GLP-1 Microdosing and Alzheimer's: What the EVOKE Trials, ELAD and the Dementia Cohort Data Show

For several years, GLP-1 medicines looked like a possible answer to one of medicine's hardest problems. Records from people with type 2 diabetes kept showing fewer dementia diagnoses among those on semaglutide and similar drugs. Animal work suggested the drugs calmed inflammation in the brain. That combination produced headlines, and it also produced a common question from people on low-dose protocols: is a GLP-1 microdose quietly protecting the brain too?

This article walks through what GLP-1 medicines and Alzheimer's disease look like in the research now that the big trials have reported. It covers the evoke and evoke+ semaglutide trials, the ELAD liraglutide trial, the observational data that started the excitement, and what any of it means for a low dose. The short version is that the randomised evidence is negative so far, and the low-dose evidence does not exist.

Does semaglutide slow Alzheimer's disease?

Semaglutide did not slow Alzheimer's disease in the two largest trials ever run on the question. Cummings and colleagues reported evoke and evoke+ in The Lancet (2026). The trials ran at 566 sites in 40 countries and randomised 3808 people aged 55-85 with amyloid-confirmed mild cognitive impairment or mild dementia due to Alzheimer's. Half took oral semaglutide up to 14 mg once daily and half took placebo, for up to 156 weeks.

The primary measure was the Clinical Dementia Rating-Sum of Boxes (CDR-SB), a clinician-scored scale of memory, judgement and daily function. At baseline the average participant was 72.2 years old with a CDR-SB of 3.7. Over 104 weeks, scores worsened by 2.3 points on semaglutide and 2.3 on placebo in evoke, and by 2.2 versus 2.1 in evoke+. The estimated differences were -0.08 (p=0.57) and 0.10 (p=0.46). Both trials were discontinued for a negative clinical outcome, and the authors wrote plainly that oral semaglutide was not efficacious in slowing clinical progression.

Why did people expect GLP-1 drugs to protect the brain?

The expectation came from people with type 2 diabetes, where GLP-1 users kept showing fewer dementia diagnoses. Nørgaard and colleagues (Alzheimer's & Dementia: TRCI, 2022) pooled three double-blind cardiovascular outcome trials covering 15,820 patients and found a lower dementia rate on GLP-1 drugs than on placebo, a hazard ratio of 0.47. In a Danish registry of 120,054 patients, each extra year of GLP-1 exposure went with a hazard ratio of 0.89.

The US record studies were larger still. Wang and colleagues (Alzheimer's & Dementia, 2024) emulated target trials in 1,094,761 people with type 2 diabetes and found semaglutide linked to fewer first Alzheimer's diagnoses over 3 years, with a hazard ratio of 0.33 against insulin and 0.59 against other GLP-1 drugs. Lin and colleagues (JAMA Network Open, 2025) matched 60,860 adults with diabetes and obesity and found a dementia hazard ratio of 0.63 for semaglutide or tirzepatide against other diabetes drugs. For a wider look at these signals, see our piece on GLP-1s, inflammation and neurodegeneration.

Why did the trials fail when the cohort data looked so good?

The two kinds of evidence asked different questions in different people. The cohort studies followed people with type 2 diabetes who had no dementia diagnosis and counted new cases. Evoke treated people who already had amyloid in the brain and early symptoms, and measured whether decline slowed. A drug can plausibly lower the chance of a diagnosis in people with diabetes, through better blood sugar, blood pressure or vascular health, without changing the course of Alzheimer's once it is under way.

Observational data also carry a known weakness. Hui and colleagues (Journal of Alzheimer's Disease, 2025) reviewed 42 studies of diabetes drugs and dementia. GLP-1 drugs were linked to a 53% lower dementia risk in three randomised trials and 27% lower risk in three case-control studies, but the authors noted that most studies were observational and limited by confounding by indication. People prescribed a newer drug often differ from people on older ones in ways the records do not capture. The three randomised trials were heart trials, where dementia was not the main outcome either.

What did the ELAD liraglutide trial find?

ELAD also missed its primary outcome, though one secondary score leaned in liraglutide's favour. Edison and colleagues published the phase 2b trial in Nature Medicine (2026). It enrolled 204 people with mild to moderate Alzheimer's and no diabetes, who took daily liraglutide injections or placebo for 52 weeks.

The primary outcome, brain glucose use measured by PET scan, did not differ between groups (difference -0.17, P = 0.14). Daily function on the ADCS-ADL (-0.58, P = 0.65) and the CDR-SoB (-0.06, P = 0.81) did not move either. Only the executive domain of the ADAS scale came out better on liraglutide (0.15, unadjusted P = 0.01). One positive secondary result among several null ones is a reason for a follow-up trial, not evidence that the drug works.

What does the pooled trial evidence say now?

Pooled across trials, GLP-1 drugs have not improved cognition or slowed Alzheimer's. Siqueira and colleagues (Alzheimer's & Dementia: TRCI, 2026) combined seven randomised trials with 4132 participants in Alzheimer's disease and mild cognitive impairment. The effect on cognition was a standardised mean difference of -0.05 and on the CDR-SB a mean difference of 0.09, neither significant.

The same meta-analysis found more adverse events with GLP-1 drugs (risk ratio 1.07) and, as expected, far more weight loss (risk ratio 5.01). Biomarker and brain-imaging results were inconsistent and did not translate into clinical benefit. The authors suggested future work should focus on metabolically selected patients or on GLP-1 drugs that cross into the brain more easily.

Were side effects higher on semaglutide in the Alzheimer's trials?

Side effects were somewhat more common on semaglutide, and the pattern matched its use elsewhere. Across evoke and evoke+, 91.2% of people on semaglutide (1729 of 1896) reported a treatment-emergent adverse event, against 84.8% on placebo (1613 of 1902). The authors judged safety and tolerability consistent with studies in other indications.

High rates in both arms are normal in long trials of older adults, because almost any new symptom over two or three years gets recorded. For older people, the weight loss that comes with a GLP-1 medicine is a consideration of its own. Our guide to GLP-1 microdosing after 65 covers muscle, falls and hydration.

Can a GLP-1 microdose protect against dementia?

Nothing in the research supports that idea. No trial has tested sub-label GLP-1 doses for dementia prevention or for Alzheimer's treatment. The closest thing to a low dose in this field is the start of the evoke titration, described in the trial design paper by Cummings and colleagues (Alzheimer's Research & Therapy, 2025): 3 mg of oral semaglutide for weeks 0-4, 7 mg for weeks 4-8, then 14 mg from week 8 onward.

Those first weeks were a ramp to the target dose, not a dose being tested, and the full 14 mg dose itself failed. If a full dose given for two years did not slow decline in early Alzheimer's, there is no basis for expecting a smaller one to do better. The cohort signals that excited people came from standard diabetes doses in people with type 2 diabetes, so they do not speak to microdoses in people without it.

What a low dose can reasonably target is the metabolic side: weight, blood sugar and blood pressure, each of which has its own evidence base. Our article on GLP-1 microdosing and prediabetes looks at what the trials show on the glucose side.

What is still open in GLP-1 and Alzheimer's research?

Prevention remains an open question, while treatment of established early Alzheimer's with semaglutide has a clear negative answer. The remaining questions are narrower than the early headlines suggested:

• Whether GLP-1 drugs lower the chance of developing dementia in people with type 2 diabetes. The pooled heart-trial result (hazard ratio 0.47) is suggestive, but dementia was not what those trials were built to measure.
• Whether tirzepatide behaves differently. The only data are observational, such as the 0.63 hazard ratio in the JAMA Network Open cohort, which pooled it with semaglutide.
• Whether drugs with better brain penetration, or patients selected for metabolic disease, would show a benefit, as the 2026 meta-analysis authors proposed.
• Whether ELAD's executive function signal holds up in a larger trial designed around it.

Frequently asked questions

Clinical references

1. Cummings JL, Atri A, Sano M, et al. 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. 2026. doi:10.1016/S0140-6736(26)00459-9
2. Cummings JL, Atri A, Feldman HH, et al. evoke and evoke+: design of two large-scale, double-blind, placebo-controlled, phase 3 studies evaluating efficacy, safety, and tolerability of semaglutide in early-stage symptomatic Alzheimer's disease. Alzheimer's Research & Therapy. 2025. doi:10.1186/s13195-024-01666-7
3. Edison P, Femminella GD, Ritchie C, et al. Liraglutide in mild to moderate Alzheimer's disease: a phase 2b clinical trial. Nature Medicine. 2026. doi:10.1038/s41591-025-04106-7
4. Nørgaard CH, Friedrich S, Hansen CT, et al. Treatment with glucagon-like peptide-1 receptor agonists and incidence of dementia: Data from pooled double-blind randomized controlled trials and nationwide disease and prescription registers. Alzheimer's & Dementia: Translational Research & Clinical Interventions. 2022. doi:10.1002/trc2.12268
5. Wang W, Wang Q, Qi X, et al. Associations of semaglutide with first-time diagnosis of Alzheimer's disease in patients with type 2 diabetes: Target trial emulation using nationwide real-world data in the US. Alzheimer's & Dementia. 2024. doi:10.1002/alz.14313
6. Lin HT, Tsai YF, Liao PL, Wei JC. Neurodegeneration and Stroke After Semaglutide and Tirzepatide in Patients With Diabetes and Obesity. JAMA Network Open. 2025. doi:10.1001/jamanetworkopen.2025.21016
7. Hui EK, Mukadam N, Kohl G, Livingston G. Effect of diabetes medications on the risk of developing dementia, mild cognitive impairment, or cognitive decline: A systematic review and meta-analysis. Journal of Alzheimer's Disease. 2025. doi:10.1177/13872877251319054
8. Siqueira RGP, Ranalli C, Yılmaz D, et al. GLP-1 receptor agonists in Alzheimer's disease and MCI: a meta-analysis. Alzheimer's & Dementia: Translational Research & Clinical Interventions. 2026. doi:10.1002/trc2.70334