Gut microbiome features associated with Bifidobacterium colonization predict personalized probiotic persistence patterns.

Abstract

Bifidobacteria are key health-associated members of the human gut microbiome and are widely used as probiotics, but their colonization success varies substantially between individuals, partly due to baseline microbiome composition. We analyzed 51,244 gut microbiomes from 149 cohorts (45 countries) to identify non-Bifidobacterial taxa associated with the Bifidobacterial features. We observed several consistent and age-/life-style-specific association patterns of different non-Bifidobacterial taxa with the different Bifidobacteria. Multiple Bifidobacteria showed positive associations with butyrate-producing Firmicutes and Collinsella; negative associations involved pathobiont-Firmicutes and specific Bacteroidota taxa. B. adolescentis and B. breve showed the strongest positive and negative associations, respectively, with health-associated adult gut microbiome members. We quantified these relationships as Association-Scores, stratified by age/life-style/sequencing-strategy/disease, which were significantly reproducible after adjusting for multiple microbiome-linked, host life-style/clinical covariates, and predictable using species-specific genomic functions. We used these Association-Scores to derive microbiome-level Receptive-Scores that quantify how permissive a baseline microbiome is to increases or persistence of a given Bifidobacterium. In an external dataset of eight Bifidobacterium interventions (n = 1633 gut microbiomes), Receptive-Scores combined with baseline abundance of the administered Bifidobacteria significantly predicted post-treatment persistence/increase in 69.23% of trial-probiotic pairs. Together, this work identifies microbiome features governing Bifidobacterial colonization and provides tools to predict personalized probiotic responses.