Across 7 randomized controlled trials enrolling 50,617 adults with type 2 diabetes at high cardiovascular risk, GLP-1 receptor agonists (GLP-1RAs) — the drug class including semaglutide and liraglutide — reduced stroke risk by 17% (pooled HR 0.83; 95% CI 0.75–0.92). Meta-regression revealed that blood pressure reduction explained only ~4 percentage points of that benefit despite a statistically significant association with systolic BP changes (β = 0.371; p = 0.043). HbA1c and body weight reductions showed no significant association with stroke protection. In four trials covering 4,206 patients with established neurodegenerative disease, no cognitive benefit was detected.
This finding is mechanistically provocative. Roughly three-quarters of the observed stroke reduction remains unexplained by the classic cardiometabolic risk factors GLP-1RAs are known to improve. Candidate mechanisms — anti-inflammatory effects, direct endothelial action, atrial remodeling, or plaque stabilization — now demand rigorous investigation. The null cognitive signal is equally important: it suggests stroke prevention may not translate to neurodegeneration benefit, possibly because established disease is too advanced for intervention, or because distinct pathways govern each outcome.
Limitations are real. Trial-level meta-regression cannot establish individual-level causality, and the cognitive trials were small and heterogeneous. Crucially, this is a preprint posted on medRxiv and has not yet undergone peer review — pooled estimates and mechanistic inferences could shift. Still, the stroke finding across 50,000+ patients is robust enough to be considered confirmatory; the mechanism gap makes this analysis genuinely paradigm-nudging for GLP-1 cardiovascular research.
