Failed Semaglutide for Early Alzheimer’s: Why the Story May Not Be Over

Detailed findings from the phase 3 EVOKE and EVOKE+ trials evaluating the oral GLP-1 receptor agonist semaglutide (Ozempic, Novo Nordisk) for early Alzheimer’s disease (AD) were presented at the 18th Clinical Trials on Alzheimer’s Disease (CTAD) Conference on December 3. Although topline results released earlier were disappointing — showing no cognitive or functional benefit — the fuller dataset, especially biomarker outcomes, offers important insights and potential avenues for future research.

No Cognitive or Functional Benefit, Despite Biological Effects

Principal investigator Jeffrey L. Cummings, MD (University of Nevada, Las Vegas), reported that semaglutide significantly reduced plasma high-sensitivity C-reactive protein (CRP), a marker of systemic inflammation. This outcome aligned with the trial’s underlying biological hypothesis: that suppressing peripheral inflammation might translate into cognitive improvement.
However, this effect failed to produce measurable cognitive or functional gains, and disease progression was not slowed.

Across both EVOKE trials, semaglutide improved several Alzheimer’s-related biomarkers, but the magnitude of change was not large enough to influence clinical outcomes. As Howard Fillit, MD, co-founder and chief science officer of the Alzheimer’s Drug Discovery Foundation, noted, the biomarker reductions — up to 10% across markers of neuroinflammation and AD pathology — were statistically significant but not clinically meaningful. He emphasized that while the trials did not meet primary goals, the studies still provide valuable biological signals that could inform future preventive strategies, especially given the broader interest in GLP-1 receptor agonists as metabolic and neuroinflammatory modulators.

Why the Hypothesis Made Sense

In his keynote address, Peter Johannsen, PhD (Novo Nordisk) explained that in 2020, when the EVOKE program was initiated, evidence supporting GLP-1 agonists in AD was compelling enough to justify a large-scale clinical investigation.

Preclinical animal studies had suggested GLP-1 agonists could:

• Improve performance in memory and maze-based tasks

• Reduce amyloid plaque accumulation

• Lower soluble amyloid-β levels

• Decrease neuroinflammatory signaling and oxidative stress

• Preserve synaptic markers and neuronal density

Human observational data had also been encouraging. Several large studies showed a reduced incidence of dementia among patients with type 2 diabetes treated with GLP-1 receptor agonists. A pooled analysis from three cardiovascular outcomes trials indicated a 53% reduction in all-cause dementia rates compared with placebo.

Even more strikingly, a target-trial emulation using real-world US data from TriNetX found that semaglutide users had 40%–70% lower risks of developing Alzheimer’s disease for the first time compared with patients on other diabetes medications — including other GLP-1 receptor agonists.

With this backdrop, the move to test semaglutide directly in early AD was scientifically reasonable, even if the final outcome fell short of expectations.

Trial Design and Primary Outcome

The EVOKE and EVOKE+ trials were randomized, double-blind, placebo-controlled studies enrolling 3808 adults aged 55–85 with:

• Mild cognitive impairment (MCI) due to AD, or

• Mild AD dementia

All participants had confirmed amyloid-positive scans.
They were randomized to receive oral semaglutide 14 mg once daily or placebo for 156 weeks (104 weeks of main treatment plus a 52-week extension).

After two years of treatment:

• EVOKE: drug-placebo difference on CDR-Sum of Boxes = 0.06 points (P = .7)

• EVOKE+: difference = 0.15 points (P = .4)

Because the primary endpoint was not met, no formal testing of secondary outcomes proceeded under the hierarchical analysis plan. Descriptive analyses across cognitive scales — including ADAS-Cog, MoCA, MMSE, Alzheimer’s Disease Composite Score, and functional measures — showed no differences between semaglutide and placebo.

A pooled analysis of progression from MCI to AD dementia also showed no delay, with a hazard ratio of 0.96, confirming lack of clinical benefit.

Encouraging Biomarker Trends, but Too Small to Matter Clinically

Despite negative clinical outcomes, semaglutide produced measurable effects on biomarkers. Across CSF assays, semaglutide was associated with up to 10% reductions in:

• pTau181

• pTau217

• npTau181

• npTau205

• YKL-40 (neuroinflammation marker)

• Total tau and neurogranin (neurodegeneration markers)

These shifts were statistically significant but not large enough to modify disease trajectory.

Conversely, several blood-based biomarkers increased modestly:

• Neurofilament light (NfL): +5% in EVOKE+

• Glial fibrillary acidic protein (GFAP): +4% in both trials

Significant reductions in hs-CRP — treatment ratios 0.76 (EVOKE) and 0.71 (EVOKE+) — confirmed semaglutide’s anti-inflammatory effect.

During a Q&A session, investigators addressed the mismatch between biomarker changes and absent clinical benefit. Johannsen explained that although the biomarker shifts were real, they were not of sufficient magnitude. For comparison, disease-modifying antibody therapies such as lecanemab and donanemab typically produce biomarker reductions of ~30%, which correlate with meaningful clinical slowing of decline.

Johannsen suggested that future phase 2 trials in AD should consider a minimum biomarker-change threshold — potentially 20%–30%, rather than 10% — before advancing to expensive phase 3 programs.

Lessons Learned and Implications for Future Research

While the results were disappointing, experts stressed that the EVOKE program still provides major scientific value. As Fillit noted, “Even negative trials move the field forward.” Early anti-amyloid trials also failed, yet they ultimately paved the way for successful therapies now on the market.

The EVOKE trials were robust, long-duration, and rigorously conducted, and they offer key lessons for how to design future studies exploring metabolic or inflammatory pathways in AD. Biomarker signals — even modest ones — help shape hypotheses for combination approaches, repurposing strategies, and preventive trials.

Fillit emphasized that oncology has long embraced this model: negative or mixed trials often inspire better-designed follow-up studies and innovative therapeutic combinations. Alzheimer’s research, he argued, should continue on a similar path, using mechanistic findings and biomarker insights to guide the next generation of interventions.

 Source

Failed Semaglutide for Early Alzheimer’s Not the End of the Road? - Medscape - December 05, 2025.