-reported, and a field where positive readouts drive stock prices and prescribing has structural pressure in one direction. A trial programme of 3,808 people over two years is a substantial investment, and its most useful output is a clear no.

Our evidence policy treats null results as evidence of the same standing as positive ones. That is easy to state and harder to practise, because a null result generates no search demand — which is exactly why it belongs on the news desk rather than only in a database.

Questions readers actually ask