COLD SPRING HARBOR, NY (GenomeWeb) – There is no one diseased gut microbiome state, according to Cornell University's Emily Davenport.
Researchers have been investigating differences between the gut microbiomes of healthy people and those who are sick to determine whether the gut microbiome enters a state of dysbiosis that contributes to disease.
In a talk at the Biology of Genomes meeting here Saturday, Davenport described how she and her colleagues developed gut microbiome co-occurrence networks for thousands of people, some of whom were healthy and some of whom had conditions like asthma. Through their analysis, though, they found no consistent community differences in the microbial networks of healthy and sick people that held across various diseases.
"There is no such thing as a healthy or disease state in the gut microbiome," Davenport said.
To examine the dysbiosis state, she and her colleagues amassed 16S rRNA data from some 2,500 people from the Twins UK cohort. They stratified the population by 53 disease states and 127 quantitative phenotypes, particularly immune-related ones such as cytokine levels and BMI.
Using a two-pronged approach, they examined whether the microbial community makeup of sick people's microbiomes differed from those of healthy people. As Davenport noted, investigators have speculated that a diseased microbiome might be highly connected, while a healthy one might be harder to perturb.
For each phenotype, Davenport and her colleagues used a combination of generalized and linear mixed models to determine which microbial taxa were more abundant in the gut microbiomes of people with various diseases, taking into account that their cohort was composed of twins. For each 180 or so phenotypes, they uncovered bacteria that were associated with each state.
At the same time, the researchers developed co-occurrence networks using the SPAR-CC software, both for people with and without disease as well as for people at different tail ends of the conditions. By merging the two, they compared these microbial networks of healthy and diseased states.
Davenport said that she and her colleagues expected to see that the microbiomes of unhealthy people had slightly more network connections than those of healthy people. In asthma, they did find this pattern, but when they looked across other diseases and phenotypes, they didn't always see the same pattern, suggesting that this was not necessarily a signature of health or of disease.
"This trend does not necessarily hold across all phenotypes," Davenport added.
Likewise, they expected the network topology to differ between healthy and diseased microbiomes, as the diseased microbiome was expected to be less modular. While they did see that in asthma — the healthy microbiome had some 86 network edges to asthma's 48 — that trend, again, wasn't necessarily present among other diseases.
According to Davenport, there is no consistent pattern across disease states.
However, she noted a number of caveats. Firstly, the data they used had been collected at one time point and thus only gave a snapshot of the microbiome. She noted that the subjects might not have been experiencing symptoms of the disease at that time and therefore the disease state might not have been reflected in their gut microbiome makeup.
Additionally, she noted that they focused on taxa common to both healthy and diseased states and that more rare species could have a role.
Davenport further suggested that taxonomy and host genetics — as gut microbe abundance appears to be partially heritable — might instead be the larger drivers of microbial co-occurrence rather than disease state.