Multivariate associative patterns between the gut microbiota and large-scale brain network connectivity

Abstract

Research on the gut-brain axis has accelerated substantially over the course of the last years. Many reviews have outlined the important implications of understanding the relation of the gut microbiota with human brain function and behavior. One substantial drawback in integrating gut microbiome and brain data is the lack of integrative multivariate approaches that enable capturing variance in both modalities simultaneously. To address this issue, we applied a linked independent component analysis (LICA) to microbiota and brain connectivity data. We analyzed data from 58 healthy females (mean age = 21.5 years). Magnetic Resonance Imaging data were acquired using resting state functional imaging data. The assessment of gut microbial composition from feces was based on sequencing of the V4 16S rRNA gene region. We used the LICA model to simultaneously factorize the subjects large-scale brain networks and microbiome relative abundance data into 10 independent components of spatial and abundance variation. LICA decomposition resulted in four components with non-marginal contribution of the microbiota data. The default mode network featured strongly in three components, whereas the two-lateralized fronto-parietal attention networks contributed to one component. The executive-control (with the default mode) network was associated to another component. We found the abundance of Prevotella genus was associated to the strength of expression of all networks, whereas Bifidobacterium was associated with the default mode and frontoparietal-attention networks. We provide the first exploratory evidence for multivariate associative patterns between the gut microbiota and brain network connectivity in healthy humans, taking into account the complexity of both systems.