Alzheimer’s disease‐driven microbiome dysbiosis is a strong predictor of cognitive performance in humans

Abstract

AbstractBackgroundThe association between Alzheimer’s Disease (AD) and dysbiosis of the gastrointestinal (GI) microbiome has been demonstrated in several studies. Here, we aimed to expand on our previous findings that in nursing home residents the microbiome and clinical variables are predictive of AD compared to no dementia (ND) by expanding our study population to include those residing in the general community. We sought to identity microbiota characteristics that stratify AD from ND individuals across environments. Further, in our community residing population, we assessed the longitudinal microbiome’s ability to predict cognitive performance (ADAS‐Cog‐13).MethodOlder individuals residing in Massachusetts nursing homes (n = 79) or in the general community (n = 71) were enrolled in our study. At each visit, fecal samples and clinical variables were collected. For those residing in the community, ADAS‐Cog‐13 testing was done. Metagenomic profiling was performed on longitudinal fecal samples. Random forest classification models were used to identify associations between residency, AD diagnosis and microbial features and metabolic pathways. Mixed‐effect random forest regression models were created to identify metagenome features informative of cognitive performance.ResultAmong AD and ND individuals residing in both the NH and community, microbial abundance and basic clinical variables were informative of AD status (F1‐score = 0.827±0.03), with Clostrida, Blautia and Neglecta species associating with AD diagnosis. In specifically the community residing individuals, increase in Bacteroidetes (F1‐score = 0.906 ± 0.03) and pathways (F1‐score = 0.885±0.03) involved in amino acid catabolism and nucleotide synthesis was informative of AD diagnosis. Additionally, mixed‐effect random forest machine learning models determined that variability in the abundance of these bacteria and pathways accurately predicts the cognitive performance (ADAS‐Cog‐13) in community residing individuals.ConclusionTaken together our study describes that the AD microbiome from both nursing home and community‐dwelling older adults is depleted in health‐associated Clostridia and that the abundance of these bacteria and their metabolites not only optimally discriminates older human according to disease status, but it also accurately reflects their cognitive ability.