GLP-1s and Dementia: 10 Studies, and the Trial That Settles It Is Not Out Yet

The only completed randomized trial of a GLP-1 in Alzheimer's: 204 participants with mild-to-moderate disease and no diabetes, on daily liraglutide or placebo for a year across UK centers. ⛔ The primary endpoint — brain glucose metabolism on PET — showed no difference. A secondary cognitive measure, the executive-function domain of the standard Alzheimer's scale, favored liraglutide (0.15, CI 0.03 to 0.28), while daily-function and overall dementia-severity scores did not move. ⚠ That p-value is unadjusted and the outcome is secondary, which is exactly the shape of result that fails to replicate. It was enough to justify the phase 3 program, and it is not a positive trial.

PMID 41326666 ↗DOI 10.1038/s41591-025-04106-7 ↗

Not a result — the enrollment description for the two trials that will actually answer this, testing oral semaglutide 14 mg against placebo in roughly 1,840 people and 1,960 people respectively, all with early symptomatic Alzheimer's. It documents who got in: mean age 72, about half carrying the highest-risk genetic variant, most already on standard Alzheimer's therapy. ★ Worth knowing this exists so that when the topline lands in late 2026 you can see whether the population it enrolled resembles the patient in front of you.

PMID 41522368 ↗NCT04777396 ↗DOI 10.1002/trc2.70200 ↗

The first randomized hint, from an exploratory analysis of a cardiovascular trial that followed 9,901 people with type 2 diabetes for a median 5.4 years. Substantial cognitive impairment occurred in 4.05% on dulaglutide against 4.35% on placebo — a 14% relative reduction that was statistically solid in a dataset this large. ⚠ Note the absolute difference: three tenths of a percentage point. It is the largest randomized cognitive dataset in the class and the effect it describes is small.

PMID 32562683 ↗NCT01394952 ↗DOI 10.1016/S1474-4422(20)30173-3 ↗

The study that launched the current wave: more than a million patients with type 2 diabetes across 60 US health systems, with semaglutide compared against seven different alternatives one at a time. First-time Alzheimer's diagnoses ran 40 to 70% lower, with the biggest gap against insulin. ⛔ That last detail is the tell — people started on insulin are sicker and further along, and the largest apparent benefit appearing against the sickest comparator is what confounding looks like. Hypothesis-generating, and it generated the phase 3 program.

PMID 39445596 ↗DOI 10.1002/alz.14313 ↗

The same group extending the outcome from Alzheimer's specifically to dementia of all kinds, including vascular and Lewy body forms, with hazard ratios of 0.40 to 0.70 holding across age, sex and weight strata. ★ The interesting wrinkle is that vascular and mixed dementia show a stronger signal than pure Alzheimer's does — which fits a boring, plausible mechanism (better blood vessels) rather than a direct effect on amyloid, and would make it a side effect of cardiovascular benefit rather than a new one.

PMID 40552638 ↗DOI 10.1177/13872877251351329 ↗

The practical safety question behind the phase 3 trials: is this drug tolerable in the people who would take it for dementia? Pooled across 14 trials and 8,464 participants aged 65 and over, including 1,463 over 75, serious adverse events came out even (14.9% against 14.5%), gastrointestinal effects were far more common (43% against 25%) but rarely caused anyone to stop, and nothing signaled trouble on falls, fractures or cognitive worsening. ★ The falls-and-fractures result is the one that matters most in this population.

PMID 40337158 ↗DOI 10.1002/trc2.70076 ↗

Three simulated head-to-head comparisons run inside a global records network: both GLP-1 drugs against SGLT2 inhibitors, and tirzepatide against semaglutide. Both beat the SGLT2 comparator on dementia incidence over three years; tirzepatide edged semaglutide without reaching significance. ★ The authors flag the subtlety that makes this useful — SGLT2 inhibitors have themselves looked protective in earlier work, so beating them is a harder test than beating insulin.

PMID 41544899 ↗DOI 10.1016/j.diabres.2026.113083 ↗

A direct tirzepatide-against-semaglutide comparison across roughly 50,000 matched pairs with type 2 diabetes followed up to three years, with tirzepatide associated with a modestly lower combined risk of mild cognitive impairment, dementia or Alzheimer's — a hazard ratio near 0.85. ⚠ Two drugs, same database, same design, and therefore the same problem: whatever distinguishes people prescribed one from people prescribed the other is not in the record. Consistent with the pattern that the dual-receptor drug does slightly more of everything.

PMID 41825212 ↗DOI 10.1016/j.jdiacomp.2026.109306 ↗

A study restricted to patients whose diabetes had already damaged their kidneys — a group whose dementia risk runs about double that of diabetes alone. GLP-1 users showed 35 to 45% lower hazards of dementia and Alzheimer's than matched non-users, with the effect larger in more advanced kidney disease. ★ That gradient — more benefit where there is more vascular damage — is the same pattern the all-dementia analysis found, and it points the same way: this may be a circulation story rather than a brain-specific one.

PMID 41697144 ↗DOI 10.1093/ndt/gfag032 ↗

The pooled randomized evidence in people with cognitive impairment but not diabetes, which amounts to roughly 350 patients across the liraglutide trial and a handful of small single-center studies. Cognitive scores improved modestly against placebo, with a standardized effect around 0.25. ⚠ Pooling a few small trials produces a number, not a finding. Its honest use is as a marker of where the evidence stood before the phase 3 trials report — worth revisiting once they do.

PMID 41524953 ↗DOI 10.1007/s10072-025-08602-z ↗