GLP-1 Receptor Agonists, Stroke, and Cognitive Outcomes: Systematic Review and Meta-Regression
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Importance
Randomised-controlled trials (RCTs) suggest glucagon-like-peptide-1 receptor agonists (GLP-1RAs) reduce the risk of stroke in type 2 diabetes (T2DM); however, the mechanisms remain unclear. If benefits were driven principally by blood pressure (BP) lowering, similar effects on cognitive impairment may be expected. Evaluating data from trials with robust ascertainment and prespecified stroke and/or cognitive outcomes may provide greater clarity.
Objective
To confirm the effect of GLP-1RAs on stroke risk in RCTs with prespecified, adjudicated stroke endpoints; to explore whether risk reduction is associated with trial-level changes in systolic BP (SBP), glycated haemoglobin (HbA1c), or body weight; to establish whether a parallel cognitive signal exists.
Data Sources
PubMed, Cochrane, Embase, and CINAHL, from inception to August 2026 (PROSPERO CRD420261305938).
Study Selection
RCTs comparing GLP-1RAs with placebo/usual care in adults, with ≥12 months follow-up, and prespecified stroke and/or cognitive outcomes.
Data Extraction and Synthesis
Hazard ratios (HRs) for stroke were pooled using inverse-variance random-effect meta-analysis. Univariable meta-regression related logHR for stroke to trial-level between-group differences in SBP, HbA1c and weight, with permutation-based inference and leave-one-out sensitivity analyses. The pooled effect was benchmarked against that predicted from achieved SBP differences. Heterogeneous cognitive outcomes were synthesized narratively.
Main Outcomes and Measures
Adjudicated stroke events and prespecified cognitive endpoints.
Results
Among 7 RCTs (n=50,617), GLP-1RA therapy was associated with 17% lower stroke risk (pooled HR 0.83; [95% CI 0.75-0.92]; p<.001). Meta regression showed an association with SBP (β = 0.371; [95% CI 0.013-0.729]; p=.043), but not HbA1c (0.432; [-0.064-0.928]; p=.088) or weight (0.109; [-0.038-0.256]; p=.148). The mean between-group SBP difference (∼1.45 mmHg) predicts a stroke reduction of roughly 4%, versus the 17% observed, leaving approximately three quarters of the log-effect unexplained by BP lowering. Four trials (n=4,206) in established neurodegenerative disease showed no cognitive benefit.
Conclusions and Relevance
GLP-1RA therapy reduces stroke risk by 17% among individuals with T2DM at high cardiovascular risk, a benefit not fully accounted for by conventional cardiometabolic risk factor change. Additional mechanisms warrant evaluation. The absence of cognitive signal suggests the benefit may have cerebrovascular origin, but whether GLP-1RAs prevent vascular cognitive impairment in cardiometabolic populations remains untested.
