We put food in. A remarkably complicated biological system happens in the middle.
And poo comes out.
Yet when we study stool, we often treat it primarily as a convenient container for something else.
We extract microbial DNA. We sequence it. We measure metabolites. We look at short-chain fatty acids. Then we try to understand what the microbiome is doing.
But what if the stool itself is part of the signal?
Food entering the gastrointestinal tract is digested, absorbed, transformed and fermented. It interacts with host secretions, bile acids, enzymes, intestinal cells and trillions of microorganisms. Some components are absorbed. Some are transformed. Some are consumed by microbes. Some leave the body.
Stool is the physical output of that entire system.
Food in → human + microbial system → stool out.
So why don’t we routinely characterise the output???
A gram of stool is not necessarily a gram of stool
How much water is in it? How much fat, protein or residual dietary material?
What is its pH? Its microbial biomass? Its short-chain fatty acid profile? Its bile acids? Its metabolites?
Two people can each provide one gram of stool, but those grams may represent quite different biological environments.
That matters because many of our measurements are normalised to stool mass.
A concentration reported per gram of wet stool quietly assumes that a gram of one person’s stool is meaningfully comparable with a gram of another’s.
Perhaps it isn’t.
And if it isn’t, some of the variability we attribute to the microbiome may actually reflect the matrix from which we extracted it.
I have started thinking about this as the Stool Matrix Hypothesis: stool composition is not simply background material surrounding the microbiome. It is an integrated biological output of diet, digestion, absorption, intestinal transit, host biology and microbial metabolism.
A gram of poo is rarely a gram of poo
Then I started thinking about faecal microbiota transplantation
This is where the idea became much more interesting to me. We call it faecal microbiota transplantation (FMT) , which naturally focuses attention on the microbes.
But we don’t transplant microbes in isolation. We transplant stool-derived material.
Depending on the protocol, that might be fresh or frozen stool, a slurry delivered by colonoscopy, an enema, material delivered through the upper gastrointestinal tract, or processed material contained in capsules.
And the clinical results are fascinatingly variable.
For recurrent Clostridioides difficile infection, FMT can be extraordinarily successful. In one randomised trial, frozen FMT delivered as 40 oral capsules achieved the same 96.2% prevention of recurrence as 360 mL of faecal slurry delivered by colonoscopy.
Move into more complex conditions, however, and things become considerably less predictable.
In ulcerative colitis trials, remission rates have varied substantially. Some studies have shown clear benefit over placebo, while others have not. There are even reports of pronounced donor effects, helping generate the idea of the so-called “super-donor”.
The metabolic studies are perhaps even more intriguing.
Transferring stool from lean donors to people with obesity or metabolic syndrome has sometimes produced short-term improvements in insulin sensitivity. Yet other studies using repeated oral FMT capsules have achieved measurable microbial engraftment without meaningful improvement in insulin sensitivity, body composition or weight.
The bugs moved in. The physiology did not necessarily follow.
That makes me wonder whether our model of FMT is sometimes too simple.
We often think:
Donor microbiome → recipient microbiome → clinical outcome
But biologically, the intervention looks more like:
Donor diet + digestion + metabolism + host biology + microbiota → stool → processing + storage + delivery → recipient ecosystem → outcome
There are an extraordinary number of things happening in that chain.
Research has quite reasonably focused on microbial diversity, donor-recipient compatibility, engraftment, donor selection, preparation methods, dose and route of administration.
But I keep coming back to another question.
What are we actually transferring?
Alongside microbial cells, donor stool contains microbial metabolites, bile acids, host-derived molecules, water, dietary residues and probably a great many biological components we rarely characterise when selecting a donor.
Could two donors with apparently desirable microbiomes produce very different biological transplant material?
Could the physical or biochemical characteristics of the stool affect microbial viability?
Could they influence engraftment?
Could metabolites or other stool-derived molecules contribute directly to an early biological response?
Could some of what we currently call a “donor effect” actually represent the wider biological system that produced the donor stool?
I don’t know.
And importantly, the FMT studies do not currently prove this.
There are already good explanations for variability, including recipient baseline microbiota, donor microbial composition, disease phenotype, treatment intensity, preparation and engraftment.
But that doesn’t make the matrix irrelevant.
It makes it something worth measuring.
Maybe the stool deserves to be phenotyped too
We phenotype humans extensively.
We record age, BMI, diet, medications, disease status and blood biomarkers.
We sequence their stool in exquisite detail.
But the physical sample sitting between the person and the sequencing machine can remain surprisingly poorly characterised.
Perhaps stool itself needs a phenotype.
Moisture, dry matter, lipid, protein, residual fibre, pH, microbial biomass, short-chain fatty acids, bile acids, and metabolites.
Potentially extracellular vesicles and other biologically active material.
Then we can begin asking whether those characteristics alter the biological signals we measure, or the effects we observe when stool-derived material is transferred into another host.
The Stool Matrix Hypothesis may turn out to explain very little.
That is entirely possible. But before we conclude that all meaningful information in stool resides in its microbial taxonomy, perhaps we should characterise the system’s output properly.
Because sometimes the poo might not just contain the answer.
It might be part of the answer.
Research that prompted this thinking
Kao D, et al. JAMA. 2017. Oral capsule versus colonoscopy-delivered FMT for recurrent C. difficile. doi:10.1001/jama.2017.17077.
Kootte RS, et al. Cell Metabolism. 2017. Lean-donor FMT and insulin sensitivity in metabolic syndrome. doi:10.1016/j.cmet.2017.09.008.
Yu EW, et al. PLOS Medicine. 2020. FMT-TRIM trial of oral FMT capsules in obesity. doi:10.1371/journal.pmed.1003051.
Smith BJ, et al. Scientific Reports. 2022. FMT delivery mode, engraftment and ulcerative colitis. doi:10.1038/s41598-022-09307-5.
Wilson BC, et al. Frontiers in Cellular and Infection Microbiology. 2019. The super-donor phenomenon in FMT. doi:10.3389/fcimb.2019.00002.