Why do different people behave differently to probiotics?

When probiotics meet the gut microbiome

Walk into any pharmacy and you will find an entire shelf devoted to probiotics. Among the most common are Bifidobacterium species, marketed for gut health, immune support, healthy aging, and an expanding range of clinical conditions. Yet probiotic benefits sharply differ among individuals. Clinical trials consistently show that while some people experience clear and lasting improvements, others derive little or no benefit.

One reason could be the persistence of the probiotic in the individual's gut, which could be influenced by the gut microbiome that already exists before a probiotic is consumed. Think of it like a new tenant moving into a crowded building: the incoming bacterium has to compete for resources and find its place among microbes that were already there. Its ability to colonize and persist depends not just on its own traits but on the resident community it walks into.

Previous studies had implicated baseline microbiome composition in probiotic engraftment, but they were limited to small cohorts and individual strains, leaving open whether microbiome features can reliably predict probiotic success. In this study, we address this question through the largest global meta-analysis of its kind, identifying the microbial features that shape Bifidobacterium colonization and introducing a framework to predict an individual's receptiveness to probiotic treatment¹.

Bifidobacteria keep selective company

Investigating 51,244 gut microbiomes (from 149 study-cohorts across 45 countries), we observed consistent sets of bacteria being associated with specific Bifidobacterium. Butyrate producers and members of the genus Collinsella tended to show consistent positive links with certain Bifidobacteria, while certain pathobiont Firmicutes and Bacteroidota behaved negatively. We captured these patterns as Association-Scores, a measure of how reliably a microbe accompanies a given Bifidobacterium.

These scores were stable: when recalculated across different diseases and compared against apparently non-diseased individuals, the associations largely held, and the same was true after adjusting for host properties like diet, medication, clinical factors, and also for certain microbiome properties. Bacteria "helping" Bifidobacterium in healthy guts tended to do so in sick ones too, which points to a real microbial relationship rather than side effects of illness itself. Mapping only the associations that passed all checks revealed a compact, consistent pattern: Collinsella and a handful of oral, Bacteroidetes, and Firmicutes groups linked reliably to specific bifidobacteria, no matter which lens was used to look for them¹.

The Receptive-Score

We then asked whether these patterns could predict something practical: whether a probiotic actually persists after someone takes it? We developed the Receptive-Score, which measures how welcoming a baseline microbiome is to a specific Bifidobacterium using its known microbial associates and baseline abundance.

Investigating eight independent probiotic trials, the Receptive-Score combined with baseline Bifidobacterium levels correctly predicted post-treatment persistence or increase in 69.23% of trial-probiotic combinations. In a B. longum trial², people who became persisters had significantly higher baseline Receptive-Scores than non-persisters. In two trials³,⁴ with clinical follow-up, higher scores also lined up with better clinical outcomes, including fewer gastrointestinal symptoms and lower inflammation markers, though sample sizes were too small for firm conclusions there.

What the genes reveal

To understand what might drive these relationships, we examined the genomes of 66 non-Bifidobacterium taxa. We identified specific genes and metabolic pathways in bacteria that showed strong positive associations with the health-associated B. adolescentis and B. longum, revealing a cooperative cross-feeding network. Many of these bacteria encoded enzymes involved in the breakdown of complex dietary carbohydrates and the production of metabolites such as acetate that can be exchanged within the microbial community, creating a nutrient-rich environment that supports Bifidobacterium persistence.

In contrast, bacteria associated with B. breve and B. dentium were enriched for genes involved in metal ion acquisition and scavenging, suggesting adaptation to resource competition and inflammatory conditions rather than metabolic cooperation. These findings indicate that the ability of different Bifidobacterium species to colonize in the gut may depend not only on their own biology but also on the metabolic capabilities of their microbial neighbours.

Predicting how your gut microbes will change

While the study still has certain gaps (limited representation of non-industrialized populations, species-level investigations while probiotics are typically characterized at strain level, limited data on clinical outcomes), this framework of capturing the receptivity of a microbiome to a given microbe can be generalized to build "digital twin" models of a person's gut community that let scientists test probiotics, dietary changes, or other microbial therapies in silico before ever trying them on an actual patient