Is the Gut Microbiome Really Causing Autism? A Scientist Breaks It Down
Neurogeneticist explains why claims that the gut microbiome causes autism are often unsupported by solid evidence.
The idea that the gut microbiome might play a causal role in autism has recently captured public imagination and fueled a wave of scientific and commercial interest.
Probiotics, dietary interventions and “gut-healing” protocols are frequently promoted as promising, or even proven, treatments.
But how much of this enthusiasm is grounded in solid evidence?
To examine that question, Technology Networks spoke with Dr. Kevin Mitchell, an associate professor of genetics and neuroscience at Trinity College Dublin. Mitchell’s research focuses on brain development, neurogenetics and the biological origins of individuality.
His recent paper critically evaluates claims about the microbiome’s role in autism, arguing that the evidence base is far weaker than many headlines suggest.
In this conversation, Mitchell discusses why the microbiome hypothesis took hold, the limitations of current research and what the data tells us about gut–brain relationships in neurodevelopmental conditions.
Why do you think the idea of a “gut cause” for autism has become so popular with the public and the media?
Kevin Mitchell, PhD (KM):
The idea that there is some simple, identifiable cause for autism is very appealing.
It’s intuitive and understandable, by contrast with the complex genetics of the condition, and it also seems to provide hope for intervention.
It also tracks an underlying idea around “wellness”, which sounds positive but can actually be quite insidious – that we are each individually responsible for our own health, especially by controlling what we eat.
For diabetes or heart disease, that’s partly true, but there’s no reason to think it applies to a genuinely neurodevelopmental condition like autism.
RLS:
What first made you question the idea that the gut microbiome might be causing autism?
KM:
The notion that the gut microbiome might be somehow causally involved in autism struck me as unlikely, given that all the evidence from known genetic causes implicates processes in fetal brain development.
I expected extraordinary evidence to be supplied to support what seemed like an extraordinary claim, but instead found obviously flawed studies.
RLS:
Your paper argues that the evidence for microbiome differences in autism is inconsistent and methodologically weak. Which issues did you find most pervasive across studies?
KM:
The observational studies claiming differences in the gut microbiome in people with autism vs control subjects suffer from numerous issues, including very small sample sizes (many in the tens when we now know they really should be in the thousands), exploratory analyses of a huge number of variables, very flexible analyses, lack of appropriate correction for so many tests and lack of a replication sample.
We know from other fields that these practices lead to spurious findings that don’t replicate. This is what we see across the microbiome studies – no consistent pattern emerges.
Published studies often present directly contradictory findings.
RLS:
When people say, “the gut microbiome is different in autistic people”, what does your review suggest is really going on?
KM:
First, we should ask what kinds of studies suggest that the gut microbiome is different in autistic people, and whether we should trust them.
Most of the studies making such claims had extremely small samples and many methodological and statistical problems, which make it very likely that any differences they observed were spurious.
This conclusion is supported by the fact that the reported patterns of differences are inconsistent – often directly contradictory – across different studies.
More recent studies with larger samples also failed to produce any specific, consistent findings.
The one thing they did agree on was that any differences in the microbiome that could be associated with autism status were very minor and that they mostly disappeared when the analyses took confounding factors (like diet) into account.
Overall, this suggests that most of the reported findings are just statistical noise that doesn’t replicate, and any real differences are very minor and likely reflect reverse causation (from behavioral differences to the microbiome, rather than the other way around).
RLS:
Were there any findings that genuinely surprised you when you went through this literature?
KM:
The one thing that surprised us was that the kinds of methodological problems that led to a highly publicised failure of reproducibility in other fields were being repeated in these microbiome studies.
These issues are very general and, by now, very well known, so it was a surprise to see them being ignored.
RLS:
Where do you see the biggest disconnect between preclinical models and the human condition?
KM:
Mouse “models” can be extremely useful for investigating the neurobiological mechanisms of neurological and psychiatric conditions. But we have to have some confidence they’re actually valid and reflect the same biology.
For something like psychosis, we can use drugs that induce a psychotic state in humans and see what state they induce in a mouse. If we can reverse that state with antipsychotics, then we can be confident it’s got some relevance to what’s going on in humans.
We don’t have any such drugs for the core symptoms of autism, which are really traits, not states. There are several behavioral tests in mice that many people take as relevant to autism, but that is based only on a superficial similarity with human behaviors, which is often quite a stretch (like marble-burying being a proxy for repetitive behaviors).
The relevance of these tests to autism has not really been validated at all.
RLS:
Do you see any legitimate role for mouse models in studying gut–brain interactions for neurodevelopmental conditions more generally?
KM:
I don’t see much good evidence that the gut-brain axis is causally involved in other neurodevelopmental conditions, either.
Gastrointestinal (GI) symptoms are more common across many such conditions, not just autism, but this does not imply a causal arrow from digestive issues to differences in brain function; both may be caused by whatever underlying genetic perturbations are present.
Of course, treating GI issues may make patients generally feel better and “improve” their behavior, but that’s quite different from thinking that the GI issues are the original cause of altered behavior.
It’s possible that mouse models could be used to study whether altering the microbiome can improve GI issues, with the important caveat that the anatomy and physiology of the digestive system of mice are different from that of humans, as is their gut microbiome.
And of course, [there is] the possibility that the GI symptoms themselves have nothing to do with the gut microbiome.
RLS:
How do conflicts of interest shape the narrative around autism treatments?
KM:
Scientific journals require a statement of commercial conflicts of interest because it means the researchers are not disinterested parties. This raises the concern that they are not just trying to objectively answer a scientific question; they also have a financial stake in how it turns out.
This is a particular concern with respect to autism and claims about the microbiome because many of the supposed treatments on offer, like probiotics, do not have to pass through the stringent regulatory processes applied to pharmaceuticals, meaning the path to commercialization is a lot faster and a lot easier.
All of which makes it reasonable to take such conflicts of interest into consideration when evaluating these kinds of claims.
