Checking poop may be unpleasant, but it might tell a scientist a lot about a person's microbiome and, importantly, whether their microbiome has any bearing on their illness. In the November 4 Nature, researchers report that simple lifestyles factors and physiological characteristics, including stool quality, have more power over the intestinal microbiome than does disease. The authors, led by Ivan Vujkovic-Cvijin and Yasmine Belkaid from the National Institute of Allergy and Infectious Disease, Bethesda, Maryland, conclude that more needs to be done to account for confounding variables before conclusions can be drawn about gut flora and disease. They specifically recommend matching cases and controls based on a list of host variables.

  • Lifestyle factors can explain microbiome differences among people.
  • These differences overshadow microbiome change caused by disease.
  • Matching cases and controls for these factors will reduce confounds.

The microbiome has become a hot topic. Research suggests that in mice, different species of bacteria in the gut can sway microglial gene expression in the brain, and either increase or decrease Aβ plaques (Apr 2020 news), or that the microbiome regulates neuroinflammation and motor deficits in Parkinson’s and neurodegeneration models (Sampson et al., 2016; Dec 2017 news). But does this hold true in people?

In this new field, researchers thus far have relied largely on cross-sectional data, comparing microbiota profiles of healthy participants with those of people diagnosed with a disease (Saji et al., 2020; Wu et al., 2020). Vujkovic-Cvijin and colleagues question that strategy, claiming that cases and controls need to be more carefully matched to avoid spurious results.

Spurious Associations? Individual and lifestyle factors can modulate your gut microbiome. Some of those variables differ (purple dots) in disease case-control studies. Could they explain microbiome changes attributed to some diseases? [Courtesy of Vujkovic-Cvijin and colleagues, Nature, 2020.]

“This study basically says that there are confounds that we weren’t really aware of before,” said Julie Andersen, Buck Institute for Age Research in Novato, California. “So how much can we trust the data that’s come out to date?” she asked. “This underscores the need to match cases and controls based on potential confounds. Going forward, everybody should be doing this,” she added.

The authors used data from the American Gut Project, the largest publicly available human gut bacterial dataset, to identify host characteristics that associate with specific types of gut microbiota (McDonald et al., 2018). After receiving stool samples from more than 11,000 volunteers, researchers at the AGP extracted the DNA and sequenced the V4 hypervariable region of the 16S rRNA gene, which allows scientists to determine what microbiomes are present in the gut and their abundance. Sequences of 16S rRNA are widely used in bacterial phylogenetics. Participants completed a questionnaire that asked about age, sex, health, lifestyle and diet, and whether they had been diagnosed with any of 19 diseases, including Type 2 diabetes, irritable bowel syndrome, cardiovascular disease, and kidney disease.

After excluding people based on factors known to skew the microbiome, such as last antibiotic use, age, and BMI, the researchers surveyed the microbiomes of 2,971 healthy controls and 5,878 participants diagnosed with at least one disease. They then used machine learning to find and quantify correlations between gut flora and each host variable. The lifestyle parameters included how frequently people drank alcohol, the quality of their bowel movements, and how often they ate meat, eggs, dairy, vegetables, whole grains, and salted snacks.

Better Match. Cases and controls that are unmatched for confounding variables (brown bars) exhibit greater microbiota differences than those that are matched (blue bars). [Courtesy of Vujkovic-Cvijin and colleagues, Nature, 2020.]

They found that how often a person drank alcohol, and features of their bowel movements such as whether their stools were solid, normal, or loose, strongly associated with the composition of their gut microbiota. For example, using a table of 16S rRNA amplicon sequence variants, the authors found that Bifidobacteria were more enriched in alcohol consumers than in nonconsumers. Correlations between other variables and the microbiome were found at the genus level, as well. These correlations often underlay ostensible disease associations. For example, cumulative drinks per week accounted for most of the differences in microbiota seen between healthy controls and people diagnosed with diabetes. People with diabetes who drank wine, beer, and cider had the greatest change from controls. Alas, once Vujkovic-Cvijin matched healthy controls with T2D patients who imbibed similarly, differences between their microbiotas collapsed. Stool quality was also among the top confounding variables across diseases, including migraine and autism spectrum disorder.

In contrast, while other dietary variables, including salt and grains, did influence the microbiome, they did not confound the effects of disease. “The alcohol and stool quality variables were surprising,” said Vujkovic-Cvijin. “Frankly, we were not expecting to see such dramatic differences when comparing disease subjects to healthy individuals.”

The findings are in line with past studies that associated both stool quality and red wine consumption with gut microbiota (Falony et al., 2016; Le Roy et al., 2020). However, this paper is the first to examine how these variables influence disease associations, Vujkovic-Cvijin told Alzforum.

Moving forward, he hopes studies will take these confounding variables into consideration when comparing controls to disease cases. He also believes studies should use the AGP questionnaire but include specific diagnoses rather than general ones, such as “kidney disease” or “skin condition.” “This [broader categorization] was a necessary first step before building larger cohorts with more detailed information,” he said. “I hope that studies like ours will stimulate more specific datasets.” 

Andersen, who studies the underlying age-related processes driving neurodegenerative diseases such as Alzheimer’s and PD, believes a more thorough database would be helpful to better understand gut-brain interactions. “For a lot of neurodegenerative diseases, the whole body is important, including how it responds to lifestyle and diet,” she said. “This study brings that home.”—Helen Santoro

Comments

No Available Comments

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. ‘Working from Home’: Do Gut Microbes Hold Sway Over Glia, Aβ?
  2. Gut Microbiome May Modify Neurodegeneration

Paper Citations

  1. . Gut Microbiota Regulate Motor Deficits and Neuroinflammation in a Model of Parkinson's Disease. Cell. 2016 Dec 1;167(6):1469-1480.e12. PubMed.
  2. . Relationship between dementia and gut microbiome-associated metabolites: a cross-sectional study in Japan. Sci Rep. 2020 May 18;10(1):8088. PubMed.
  3. . The Gut Microbiota in Prediabetes and Diabetes: A Population-Based Cross-Sectional Study. Cell Metab. 2020 Sep 1;32(3):379-390.e3. Epub 2020 Jul 10 PubMed.
  4. . American Gut: an Open Platform for Citizen Science Microbiome Research. mSystems. 2018 May-Jun;3(3) Epub 2018 May 15 PubMed.
  5. . Population-level analysis of gut microbiome variation. Science. 2016 Apr 29;352(6285):560-4. Epub 2016 Apr 28 PubMed.
  6. . Red Wine Consumption Associated With Increased Gut Microbiota α-Diversity in 3 Independent Cohorts. Gastroenterology. 2020 Jan;158(1):270-272.e2. Epub 2019 Aug 28 PubMed.

Further Reading

No Available Further Reading

Primary Papers

  1. . Host variables confound gut microbiota studies of human disease. Nature. 2020 Nov 4; PubMed.