Home / Publications

GLP-1 Receptor Agonists and Alzheimer’s Disease: What ELAD and EVOKE Actually Tell Us

Audio

Listen to this article

Article image

GLP-1 receptor agonists (GLP-1RAs) have transitioned from initial epidemiological observations to comprehensive Phase 3 trials for Alzheimer's, verified by biomarkers. Trials with semaglutide showed biomarker changes in symptomatic Alzheimer's, but no clinical improvement. The Phase 2b ELAD study on liraglutide revealed some neurobiological signals despite a negative primary outcome.

This article explores current knowledge, public findings not yet fully published, and their implications for clinicians considering GLP-1 drugs for Alzheimer's.

Focus on GLP-1RAs in Dementia Research

GLP-1RAs, initially for type 2 diabetes and obesity, became of interest in dementia research due to studies suggesting lower dementia rates among users, especially semaglutide, compared to other glucose-lowering treatments. These findings come from large comparative analyses replicating randomized designs using claims or EHR data.

The biological plausibility of these observational signals is supported by links between cardiometabolic risk, inflammation, endothelial dysfunction, insulin resistance, cognitive decline, and mixed pathology. GLP-1RAs improve weight, glycemic control, and cardiovascular outcomes, potentially enhancing brain resilience through inflammation and vascular biology modulation. The clinical question is whether they can slow Alzheimer's progression once amyloid-positive symptoms are established, which was the focus of the EVOKE and EVOKE+ trials.

EVOKE and EVOKE+: Insights into Semaglutide's Phase 3 Trials

Trial Design Overview

The EVOKE and EVOKE+ trials were large, randomized, double-blind, placebo-controlled Phase 3 studies testing oral semaglutide in adults with early symptomatic Alzheimer's confirmed by amyloid pathology. The primary endpoint was the change in CDR-Sum of Boxes (CDR-SB) from baseline to week 104, with secondary outcomes including ADCS-ADL-MCI and progression to dementia (CDR global >=1.0) among participants with MCI at baseline.

Overall Outcome: No Clinical Slowing

On November 24, 2025, Novo Nordisk announced that EVOKE and EVOKE+ did not show semaglutide's superiority over placebo on the primary endpoint (CDR-SB change), and biomarker improvements did not translate into delayed clinical progression.

Conference reports at CTAD 2025 added nuances: EVOKE showed a slightly favorable direction, while EVOKE+ was slightly unfavorable, leading to a null pooled estimate. The CTAD slides offer detailed data on effect direction and safety, though a full peer-reviewed outcomes manuscript is pending.

Biomarkers: A Modest, Clinically Disconnected Signal

Conference summaries note that semaglutide modestly improved selected Alzheimer's-related biomarkers, including CSF measures, without changing clinical outcomes. This highlights EVOKE's paradox: biological changes occurred, but patient conditions remained unchanged.

This pattern is not new in Alzheimer's treatments. Small biomarker shifts, especially without significant amyloid plaque reduction in amyloid-positive symptomatic cases, often fail to show a CDR-SB signal over 18-24 months. EVOKE suggests that semaglutide's effect might be too small, too late, or too unrelated to the main drivers of symptomatic progression in this context.

Safety and Tolerability: Typical GLP-1RA Effects with Dementia-Specific Concerns

Public CTAD materials indicate higher adverse event rates and discontinuations in the semaglutide group, consistent with gastrointestinal intolerance and weight loss. These effects, manageable in younger obesity cohorts, pose challenges for older adults with cognitive impairment, frailty, or borderline nutrition.

The key takeaway for clinicians is not that semaglutide is "dangerous" for Alzheimer's, but that the risk-benefit ratio shifts when the expected benefit is zero clinical slowing. In this context, tolerability issues like nausea, reduced appetite, weight loss, and dehydration risk become more significant.

Did Any Subgroup "Get Worse"?

Currently, public EVOKE materials support a cautious statement: EVOKE+ showed a worse trend than placebo at the trial level, but available sources do not yet show a definitive, prespecified subgroup analysis demonstrating a consistent "harmful phenotype" (e.g., APOE4 carriers, higher frailty, lower BMI) with statistically significant worsening. The best-supported claim is trial-level directionality, not a proven harmful subgroup.

ELAD: Liraglutide's Phase 2b Story is Nuanced and Peer-Reviewed

ELAD (Evaluating Liraglutide in Alzheimer’s Disease) was a multicenter Phase 2b randomized trial of daily liraglutide in mild-to-moderate Alzheimer's syndrome. Unlike EVOKE, ELAD is fully published in Nature Medicine, providing clarity on endpoints and interpretation.

Primary Endpoint: Negative

ELAD did not meet its primary outcome, which focused on cerebral glucose metabolism (FDG-PET), the prespecified "win condition."

Secondary and Exploratory Signals: Cognition and Brain Volume

Despite the negative primary endpoint, the publication reports favorable secondary signals on an executive composite and slower MRI brain volume loss compared to placebo. These findings align with earlier conference narratives about reduced atrophy. These results are best seen as hypothesis-generating, warranting mechanistic interest and potentially better-targeted trials, but they do not establish liraglutide as a disease-modifying therapy for Alzheimer's.

Reconciling the Discrepancy: Why GLP-1RAs Appeared Promising but Failed in Phase 3 AD

EVOKE prompts reconsideration. The epidemiology might still be accurate in another sense: GLP-1RAs could lower dementia incidence by modifying systemic risks like obesity, diabetes, and vascular disease, or by delaying mixed-pathology trajectories over time - effects beyond the scope of a two-year symptomatic Alzheimer's trial. Observational cohorts face confounding and reverse causation (patients with early cognitive decline may be less likely to start or adhere to complex therapies), potentially exaggerating protective effects.

In essence, GLP-1RAs might still be relevant for "dementia" broadly while not affecting biomarker-confirmed symptomatic Alzheimer's progression in the short term. EVOKE suggests that if GLP-1RAs have a future role in neurodegeneration, it is likely in prevention, in mixed vascular-neurodegenerative phenotypes, as an adjunct to target-engaged therapies, or in carefully selected subpopulations - none of which has yet been proven.

Blood-Based Biomarkers and Their Impact on GLP-1 and Other Therapies

A significant breakthrough for prevention strategies is the advancement of blood biomarkers. In May 2025, the FDA approved the first blood test for aiding Alzheimer’s diagnosis: the Lumipulse G p-tau217/beta-amyloid 1-42 plasma ratio, for symptomatic adults as an adjunct, not a screening test or stand-alone diagnostic.

For trials, this development reduces logistical barriers for identifying biomarker-appropriate participants earlier, at scale, and potentially in more diverse settings compared to traditional CSF/PET-dependent recruitment. Clinically, it shifts practice towards using blood biomarkers to guide more expensive or invasive confirmatory testing.

For GLP-1RAs, blood biomarkers are a double-edged sword. They could facilitate prevention trials in earlier biological stages, but they also challenge reliance on "soft" cognitive endpoints alone, as biomarker trajectories can be more precisely tracked, and null clinical results will increasingly be contextualized by the degree to which a therapy engages the targeted biology.

Clinical Implications for Physicians

Based on current Phase 3 outcomes, semaglutide should not be considered a disease-modifying treatment for symptomatic, amyloid-positive Alzheimer's disease.

However, GLP-1RAs remain significant in dementia care as they address cardiometabolic and vascular risks, which substantially influence cognitive trajectories in real patients. The practical challenge is distinguishing two conversations that patients often conflate: whether GLP-1RAs are suitable for diabetes, obesity, or cardiovascular risk in someone with cognitive impairment, versus whether GLP-1RAs treat Alzheimer’s itself. EVOKE answers the latter negatively for the tested population and timeframe.

If a patient with MCI or early dementia is a strong candidate for a GLP-1RA due to cardiometabolic reasons, the clinician's role is risk management in a more vulnerable physiology, including monitoring weight trajectory, nutrition, sarcopenia/frailty, dehydration, and tolerability, with clear expectations that the medication is used for metabolic/vascular benefits, not as a proven Alzheimer’s therapy.

The Way Forward: What a "GLP-1 in Dementia" Program Might Become

The most plausible next step is not to "repeat EVOKE," but to focus on tighter biological targeting: prevention-stage designs, mixed pathology enrichment, combination approaches with anti-amyloid or anti-tau strategies, and hypotheses about biomarker-defined responders that can be tested swiftly. EVOKE/EVOKE+ also highlight the importance of managing unintended effects in older adults, like weight loss and appetite suppression, when the brain outcome benefit is uncertain.

For clinicians, the immediate value lies in maintaining conceptual clarity: GLP-1RAs remain promising dementia-risk modifiers at the population level and may eventually find a place in prevention or combination paradigms. However, semaglutide's Phase 3 results argue against using GLP-1RAs as stand-alone disease-modifying therapies for symptomatic Alzheimer's.

References

About

Championing neurocognitive health and wellness within the Gulf Coast community through the forefront of precision medicine, innovative solutions, and dedicated community advocacy.

Company
About us
Our Mission
Our Team
Copyright © Mind the Gulf 2019-2024. All rights reserved.