Is human obesity related to the composition and richness of our gut microbiota – or not?

the guardian.com

There is a sobering paper in the journal of the American Society for Microbiology, mBio, this week, written by Marc Sze and Patrick Schloss from the Department of Microbiology and Immunology at the University of Michigan. As they report, obesity is a growing health concern throughout the western world. 20% of the youth age group in the US from 2 to 19 years is either overweight or obese, they say, and that value exceeds one third in adults aged 20 or over. Since obesity is reliably linked to a host of life-limiting diseases like diabetes, liver disease and heart disease, and since obese individuals and their health problems are increasingly clogging up our healthcare provision systems, we need clear direction on where to place research emphasis to stem this current epidemic. A view that has been gaining in popularity recently is that the richness (in number and species number) of our gut microbes is directly associated with obesity; that obese individuals have an impoverished, dysbiotic, gut microbiota; that they can pass these settings onto their children; and that you can predict obesity from gut microbial species composition. However, Sze and Schloss trained powerful statistical analysis on the pooled data from a number of studies relating the gut microbiota and obesity and concluded that only one of them had sufficient power to detect even a 5% difference in diversity and that “although there was support for a relationship between the microbial communities found in human feces and obesity status, this association was relatively weak and its detection confounded by large interpersonal variation and insufficient sample size”.

Sze and Schloss identify Jeffrey Gordon’s lab at WUSTL as classic leaders in this field that is establishing links between gut microbes and under- and over-nutrition and obesity, and cite two of their studies which found that the species composition was different between obese and lean mice, specifically the ratio between Bacteriodetes and Firmicutes, which was low in obese mice and high in lean mice. Extrapolating that research to humans, Gordon’s lab could not repeat the ratio effect seen in mice but did find that obese individuals had a lower alpha diversity than lean individuals and that, by manipulating the diet of obese human individuals by restricting either fats or carbs, they could change the relative abundance of Bacteriodetes and Firmicutes in the gut. The Gordon lab’s website claims claims that these sorts of results can be pressed into service worldwide: “Preclinical models are providing new therapeutic candidates; notably (i) microbiota-directed therapeutic foods (MDTFs) and human gut-derived microbial strains (next generation probiotics) designed to repair developmental abnormalities we have documented in the gut communities of children with undernutrition, and (ii) diet ingredients and consortia of human gut-derived microbes to rectify the perturbed functioning of the microbiota in individuals with obesity and its related metabolic abnormalities.”

However, say Sze and Schloss, two recent meta-analyses of microbiome and obesity data, by Walters et al and Finucane et al, concluded that these previously reported differences in community diversity and Bacteriodetes/Firmicutes ratio among lean and obese people could not be generalised. Given the importance being attached to the role of our gut microbes in health and disease (see Ed Yong’s “I Contain Multitudes”) they decided to give the matter a closer look. They identified 187 relevant studies in Pubmed and finally winnowed them down to 10 which included both body mass index and 16S rRNA data. They discovered that the B/F ratio and the relative abundance of Firmicutes were not significantly associated with obesity in any study, and that, although there was a significant relationship between decreased microbial richness, evenness, and diversity among obese individuals, the effect size was small. Obese individuals averaged 7.4% lower richness, for example, and 2% lower diversity. As they conclude: “These results indicate that obese individuals do have statistically significantly lower diversity than non-obese individuals; however, it is questionable whether the difference is biologically significant”. Meaning that, by their results, it is not practically possible to classify individuals as obese on the composition of their microbiome. Such differences as may be present are certainly beyond the power of all studies to date to detect, they say.

I’m quoting their final conclusions because they suggest that microbial diversity may not be the right indicator of a microbial effect on obesity. Microbiomes that are structurally very different in terms of their species composition could nevertheless produce common sets of gene transcripts or metabolites that either protect or expose individuals to obesogenic effects. “Although we failed to find an effect, this does not necessarily mean that there is no role for the microbiome in obesity. There is strong evidence in murine models of obesity that the microbiome and level of adiposity can be manipulated via genetic manipulation of the animal and manipulation of the community through antibiotics or colonization of germfree mice with diverse fecal material from human donors. These studies appear to conflict with the observations based on human subjects. Recalling the large interpersonal variation in the structure of the microbiome, it is possible that each individual has his or her own signatures of obesity. Alternatively, it could be that the involvement of the microbiome in obesity is not apparent based on the taxonomic information provided by 16S rRNA gene sequence data. Rather, the differences could become more apparent at the level of a common set of gene transcripts or metabolites that can be produced from different structures of the microbiome.”

I would certainly be interested to hear, given his impending arrival at CEM, what Ed Yong has to say about this research and would also like to hear Marty Blaser’s reaction, given that several of his studies are cited in this paper, and that of Joe Alcock, who has probably the strongest grasp of the microbiome and health literature of anyone at the center of the evmed community. The take-home message, as I see it, is that you need, in the first place, very large numbers of samples to get the statistical power currently lacking, and that the microbial signature, or microbial metabolite signature, that makes me obese might be different from yours. Which, if true, could make nonsense of any cure-all probiotic or fecal transplant therapy and points towards an arduous route to personalized medicine.