People with type 2 diabetes mellitus (DM2) have an elevated risk of developing Alzheimer disease (AD) and other dementias. Pharmacologic agents developed for glycemic control—specifically GLP‑1 receptor agonists (GLP1RAs) and SGLT2 inhibitors (SGLT2i)—have been proposed as interventions that might modify dementia risk and also provide mechanistic insights into neurodegeneration. The authors conducted a systematic review and meta‑analysis to compare the association of GLP1RAs and SGLT2i with incident AD and dementia in adults with DM2 who had no cognitive impairment at baseline.
The review searched PubMed, Embase, and the Cochrane Library for randomized controlled trials, cohort studies, and case‑control studies published between January 1, 2006, and April 15, 2025. Eligible studies compared GLP1RAs and/or SGLT2i to other glucose‑lowering therapies in adult populations with DM2 and without baseline cognitive impairment. From this search, the authors identified 12 studies that met inclusion criteria: 1 randomized controlled trial (RCT) and 11 cohort studies.
Included evidence consisted predominantly of observational cohort studies with one RCT. The prespecified primary outcome across studies was incident Alzheimer disease and dementia, most commonly reported as adjusted hazard ratios (HRs). Secondary measures reported in some studies included glycemic indices such as HbA1c and analyses exploring the relationship between glycemic control and cognitive outcomes. The authors performed regression analyses to assess potential explanatory variables for observed effects.
Pooled results reported by the authors indicated that use of GLP1RAs and SGLT2i, when compared with other glucose‑lowering therapies, was associated with a lower incidence of AD and dementia. The summary estimates supplied in the preprint were:
Incident Alzheimer disease: pooled HR = 0.72 (95% CI, 0.59 to 0.87)
Incident dementia: pooled HR = 0.84 (95% CI, 0.76 to 0.92)
The authors note moderate to high heterogeneity across studies and judged the overall certainty of evidence as moderate. The primary outcome measure across included studies was the adjusted hazard ratio, reflecting time‑to‑event analyses controlling for covariates where available.
When directly comparing the two drug classes, the authors report a modest advantage for SGLT2i over GLP1RAs in protection against AD. The preprint provides a comparative estimate reported as HR = 1.07 (95% CI, 1.00 to 1.14; p = 0.0496). The authors present this result as evidence of a small differential effect favoring SGLT2i in the context of the pooled data from included studies.
Secondary analyses examined the role of glycemic control, particularly baseline HbA1c, in modulating the association between these drug classes and AD risk. The authors found that baseline HbA1c was associated with AD risk estimates among SGLT2i users but not among GLP1RAs users. A regression analysis was performed to probe variables that might explain the observed associations, although full regression details and covariates assessed are presented in the source manuscript and supplementary material.
The pooled analyses exhibited moderate to high heterogeneity. The authors rated the certainty of evidence as moderate. Important limitations inherent to the included literature stem from the observational design of most studies (11 of 12), potential residual confounding, and between‑study variability in outcome ascertainment and covariate adjustment. The report is a preprint and has not undergone peer review; the authors explicitly state that findings should not be used to guide clinical practice until validated by further peer‑reviewed research.
The authors conclude that current pooled evidence reinforces the potential of GLP1RAs and SGLT2i as therapeutic options that may slow cognitive decline in people with DM2, with a possibility that effects depend on glycemic control and central nervous system modulation. They report no competing interests. All data used in the study are available from the authors on reasonable request. The manuscript is a medRxiv preprint and has not been certified by peer review; therefore, its findings are preliminary and intended to inform further investigation rather than immediate clinical change.