Semaglutide and Liraglutide Do Not Treat Alzheimer’s Dementia
STUDIES 1-2: Cummings JL et al, Lancet 2026; published online March 19
STUDY 3: Edison P et al, Nature Medicine 2026;32:353–361, PMID: 39613937
STUDY TYPE: Randomized, double-blind, placebo-controlled trials
FUNDING: Industry (Novo Nordisk) and Alzheimer’s Society UK; Alzheimer’s Drug Discovery Foundation
Background
GLP-1 receptor agonists lower rates of dementia in large observational trials, and animal studies tell us the mechanism is there. They reduce neuroinflammation, amyloid, and tau. But in 2026 they failed to work, negating their chance of FDA approval.
Studies 1-2: Evoke Trials of Semaglutide
These are two trials published in a single paper. 3,808 adults aged 55–85 with amyloid-confirmed mild cognitive impairment or mild Alzheimer’s dementia were randomized across 566 sites in 40 countries to oral semaglutide 14 mg daily or placebo for up to 3 years, added to standard care. The evoke+ trial additionally enrolled patients with significant cerebrovascular small vessel disease. The primary outcome was change in CDR-SB (a global scale of cognition and function) at 104 weeks.
Semaglutide did not slow clinical decline. Cognitive and functioning scores worsened by 2.3 points with both semaglutide and placebo in evoke (estimated difference −0.08, p = 0.57) and by 2.2 vs. 2.1 points in evoke+ (difference 0.10, p = 0.46). No secondary clinical outcomes — daily function, global disease progression, time to dementia — differed between groups.
Part of the study looked at cerebral spinal fluid in 199 participants. Here, semaglutide reduced several tau biomarkers and neuroinflammation markers by 5–10%, and an inflammatory marker (C-reactive protein) also fell — but neither translated into clinical benefit.
Gastrointestinal side effects were common; weight loss averaged 5.8% with semaglutide vs. 0.6% with placebo. No new safety signals emerged.
Study 3: ELAD trial of Liraglutide
Two hundred four adults with mild to moderate Alzheimer’s disease (no diabetes) were randomized to daily subcutaneous liraglutide 1.8 mg or placebo for 52 weeks at 24 UK sites. The primary outcome was change in cerebral glucose metabolism on FDG-PET. Secondary outcomes included cognitive change on the ADAS-Exec (a composite of the standard Alzheimer’s Disease Assessment Scale-Cognitive Subscale and executive function testing), daily functioning (ADCS-ADL scale), and overall disease severity (CDR-SoB scale).
Liraglutide did not improve cerebral glucose metabolism compared to placebo (difference = −0.17, p = 0.14). On the secondary cognitive outcome, liraglutide-treated patients showed slower decline on the ADAS-Exec at 52 weeks (difference = 0.15, p = 0.01), though this was not a statistically reliable result (not corrected for multiple comparisons). It did not improve daily functioning (ADCS-ADL) or overall dementia severity (CDR-SoB).
The study also explored changes on brain imaging (MRI). Here, liraglutide led to less gray matter and temporal lobe volume loss — in line with the drug’s proposed neuroprotective mechanism.
Gastrointestinal side effects were more common with liraglutide, but serious adverse events were actually more frequent in the placebo group.
These trials close the door on dementia treatment, but we don’t yet know if GLP-1 agonists prevent dementia.
Practice Implications
- GLP-1 agonists for weight loss? Yes.
- For diabetes? Yes.
- For addiction and depression? Not yet.
- For dementia? No,
—Chris Aiken, MD
Director, Psych Partners
Editor in Chief, Carlat Psychiatry Report
Share Your Input in Comments
- What is your take on these findings?
