The research, which comes from the Gibbons Lab at the Institute for Systems Biology (ISB) in Seattle, is also helping to explain why probiotics work for some people and not others.
In work published in PLOS Biology, ISB associate professor Sean Gibbons and his team used data from two clinical trials to test whether computer modeling programs could predict whether probiotics would colonize and engraft in an individual’s gut.
The findings suggest that the modeling can predict whether certain bacterial strains will ‘stick’ with an accuracy of 75% to 80%, opening new opportunities for biotic personalization.
“By combining longitudinal data, mechanistic modeling and increasingly rich biosensing tools, we are beginning to sketch out what truly personalized microbiome medicine might look like,” Gibbons said. “Not just trying things and hoping they stick, but knowing, ahead of time, what’s likely to take.”
Predicting colonization
Probiotics are notoriously fickle, with effects varying from person to person. Research suggests that this variation is often caused by differences in people’s individual gut microbiome composition and diets, but scientists have historically lacked the tools to predict whether a probiotic will engraft and importantly, whether it will confer a benefit to the host.
To address this knowledge gap, the Gibbons Lab leveraged microbial community-scale metabolic models (MCMMs), which act like digital twins of an individual’s gut microbial metabolism, simulating the metabolic functions of the gut in the context of a given dietary input.
In addition to predicting whether probiotics might engraft (which can lead to consequential changes in gut microbiota composition and function), the team also assessed short-chain fatty acid (SCFA) production in response to probiotic and prebiotic interventions.
In the first phase of the research, the team used data from two intervention studies. One administered a synbiotic in people with type 2 diabetes and another used an eight-strain probiotic cocktail in people with recurrent Clostridioides difficile infections.
The researchers then evaluated how well MCMM-derived predictions aligned with observed shifts in microbial composition and clinical markers. The model demonstrated substantial accuracy, showing over 84% agreement with empirical engraftment measures in the first study and over 75% in the second study. However, the model was not always successful in its predictions.
“Poor predictions might be driven by poorly-curated genome-scale metabolic models of gut strains or by mismatches between the diet consumed by the patient and the diet applied to their model,” Gibbons explained. “Future work will look at ways to reduce these sources of error. We could only look at short-term engraftment, because that’s all we had data for. We’ll need new data sets to look at long-term engraftment predictions.”
What is the difference between colonization and engraftment?
Colonization typically refers to long-term establishment of probiotic in a person’s gut. Colonization can eventually lead to engraftment: when microbes establish a permanent, stable presence, competing with existing microbial communities.
Are colonization and engraftment necessary for health benefits?
Considering whether probiotic colonization is necessary for health benefits, Maria Marco, a professor in the department of food science and technology at the University of California, said probiotics often do not colonize the digestive tract, but this does not mean that they do not confer a benefit.
Professor Marco, who is also president of the International Scientific Association for Probiotics and Prebiotics (ISAPP) Board of Directors and has built an internationally-recognized research program on probiotics, fermented foods and dietary modulation of the gut microbiome, noted that even when probiotics do not permanently establish themselves in the gut, they can still alter the digestive tract by modulating the activity of the gut microbiota or stimulating the intestinal epithelium (the gut barrier).
“Colonization is not a requirement for a probiotic to confer a health benefit,” she said. “However, in some instances, longer-term persistence, and in some cases, colonization, may be desired.”
In fact, colonization can come with risks and unintended consequences, as well as a lack of ability to control the dose, frequency and duration of exposure to that particular microbe, Professor Marco added.
“Colonization is not necessarily desirable for all applications, and there is a possibility of unintended negative consequences with strain engraftment,” she said. “We are still learning about the gut microbiome and what a healthy gut microbiome looks like for different individuals. Also, assessments of the gut microbiome using stool samples may miss the probiotic persistence (colonization) levels of microorganisms in the small intestine and higher up in the digestive tract.”
What is the official definition of a probiotic?
In 2001, the Food and Agriculture Organization (FAO) of the United Nations and the World Health Organization (WHO) defined probiotics as “Live microorganisms that, when administered in adequate amounts, confer a health benefit on the host”. This definition is used by governments and organizations such as the International Scientific Association for Probiotics and Prebiotics (ISAPP) and the International Probiotics Association (IPA).
Conversely, Jens Walter, professor of ecology, food and the microbiome at University College Cork said he is not aware of a single instance in which colonization by a non-pathogen due to probiotics has caused harm in humans.
Professor Walter, who has authored more than 140 peer-reviewed publications primarily focusing on the evolutionary and ecological processes that have shaped host-microbiome symbiosis, said there has been a lot of confusing language around colonization, both in the literature and towards consumers.
“There were and probably still are many unrealistic claims of probiotics doing all sort of ecological effects (seeing of beneficial organisms, colonization resistance, competition with pathogens, etc.),” he explained. “If well designed, probiotics could perhaps do those in specific settings (after antibiotics, early in life, etc.), and for those to happen, colonization would be absolutely necessary.”
In a paper published in Current Opinion in Biotechnology, Professor Walter and colleagues argued that even transient bacteria can result in small temporary alterations in the composition and/or metabolic output of the bacterial community. However, to have long-standing effects on the microbiome, they stated that the microbes must be able to consistently grow and compete with resident taxa, altering community composition and function.
“Colonization is not always ‘needed’, but if you want effects on the microbiome, or require the probiotic to be metabolically active (to, for example, produce a metabolite), colonization would likely be either required or at least beneficial,” he said. “If the mechanisms do not require metabolic activity or ecological competitiveness, like direct immune mediated effects, then colonization is not needed.”
This debate highlights the complexities of defining what makes a probiotic effective and raises questions surrounding whether benefits can arise from transient activity as well as true colonization.
Analyzing functional impact and barriers
Gibbons and his team assessed whether the MCMMs could predict the functional impact of probiotic administration. To do this, they analyzed the production of butyrate and propionate—two SCFAs known for their beneficial effects on metabolic health, inflammation and gut barrier integrity.
Using a cross-sectional cohort of healthy individuals who were coached to improve their diet and lifestyle, the researchers used MCMMs to simulate a switch from a low-fiber diet to a high-fiber diet and compared how changes in predicted butyrate were associated with longitudinal changes in cardiometabolic and immune activation markers.
Responses varied across individuals, and not all subjects benefited equally from the high fiber diet. While a shift to a high-fiber diet generally increased butyrate, the magnitude of this increase varied substantially.
Gibbons and his team then analyzed the outcomes based on different types of fiber and the addition of a probiotic cocktail, and the MCMM was then able to identify the optimal prebiotic–probiotic combination for maximizing butyrate or propionate production.
The model therefore seems able to predict how microbial communities interact with one another and with what a person eats, which could enable personalized interventions to be developed in the future, Professor Marco commented.
“Understanding the metabolic capacities of the gut microbiome as demonstrated in this paper could lead to a step change in personalization of biotics like probiotics, prebiotics and other microbiome-related approaches,” she said. “Presently, there is significant inter-individual variation in responses to microbiome-targeted interventions. Although there is much more that needs to be done, having the capacity to predict whether a probiotic engrafts will lead to personalized approaches that are better suited for use, whether colonization is the goal or not.”
The Gibbons Lab, however, noted a number of limitations, including the model’s use of simplified “steady-state” assumptions that do not fully capture how the gut microbiome changes over short periods of time. The models also currently group bacteria by species, which might overlook important differences between individual strains.
“The largest barrier to applying these tools more widely to human populations is a lack of clinical trial data that validates whether or not MCMM-predicted personalized interventions perform better than standard biotic interventions,” Gibbons said. “We are working to fund prospective, controlled, human trials that provide this validation data. If it works, these interventions could be commercialized and made available to nutritionists, clinicians or individuals.”
