GLP-1 drugs and Alzheimer's: what the EVOKE trials actually found

The brain hypothesis behind Alzheimer's GLP-1 research

GLP-1 receptors are not only in the gut. They appear in the hippocampus, hypothalamus, and cortex. That discovery prompted a reasonable question: could drugs that activate those receptors slow the neurodegeneration that drives Alzheimer's?

The proposed mechanisms were specific. GLP-1 activation reduces neuroinflammation. It improves glucose metabolism in brain tissue, which matters because Alzheimer's brains show severely impaired glucose uptake decades before symptoms appear. It lowers oxidative stress. Some animal studies suggested it could reduce amyloid accumulation. These were not wild speculations. They were grounded in receptor biology and preclinical findings.

Observational data added to the interest. Pooled analyses from large cardiovascular outcomes trials found that people on liraglutide or semaglutide had lower rates of subsequent dementia diagnosis than controls. The signal was consistent enough to justify a dedicated RCT. That was EVOKE.

What the EVOKE trials were designed to test

Novo Nordisk enrolled 3,808 adults aged 55 to 85 with early-stage symptomatic Alzheimer's disease, confirmed by amyloid biomarkers. Every participant had the biological hallmark of the disease, not just symptoms. Half received once-daily oral semaglutide with doses titrated up to 14mg. Half received placebo. The planned treatment period was two years, followed by a 12-month blinded extension.

This was, by design, the most rigorous test of the GLP-1 brain hypothesis ever run. Topline results appeared in November 2025 and were presented at the Clinical Trials on Alzheimer's Disease (CTAD) conference. Full results published in The Lancet in 2026 after presentation at the AD/PD conference in March.

The result: no difference on either measure that matters

The primary endpoint was the CDR-SB score at week 104. The CDR-SB is the standard clinical scale for tracking dementia severity across six domains: memory, orientation, judgment, community affairs, home activities, and personal care. Higher numbers mean worse function. Both groups entered the trials with similar scores. Both groups worsened at nearly the same rate.

Secondary endpoints, including additional cognitive tests and daily functioning scales, showed no meaningful differences. The 12-month extension was discontinued. There was no efficacy signal to extend.

Biomarkers moved; the disease did not

One detail that complicates a flat "failed" reading: some biomarkers shifted. Reductions of approximately 10% appeared in certain Alzheimer's biomarkers consistent with semaglutide's proposed mechanisms. The drug was doing something biologically. It was not doing nothing.

The problem is that biomarker movement did not translate into clinical benefit within the trial window. This pattern has appeared before in Alzheimer's research. Drugs that moved amyloid levels sometimes failed to preserve cognition. The brain does not respond simply to the removal of one pathological feature. Moving a marker is not the same as moving the disease.

Whether a longer trial or higher dose would have shown clinical benefit is an open question. Novo Nordisk chose to stop rather than find out.

Why oral semaglutide specifically may have been the wrong tool

The drug that failed here matters. EVOKE used oral semaglutide at doses up to 14mg. That is not the same molecule, route, or exposure as injectable semaglutide.

Semaglutide is a large peptide, roughly 4,113 daltons. Large molecules cross the blood-brain barrier poorly. Oral formulations face an additional problem: bioavailability is limited, and systemic exposure is considerably lower than injectable semaglutide at any comparable dose. Brain concentrations for oral semaglutide at 14mg are likely far below what the preclinical neuroprotection literature tested.

A 2025 research review found that neuroprotective effects across the GLP-1 drug class correlate directly with blood-brain barrier penetration ability. Exenatide, a smaller peptide, crosses the BBB more readily and has shown promising signals in Phase 2 Parkinson's trials. Liraglutide penetrates at low but non-zero rates and restored impaired glucose transport in the brains of early Alzheimer's patients in a smaller study.

The EVOKE failure tells us specifically that oral semaglutide at 14mg did not help early Alzheimer's patients over two years. It does not tell us that no GLP-1 drug at any dose or route will. Those are different claims, and conflating them overstates what the trial actually tested.

What connects this to the addiction and reward research

The EVOKE failure sits in an interesting context. GLP-1 receptors in the brain do clearly affect behavior, just not in the way Alzheimer's researchers hoped. The addiction research is the clearest example: a 2025 Lancet meta-analysis of 5.26 million participants found semaglutide users had 50 to 75% lower rates of alcohol use disorder diagnosis. That effect appears to work through dopamine signaling in the nucleus accumbens and ventral tegmental area, brain regions that are well-perfused and accessible to systemic drugs.

Alzheimer's involves different brain structures and different mechanisms. The hippocampus, where early memory loss originates, may be harder to reach with oral semaglutide at this dose. The fact that GLP-1 receptors matter in the reward system does not guarantee they matter in the same way in the degenerating Alzheimer's brain. Same receptor family, different context, different result.

Where the field goes next

Alzheimer's researchers are not walking away from the GLP-1 hypothesis. They are refining it.

The observational signal that started all of this, lower dementia rates in people on GLP-1s in cardiovascular trials, came from populations who did not have Alzheimer's at baseline. That is a prevention question, not a treatment question. EVOKE was a treatment trial. The prevention question has not been answered by these results.

What this means for people on GLP-1s now

If you are taking semaglutide or tirzepatide for obesity, type 2 diabetes, or cardiovascular risk, the EVOKE result is not a reason to stop. Kidney protection, heart disease prevention, and weight loss each have their own large, well-powered trial evidence. None of that is changed by this Alzheimer's finding.

What the EVOKE data removes is the assumption that your GLP-1 is simultaneously protecting your brain from Alzheimer's. That was always a hypothesis, not a proven benefit. It is harder to hold now, at least for oral semaglutide at this dose in established disease.

The GLP-1 receptor matters in the brain. The addiction data shows that clearly. The Alzheimer's question is: does it matter in the right way, with the right drug, at the right exposure, to slow neurodegeneration? EVOKE gave one clean answer to one version of that question. Several others remain open. That is where the research goes next.

For context on where the next generation of GLP-1 candidates stands, including the triple-agonist retatrutide and its Phase 3 data, the pipeline beyond semaglutide is moving quickly on multiple indications. The brain is one of them, even after this setback.