What the EVOKE Alzheimer's trials found

Oral semaglutide did not slow cognitive or functional decline in people with early Alzheimer's disease. That is the headline finding from EVOKE and EVOKE+, the two phase 3 trials Novo Nordisk ran in more than 3,800 adults, with topline results reported in November 2025^[1]. A review of the programme by Alzheimer's researchers Jeffrey Cummings and Philip Scheltens, published in September 2026, confirms the miss and asks a more useful question: what did two years of data on a well-studied drug in a new disease actually teach the field^[2]?

What the trials tested

EVOKE and EVOKE+ were identically designed, randomised, placebo-controlled trials in adults aged 55 to 85 with mild cognitive impairment or mild dementia caused by Alzheimer's disease, confirmed by amyloid testing. EVOKE enrolled 1,855 participants and EVOKE+ enrolled 1,953, for a combined 3,808 people^[1][3]. Both trials escalated participants to an oral semaglutide dose of 14 mg once daily, or matching placebo, on top of standard care. The primary endpoint was the change in Clinical Dementia Rating Sum of Boxes (CDR-SB), a standard scale of cognition and daily function, from baseline to week 104, roughly two years in^[4][5].

The result: cognitive decline tracked placebo

CDR-SB scores worsened at close to the same rate in both arms for the length of the study. Novo Nordisk reported that semaglutide did not demonstrate statistically significant superiority over placebo on the primary endpoint^[1]. Cummings and Scheltens put it more bluntly in the review, describing the trials as producing no clinical benefit of treatment in symptomatic early Alzheimer's disease^[2]. Novo Nordisk said the planned one-year extension phase of both trials will be discontinued based on the primary-analysis results, rather than keep participants on treatment that was not working^[1][3]. Secondary cognitive and functional measures showed no meaningful difference between the semaglutide and placebo groups either^[3].

What the biomarkers showed anyway

The trials were not a total blank. Semaglutide produced a statistically significant drop in high sensitivity C-reactive protein, a marker of body-wide inflammation, in both studies^[2]. In cerebrospinal fluid, two markers linked to brain inflammation, YKL-40 and glial fibrillary acidic protein, fell by 7 to 10 percent, a small but statistically real change consistent with reduced activity in the brain's support cells^[2][3]. None of it mattered clinically. The review authors found that these biomarker shifts had no correlative effect on Alzheimer's progression^[2].

A plausible reason it did not work

Diabetes researcher Daniel Drucker has pointed to the drug's route into the brain as a likely reason. GLP-1 medicines have delivered strong results in other diseases, he told Scientific American, but brain disorders have proven a harder target, and the oral pill's fatty-acid structure may keep it from crossing into brain tissue as effectively as the injectable formulations used in some earlier, smaller studies that hinted at a cognitive benefit^[7]. Cummings and Scheltens reach a similar conclusion from the trial data. Reducing peripheral and even some central inflammation was not enough on its own to change the disease's course, and they point to two directions worth testing next: brain-penetrant GLP-1 compounds, and patient groups selected for higher baseline inflammation^[2].

What this does not change

This result is specific to one dose, one formulation, and one disease population. It does not reopen semaglutide's record in type 2 diabetes, obesity, or cardiovascular risk reduction, where the trial evidence is large, separate, and unaffected by an Alzheimer's readout. Novo Nordisk's own framing of the results kept that distinction explicit, and the regulatory status of semaglutide for its approved indications has not changed^[1]. Country-specific detail on how semaglutide is regulated for its licensed uses is on our regulation pages. That separation matters for anyone currently taking semaglutide for diabetes or weight management: the EVOKE and EVOKE+ result speaks only to this specific attempt to repurpose the drug for early Alzheimer's disease, not to the indications it is actually licensed for.

The safety picture in EVOKE and EVOKE+ also matched what is already known about the drug. Participants on semaglutide reported more gastrointestinal symptoms and more weight loss than participants on placebo, consistent with the pattern already documented in semaglutide's diabetes and obesity trials, a side effect in this population rather than a goal^[3].

What we still do not know

The Alzheimer's Association called the result disappointing but said negative trial data still add to the field's understanding of the disease and of this drug class, and noted that its tracked pipeline still includes well over a hundred other candidate treatments in active trials^[6]. What EVOKE and EVOKE+ cannot answer is whether semaglutide, or a brain-penetrant successor, might do more if given earlier, before symptoms appear, or in people selected for elevated inflammatory markers rather than an unselected amyloid-positive population. Those are hypotheses the review raises, not results anyone has yet reported^[2]. Anyone weighing what this trial means for their own Alzheimer's risk or treatment plan should raise it with a neurologist or their prescribing clinician, not read a two-year negative trial as the final word on the drug class.