rethinkPeptides Search
Menu
Study breakdown

Diabetes Drugs Compared for Dementia Risk: SGLT2 Inhibitors Show Strongest Brain Protection

CohortStrong evidence
The takeaway

SGLT2 inhibitors were linked to a 14–30% lower dementia risk compared to DPP4 inhibitors in older adults with type 2 diabetes, while GLP-1 receptor agonists showed benefits only with continuous use.

30% lower dementia risk

Continuous SGLT2 inhibitor use was associated with a 30% reduction in dementia risk compared to DPP4 inhibitors in adults 60+ with type 2 diabetes

What the researchers found

In this large UK cohort of adults aged 60+ with type 2 diabetes, SGLT2 inhibitors were associated with a 14% lower dementia risk compared to DPP4 inhibitors in the intention-to-treat analysis (HR 0.86, 95% CI 0.79–0.94), increasing to a 30% reduction with continuous use (HR 0.70, 95% CI 0.60–0.82). GLP-1 receptor agonists showed no significant difference from DPP4 inhibitors in the intention-to-treat analysis (HR 0.95, 95% CI 0.87–1.04), but continuous GLP-1RA use was associated with a 21% lower risk (HR 0.79, 95% CI 0.64–0.97).

GLP-1RAs and SGLT2 inhibitors showed comparable dementia risk in head-to-head comparison (HR 0.98, 95% CI 0.87–1.11).

Why it matters

With the global rise of both type 2 diabetes and dementia, identifying diabetes drugs that also protect brain health could be transformative. This study provides real-world evidence that SGLT2 inhibitors — and possibly GLP-1 receptor agonists with sustained use — may reduce dementia risk in older adults with diabetes. These findings could influence treatment selection for millions of patients, especially since these drugs are already widely prescribed.

The numbers in context

n=13,965 GLP1-RA vs DPP4i pairs · n=25,533 SGLT2i vs DPP4i pairs · n=14,214 GLP1-RA vs SGLT2i pairs · SGLT2i vs DPP4i ITT: HR 0.86 (0.79–0.94) · SGLT2i continuous use: HR 0.70 (0.60–0.82) · GLP1-RA continuous use vs DPP4i: HR 0.79 (0.64–0.97) · Mean follow-up: 4.9–6.5 years

How the study worked

This was a target trial emulation cohort study using UK Clinical Practice Research Datalink electronic health records. Three pairwise comparisons were made between new users of GLP-1RAs, SGLT2 inhibitors, and DPP4 inhibitors in adults aged ≥60 with type 2 diabetes and no pre-existing cognitive impairment. Propensity score overlap weighting was used to balance groups. Both intention-to-treat and as-treated (continuous use) analyses were performed. The primary outcome was incident all-cause dementia.

Who was studied

UK adults aged ≥60 with type 2 diabetes and no pre-existing cognitive impairment

What this study cannot tell us

This is an observational study, so it cannot prove causation — unmeasured confounders may explain the associations. The as-treated analysis showing stronger effects with continuous use had shorter follow-up (~2.4–2.7 years) and may be affected by adherence bias. DPP4 inhibitors served as the comparator, but they may also have neuroprotective effects, potentially underestimating the benefits of the other drugs. The study used UK data, which may not generalize globally.

How to read the evidence

This is a large, well-designed population-based cohort study using a target trial emulation framework and propensity score weighting to reduce confounding. The large sample size (53,000+ patient pairs), long follow-up (up to 6.5 years), and multiple sensitivity analyses support strong evidence, though as an observational study it cannot establish causation.

When this study was published

Published in 2025, this study uses UK health records through 2022 and reflects the most current real-world evidence on diabetes drug classes and dementia risk.

The bigger picture

The link between type 2 diabetes and dementia is well established — diabetic patients face roughly double the risk. As the search for Alzheimer's treatments continues to disappoint, the idea that existing diabetes drugs might protect brain health has gained significant traction. This study adds to growing evidence that metabolic drugs — particularly SGLT2 inhibitors and GLP-1RAs — may have neuroprotective effects beyond glucose control. If confirmed in randomized trials, these findings could reshape how clinicians choose diabetes medications for older patients, factoring cognitive protection into prescribing decisions.

Questions still open

  • Would a randomized clinical trial of SGLT2 inhibitors specifically for dementia prevention confirm the observational findings?
  • What biological mechanisms might explain the neuroprotective effects of SGLT2 inhibitors and GLP-1 receptor agonists?
  • Does the dementia risk reduction increase with longer durations of continuous use beyond the 2–3 year follow-up in the as-treated analysis?

Common questions

Does this mean GLP-1 drugs like semaglutide prevent dementia?
Not definitively. This observational study found that continuous GLP-1RA use was associated with a 21% lower dementia risk, but the overall intention-to-treat result was not statistically significant. Observational studies can identify associations but can't prove cause and effect. Randomized clinical trials would be needed to confirm a true protective effect.
Why might diabetes drugs protect against dementia?
Several theories exist. Diabetes damages blood vessels in the brain and promotes inflammation, both of which contribute to dementia. SGLT2 inhibitors and GLP-1RAs improve cardiovascular health, reduce inflammation, and may improve brain blood flow. GLP-1 receptors also exist in the brain, where they may directly support neuron survival and function.

Read the original research

Comparative dementia risk with GLP1 receptor agonists, SGLT2 inhibitors, or DPP4 inhibitors: a population-based cohort study.

Alzheimer's research & therapy, 17(1), 269

Citation

Wu, Che-Yuan; Alkabbani, Wajd; Shah, Baiju R; Kapral, Moira K; Edwards, Jodi D; Maxwell, Colleen J; Swardfager, Walter. (2025). Comparative dementia risk with GLP1 receptor agonists, SGLT2 inhibitors, or DPP4 inhibitors: a population-based cohort study.. Alzheimer's research & therapy, 17(1), 269. https://doi.org/10.1186/s13195-025-01929-x