Depression and the gut-brain axis: what the microbiome has to do with it

You are sitting on the edge of the bed and the day lies ahead of you like a mountain. Not sad in the sense of crying. More like switched off. And your gut is not playing along either, and has not been for weeks.

At some point you type three words into your phone at night. Gut, serotonin, depression. And the first page tells you that 95 percent of your happiness hormone sits in your gut.

Many people know this pattern. The sentence sounds like an explanation. It sounds like a door nobody had opened before. And it is even half right.

It just does not hold. The serotonin from your gut never arrives in your head. There is a connection between gut and mood, quite an elegant one in fact, but it runs differently. And that matters, because a great deal of product advertising has been built on that one wrong shortcut.

I am writing this article as a physician who takes both seriously: psychiatry with its restraint and microbiome research with its curiosity. Both have good reasons. What I give you here are the numbers, where the numbers come from, and the points at which they stop holding.

Four lines between gut and head

Do you know the feeling that your gut already knows before your head has put it into words? Before an exam, before a difficult conversation, before a message you do not want to open.

That is neither imagination nor metaphor. Between gut and brain there are four established communication routes, and they all work in both directions.

Bidirectional is the decisive word here. Stress changes the composition of your gut flora. And the composition of your gut flora changes how your stress axis answers.

That is uncomfortable, because it destroys the simple narrative. There is no clean cause and no clean consequence. There is a circle, and you can enter it at several points.

How the axis works in detail and how the vagus nerve can be addressed deliberately, I have written up separately. You will find the mechanism together with breathing, cold and transcutaneous stimulation in Gut-brain axis and vagus stimulation.

And now you know why the topic of gut and mood so rarely turns on a single cause. It turns on four lines that are all on the phone at once.

The correction: serotonin from the gut does not reach the brain

Now to the sentence that probably brought you here. It runs roughly like this: 95 percent of serotonin is made in the gut, so mood sits in the belly.

The first part is right. The second part does not follow from it.

And here comes the point that is missing from almost every popular account. For serotonin from enterochromaffin cells there is to this day no evidence that it crosses the blood-brain barrier. That is basic physiology, and the review by the group around Mayer in Los Angeles describes the blood-brain barrier as exactly the place that regulates how much information from the gut arrives upstairs at all.

Picture the blood-brain barrier as the bouncer at a club. Serotonin is on no guest list. It does not get in, no matter how much of it waits outside.

The serotonin in your brain is therefore made in the brain itself. The raw material for it is tryptophan, an amino acid from food. And tryptophan does come through the door.

The same applies to a second candidate. Bacteria in the gut make GABA, the most important damping messenger of the brain. By current understanding, this GABA does not reach the brain directly either. It might take influence through the enteric nervous system and possibly through the vagus, but not as a messenger that travels upstairs itself.

And now you know why a product advertising the happiness hormone from the gut does you no harm, but also does not explain what it claims to explain.

What mouse experiments show, and what they do not

When you read a dramatic sentence about gut and mind online, it very probably comes from an animal experiment. That is not an accusation. The strongest causal evidence in this whole field of research comes from the lab, and it really is impressive.

It just happens to be from the lab. That is why I mark every single one of these papers here as what it is.

And then comes the paper everything revolves around. It is from 2011, it stands in the Proceedings of the National Academy of Sciences, and it is the most elegant piece of evidence in this literature.

The third building block are the transfer experiments. They sound like science fiction and are nonetheless cleanly done.

Why a mouse does not have depression

Here comes the paragraph almost nobody writes. A mouse cannot be depressed. It can behave as though it were.

The best known test is called the forced swimming test. A mouse is placed in a water container it cannot get out of, and the time it spends no longer moving is measured. Less movement counts as depression-like.

That measures behaviour. It does not measure experience. No animal model captures hopelessness, guilt, the loss of meaning, or the thought that things would be better without oneself. Yet precisely these things make up the illness this article is about.

And now you know why I change the tone in the next section. In humans it gets quieter.

What studies in humans have found, and where they contradict each other

In humans you cannot take bacteria away and then look at what happens. You can count, compare and search for connections. That is weaker, but it is what we have.

And now the brake block

At this point you could write an enthusiastic article. Two large cohorts, a replicated finding, a plausible mechanism. There is only one paper that gets in the way, and it is the largest of them all.

Source: Nikolova VL et al. Perturbations in Gut Microbiota Composition in Psychiatric Disorders. JAMA Psychiatry. 2021;78(12):1343-1354. DOI: 10.1001/jamapsychiatry.2021.2573 [Meta-analysis, k=59]

Which test can do it then, and which values say anything?

Briefly: for the question of depression, none. A stool test can answer sensible questions about your digestion, about signs of inflammation in the gut and, in certain situations, about pathogens. What such a test can and cannot do is set out at length in Stool testing and PCR diagnostics.

For mood, by contrast, there is no value that carries a statement. If you are offered a microbiome test meant to classify your mental state, ask to be shown which study supports that classification. From what I can find, it does not exist. What you get is an interpretation and not a finding, and that holds regardless of who offers the test.

And now you know why the next section contains so many numbers. With probiotics it pays to look closely.

Probiotics and prebiotics: the numbers without the marketing

You are standing in the chemist in front of a shelf. One box says psychobiotic. Another says something about the gut-brain axis. And you think: worth a try.

That is a completely reasonable reaction. I only want to give you the numbers to go with it, so you know what you are getting into.

The last row is the most honest data point in this whole field. Six studies, 341 people, all without psychotropic medication. A small but measurable advantage. And when you take out the studies funded by manufacturers or containing added substances, a value remains that statistically says nothing any more.

Which strain is the right one, and what about psychobiotics?

The term psychobiotic comes from research and describes bacterial strains credited with an influence on mood, anxiety or the stress response. It is not a quality seal and not a regulatory classification.

On the strain question there is an unsatisfying but honest answer. Goh 2019 found multi-strain preparations more favourable on average than single strains. Robust direct comparisons of individual strains in depression do not exist. Everything claimed beyond that is extrapolation.

One figure circulates particularly stubbornly online: a 41 percent symptom reduction through a particular Bifidobacterium strain. I know of no primary study from which exactly this figure for this endpoint follows.

What does exist is a small Canadian study in 44 people with irritable bowel syndrome and mild to moderate anxiety or mood symptoms. Under Bifidobacterium longum NCC3001, the depression score on the HAD scale fell by at least two points in 14 of 22 people after six weeks, against 7 of 22 under placebo, at p equal to 0.04. On anxiety and on bowel symptoms the preparation had no significant effect. This is a pilot study in people with irritable bowel syndrome and not a depression study, and staff of a manufacturer were involved as co-authors (Pinto-Sanchez 2017, DOI 10.1053/j.gastro.2017.05.003). So if a percentage figure crosses your path, ask for the study behind it before it convinces you.

Whether a preparation even arrives alive where it is supposed to go is a question of its own. Spore form, capsule, passage through stomach acid and storage are covered in Probiotics: spore form or capsule.

Prebiotics: the second route, and what fibre has to do with it

A second human finding comes from imaging. A group around Tillisch had healthy women eat a fermented milk product with probiotic twice daily for four weeks in 2013, compared with a non-fermented product and with no intervention. In functional magnetic resonance imaging, the probiotic group showed an altered response in a network of affective, viscerosensory and somatosensory cortical areas. There were twelve women in the intervention group, and what was measured was brain activity, not an improvement in mood.

What the microbiome gets to eat at all is covered in Prebiotics and resistant starch. And what is established for sauerkraut, kefir and kimchi in Fermented foods.

And now you know why I get louder in the next section. With food the data look different.

The nutrition trail carries furthest

If I were allowed to take one single thing from this whole field, it would be this section. Not because food were magic, but because here there are randomised trials with a clinical endpoint.

What I take from this for practice, and what I do not

I deliberately give you no meal plan here and no food list with amounts. That would be a prescription, and prescriptions belong in a conversation with a history taken, not in a blog article.

What can be derived from the literature is a direction. SMILES was about a way of eating with more unprocessed foods, more plant variety and fewer heavily processed products. The mechanism that fits it runs through fibre, through the bacteria that turn fibre into short-chain fatty acids, and through inflammatory signals.

Which foods are related to inflammatory markers in studies is covered in Inflammatory foods. On the ketogenic diet in mental illness there is a separate text with the current, still thin body of evidence: Ketogenic diet in mental illness.

And now you know why the nutrition trail is the only one where I speak of evidence without a knot in my stomach.

The inflammation trail and the subgroup idea

There is an observation many people know from their own experience. In a heavy bout of flu, not only strength disappears. Interest disappears too. You want to see nobody, plan nothing, feel nothing.

This behaviour has a name in research: sickness behaviour. It is triggered by inflammatory messengers and it looks astonishingly similar to a depressive episode. From that arose the question of whether inflammation plays a part in some people even without an infection.

And still the subgroup idea stays in the race, because there is a randomised trial on it. It is small, it is old and it remains to this day the most interesting paper in this area.

So does my CRP value tell me anything?

On its own, no. A CRP value can neither confirm nor rule out a depression. It can raise a question, particularly when it stays raised over a longer period and nobody has looked for the reason.

In clinical practice I see that with longer-standing exhaustion accompanied by abdominal complaints, a look at inflammatory values sometimes opens a trail nobody had followed before. That is my observation and not a study result, and it does not speak against questionnaires, which are well established for tracking the course of a depression. I know the counterexamples from my own consulting room too.

Where inflammation, gut and exhaustion run together I have described separately: Burnout, gut, inflammation and mitochondria. On the stress axis itself there is Cortisol and the HPA axis. And if you want to know how your omega 3 status is actually measured rather than estimated: Measuring the omega 3 index.

And now you know why the next section is the most important one for many people. It is about the combination that gets overlooked most often.

When the gut is chronically ill: irritable bowel syndrome and IBD

Imagine planning your day around toilets. You cancel arrangements because you do not know how your gut will decide tomorrow morning. And at some point you are not only exhausted but also sad.

Many people know this pattern. And it is not a weakness of character but well documented.

Gut-directed hypnotherapy: what it has shown, and what it has not

At this point there is a German guideline recommendation, though not where you would expect it. In its 2021 version, the S3 guideline on irritable bowel syndrome from DGVS and DGNM listed gut-directed hypnosis as an evidence-based effective procedure in irritable bowel syndrome [Guideline]. That version is approaching the end of its regular validity period and a revision is due. The statement therefore comes from the most recently published version and may change with an update. On depression it says nothing, because that is not what it is there for.

Source: Devenney J et al. Clinical trial: predictive factors for response to gut-directed hypnotherapy for refractory irritable bowel syndrome, a post hoc analysis. Aliment Pharmacol Ther. 2024;59(2):269-277. DOI: 10.1111/apt.17790 [RCT, post hoc analysis, n=448]

What gut-directed hypnotherapy looks like in practice is covered in Gut-directed hypnotherapy in irritable bowel syndrome. On the search for causes in irritable bowel syndrome there is Irritable bowel: finding causes, and on inflammatory bowel disease IBD viewed integratively.

And now you know why at the end of this article I do not come with a solution but with limits.

The honest limits, and why the restraint has good reasons

Three sentences I consider so important that I write them out separately. They are the core of what this article promises.

Why no test can do it

Two reasons, both established. The first is the transdiagnostic pattern from the meta-analysis by Nikolova 2021: the same shifts in depression, bipolar disorder, psychosis and anxiety. A marker that is conspicuous in everything is good for nothing in particular.

The second reason is a confounder that often gets overlooked.

That antidepressants and other drugs can produce gastrointestinal side effects belongs in the same picture. It is a reason to talk about it, not a reason to stop anything. Which drug classes can do what to the gut is covered in Medication and the gut.

What the guidelines say, and why they are so cautious

I like listing the reasons for this restraint, because they convince me too. No diagnosis-specific pattern. Confounding by medication. Closeness to industry, visible in the fact that one meta-analysis lost its significance as soon as industry-funded studies were excluded. Small, short studies with self-acknowledged low certainty of evidence. And an open direction of causality.

Which came first?

In humans this question is almost never answerable. There are data with a temporal order, and they are interesting, but they prove no cause.

A group around Lurie examined 202,974 people with depression in a large British primary care database in the Journal of Clinical Psychiatry in 2015, each with four matched controls. A single course of antibiotics was associated with an adjusted odds ratio of 1.23 for penicillins, and with more than five courses the value rose to 1.56. For psychosis there was no association. The authors adjusted for the number of infection events, and even so the obvious explanation remains that whoever is ill more often also becomes depressed more often, for many reasons.

The most radical approach to clarifying the direction would be faecal transplantation. An Australian group around Green randomised 15 adults with moderate to severe depression in the Canadian Journal of Psychiatry in 2023, ten receiving a transfer by enema and five a placebo enema. There were no serious side effects and not a single dropout. The study was explicitly not designed to measure efficacy. Whoever reads somewhere that faecal transplantation is established in depression is reading past that sentence.

And now you know why I do not end this article with a promise but with questions you can ask.

Frequently asked questions

Where this topic connects to the rest of the body

Gut and mood do not stand alone. They hang on sleep, iron, inflammation, the nervous system and on the question of how permeable a mucous membrane happens to be. These texts belong with it, in my view.