GLP-1 Receptor Agonists, Stroke, and Cognitive Outcomes: Systematic Review and Meta-Regression

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 (? = 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.