Associations of gut bacterial classes Clostridia and Deltaproteobacteria with type 2 diabetes and Alzheimer's disease: A two-sample Mendelian randomization study.

Using a 2-sample Mendelian randomization (MR) approach, we evaluated MR evidence for genetically proxied effects of 10 pre-specified gut bacterial classes on type 2 diabetes mellitus (T2DM) and Alzheimer disease (AD), and assessed MR evidence for a direct genetically proxied effect of T2DM on AD. Mediation by T2DM was explored where feasible. We performed 2-sample MR using MiBioGen gut microbiota GWAS summary statistics as exposure, and DIAMANTE (T2DM) and European AD GWAS data as outcomes. Inverse-variance weighted (IVW) MR was the primary analysis, with MR-Egger, weighted median, and weighted mode as sensitivity analyses. Heterogeneity and directional pleiotropy were assessed, and IVW results were false discovery rate (FDR) corrected using Benjamini-Hochberg. MR analysis using 184 independent SNPs provided no robust evidence for a direct genetically proxied effect of T2DM on AD risk (IVW odds ratio [OR] = 0.98, 95% confidence interval [CI]: 0.94-1.02, P = .28). After FDR correction, higher genetically predicted Clostridia abundance was associated with lower AD risk (IVW OR = 0.86, 95% CI: 0.77-0.96, FDR < 0.05), and higher genetically predicted Deltaproteobacteria abundance was associated with higher T2DM risk (IVW OR = 1.11, 95% CI: 1.04-1.20, FDR < 0.05). Other classes were not significant after FDR correction. Reverse MR and multivariable MR were not testable due to insufficient SNP overlap after harmonization (effective SNPs < 3). These findings provide MR evidence consistent with genetically proxied effects of Clostridia on AD and Deltaproteobacteria on T2DM and do not support robust MR evidence for a direct genetically proxied effect of T2DM on AD. Results are hypothesis-generating and require replication in diverse ancestries and follow-up mechanistic/experimental studies, given that included GWAS sources are predominantly of European ancestry.