Oral semaglutide does not slow early Alzheimer's in two large trials, Lancet reports

The Lancet published the full results of the evoke and evoke+ trials on March 19, 2026, the same day the data were presented at the AD/PD 2026 conference. In 3,808 people with early Alzheimer's disease, once-daily oral semaglutide at doses up to 14 mg did not slow cognitive or functional decline compared with placebo ^[1][2][3]. Both trials and their planned one-year extensions have been discontinued ^[1][3].

What the trials found

evoke and evoke+ were phase 3, randomized, double-blind, placebo-controlled trials run at 566 sites in 40 countries ^[1]. Researchers screened 9,981 people and randomly assigned 3,808 of them: 1,855 in evoke and 1,953 in evoke+ ^[1]. Participants were 55 to 85 years old with amyloid-confirmed Alzheimer's disease and either mild cognitive impairment or mild dementia. Their mean age was 72.2 years, and their mean CDR-SB score at the start was 3.7 ^[1]. Enrollment ran from May 18, 2021, to Sept. 8, 2023, and people were treated for up to 156 weeks ^[1]. The dose started at 3 mg, rose to 7 mg at four weeks and up to 14 mg at eight weeks; participants who had side effects could drop back to a lower dose for the rest of the trial ^[4].

The primary endpoint was change in the Clinical Dementia Rating-Sum of Boxes (CDR-SB) score from baseline to week 104 ^[1]. In evoke, CDR-SB rose by a mean of 2.3 points in both the semaglutide and placebo groups, an estimated difference of -0.08 (95% CI -0.35 to 0.20, p=0.57). In evoke+, the mean change was 2.2 points with semaglutide and 2.1 with placebo, a difference of 0.10 (95% CI -0.17 to 0.38, p=0.46) ^[1]. Higher CDR-SB scores mean worse function. There was also no difference on secondary measures, including a daily-living scale (ADCS-ADL-MCI), the MMSE and the ADAS-Cog ^[3]. A pooled analysis found no delay in progression from mild cognitive impairment to mild dementia, and exploratory subgroup analyses by sex, age, race, body mass index and type 2 diabetes status showed no clear signal of benefit ^[3].

"When you look at the curves, they are exactly on top of each other," Novo Nordisk's Peter Johannsen said when asked at an earlier meeting whether any group appeared to benefit ^[4].

Treatment-emergent adverse events were reported in 1,729 of 1,896 participants (91.2%) on semaglutide and 1,613 of 1,902 (84.8%) on placebo ^[1]. Investigators considered five deaths related to treatment: one in the semaglutide group and four in the placebo group ^[1]. Gastrointestinal side effects were the main reason people lowered their dose or stopped treatment. More than a third of participants on semaglutide lost weight, an average of 5.8% of body weight over two years, compared with an average gain of 0.6% on placebo ^[4]. Nausea was reported by nearly a quarter of those on semaglutide, diarrhea by 14% and vomiting by 12% ^[4].

Some cerebrospinal fluid markers of Alzheimer's pathology, including p-tau181, fell by about 10% or less in the semaglutide group at week 78, but the same markers did not shift in blood ^[3][4]. Plasma GFAP rose 4% in both trials, plasma NfL rose 5% in evoke+ only, and hs-CRP, a marker of inflammation, fell about 30% ^[4]. Researchers at the conference said the spinal fluid changes were too small to explain any patient benefit, and that the drug's safety and tolerability in this population matched other semaglutide studies ^[1][3].

Why it matters for patients

Semaglutide is the molecule in Ozempic, Wegovy and Rybelsus, which are used for type 2 diabetes and weight loss. These results apply to Alzheimer's disease, not to those uses, and do not change what is known about the drug for diabetes or obesity. The trials also do not show that the drug is unsafe for its approved uses; the safety profile reported here was consistent with studies in other conditions ^[1].

The bigger question is whether GLP-1 drugs can prevent dementia, and that remains unanswered. The observational studies that prompted the trials followed people with type 2 diabetes or obesity and often tracked whether a dementia diagnosis occurred, while evoke and evoke+ required confirmed amyloid pathology and measured whether existing disease slowed ^[3]. Conference panelists described those as different questions in different patient groups ^[3]. Whether a GLP-1 medication protects memory in people who do not already have Alzheimer's pathology has not been established.

What happens next

Novo Nordisk discontinued the one-year extension periods in both trials and ended the program after topline results were announced in November 2025 ^[3]. The full paper is scheduled for The Lancet's May 30, 2026 print issue ^[2]. Panelists at AD/PD 2026 said the failure should not end research into GLP-1 receptor agonists in Alzheimer's, pointing to possible studies in people who have both Alzheimer's and metabolic or vascular conditions, or studies of prevention ^[3]. Any such work would require significant time and investment, and no further Alzheimer's trial of semaglutide has been disclosed ^[3].