If you read media headlines or watch television programs like "The Dr. Oz Show" you might be convinced that an out-of-whack balance of microbes causes obesity, and that stool implants or fancy yogurt will cure it.

No, you got obese because you eat too much. Every other claim is selling you something.

A new paper in mBio finds there's no clear common characteristic of the microbe populations, or microbiomes, in the digestive systems of obese people that makes them different from the microbiomes of those with a healthy weight.  This lack of a clear 'signature' across more than 1,000 volunteers in 10 of the largest studies done to date may not please overweight people. It may also disappoint the companies that sell them products aimed at altering the gut's microscopic population through fiber, nutrients and 'good' bacteria.

In other words, with trillions of these things, trying to game your microbiome could be doing good, it could be doing harm, but almost certainly it was doing nothing at all. It can't hurt, but it's an expensive placebo.

Credit: Flickr

If you're one of the scientists who latched onto this bandwagon, you can always argue that flaws in existing knowledge mean more complexity, and even more reason you should get another R01 grant from the NIH.

"In the end, we found that there are no clear signatures or predictors of obesity across the microbiome data reported thus far, and that if there is any signature at all, related to diversity of microbes it's not biologically useful. This is a cautionary tale that points to the need to do more work to clarify what we know and don't know," said  Marc Sze, Ph.D., University of Michigan postdoctoral research fellow and lead author of the paper.

Pooling the data

The idea that obesity and imbalanced microbiomes are linked started with research in animals, and the observation that obese animals tended to have a certain ratio of two groups of bacteria species. That led to the studies in humans -- some of them with just a few dozen participants -- that grabbed headlines in recent years.

But people and rodents bred for laboratory use are very different, and large samples of people are needed to say anything definitive about human health and its link to the microbiome.

The authors used machine-learning computing tools developed in Schloss's lab over recent years to perform the analysis. This included a program called mothur that gives microbiome researchers a free open-source tool for studying the vast amount of data that can emerge from microbiome studies. 

They created a classification model that takes into account all the data about the different microbe species present in an individual's microbiome, and other information. They then tried to use it to predict whether that individual was obese or not.

While the tool worked somewhat when they used it on data from one of the 10 studies done by others, it no longer predicted obesity well at all when they used it on data from the other studies.

The team is developing a similar tool for use in evaluating the gut microbiome's possible links to colon cancer. Earlier this year, they published results from an analysis of several hundred people, and showed that their tool has reasonable accuracy in detecting microbiome signatures that are more common in people with colon cancer than those without.