GLP-1 and Alzheimer’s: Why EVOKE Was a Letdown (and What It Still Means)

If you followed peptide and GLP-1 news in 2024–2025, you saw a steady drumbeat of headlines: “Ozempic may prevent Alzheimer’s.” Observational data from large registries consistently showed GLP-1 users had lower dementia risk—in some studies, less than half the rate. Novo Nordisk bet big on this signal, running two massive Phase 3 trials.

In late 2025, the results came in. Semaglutide did not slow cognitive decline in people with early Alzheimer’s disease.

This is the honest breakdown of what happened, what the data actually shows, and why the story isn’t over.

The EVOKE Trials: Design and Scale

EVOKE (NCT04777396) and EVOKE+ (NCT04777409) were two Phase 3, randomized, double-blind, placebo-controlled trials enrolling a combined 3,808 participants with mild cognitive impairment or mild dementia due to Alzheimer’s disease, confirmed by amyloid biomarkers.

• Drug: Oral semaglutide 14 mg daily
• Duration: 104-week main treatment phase, plus 52-week extension
• Primary endpoint: Change in Clinical Dementia Rating–Sum of Boxes (CDR-SB)

Both trials missed their primary endpoints. Semaglutide did not produce a statistically significant difference in CDR-SB scores compared to placebo at 104 weeks.

What the Biomarkers Showed

Here’s where it gets interesting. While semaglutide failed on cognition, it did move several Alzheimer’s-related biomarkers:

• CSF inflammatory markers improved
• Plasma high-sensitivity C-reactive protein (hs-CRP) dropped ~30% in both trials
• Some neurodegeneration markers showed modest improvement

But critically, plasma GFAP (a marker of astrocyte reactivity) rose ~4% in the semaglutide group, and plasma NfL (neurofilament light, a marker of neuronal damage) rose ~5% in EVOKE+. These went the wrong direction.

The bottom line: semaglutide engaged some inflammatory pathways but could not overcome the neuronal destruction already underway in symptomatic Alzheimer’s.

Why Prevention ≠ Treatment

The disconnect between the observational data (GLP-1 users get less dementia) and the trial data (semaglutide didn’t treat Alzheimer’s) likely comes down to timing. UK experts commenting on the results noted that by the time mild cognitive impairment or early dementia is present, years of neuronal loss and circuit disruption have already occurred. A drug that reduces neuroinflammation may slow the development of pathology over decades without being able to reverse damage already done.

This is the “prevention vs treatment” gap that has plagued Alzheimer’s research for thirty years. Anti-amyloid antibodies (lecanemab, donanemab) slow decline by ~30%—but that’s in the same early-stage population where semaglutide showed no benefit on cognition.

What Comes Next

The Alzheimer’s Drug Discovery Foundation called the EVOKE results disappointing but valuable, noting that the completion of two large Phase 3 trials targeting a non-amyloid pathway represents real progress toward combination approaches. The GLP-1/Alzheimer’s hypothesis isn’t dead—but it likely belongs in the prevention column, not the treatment column. Future research will probably focus on GLP-1 use in pre-symptomatic, at-risk populations rather than diagnosed Alzheimer’s patients.

Research Peptide Sources

The following vendors ship research-grade peptides with published certificates of analysis. Use code where shown for reader pricing. All products are sold for laboratory and research use only.

Frequently Asked Questions

Related Articles