One of the most anticipated clinical trials in recent memory has reported its results, and they are sobering. The evoke and evoke+ trials, published in The Lancet on March 19, 2026, tested whether oral semaglutide could slow the clinical progression of early-stage Alzheimer’s disease in nearly 3,800 patients over two years. The answer, at least for this formulation and population, is no.
The result doesn’t close the book on GLP-1 receptor agonists in neurodegeneration, but it does recalibrate what we should expect from them and why.
What They Found
The evoke and evoke+ trials enrolled 3,778 people with early-stage symptomatic Alzheimer’s disease across multiple countries and randomized them to receive either oral semaglutide 14 mg or placebo once daily for 104 weeks (two years). The primary endpoint was change in Clinical Dementia Rating sum of boxes (CDR-SB) score from baseline to week 104.
The CDR-SB is a well-established composite measure of Alzheimer’s severity, scoring memory, orientation, judgment, social interactions, and daily function across six domains. A higher score means worse disease.
In both trials, mean CDR-SB change from baseline to week 104 was essentially identical between the semaglutide and placebo groups: 2.3 vs. 2.3 in evoke, and 2.2 vs. 2.1 in evoke+. There was no statistically significant difference on the primary endpoint, and no significant differences on the key secondary endpoints either, including activities of daily living.
Semaglutide did produce statistically significant effects on certain cerebrospinal fluid biomarkers of Alzheimer’s pathology. Participants on semaglutide showed approximately a 10% reduction in phosphorylated tau 181 (p-tau181) compared with placebo at week 78. But that effect was too small, and too disconnected from the clinical endpoints, to suggest meaningful biological impact on disease course.
Safety was consistent with what’s known about semaglutide: nausea, vomiting, and gastrointestinal side effects were more common in the semaglutide arm, as expected.
Why This Trial Mattered
GLP-1 receptor agonists became one of the most exciting therapeutic hypotheses in Alzheimer’s research over the past several years. The hypothesis was grounded in multiple independent lines of evidence.
Epidemiological data from large registries consistently found that people with type 2 diabetes taking GLP-1 agonists had lower rates of dementia and cognitive decline compared to those on other diabetes medications. While this correlation could reflect confounding (healthier, more metabolically controlled patients tend to get GLP-1s), the signal was consistent and attracted significant attention.
Animal model work reinforced the interest. Studies in rodents demonstrated that GLP-1 receptor signaling could reduce neuroinflammation, amyloid accumulation, and tau phosphorylation. Several preclinical studies also showed cognitive improvement in Alzheimer’s-like mouse models after GLP-1 treatment.
Together, these data were compelling enough that Novo Nordisk invested in a multi-thousand-patient, multi-year phase 3 program. It was by far the largest and longest GLP-1 trial ever run in neurodegeneration. The scientific community was genuinely hopeful.
The negative result is therefore significant. It’s not a small signal that might be salvaged by a larger trial; the evoke program was large enough that a meaningful effect would have been detectable.
How the Trial Was Conducted
Both evoke and evoke+ were randomized, double-blind, placebo-controlled phase 3 trials. Participants had a confirmed diagnosis of early-stage symptomatic Alzheimer’s disease (CDR 0.5 or 1.0) and documented evidence of Alzheimer’s pathology through CSF or PET imaging. They received oral semaglutide 14 mg or matching placebo once daily for 104 weeks.
The trials were conducted across sites in North America, Europe, and Asia Pacific, and were powered to detect a clinically meaningful difference of approximately 0.5 points on CDR-SB. The combined enrollment of roughly 3,800 patients provided substantial statistical power.
An accompanying editorial in The Lancet by Semaglutide for Alzheimer’s disease investigators noted that the biomarker signal, while present, likely reflects an effect on peripheral metabolism rather than direct CNS target engagement at the doses achieved with the oral formulation.
Limitations and Caveats
Several limitations of the evoke program are worth understanding before drawing broad conclusions.
The trials used oral semaglutide (Rybelsus), not the injectable formulations (Ozempic or Wegovy). Oral and injectable semaglutide have different pharmacokinetics and bioavailability. Oral semaglutide achieves meaningfully lower plasma concentrations than the injectable form, and CNS penetration may be limited. It is not yet clear whether higher-dose injectable semaglutide would produce a different result.
The patient population had early-stage but symptomatic Alzheimer’s. Whether GLP-1 intervention in presymptomatic individuals with elevated amyloid burden but no cognitive symptoms would produce a different result remains unknown. Many researchers believe Alzheimer’s interventions need to start earlier in the disease course, when synaptic and neuronal loss has not yet accumulated.
The CDR-SB is a sensitive endpoint for established disease but may not capture subtle early benefits or proxy the right biological changes in a population where treatment begins relatively early.
Finally, two years may not be long enough for a neuroprotective effect to manifest or diverge from placebo in this population.
None of these caveats rescue the clinical results. But they suggest the question of GLP-1s in Alzheimer’s is not fully closed.
What This Means in Practice
For researchers studying neurodegeneration: the negative evoke result will redirect attention toward whether higher brain exposure (injectable, intrathecal) or earlier intervention changes the calculus. The biomarker signal, while modest, does suggest semaglutide reaches and affects Alzheimer’s-related pathways to some degree. Multiple other GLP-1 trials in Parkinson’s and Alzheimer’s are ongoing, and the evoke result will inform how those are designed and interpreted.
For clinicians: there is currently no basis for prescribing semaglutide to patients for cognitive benefit or Alzheimer’s prevention. The evoke results should not be read as any kind of provisional green light. The question of whether metabolic control using GLP-1s incidentally benefits cognition in people with diabetes and early dementia risk remains open, but that’s different from an Alzheimer’s therapeutic indication.
For science-watchers more broadly: this is a useful reminder that epidemiological signals and animal model data, however compelling, do not reliably predict phase 3 outcomes in neurodegenerative disease. The translation failure rate in Alzheimer’s remains high, and the EVOKE result fits a familiar pattern: a credible biological hypothesis, robust preclinical support, a large well-run clinical trial, and a null result.
Source
The evoke and evoke+ investigators. “Efficacy and safety of oral semaglutide 14 mg (flexible dose) in early-stage symptomatic Alzheimer’s disease (evoke and evoke+): two phase 3, randomised, placebo-controlled trials.” The Lancet, March 19, 2026. DOI: 10.1016/S0140-6736(26)00459-9
Science AAAS news coverage: Popular obesity drug fails in hotly anticipated Alzheimer’s trials
For context on other GLP-1 receptor agonist research across different disease areas, see Retatrutide and TRIUMPH-4: What the Obesity and Osteoarthritis Results Mean.