Comparison on cognitive outcomes of antidiabetic agents for type 2 diabetes: A systematic review and network meta-analysis

We aimed to summarise current evidence on different antidiabetic drugs to delay cognitive impairment, including mild cognitive impairment, dementia, Alzheimer's disease (AD) and vascular dementia, among subjects with type 2 diabetes mellitus (T2DM). Medline, Cochrane and Embase databases were searched from inception to 31 July 2022. Two investigators independently reviewed and screened trials comparing antidiabetic drugs with no antidiabetic drugs, placebo, or other active antidiabetic drugs on cognitive outcomes in T2DM. Data were analysed using meta-analysis and network meta-analysis. Twenty-seven studies met the inclusion criteria, including 3 randomised controlled trials, 19 cohort studies and 5 case-control studies. Compared with non-user, SGLT-2i (OR 0.41 [95% CI 0.22-0.76]), GLP-1RA (OR 0.34 [95% CI 0.14-0.85]), thiazolidinedione (OR 0.60 [95% CI 0.51-0.69]), and DPP-4i (OR 0.78 [95% CI 0.61-0.99]) users had a decreased risk of dementia, whereas sulfonylurea (OR 1.43 [95% CI 1.11-1.82]) increased dementia risk. Network meta-analysis showed that SGLT-2i was most likely to rank best (SUCRA = 94.4%), GLP-1 RA second best (SUCRA = 92.7%), thiazolidinedione third best (SUCRA = 74.7%) and DPP-4i fourth best (SUCRA = 54.9%), while sulfonylurea second worst (SUCRA = 20.0%) for decreasing dementia outcomes, by synthesising evidence from direct and indirect comparisons of multiple intervention. Evidence suggests the effects of SGLT-2i approximate to GLP-1 RAs > thiazolidinedione > DPP-4i for delaying cognitive impairment, dementia and AD outcomes, whereas sulfonylurea was associated with the highest risk. These findings provide evidence for evaluating the optional treatment for clinical practice. PROSPERO RegistrationRegistration no. CRD42022347280.