Abstract
Numerous observational studies suggest associations between proton pump inhibitors (PPIs) and dementia, but causal relationships remain uncertain. Using large-scale genome-wide association study (GWAS) data, we performed univariable Mendelian randomization (UVMR) analysis to assess the causality between five PPI types, and all-cause dementia and its five subtypes. Confounders were controlled through multivariable MR (MVMR) analysis to isolate PPIs' direct effects on dementia. Heterogeneity and pleiotropy assessments, and leave-one-out analysis, were conducted to validate the robustness of our results. Initial UVMR estimates suggested that lansoprazole (odds ratio [OR] 1.291; 95%confidence interval [CI] 1.001-1.665; p = 0.049) and pantoprazole (OR 1.118; 95% CI 1.014-1.233; p = 0.025) potentially increased VD risk, with their respective direct associations also discovered in MVMR. Additionally, FTD was found to reversely increase rabeprazole use (OR 1.086; 95% CI 1.011-1.167; p = 0.023). However, after adjustment for the false discovery rate (FDR), none of these associations remained statistically significant (p(FDR)> 0.05). The robustness of our results is supported by consistent estimates across complementary MR methods and the absence of pleiotropy. Our study indicates no robust causality between PPI use and increased dementia risk. Thus, it is inappropriate to restrict clinically justified PPI prescriptions merely due to potential cognitive risks.