Investigating Causal Associations between the Gut Microbiota and Dementia: A Mendelian Randomization Study

Abstract

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.

Keywords: dementia; gut microbiota; mendelian randomization.

Figure 1

The study design of the MR analysis (A) and the overall workflow (B). GWAS: genome-wide association study; MR: Mendelian randomization; IVs: instrument variables; LD: linkage disequilibrium; SNP: single nucleotide polymorphism; IVW: inverse-variance-weighted; MR-PRESSO: MR pleiotropy residual sum and outlier; AD: Alzheimer’s disease; VaD: vascular dementia; FTD: frontal-temporal dementia.

Figure 2

Forest plots of estimates identified with inverse-variance weighted. (A) Dementia, (B) Alzheimer’s disease, (C) vascular dementia, and (D) frontal-temporal dementia. The odds ratios (95% confidence interval) for dementia and its three subtypes per relative abundance increase in gut microbiota as estimated in the inverse-variance weighted MR analysis.

Figure 3

Multivariable MR and MR-BMA results for dementia and its three subtypes. The odds ratios (95% confidence interval) for dementia and its three subtypes per relative abundance increase in gut microbiota as estimated in the inverse-variance weighted MR analysis. The last column illustrates the MIP and MACE (MR-BMA output) to prioritize risk factors for dementia. MIP, marginal inclusion probability; MACE, model-averaged causal estimate.