08.25.2016
Gut Bacteria and Obesity: How Strong a Link?
Meta-analysis finds variation, statistical insignificance
Popular hypotheses suggesting actionable links between obesity and the bacterial makeup of the human digestive system are difficult to support, a new analysis concluded.
The analysis examined 10 previous studies on the topic, and raises doubts over those studies' suggestions that a certain "signature" in the digestive systems or microbiomes of obese mammals might increase a person's likelihood of weight gain or obesity, reported Patrick Schloss, PhD, and Marc A. Sze, PhD, both of the University of Michigan in Ann Arbor, in mBio.
"While there are differences in microbial diversity, those differences are really, really small," Schloss told MedPage Today. "It's really hard to believe it's biologically significant."
Following the PRISMA guidelines for meta-analysis, Schloss and Sze curated the 10 studies, then analyzed the studies' microbiome sample data for ecological diversity, evenness, richness, and the relative abundance and ratios of Bacteroidetes and Firmicutes (B/F) -- the two intestinal bacteria most often singled out for connections to obesity.
Prior studies, conducted in mice and humans, observed that B/F ratios and diversity levels were lower in obese mice than in lean mice, and that relative abundance of Bacteroidetes and Firmicutes increased and decreased as individuals dieted to lose weight.
After normally distributing the six alpha diversity metrics and using a t test to identify significant associations between those metrics and whether individual patients were obese, Schloss and Sze concluded that there was too much variation and uncertainty across studies to firmly establish connections.
"The B/F ratio and relative abundance of Firmicutes were not significantly associated with obesity in any study," they wrote. "We identified seven P-values that were 0.05; three studies indicated that obese individuals had a lower richness, two studies indicated a significantly lower diversity, one study indicated a significantly lower evenness, and one study indicated a significantly higher relative abundance of Bacteroidetes ... Interestingly, although only 2 of the 10 studies observed the previously reported association between lower diversity and obesity, the other studies appeared to have the same trend, albeit the differences were not statistically significant."
Building on the initial alpha diversity analysis, the authors calculated the relative risk of being obese based on an individual's alpha diversity metrics relative to a median. Across the 10 studies and six metrics, only one study and three metrics were found to have significant relative risk values.
At the same time, relative risk values tended to be above 1. Using a random-effects model, relative risks associated with having below-median metrics in a microbiome's richness, evenness, or diversity was significantly associated with obesity. The relative risk of having low richness was 1.30 (95% CI 1.13 to 1.49), that of having low evenness was 1.20 (95% CI 1.06 to 1.37), and that of having low diversity was 1.27 (95% CI 1.09 to 1.48).
Nevertheless, those results, "indicate that although individuals with a lower richness, evenness, or diversity are at statistically significantly increased risk of being obese, it is questionable whether that risk is biologically or clinically relevant," the authors wrote.
Schloss said there appears to be a causal effect between the microbiome and body weight, but that only wider studies involving many more patients can reveal those connections with any certainty.
"In the microbiome field in general, we need larger studies," Schloss said. "The human microbiome is so diverse across individuals, and the signal is so different and there is so much variation. We need to look at not just the types of microbes but their functions. There is not currently enough statistical power to find differences."
"Nothing nefarious happened here; this is just the scientific method playing out," he added.
The study was funded by the U.S Department of Health and Human Services and the NIH.
Schloss and Sze disclosed no relevant relationships with industry.
Gut Bacteria and Obesity: How Strong a Link?
Meta-analysis finds variation, statistical insignificance
- by Scott Harris
Contributing Writer, MedPage Today
Popular hypotheses suggesting actionable links between obesity and the bacterial makeup of the human digestive system are difficult to support, a new analysis concluded.
The analysis examined 10 previous studies on the topic, and raises doubts over those studies' suggestions that a certain "signature" in the digestive systems or microbiomes of obese mammals might increase a person's likelihood of weight gain or obesity, reported Patrick Schloss, PhD, and Marc A. Sze, PhD, both of the University of Michigan in Ann Arbor, in mBio.
"While there are differences in microbial diversity, those differences are really, really small," Schloss told MedPage Today. "It's really hard to believe it's biologically significant."
Following the PRISMA guidelines for meta-analysis, Schloss and Sze curated the 10 studies, then analyzed the studies' microbiome sample data for ecological diversity, evenness, richness, and the relative abundance and ratios of Bacteroidetes and Firmicutes (B/F) -- the two intestinal bacteria most often singled out for connections to obesity.
Prior studies, conducted in mice and humans, observed that B/F ratios and diversity levels were lower in obese mice than in lean mice, and that relative abundance of Bacteroidetes and Firmicutes increased and decreased as individuals dieted to lose weight.
After normally distributing the six alpha diversity metrics and using a t test to identify significant associations between those metrics and whether individual patients were obese, Schloss and Sze concluded that there was too much variation and uncertainty across studies to firmly establish connections.
"The B/F ratio and relative abundance of Firmicutes were not significantly associated with obesity in any study," they wrote. "We identified seven P-values that were 0.05; three studies indicated that obese individuals had a lower richness, two studies indicated a significantly lower diversity, one study indicated a significantly lower evenness, and one study indicated a significantly higher relative abundance of Bacteroidetes ... Interestingly, although only 2 of the 10 studies observed the previously reported association between lower diversity and obesity, the other studies appeared to have the same trend, albeit the differences were not statistically significant."
Building on the initial alpha diversity analysis, the authors calculated the relative risk of being obese based on an individual's alpha diversity metrics relative to a median. Across the 10 studies and six metrics, only one study and three metrics were found to have significant relative risk values.
At the same time, relative risk values tended to be above 1. Using a random-effects model, relative risks associated with having below-median metrics in a microbiome's richness, evenness, or diversity was significantly associated with obesity. The relative risk of having low richness was 1.30 (95% CI 1.13 to 1.49), that of having low evenness was 1.20 (95% CI 1.06 to 1.37), and that of having low diversity was 1.27 (95% CI 1.09 to 1.48).
Nevertheless, those results, "indicate that although individuals with a lower richness, evenness, or diversity are at statistically significantly increased risk of being obese, it is questionable whether that risk is biologically or clinically relevant," the authors wrote.
Schloss said there appears to be a causal effect between the microbiome and body weight, but that only wider studies involving many more patients can reveal those connections with any certainty.
"In the microbiome field in general, we need larger studies," Schloss said. "The human microbiome is so diverse across individuals, and the signal is so different and there is so much variation. We need to look at not just the types of microbes but their functions. There is not currently enough statistical power to find differences."
"Nothing nefarious happened here; this is just the scientific method playing out," he added.
The study was funded by the U.S Department of Health and Human Services and the NIH.
Schloss and Sze disclosed no relevant relationships with industry.
