After my call for a social network analysis of bacterial populations in the gut, I’ve been on the lookout for such a study. I got excited when I saw this network analysis by Kelder et al. that was published yesterday. While it’s not a social network, they used network methods to look at correlations between microbe levels and various biomarkers and measurements in a set of men prescribed a high fat, high calorie diet for one month. They constructed a network based on correlations between the change in levels of biological characteristics/microbes from the beginning to the end of the month.
Figure 3 from Kelder et al. Triangles indicate human attributes and circles indicate microbial species.
Unfortunately the statistical results didn’t impress me. With a sample size of 10 and no control group, it’s hard to draw any meaningful conclusions from the data. Additionally, there was no mention of a correction for multiple tests, so some of these correlations that are significant at the 0.05 level are probably spurious.
I’ll have to keep an eye out for my social network analysis. Has anyone seen something similar done in the past that I missed?
After Monday’s post on the gut microbiota, I found a study from a few weeks ago that links exercise and microbiota diversity. Specifically, the authors found that athletes in their study had greater diversity of gut microbe species than controls, suggesting that they have better digestion and metabolism. The authors compared the diet and exercise of male rugby players to a sample of controls. Their main results included:
- Athletes had higher levels of plasma creatine kinase, a biomarker for exercise, than controls
The gut microbiota of athletes was more diverse than that of controls, and diversity was correlated with creatine kinase levels
- Athletes’ diets were comprised of more protein and supplements than controls, and gut microbiota diversity was correlated with protein intake
So exercise is good for our gut? Not necessarily. Diet and exercise go hand in hand. In this case, athletes had both increased creatine kinase and increased protein consumption, and both were correlated with microbial diversity. It has been shown before that diet directly affects microbiota diversity, but from this study we can’t pick apart the causal relationship between these three things. To say that exercise has an impact on diversity suggests that the first diagram below is the causal model, but really it could be any of the three, or a more complicated one entirely.
Fortunately, the authors acknowledge this limitation:
Further, intervention-based studies to tease apart this relationship will be important and provide further insights into optimal therapies to influence the gut microbiota and its relationship with health and disease.
It will be interesting to follow this lead. In the mean time, go for a run.
Causal diagrams were made with DAGitty v2.0
We’ve all been told to eat more yogurt to “aid digestion”. That’s because yogurt and other fermented foods contain common strains of bacteria like Lactobacilli and Streptococcus thermophilus, which naturally occur in our GI tract. But our gut microflora might play an even greater role in staying healthy than just supplementing our digestive system. A new study compared the gut microbiota of healthy individuals to obese and diabetic individuals. Results indicate that the unhealthy individuals are significantly lacking certain strains of bacteria compared to healthy individuals. This would seem to suggest an association between metabolic disorders and gut microflora. If this is the case, then the role that our gut microbes play is more complex than just helping break down food our GI tract.
I can’t comment on the study’s internal validity without seeing full results and statistical methods. However, I’d guess that the study was sufficiently powered (n=81) to find associations. Given the abundance of recent research on how diet influences gut microflora and advances in treating inflammatory bowel disease by replenishing gut microbes, I am inclined to believe these results.
I’d like to see causal inference methods done to tease apart the relationship between bacterial strains and diseases like diabetes. As it stands, the study provides no causal evidence for the associations they find. A third variable could be confounding the association and might provide a better therapeutic target than the intestinal microflora.
Some work has been done modeling the gut microbiome as a social network, where nodes in the network represent a species of bacteria and edges represent co-occurence of species in an individual. I’d like to see more of this line of thought. Species of bacteria may have symbiotic relationships, for example, if one produces food for another. Others may have antagonistic relationships. These relationships between bacteria manifest as correlations in the data, which can cause problems in parametric analyses. Network-based methods can better handle this type of data and give greater insight into how groups of bacterial species might be involved with obesity and diabetes.