GLP-1 Drugs and Alzheimer's: Why the Largest Trials Failed Despite Promising Signs
The EVOKE trials testing semaglutide for Alzheimer's disease failed their primary endpoints in November 2025. But the story isn't that simple—and researchers haven't given up.
The Bottom Line
Semaglutide did not slow cognitive decline in early Alzheimer's patients over 2 years in Phase 3 trials. However, biomarkers improved—suggesting the drug engaged relevant pathways. Meanwhile, observational data continues to show 40-70% lower dementia risk in diabetic patients taking GLP-1s. The disconnect between trial failure and real-world signals remains unexplained.
The EVOKE Trials: What Happened
In November 2025, Novo Nordisk announced that EVOKE and EVOKE+ failed their primary endpoints. These were the first large-scale trials testing whether GLP-1 drugs could slow Alzheimer's disease progression.
EVOKE Trial Design
Population: Adults aged 55-85 with mild cognitive impairment or mild dementia due to Alzheimer's, with confirmed amyloid positivity
Treatment: Oral semaglutide 14mg daily vs placebo for 156 weeks
Primary Endpoint: Change in Clinical Dementia Rating – Sum of Boxes (CDR-SB) score from baseline to week 104
Result: No significant difference between semaglutide and placebo
What the Data Actually Showed
While the primary endpoint failed, the trials weren't entirely negative. Novo Nordisk reported that Alzheimer's-related biomarkers improved during treatment—they just didn't translate into slower clinical decline.
"This is not the first time that we have seen well-conducted AD clinical trials where biomarkers have moved in a healthy direction through the intervention, yet clinical—for example, cognitive responses—did not see a statistically significant benefit during the duration of the trial."
— Alzheimer's Association statement, November 2025The Biomarker Paradox
Researchers noted improvements in markers of blood-brain barrier integrity, oxidative stress, vascular integrity, and neuroinflammation. This suggests semaglutide engaged relevant biological pathways—but by the time patients have symptoms, too much damage may already be done.
"By the time individuals have mild cognitive impairment or mild dementia, much of the neuronal loss and circuit disruption is already entrenched. At this point in the disease course, even meaningful shifts in fluid biomarkers may not translate into measurable improvements in cognition or daily functioning over a two-year trial."
— Ivan Koychev, PhD, Imperial College LondonThe Observational Evidence: A Different Story
The trial failure is puzzling because real-world data consistently shows protective effects. Large studies of diabetic patients found dramatically lower dementia rates among GLP-1 users.
| Study | Population | Risk Reduction |
|---|---|---|
| Wang et al. 2024 (Alzheimer's & Dementia) | T2D patients, US EHR data | 40-70% lower AD risk vs other diabetes drugs |
| Tang et al. 2024 (EClinicalMedicine) | Swedish T2D patients 65+ | Lower dementia vs DPP-4i or sulfonylureas |
| Journal of Alzheimer's Disease 2025 | 1.7M T2D patients | 46% lower ADRD vs insulin, 33% vs metformin |
The most recent analysis, published in the Journal of Alzheimer's Disease in December 2025, found semaglutide associated with significantly lower risk across multiple dementia subtypes compared to other diabetes medications.
Why the Disconnect?
Several theories explain why observational data shows benefit while the RCT failed:
1. Too Late in Disease Course
EVOKE enrolled patients with symptomatic Alzheimer's. By this stage, irreversible damage may have occurred. The observational studies captured people who started GLP-1s years earlier, potentially preventing progression rather than reversing it.
2. Oral vs Injectable Formulation
EVOKE used oral semaglutide. Some researchers speculate injectable formulations may achieve better brain penetration. The observational signals largely came from injectable GLP-1 users.
"We may need to pursue similar agents with stronger brain penetration and more potent metabolic or anti-inflammatory effects."
— Riccardo De Giorgi, MD, University of Oxford3. Prevention vs Treatment
GLP-1s might prevent dementia through metabolic benefits (weight loss, improved insulin sensitivity, reduced inflammation) that take years to manifest—effects that wouldn't show up in a 2-year treatment trial of people who already have the disease.
4. Confounding in Observational Studies
People prescribed GLP-1s may differ systematically from those on other diabetes medications. Despite statistical adjustments, unmeasured confounders could inflate the apparent benefit.
What Happens Next
Full EVOKE data presented at Clinical Trials on Alzheimer's Disease (CTAD) conference
Complete data presentation at Alzheimer's and Parkinson's Diseases Conference
Researchers analyzing biomarker data to understand pathway engagement
Potential trials in earlier disease stages or with stronger brain-penetrating formulations
Novo Nordisk discontinued the planned 52-week extension of EVOKE. However, the field isn't abandoning GLP-1s for brain health entirely.
"While these results are not what we had hoped for, they will contribute to our understanding of this devastating and fatal disease. These results will help us refine our understanding of this class of drugs."
— Joanne Pike, DrPH, Alzheimer's Association President and CEOThe Broader Context
Alzheimer's drug development has a brutal track record—over 99% of candidates fail. The amyloid-targeting antibodies like lecanemab and donanemab that recently gained approval show modest benefits (27-35% slowing of decline) with significant safety concerns.
GLP-1 drugs offered hope for a different approach: targeting metabolic dysfunction and inflammation rather than amyloid directly. That hypothesis hasn't been validated in clinical trials—yet.
What This Means for Current GLP-1 Users
If you're taking semaglutide or another GLP-1 for diabetes or obesity, this trial doesn't change your treatment. The drugs still work for their approved indications. The observational data suggesting brain benefits remains intriguing but unproven.
No one should take GLP-1 medications specifically for dementia prevention outside of clinical trials—there's no evidence they work for this purpose in people who don't have diabetes or obesity.
Primary Sources
- Alzheimer's Research & Therapy (Jan 2025) - EVOKE/EVOKE+ trial design: doi: 10.1186/s13195-024-01666-7
- Alzheimer's & Dementia (Dec 2024) - Wang et al. semaglutide and AD risk: doi: 10.1002/alz.14313
- Journal of Alzheimer's Disease (Jan 2025) - Semaglutide vs other diabetes drugs: doi: 10.1177/13872877251351329
- Novo Nordisk Press Release - EVOKE results announcement: November 24, 2025
- Medscape (Nov 2025) - Expert commentary on trial failure
- ClinicalTrials.gov - NCT04777396 (EVOKE), NCT04777409 (EVOKE+)
Medical Disclaimer: This article is for informational purposes only and does not constitute medical advice. GLP-1 medications are not approved for Alzheimer's treatment or prevention. Consult your healthcare provider about dementia risk factors and treatment options.