Ecological Modeling Of Gut Microbiome Points To Probiotic Cures
By Deborah Borfitz
September 24, 2020 | Fecal microbiota transplantation (FMT), the idea of using fecal material from a healthy person to treat others who are sick, has been proposed for challenging conditions ranging from inflammatory bowel disease to autism and obesity. But to date, robust evidence from randomized clinical trials exists only for the use of FMT to treat recurrent C. difficile infection (rCDI), according to Yang-Yu Liu, PhD, an assistant professor in the department of medicine at Brigham and Women's Hospital and Harvard Medical School.
Although in western medicine the first use of FMT (in treating fulminant pseudomembranous colitis) was published in 1958, the history of FMT dates back to ancient China when fecal material was used to treat food poisoning with some apparent success, says Liu. The recent explosion in the number of microbiome and FMT studies was prompted by the development of next-generation sequencing along with advances in bioinformatics used to catalogue and characterize the thousands of identified bacterial species.
Among rCDI patients who don’t respond to standard antibiotics, FMT has an impressive 80% cure rate, says Liu. But, until now, no one could explain why FMT would sometimes work and other times would not.
What underpins FMT success and failure for rCDI patients was the subject of a recently published paper in Nature Communications (DOI: 10.1038/s41467-020-17180-x), co-authored by Liu, which offers an explanation using tools and concepts borrowed from community ecology and network science. The study validates the theoretical framework by leveraging the network of interacting bacterial species inferred from mice experiments.
Mapping the ecological network of the human gut microbiome is the ultimate goal, Liu says, and the methods for doing so (DOI: 10.1038/s41467-017-02090-2) already exist. What’s missing is the requisite data—metagenomic sequencing of gut microbiome samples from at least 5,000 healthy individuals. Comprehensively cataloging individual members comprising a specific microbiome and their influence on one another is costly and logistically challenging work, albeit with enormous potential to improve human health.
As Liu and his colleagues have just demonstrated, the possibilities include predicting the effectiveness of FTM as well as designing personalized probiotic cocktails suited to the microbiota of individual patients, and perhaps more generically to large groups of them. Dozens of microbiome-based companies have emerged over the past few years endeavoring to replace the need for FMT with probiotic cocktails designed with a consortium of bacterial species to treat disease such as rCDI.
Designing A Cocktail
The ecological modeling approach is a significant departure from the usual focus on a particular bug, pathway, or function, explains Liu, and acknowledges the fact that “microbes interact with each other in many different ways.” This complex ecosystem includes bacterial species that directly inhibit the growth of C. diff but at the same time indirectly promote its growth by virtue of their encounters with other types of bacteria inhabiting the same community.
In the just-completed study, researchers considered the interplay between a relatively small pool of 100 possible bacterial species (almost 2,000 species have been found in the human gut) and simulated the FMT process of treating rCDI, he says. They then estimated how effective FMT would be at restoring the recipient’s healthy gut microbiome, validating their modeling with the real-world data from preclinical experiments in mice and human participants in an FMT clinical trial.
The theoretical model helped the team predict what factors determine the efficacy of FMT, which notably include guts with less bacterial diversity. They also used what they learned about interactions between microbial inhabitants to design a personalized probiotic cocktail to decolonize C. diff using only the effective components of fecal bacteria. “Our design of the probiotic is based on the network structure signature so you can check if the cocktail will work or not [before administered to patients with rCDI],” Liu says.
He is quick to add that dietary intake was not factored into the current modeling framework, which would have complicated the ecological modeling. However, the research team is now taking that on that challenge for the next iteration of their modeling framework that may enhance its predictive power.