Multimodal neuroimaging-microbiota integration identifies Akkermansia as a modulator of alcohol-induced gut-liver-brain pathology

Abstract

Alcohol use disorder (AUD) disrupts the gut-liver-brain axis, yet mechanistically grounded and therapeutically actionable targets within this network remain poorly defined. To identify microbial modulators of alcohol-induced tissue pathology, longitudinal advanced diffusion MRI and fecal 16S rRNA profiling were integrated across Marchigian Sardinian alcohol-preferring rats evaluated at baseline, after four weeks of voluntary alcohol intake, and following six weeks of abstinence. Machine learning, specifically random forest models combining neuroimaging and microbiota data, improved phase classification and identified Akkermansia as the microbial feature most strongly associated with alcohol-related white matter microstructural abnormalities. Alcohol exposure induced widespread white matter alterations alongside gut dysbiosis characterized by reduced microbial diversity. To evaluate functional relevance, Akkermansia muciniphila was administered during the abstinence phase. Supplementation with A. muciniphila restored intestinal mucus, reduced liver injury markers, and elevated myelin basic protein levels within affected white matter regions. Collectively, these findings highlight Akkermansia as a critical modulator of alcohol-induced gut-liver-brain pathology and provide experimental support for a causal contribution of specific gut bacteria to persistent white matter damage in AUD. More broadly, this work establishes a robust multimodal framework for microbiome-based target discovery with clear translational relevance for disorders characterized by dysfunction along the gut-liver-brain axis. Research in contextO_ST_ABSEvidence before this studyC_ST_ABSAlcohol use disorder (AUD) is associated with gut dysbiosis, impaired intestinal barrier function, liver injury, and persistent white matter abnormalities. Previous studies in patients and animal models have linked alcohol exposure to reduced microbial diversity, altered gut permeability, and white matter microstructural damage, particularly during abstinence. Other work has shown that microbiota-derived interventions can ameliorate peripheral consequences of alcohol exposure, especially in the gut and liver. However, the specific microbial features linked to alcohol-induced brain pathology remain poorly defined, and no prior study has integrated longitudinal microbiota and neuroimaging data to identify candidate microbial modulators of alcohol-related white matter damage and then functionally test them in vivo across the gut-liver-brain axis. Added value of this studyWe developed a multimodal framework that integrates longitudinal advanced diffusion MRI with fecal microbiota profiling and machine learning in alcohol-preferring rats. This approach identified Akkermansia as the microbial feature most strongly associated with alcohol-induced white matter abnormalities. Guided by this result, we administered Akkermansia muciniphila during abstinence and observed coordinated beneficial effects across multiple organs, including restoration of intestinal mucus, reduction of liver injury markers, and recovery of myelin basic protein in affected white matter regions. To our knowledge, this is the first study to combine longitudinal microbiota-MRI integration with experimental validation of a microbiota-based intervention that mitigates alcohol-induced pathology across the gut-liver-brain axis while restoring central white matter integrity. Implications of all the available evidenceOur findings support a mechanistic contribution of specific gut bacteria to persistent alcohol-induced tissue damage and identify Akkermansia as a candidate modulator of gut-liver-brain axis dysfunction in AUD. More broadly, this study establishes a generalizable strategy for integrating microbiota and neuroimaging data to discover biologically meaningful and therapeutically actionable targets in complex disorders involving coordinated peripheral and central pathology.