A million-patient real-world study suggested semaglutide could cut Alzheimer's risk by up to 70%. Then Novo Nordisk ran the actual randomized trial — the largest GLP-1 study ever done in Alzheimer's disease.
The case for testing semaglutide in Alzheimer's disease didn't start with a hunch — it started with a signal in the data. A 2024 target trial emulation published in Alzheimer's & Dementia, drawing on electronic health records from more than one million people with type 2 diabetes, found that those treated with semaglutide had a markedly lower three-year risk of a first Alzheimer's diagnosis than those on insulin (hazard ratio 0.33) or other GLP-1 receptor agonists (hazard ratio 0.59) — a 40–70% relative risk reduction depending on the comparator. A separate analysis in JAMA Neurology found GLP-1 drugs and SGLT2 inhibitors both outperformed older diabetes medications on dementia risk.
That kind of observational signal is exactly what a randomized trial exists to test, because people who stay on a GLP-1 drug for years differ from people who don't in ways a database can't fully capture. Novo Nordisk ran two identically designed phase 3 trials, EVOKE (NCT04777396) and EVOKE+ (NCT04777409), enrolling a combined 3,808 adults aged 55–85 with mild cognitive impairment or mild dementia due to Alzheimer's disease, confirmed by amyloid biomarkers. Baseline severity was closely matched between arms — mean CDR-SB score of 3.7 in both trials.
Participants took once-daily oral semaglutide (14mg) or placebo, both on top of standard AD care, for a planned 104-week main treatment phase with a further extension to look for slower, disease-modifying effects. Enrollment spanned nearly 40 countries, making this the largest randomized evaluation of a GLP-1 drug in Alzheimer's disease to date.
Semaglutide did not slow the disease. Topline results presented at the CTAD conference in December 2025 showed no statistically significant reduction in clinical progression on the CDR-SB — the standard measure of cognitive and functional decline used across Alzheimer's trials — in either EVOKE or EVOKE+. Secondary measures of cognition and daily functioning told the same story: no meaningful separation from placebo.
The biomarker data moved, just not enough to matter clinically. Novo Nordisk reported reductions of roughly 10% in certain biomarkers tied to neuroinflammation and Alzheimer's pathology — a statistically real effect that didn't translate into any measurable benefit in cognition or function within the trial's timeframe.
Based on the overall efficacy readout, Novo Nordisk discontinued the trials' planned one-year extension. Martin Holst Lange, the company's chief scientific officer, said in a statement that the company felt "a responsibility to explore semaglutide's potential, despite a low likelihood of success," and that despite the negative result, "the extensive body of evidence supporting semaglutide continues to provide benefits for individuals with type 2 diabetes, obesity, and related comorbidities." Safety and tolerability were consistent with semaglutide's established profile in other approved uses — gastrointestinal side effects and weight loss, no new signals specific to this population.
Real-world data had suggested people on semaglutide were up to 70% less likely to be diagnosed with Alzheimer's disease. In the largest randomized trial ever run to test that directly, semaglutide showed no significant slowing of cognitive or functional decline over two years — only a small, sub-clinical shift in disease biomarkers.
Sources: Novo Nordisk A/S, "Evoke phase 3 trials did not demonstrate a statistically significant reduction in Alzheimer's disease progression," news release, November 24, 2025 · Wang W, Wang Q, Qi X, et al. Associations of semaglutide with first-time diagnosis of Alzheimer's disease in patients with type 2 diabetes: target trial emulation using nationwide real-world data in the US. Alzheimer's & Dementia, 2024;20(12):8661-8672 (doi:10.1002/alz.14313) · Tang H, Donahoo WT, DeKosky ST, et al. GLP-1RA and SGLT2i medications for type 2 diabetes and Alzheimer disease and related dementias. JAMA Neurology, published online April 7, 2025 (doi:10.1001/jamaneurol.2025.0353). Novo Nordisk press release · ClinicalTrials.gov NCT04777396 (EVOKE) · ClinicalTrials.gov NCT04777409 (EVOKE+).
This is a textbook example of why randomized trials exist. The real-world signal that motivated EVOKE was built on people who were already prescribed and stayed on semaglutide for diabetes — a population that differs from a randomized cohort in ways a database can't fully adjust for, including that people who tolerate a drug well and keep taking it for years tend to be healthier to begin with. Comparing against insulin specifically, as one of the two hazard ratios did, also compares against a group whose diabetes was typically more advanced.
A biomarker move without a clinical benefit is a familiar pattern in Alzheimer's research. Several drugs targeting amyloid and related pathways have shown measurable biological effects that didn't clear the bar of a meaningful difference in how patients actually think or function. Alzheimer's Drug Discovery Foundation chief science officer Howard Fillit noted after the readout that the biomarker signal "may suggest a path forward for semaglutide as part of a combination therapy approach," rather than as a therapy on its own.
This result doesn't change what semaglutide is approved for. It remains approved for type 2 diabetes, weight management, and — following the SELECT trial — reduction of major cardiovascular events in adults with cardiovascular disease and obesity. EVOKE tested a specific, separate question: whether the drug can slow an already-diagnosed neurodegenerative disease. The answer, at this dose and in this trial, was no.
This article summarizes published, peer-reviewed and company-reported clinical trial data for general education. It is not medical advice. It should not be read as guidance to start, stop, or continue any GLP-1 medication for cognitive or brain-health reasons. Any decision about your treatment should be made with your prescriber.