Semaglutide Fails to Slow Alzheimer Disease in EVOKE

Two large Phase 3 trials testing oral semaglutide in early Alzheimer disease have returned a clean miss: the drug did not outperform placebo on cognitive or functional outcomes at two years. The results, published in The Lancet on March 19, 2026 and presented at the AD/PD 2026 International Conference, add an important cautionary note to the considerable optimism that observational data had generated about GLP-1 drugs and brain health.

What happened

EVOKE and EVOKE+ were randomized, double-blind, placebo-controlled Phase 3 trials across 566 sites in 40 countries, conducted by Novo Nordisk. Together they enrolled participants aged 55–85 with amyloid-confirmed early Alzheimer disease — specifically mild cognitive impairment (MCI) or mild dementia — and treated them with oral semaglutide 14 mg (flexible dose, titrated to 7 or 14 mg) for 104 weeks (two years).

The primary endpoint was change from baseline in the Clinical Dementia Rating – Sum of Boxes (CDR-SB) at week 104 — the most-used global scale for staging Alzheimer progression. The secondary endpoint was change in ADCS-ADL-MCI (activities of daily living).

Results on both endpoints: no statistically significant difference between oral semaglutide and placebo.

Why this matters — and why it isn't the whole story

The negative result is significant, but it doesn't simply mean GLP-1 drugs don't affect the brain. Several layers of context:

Observational data is not predictive of RCT outcomes. Multiple large real-world analyses had shown that semaglutide and other GLP-1 agonists were associated with 20–70% lower rates of Alzheimer diagnosis in people with diabetes compared with other antidiabetic drugs. Observational signals this strong generated enormous expectations. The EVOKE trial is a reminder that confounders — health-seeking behavior, metabolic health differences, differential diagnosis rates — can produce dramatic observational associations that don't survive randomization.

Treatment vs. prevention is a different question. EVOKE enrolled patients with existing, amyloid-confirmed Alzheimer disease. The observational studies mostly followed people who did not yet have a dementia diagnosis. It's biologically plausible that GLP-1 agonists might reduce risk of developing AD (via metabolic, vascular, and inflammatory mechanisms) without meaningfully treating established disease. The trial can't speak to prevention, only to treatment.

Oral bioavailability may be a factor. Oral semaglutide has much lower bioavailability than subcutaneous semaglutide. If GLP-1 receptor engagement in the brain is part of the mechanism, higher CNS drug exposure from injectable administration might produce a different result. An injectable semaglutide trial in Alzheimer disease would test a different pharmacokinetic scenario.

The dose and population may not have been optimal. A 2-year trial in already-established Alzheimer disease may be too late in the disease course for a metabolic intervention to show functional benefit, and 14 mg oral is a lower systemic exposure than 2.4 mg subcutaneous Wegovy.

What to watch

Novo Nordisk has not announced plans to pursue an Alzheimer indication following these results — and on a clinical basis, the results don't support one. However, prevention trials (enrolling people with metabolic risk factors before cognitive decline) remain a scientifically plausible next step. Several smaller trials with injectable formulations are ongoing.

For GLP-1 users who were hoping this represented an add-on brain benefit: the evidence is not there for treatment, and shouldn't be a factor in treatment decisions for established Alzheimer disease. The metabolic, cardiovascular, and weight-loss benefits of semaglutide stand on their own extensive evidence base.

See our semaglutide guide for evidence-backed indications, and GLP-1 side effects overview for the full safety picture.

What happened

Why this matters — and why it isn't the whole story

What to watch

Sources