Dementia progression with GLP-1 receptor agonists in people with mild cognitive impairment and type 2 diabetes: target-trial emulation study.

INTRODUCTION: In people with type 2 diabetes (T2D) without dementia, glucagon-like peptide-1 receptor agonists (GLP-1 RAs) improved cardiovascular, kidney, and mortality outcomes, with multiple real-world studies suggesting reduced risk of dementia onset. However, the EVOKE/EVOKE+ randomized-controlled trials (RCTs) found that oral semaglutide did not slow cognitive decline in individuals with early Alzheimer's disease (AD), most (86%) without T2D. It is unclear whether GLP-1 RA therapy influences progression to dementia in people with both T2D and mild cognitive impairment (MCI). METHODS: In a target-trial emulation hypothesis-generating study using TriNetX global collaborative network data, we compared incident dementia (including AD) in adults with MCI and T2D without dementia, who were new users of GLP-1 RAs versus dipeptidyl peptidase-4 inhibitors (DPP4i). RESULTS: We included 1320 propensity-score-matched (1:1) individuals who initiated GLP-1 RAs or DPP4i in the period 2010 to 2021 (702 women; mean age 67.2 years). Over up to 5 years of follow-up, incident dementia occurred in 101 and 132 participants initiating GLP-1 RAs and DPP4i, respectively (Cox proportional hazard ratio 0.74 [95% confidence interval [CI]: 0.57 to 0.95]). At a mean follow-up of 3.9 years, the respective Kaplan-Meier-derived cumulative incidences were 15.5% and 20.4% (absolute difference -4.9% [95% CI -9.4 to -0.4], number needed to treat of 21 [11 to 234] individuals to prevent one event). Various sensitivity analyses supported the primary analysis: comparing GLP-1 RAs with basal insulin, using alternative outcome definitions, repeating the analyses across baseline subgroups, and using a negative control outcome. CONCLUSION: In this hypothesis-generating study, initiation of GLP-1 RAs versus DPP4i in people with T2D and MCI without dementia was associated with a lower risk of incident dementia.