Why semaglutide failed in two Alzheimer's trials

The EVOKE and EVOKE+ phase 3 trials, which tested oral semaglutide in 3,808 people aged 55 to 85 with early Alzheimer’s disease, found no benefit over placebo on the primary measure of disease progression after 104 weeks. Novo Nordisk reported the topline result in November 2025, and the Lancet has since published the trials. Search interest in GLP-1 drugs is up 159% year over year, so the result matters to a lot of people.

What did the EVOKE trials test, and what did they find?

EVOKE and EVOKE+ were randomised, double-blind, placebo-controlled trials across nearly 40 countries. Participants had mild cognitive impairment or mild dementia due to Alzheimer’s disease. The main phase ran 104 weeks. The primary endpoint was change on the Clinical Dementia Rating Scale Sum of Boxes (CDR-SB), on which participants started at a mean of 3.7 in both trials.

According to NeurologyLive, semaglutide did not show superiority over placebo on that endpoint. Analysts writing for Clinical Trials Arena add that secondary endpoints, including activities of daily living and cognitive tests such as the Mini-Mental State Examination, also showed no benefit.

There was a biological signal. Semaglutide lowered cerebrospinal fluid biomarkers of Alzheimer’s pathology, such as p-tau181, by roughly 10% at week 78. Experts judged that too small to be clinically meaningful given the negative outcomes.

Why does this matter beyond Alzheimer’s?

Because a randomised, placebo-controlled trial is the design that separates a drug’s effect from the differences between people who do and do not take it. Hopes for a treatment often begin with observational or biological signals, and a trial is where they are tested. Here the trial did not confirm a clinical benefit.

Two limits on the conclusion apply. These trials enrolled people who already had symptomatic disease, so they do not speak directly to whether a GLP-1 drug could affect dementia risk earlier in life. And the panellists Clinical Trials Arena quotes suggested future work should look at semaglutide in patients with metabolic comorbidities rather than abandoning the mechanism entirely.

The result also sits alongside better news for the same drug class elsewhere. In the SELECT trial, semaglutide’s heart benefit did not track weight loss, a randomised result in a different condition. The lesson is that each claim needs its own trial: the same drug can work in one condition and not in another. Related: how much GLP-1 weight loss is muscle.

The bottom line

• In 3,808 adults aged 55 to 85 with early Alzheimer’s, oral semaglutide did not beat placebo on CDR-SB at 104 weeks in EVOKE and EVOKE+.
• Secondary endpoints also showed no benefit, although CSF p-tau181 fell by about 10% at week 78.
• A randomised trial did not confirm clinical benefit, and the results do not address prevention earlier in life.