"There was some concern that we were 'over-fitting' the model," Spencer noted, "so we tested it out and ran a million permutations, altering the bug abundance and subject association, to see if we could identify how many actually showed a higher correlation by chance. What we found is that the p values still held up. We can have a lot of confidence in the result."
The big question that remains for the team is why the two bacterial populations correlate so strongly to the development of fatty liver. Anthony Fodor, UNC Charlotte assistant professor of bioinformatics and the project's director, sees a possible explanation, while warning against drawing specific conclusions without further study.
"We cannot yet assign cause and effect, but it implies that some bacteria are doing something that is making it easier for people to deal with a choline deficiency and for the liver to metabolize fat."
Conversely, the bacteria whose high population levels correlate with disease may be somehow removing available forms of choline from digested food. Fodor explains that further study will be needed to answer those questions.
"We're debating what the next step is," he said. "In some ways, this is a very specialized experiment because we are inducing fatty liver in a very specific way. In the general population, fatty liver is induced in many, many ways and not everyone who has fatty liver has low choline.
"It's probably like Alzheimer's or cancer, where there are many different causes for a disease that displays a common phenotype. More research will be required to determine the extent to which bacterial populations play a role in fatty liver development in the general population, but our results strongly suggest that there may be a link in some people."
|Contact: James Hathaway|
University of North Carolina at Charlotte