Semaglutide shows no benefit in early Alzheimer's disease, full trial results confirm

Oral semaglutide did not slow early Alzheimer's disease, according to the detailed topline results presented December 3 at the Clinical Trials on Alzheimer's Disease (CTAD) meeting in San Diego ^[1][2]. On the primary measure and every secondary cognitive and functional measure the researchers displayed, the treatment and placebo groups followed the same path ^[1].

Between them, the two trials randomized 3,808 people aged 55 to 85 who had mild cognitive impairment or mild dementia due to Alzheimer's disease, confirmed by amyloid testing — 1,855 in EVOKE and 1,953 in EVOKE+, across 40 countries. Nearly 10,000 people were screened ^[1]. Participants took either a once-daily oral semaglutide pill or a placebo. The dose began at 3 mg, rose to 7 mg after four weeks and reached 14 mg after eight weeks, a regimen on par with type 2 diabetes treatment ^[1]. The main yardstick was the change in Clinical Dementia Rating–Sum of Boxes (CDR-SB) score from the start of the study to week 104 ^[1][2].

Jeffrey Cummings of the University of Nevada, Las Vegas presented the efficacy data. The decline curves for semaglutide and placebo sat on top of one another throughout the three-year studies, at a pace typical of early Alzheimer's, he said ^[1]. Secondary scales — ADCS-ADL-MCI, MoCA, ADAS-Cog13, MMSE and ADCOMS — behaved the same way, as did the time it took for participants with mild cognitive impairment to progress to dementia ^[1]. Subgroup analyses were not shown. Asked whether any group appeared to do better, Novo Nordisk's Peter Johannsen said that if he had to name one, it would be a very small one ^[1]. About 85 percent of participants completed the second year, and 30 percent had reached the end of the third year when the data were locked for analysis ^[1].

A cerebrospinal fluid (CSF) substudy produced the only signal in the drug's favor. Seven markers fell by 10 percent or less among people taking semaglutide: p-tau181, p-tau217, np-tau181, np-tau205, YKL-40, total tau and neurogranin ^[1]. The Alzheimer's Drug Discovery Foundation described those reductions as statistically significant but too small to have a clinical impact ^[3]. In blood, none of the same markers shifted. Plasma GFAP climbed about 4 percent in both trials, and plasma NfL rose 5 percent in EVOKE+ only — findings Cummings said he could not explain ^[1]. High-sensitivity C-reactive protein, an inflammation marker, dropped roughly 30 percent ^[1]. Johannsen noted that semaglutide is not expected to act on amyloid, and that the CSF shifts were in the same range as changes seen with other drugs that turned out not to work ^[1]. A separate poster reported that after 12 weeks the median semaglutide concentration was 0.14 nmol/L in CSF versus 34.8 nmol/L in plasma, about 0.4 percent, indicating the drug does reach the brain ^[1].

On safety, total adverse events were more common with semaglutide, though serious, severe and fatal events occurred at similar rates in both groups ^[1]. More people on the drug lowered their dose or stopped treatment, mostly because of mild to moderate gastrointestinal effects. More than a third lost weight; about 23 percent reported nausea, 14 percent diarrhea and 12 percent vomiting ^[1]. Over the first two years, semaglutide users lost an average of 5.8 percent of body weight, while placebo users gained 0.6 percent on average ^[1].

Why it matters for patients

This result answers a specific question: whether semaglutide can treat Alzheimer's disease. It does not change the approved uses of Ozempic and Rybelsus for type 2 diabetes or Wegovy for chronic weight management, and the trials turned up no new safety problems beyond the gastrointestinal effects and weight loss already tied to the drug ^[1][2]. Novo Nordisk says more than 37 million patient-years of semaglutide exposure have been recorded across other populations ^[2].

What the data do not show is any protection of memory or daily functioning in people who already had early symptoms ^[1][3]. The biomarker changes were measurable but small, and they did not translate into help for patients ^[3]. Some researchers still argue that GLP-1 drugs might postpone dementia if given before symptoms begin, an idea supported by registry studies linking the drugs to lower dementia risk. That question has not been tested in a trial designed to answer it ^[1][3]. Novo Nordisk said it launched the program on the strength of real-world evidence, animal models and post-hoc analyses of diabetes and obesity trials ^[2], and its chief scientific officer, Martin Holst Lange, said the company felt a responsibility to explore the drug's potential despite a low likelihood of success ^[2].

What happens next

The planned 52-week extension of both trials is being stopped because of the overall results ^[2]. The full dataset is scheduled for presentation at the Alzheimer's and Parkinson's Diseases conference in Copenhagen in March 2026 ^[2][3]. Neither Novo Nordisk nor any other sponsor has disclosed plans for a prevention trial ^[1][3].