The promise of GLP-1 receptor agonists for brain protection has been one of the most closely watched hypotheses in dementia medicine — and these two large trials represent the most rigorous test of that idea to date. The outcome matters profoundly: if semaglutide had worked, it would have immediately changed the treatment calculus for the hundreds of millions of adults already taking or considering GLP-1 drugs for metabolic reasons.
The evoke and evoke+ trials enrolled 3,808 participants aged 55–85 with amyloid-confirmed mild cognitive impairment or mild Alzheimer's dementia across 566 sites in 40 countries — one of the largest Alzheimer's intervention trials ever conducted. Participants received oral semaglutide titrated to 14 mg once daily or placebo for up to 156 weeks. The primary endpoint was change in the Clinical Dementia Rating-Sum of Boxes (CDR-SB) score at week 104. evoke+ additionally enrolled participants with significant small-vessel cerebrovascular pathology. Both trials were discontinued early due to negative clinical outcomes, indicating semaglutide did not meaningfully slow cognitive decline on the primary measure.
This result is a significant setback for the GLP-1-as-neuroprotectant hypothesis, though it does not fully extinguish it. Observational data from diabetic and obese populations showing lower dementia incidence among GLP-1 users may reflect metabolic confounding — people on these drugs have better-controlled glucose, weight, and vascular risk factors — rather than direct neuronal protection. The evoke trials tested a relatively pure neurological intervention in an already-symptomatic Alzheimer's population, where amyloid burden is established and neurodegeneration is underway. It is biologically plausible that GLP-1 signaling could reduce dementia risk earlier in the disease cascade without reversing existing damage. The oral formulation also raises bioavailability questions versus injectable semaglutide. Nonetheless, two well-powered, placebo-controlled phase 3 trials failing their primary endpoints carries substantial evidential weight. This finding is paradigm-clarifying: GLP-1 agonists appear unlikely to benefit patients with established symptomatic Alzheimer's, even if preventive roles remain an open question.