Skip to content
This GLP
← Research
Evidence

Alzheimer's disease: the trial that closed the question

Two phase 3 trials randomized 3,808 people with confirmed early Alzheimer's disease. At 104 weeks the estimated difference on the primary endpoint was -0.08. Both trials were discontinued.

Glenn Torres6 min read
Dementia rating, change over 104 weeks (higher is worse)0semaglutide 2.3placebo 2.33,808 participants. Estimated difference −0.08. Trials stopped.

Most of what this library covers is a question the evidence has not settled. This one it has. A drug class widely reported to protect the brain was tested in the population that matters, at scale, for two years, and it did nothing measurable. That is a more useful result than a positive one, and it is also the kind that gets far less coverage — which is itself a pattern worth recognizing when reading any claim on a seller’s page.

What was tested

evoke and evoke+ ran across 566 sites in 40 countries [1]. Participants were aged 55 to 85 with mild cognitive impairment or mild dementia due to Alzheimer’s disease, confirmed by amyloid imaging or cerebrospinal fluid analysis — not self-reported memory complaints. They were randomized 1:1 to once-daily oral semaglutide at up to 14 mg or placebo for up to 156 weeks. evoke+ additionally included people with significant small vessel pathology.

Between May 2021 and September 2023, 9,981 people were screened and 3,808 randomized: 1,855 in evoke and 1,953 in evoke+. Mean age was 72.2 years and the mean baseline dementia rating score was 3.7.

What they found

The primary endpoint was the change in the Clinical Dementia Rating – Sum of Boxes score from baseline to week 104. In evoke, scores worsened by 2.3 points on semaglutide and 2.3 on placebo, an estimated difference of −0.08 (95% CI −0.35 to 0.20, p=0.57). In evoke+, 2.2 against 2.1, a difference of 0.10 (95% CI −0.17 to 0.38, p=0.46).

Both intervals sit symmetrically around zero. This is not an underpowered trial that missed a small effect; it is a very large trial that found no effect. The investigators’ own conclusion is that oral semaglutide was not efficacious in slowing clinical progression, and both trials were discontinued.

What it did not find, either

Safety was unremarkable. Treatment-emergent adverse events were reported in 91.2% of 1,896 participants on semaglutide against 84.8% of 1,902 on placebo, and the investigators judged five deaths treatment-related — one in the semaglutide group and four in the placebo group. Tolerability was consistent with what the drug does in its other indications, described in titration and nausea.

How to use a negative trial

A well-run trial that finds nothing closes a question, and closing questions is the scarcer service. It means anyone still telling you these drugs help with memory is working from evidence that has been superseded — and, on current publishing incentives, may not know it.

It also tells you what to do with the next observational headline in this category. The pattern here is exactly the pattern that ran for other outcomes: a real correlation, a plausible mechanism, and no effect when tested. That is worth holding onto while reading what the semaglutide trials are worth, and while reading any seller’s summary of them — most publish nothing about which claims rest on trials and which on cohorts, a silence counted in what sellers will not tell you.

Frequently asked

Does semaglutide slow Alzheimer's disease?
No. Two phase 3 trials in 3,808 people with amyloid-confirmed early Alzheimer's disease found estimated differences of -0.08 and 0.10 on the primary endpoint at 104 weeks. Both trials were discontinued for a negative clinical outcome.
Why did earlier studies suggest it might?
Animal work and observational data in people with diabetes or obesity had shown lower dementia rates after exposure. People prescribed a drug differ from people who are not in ways that can also predict the outcome, which is what a randomized trial exists to rule out.
Was the dose too low?
The trials used oral semaglutide at up to 14 mg once daily for up to 156 weeks, with an eight-week escalation. Both confidence intervals sit symmetrically around zero rather than pointing toward an effect the trial narrowly missed.
Were there safety problems?
Nothing unusual. Treatment-emergent adverse events occurred in 91.2% on semaglutide against 84.8% on placebo, and of five deaths judged treatment-related, four were in the placebo group.

Sources

  1. [1] Cummings JL, et al. (2026). Efficacy and safety of oral semaglutide 14 mg (flexible dose) in early-stage symptomatic Alzheimer's disease (evoke and evoke+): two phase 3, randomised, placebo-controlled trials The Lancet. PMID 41865758
  2. [2] Cummings JL, et al. (2025). evoke and evoke+: design of two large-scale, double-blind, placebo-controlled, phase 3 studies evaluating efficacy, safety, and tolerability of semaglutide in early-stage symptomatic Alzheimer's disease Alzheimer's Research & Therapy. PMID 39780249

Where to get it

Best GLP-1 injections

Every injectable seller we can verify, with the price each one publishes and an honest read of what the trials measured.

Compare providers →

More in Evidence