Detailed results from the phase 3 EVOKE and EVOKE+ trials of the oral GLP-1 receptor agonist semaglutide (Ozempic, Novo Nordisk) in early Alzheimer’s disease (AD) have been presented.

The initial negative top-line results announced last week by the company were met with disappointment in the AD research community, but the newly presented biomarker data offer some grounds for optimism and may point to a potential path forward.

The full results were presented on December 3 at the 18th Clinical Trials on Alzheimer’s Disease (CTAD) Conference.

No Cognitive or Functional Benefit

One notable finding was that semaglutide led to a significant decrease in plasma high-sensitivity C-reactive protein (CRP), a marker of inflammation, principal investigator, Jeffrey L. Cummings, MD, Department of Brain Health, University of Nevada, Las Vegas, reported.

“Our underlying biological hypothesis was, if we can suppress peripheral inflammation, we will have a beneficial impact on cognition. So we did significantly suppress peripheral inflammation, as measured by the high-sensitivity CRP, but we did not have the corresponding benefit on the cognition that we had hoped for,” Cummings told conference delegates.

However, although semaglutide improved several AD-related biomarkers in both trials, these changes did not translate into cognitive or functional benefit or slowed disease progression. 

While “disappointing,” the results demonstrated “reductions of up to 10% in biomarkers linked to neuroinflammation and AD — a statistically significant change, but not large enough to have a clinical impact,” Howard Fillit, MD, co-founder and chief science officer of the Alzheimer’s Drug Discovery Foundation, said in a statement. 

Fillit believes that exploring GLP-1 drugs as preventive therapy may still hold promise within a broader strategy targeting AD pathways beyond amyloid.

In a keynote address at CTAD, Peter Johannsen, PhD, with Novo Nordisk, said the case for investigating semaglutide in early AD was strong.

Given the evidence available in 2020, the decision to proceed with the research was justified, said Johannsen. “We still think this was a scientific question that needed an answer, although we didn’t get the answer we had hoped for,” he said.

Primary Endpoint Not Met

In animal models of AD, GLP-1 receptor agonists have been shown to improve performance on maze and memory tasks, reduce amyloid plaque burden and soluble amyloid-beta levels, lower neuroinflammatory markers and oxidative stress, and preserve synaptic markers and neuronal density.

Multiple observational studies have suggested a reduced risk for all-cause dementia or AD with GLP-1 receptor agonist exposure. For example, pooled data from three large cardiovascular outcomes trials showed that GLP-1 receptor agonists reduced the rate of all-cause dementia by 53% (vs placebo) in patients with type 2 diabetes. 

In addition, in a target trial emulation study using nationwide real-world data from TriNetX US, semaglutide was associated with 40%-70% reduced risks for first-time AD diagnosis in patients with type 2 diabetes compared with other antidiabetic medications, including other GLP-1 receptor agonists.

The EVOKE and EVOKE+ randomized, double-blinded trials enrolled a total of 3808 adults aged 55-85 years with mild cognitive impairment (MCI) or mild dementia due to AD with confirmed amyloid positivity.

Participants were randomly allocated to once-daily oral semaglutide 14 mg or placebo for 156 weeks (104-week main treatment phase and 52-week extension) in addition to standard care.

At 2 years, both trials showed no drug-placebo difference on the Clinical Dementia Rating-Sum of Boxes, the primary outcome, with an estimated 0.06-point difference between semaglutide and placebo (P = .7) in EVOKE and a 0.15-point difference between semaglutide and placebo in EVOKE+ (P = .4).

“Because the primary endpoint was not met, statistical testing for secondary outcomes was not pursued under the hierarchical analysis plan,” Cummings explained.

Descriptive analyses revealed no significant difference between semaglutide and placebo on key secondary outcomes. These included the Alzheimer’s Disease Cooperative Study Activities of Daily Living Scale for MCI, the Montreal Cognitive Assessment, the Alzheimer’s Disease Assessment Scale-Cognitive Subscale, the Mini-Mental Health Examination, and the Alzheimer’s Disease Composite Score — collectively demonstrating no impact of semaglutide on cognition or function in the two trials.

Encouraging Biomarker Data

A preplanned pooled analysis assessed the transition from MCI to AD dementia. There were 717 progression events in the semaglutide group and 757 in the placebo group. Across both trials, semaglutide failed to delay the time to progression to dementia in those with MCI (hazard ratio, 0.96 in pooled event analysis across both trials).

Encouragingly, semaglutide was associated with reductions of up to 10% in several AD-relevant cerebrospinal fluid biomarkers, including pTau181, pTau217, npTau181, and npTau205, as well as in the neuroinflammation marker YKL-40 and two neurodegeneration markers (total tau and neurogranin).

Across exploratory blood-based biomarkers, there was a significant increase of about 5% in neurofilament light in the EVOKE+ trial and a significant increase of about 4% in glial fibrillary acidic protein in both trials, Cummings reported. Significant decreases in plasma high-sensitivity CRP with semaglutide were also noted (the estimated treatment ratio was 0.76 in EVOKE and 0.71 in EVOKE+).

During the Q&A portion of the presentation, one audience member asked the investigators how they reconcile the discrepancy between the biomarker changes and the lack of clinical benefit.

Johannsen noted that while semaglutide produced “measurable biomarker shifts,” the magnitude was likely too small to meaningfully influence disease progression.

The level of biological effect — roughly 10% change of AD-relevant biomarkers, plus neurodegenerative biomarkers — is not comparable to drugs that do show clinical benefit, he explained.

“If we look across other trials, this level is in the ballpark of what was more or less seen with certain anti-amyloid monoclonals that translated into clinical efficacy. The two that are on the market — lecanemab and donanemab — change in biomarkers is around 30%,” Johannsen said.

Because of this gap in effect size, he suggested that future phase 2 trials should require larger biomarker changes before advancing to phase 3. “Probably you want more than 10% change — whether that is 20% or 25% or it really has to be 30%, I don’t know,” Johannsen said.

Valuable Lessons

Fillit acknowledged the CTAD results were disappointing but emphasized that high-risk, high-reward research remains essential for advancing understanding of AD and moving the science forward.

“Even negative trials move the field forward, as they still teach us something. If we look to the early anti-amyloid studies, which were also negative, they offered critical lessons that informed later trials and ultimately helped bring drugs like Leqembi and Kisunla to market,” he noted.

The EVOKE trials were “robust and rigorous” and provide “crucial learnings” that will help refine future study designs and guide the evaluation of other drugs targeting metabolic and inflammatory pathways, Fillit said.

He pointed out that even when trials miss their primary endpoints, biomarker and mechanistic findings can guide how therapies are refined, repurposed, or combined. He pointed out that this approach has driven progress in oncology and said AD research should continue along a similar path.

Full results from the EVOKE and EVOKE+ trials will be presented at the International Conference on Alzheimer’s and Parkinson’s Diseases and Related Neurological Disorders in Copenhagen, Denmark, in March 2026.

The EVOKE studies were funded by Novo Nordisk. Several authors reported having relationships with the company. Fillit reported having no relevant disclosures.