Dysbiosis, an imbalance in the microbiota, has been a major organizing concept in microbiome science. Here, we discuss how the balance concept, a holdover from prescientific thought, is irrelevant to — and may even distract from — useful microbiome research.
Throughout most of Western medical history, it was believed that the digestion of foods produced humours: mostly invisible liquids that were transformed into the materials that make up the body. Humoural theory dictated that diseases were caused by imbalances in these humours. A doctor's role was to help the body correct this imbalance, either by altering a patient's diet — thus changing the intake of humours — or by expelling excess humours, for example, by bloodletting1.
Now, with modern DNA sequencing technology, we know our digestive systems are actually filled with mostly invisible microorganisms that respond to our diet. We have found that altering diet, ingesting probiotics, and wholesale replacement of the microbial community can improve our health. A common explanation for the effectiveness of these therapies is that they ameliorate ‘dysbiosis’, an imbalance in the microbiota2–4.
The word ‘imbalance’ is crucial, because it implies a dysfunction of some complex set of processes, that is, that dysbiosis causes disease. This is not implausible; there is accumulating evidence of how imbalances in the microbial ecology of the gut or in the host immune system could cause disease. However, at the moment, the concept of balance is supported by evidence from only a few model systems and model interactions.
In terms of microbial ecology, dysbiosis can be conceptualized as a disruption in the many potential ecological links between microorganisms: competition, inhibition, commensalism, and perhaps symbiosis. In theory, a shift away from the ‘healthy’ ratios of species in the microbiota could lead to a disrupted ecosystem that harms the host. This explanation, though theoretically sound and intellectually engrossing, ignores the wide gap between what we actually know about the microbiome and how we suspect it may be. In general, the signal of microbe–microbe interactions in the human gut is not strong. Only a small fraction of taxa in the gut are correlated5, and there are only a few examples of known ecological relationships among microbiota constituents (for example, ref. 6). There are few or no examples of instances in which breakdowns in ecological relationships between gut microorganisms are proven to affect host health. We may eventually discover a vast network of strong ecological interactions, but we may also discover that such a network does not exist. So far the data have not provided unequivocal support for either scenario.
Dysbiosis can also be tied to the immune system's need to balance tolerance, which prevents attacks against the host's own tissues and its healthy commensal microorganisms, against inflammation, which protects against infection. For example, Lee and Mazmazian7 explain how an imbalance between the abundance of pro-inflammatory segmented filamentous bacteria (SFB), which causes the generation of Th17 cells, and the abundance of anti-inflammatory Bacteroides fragilis, which causes the generation of Treg cells, can lead to disease. This is a compelling argument, and it may be that many microbial species beyond B. fragilis and SFB are ‘balanced’ in this way. However, only a small number of model systems have been studied, and it is premature to assert that a balance between any particular microbial species is important to host health. Notably, the relevance of SFB to the human microbiome is still unclear8,9.
Given that the possible mechanisms by which dysbiosis could cause disease are still under investigation and that the relevance of most microbiota compositions to disease remains speculative, we believe researchers may be using the word ‘dysbiosis’ to refer to the result of disease, rather than the cause. Indeed, dysbiosis has such varying definitions in the literature that the term could apply to either cause or effect. The most common definition of dysbiosis is an imbalance between beneficial microorganisms like Lactobacillus or Bifidobacterium and harmful microorganisms like Escherichia coli2–4. However, the range of definitions is broad enough to capture any difference in microbial composition: dysbiosis has been defined as a change in the abundance or diversity of some groups of microorganisms10, a change in community composition caused by lifestyle7, a microbiota composition in which harmful organisms outweigh beneficial ones11, a bloom of a single pathobiont12, and as a reduction in bifidobacteria13.
This ambiguity in definition means that any measured difference in microbial composition, whether the cause or effect of disease, can be called dysbiosis. It is a ‘mechanism-free’ cause of disease to which we can retreat when plausible mechanistic explanations are discounted. For example, although there has been suspicion that a single, unknown pathogenic species causes inflammatory bowel disease (IBD), an autoimmune disorder characterized by chronic inflammation of the parts of the digestive system2, such a microorganism has not been found. Now most discussion about the role of the microorganisms in IBD focuses on dysbiosis.
The slipperiness of the dysbiosis concept means that a study can claim a ‘positive’ result — that dysbiosis was discovered — without actually accomplishing anything useful. For example, if you intend to design a diagnostic using microbiome data, then profiling the microbiome of healthy and ill people is not sensible because a diagnostic would not be needed to determine health status. A questionnaire is usually sufficient for this. Microbiome data can still add value to diagnostics, however. For example, it is easy to determine that someone might have IBD based on their symptoms. The more difficult challenge is to determine if someone who appears to have IBD is suffering from IBD or from a disease with related symptoms such as irritable bowel syndrome (IBS). Thus, designing a diagnostic requires comparison of people who have IBD and people who have diseases that might be mistaken for IBD.
A microbiome diagnostic must also add value beyond other diagnostics. For example, faecal calprotectin, a biochemical assay on patient stool, can help diagnose IBD. If microbiome data has no explanatory power beyond faecal calprotectin measurements, then a microbiome-based diagnostic is essentially an expensive and complicated way to estimate faecal calprotectin.
Like IBD, obesity is probably related to the gut microbiome. Microbiome diagnostics for obesity make no sense when we can just use a weighing scale. The more important possibility, of course, is that the microbiome causes obesity. Microbiome surveys comparing healthy and ill subjects are very valuable for addressing the causality of a disease, but only insofar as those surveys are connected with experiments that can verify causality. For example, if an organism is depleted in the microbiome of ill subjects, that organism could be re-introduced as a probiotic14 or in a gnotobiotic animal to test its effect on disease state, host immunity, or gut function. Even if it remains unclear whether the depletion of a beneficial microorganism interferes with host health, and even if the mechanism by which the addition of a microorganism affects health remains unclear, a simple experiment can show that the addition of the microorganism improves host health. Causality without mechanism is still important: a therapeutic requires causality, not necessarily mechanism.
Developing diagnostics and addressing causality are difficult problems (indeed, even the causative role of the microbiome in obesity remains debated)15. Studies designed to do either of these things can have a high probability of failure. However, negative results about diagnostics or therapeutics are more useful than positive results about dysbiosis. We already have two thousand years of tradition supporting the idea that balance is related to health, and the fact that healthy and ill people have different microbiomes is no longer a novel or useful observation. We need to show that differences in the microbiota can be used to predict or ameliorate disease, and not just show that differences exist.
Magner, L. N. A History of Medicine (CRC Press, 2005).
Tamboli, C. P., Neut, C., Desreumaux, P. & Colombel, J. F. Gut 53, 1–4 (2004).
Bäumler, A. J. & Sperandio, V. Nature 535, 85–93 (2016).
Mazmanian, S. K., Round, J. L. & Kasper, D. L. Nature 453, 620–625 (2008).
Friedman, J. & Alm, E. J. PLoS Comput. Biol. 8, e1002687 (2012).
Fischbach, M. A. & Sonnenburg, J. L. Cell Host Microbe 10, 336–347 (2011).
Lee, Y. K. & Mazmanian, S. K. Science 330, 1768–1773 (2010).
Sczesnak, A. et al. Cell Host Microbe 10, 260–272 (2011).
Caselli, M., Tosini, D., Gafà, R., Gasbarrini, A. & Lanza, G. Am. J. Gastroenterol. 108, 860–861 (2013).
Henao-Mejia, J. et al. Nature 482, 179–185 (2012).
Roberfroid, M. et al. Br. J. Nutr. 104, S1–S63 (2010).
Devkota, S. et al. Nature 487, 104–108 (2012).
Hopkins, M. J., Sharp, R. & Macfarlane, G. T. Gut 48, 198–205 (2001).
Sokol, H. et al. Proc. Natl Acad. Sci. USA 105, 16731–16736 (2008).
Harley, I. T. & Karp, C. L. Mol. Metab. 1, 21–31 (2012).
About this article
Integrating the Ecosystem Services Framework to Define Dysbiosis of the Breastfed Infant Gut: The Role of B. infantis and Human Milk Oligosaccharides
Frontiers in Nutrition (2020)
Interactions Between Gut Microbiota, Host, and Herbal Medicines: A Review of New Insights Into the Pathogenesis and Treatment of Type 2 Diabetes
Frontiers in Cellular and Infection Microbiology (2020)
Biology & Philosophy (2020)
Microbial Biotechnology (2020)