Search

Background: The causal association of specific gut microbiota with dementia remains incompletely understood. We aimed to access the causal relationships in which one or more gut microbiota account for dementia. Method: Using data from the MiBioGen and FinnGen consortia, we employed multiple Mendelian randomization (MR) approaches including two-sample MR (TSMR), multivariable MR (MVMR), and Bayesian model averaging MR to comprehensively evaluate the causal associations between 119 genera and dementia, and to prioritize the predominant bacterium. Result: We identified 21 genera that had causal effects on dementia and suggested Barnesiella (OR = 0.827, 95%CI = 0.722–0.948, marginal inclusion probability [MIP] = 0.464; model-averaged causal estimate [MACE] = −0.068) and Allisonella (OR = 0.770, 95%CI = 0.693–0.855, MIP = 0.898, MACE = −0.204) as the predominant genera for AD and all-cause dementia. Conclusions: These findings confirm the causal relationships between specific gut microbiota and dementia, highlighting the necessity of multiple MR approaches in gut microbiota analysis, and provides promising genera as potential novel biomarkers for dementia risk.

The causal association of circulating metabolites with dementia remains uncertain. We assessed the causal association of circulating metabolites with dementia utilizing Mendelian randomization (MR) methods. We performed univariable MR analysis to evaluate the associations of 486 metabolites with dementia, Alzheimer’s disease (AD), and vascular dementia (VaD) risk. For secondary validation, we replicated the analyses using an additional dataset with 123 metabolites. We observed 118 metabolites relevant to the risk of dementia, 59 of which were lipids, supporting the crucial role of lipids in dementia pathogenesis. After Bonferroni adjustment, we identified nine traits of HDL particles as potential causal mediators of dementia. Regarding dementia subtypes, protective effects were observed for epiandrosterone sulfate on AD (OR = 0.60, 95% CI: 0.48–0.75) and glycoproteins on VaD (OR = 0.89, 95% CI: 0.83–0.95). Bayesian model averaging MR (MR-BMA) analysis was further conducted to prioritize the predominant metabolites for dementia risk, which highlighted the mean diameter of HDL particles and the concentration of very large HDL particles as the predominant protective factors against dementia. Moreover, pathway analysis identified 17 significant and 2 shared metabolic pathways. These findings provide support for the identification of promising predictive biomarkers and therapeutic targets for dementia.