Gut-derived metabolites, neuropsychiatric health, and the markers worth reading on the iollo panel
Patients rarely walk into my office saying their gut bacteria are making them anxious. They say they cannot focus, that their mood has flattened, that sleep has become unreliable, or that a parent developed dementia and they want to know what their own trajectory looks like. Increasingly, part of the answer to those questions is being written in the colon and delivered to the brain through the bloodstream.
The microbiota-gut-brain axis is no longer a fringe idea. What has changed in the last several years is not the concept but the measurement. Blood metabolomics now lets us quantify the specific molecules that gut bacteria manufacture and release into circulation, several of which cross the blood-brain barrier or act on receptors that shape neuroinflammation, neurotransmitter availability, and neuronal energy metabolism. The iollo panel, which quantifies 624 metabolites by mass spectrometry from a small blood sample, includes a meaningful subset of these gut-derived signals. I have written separately about why I use this panel in practice. Below is how I read them.
How the gut talks to the brain
There are four well-characterized channels. The vagus nerve carries signals from enteric sensory neurons and enteroendocrine cells directly to the brainstem. The immune system responds to bacterial products crossing a compromised intestinal barrier, generating cytokines that alter brain function. The hypothalamic-pituitary-adrenal axis is calibrated in part by early microbial colonization and remains responsive to gut inflammation throughout life. And finally, bacteria produce neuroactive metabolites, some beneficial and some frankly neurotoxic, which enter portal circulation, pass through the liver, and reach the brain.
That fourth channel is the one metabolomics captures. Stool sequencing tells us which organisms are present and what they could theoretically do. Blood metabolomics tells us what actually reached systemic circulation. Those are different questions, and for neuropsychiatric concerns the second one is often more clinically useful.
Tryptophan and the kynurenine pathway
Tryptophan is the shared substrate for three competing fates: serotonin and melatonin synthesis, bacterial conversion to indoles, and the kynurenine pathway. Inflammation shifts the balance decisively toward kynurenine by inducing indoleamine 2,3-dioxygenase, and the downstream metabolites are not neutral. Quinolinic acid is an NMDA receptor agonist and excitotoxin, while kynurenic acid is an NMDA antagonist with neuroprotective properties.
The iollo panel reports tryptophan and the kynurenine-to-tryptophan ratio, which is the clinical proxy for how aggressively inflammation is diverting tryptophan away from serotonin. A meta-analysis of 35 studies covering more than 4,600 subjects found significantly lower tryptophan and a lower tryptophan-to-competing-amino-acid ratio in severe depression, with quinolinic acid elevated and kynurenic acid reduced. The kynurenine-to-tryptophan ratio itself rose significantly in patients with psychotic features. In ADHD, a separate meta-analysis found higher kynurenine and lower kynurenic acid compared with controls, a pattern the authors interpreted as reduced neuroprotection.
One limitation matters here. iollo reports the kynurenine-to-tryptophan ratio directly, but the panel does not measure quinolinic acid, kynurenic acid, or 3-hydroxykynurenine. So it tells you that tryptophan is being diverted down this pathway without telling you whether the traffic is heading toward the excitotoxic branch or the neuroprotective one. That distinction is where much of the clinical meaning lives, and reading it requires a different assay.
One nuance the panel supports well: tryptophan competes with the large neutral amino acids, including phenylalanine, tyrosine, leucine, isoleucine, and valine, for the same transporter into the brain. Because iollo quantifies those amino acids alongside tryptophan, the ratio can be calculated rather than assumed. A normal tryptophan level in the setting of high competing amino acids is not the same thing as adequate delivery to the brain.
Indole-3-propionic acid
If there is a single marker on this panel I look at first for gut-brain concerns, it is indole-3-propionic acid. IPA is produced almost exclusively by gut bacteria, most notably Clostridium sporogenes, from tryptophan, and its production depends heavily on fermentable fiber intake. It is a potent antioxidant, an aryl hydrocarbon receptor and pregnane X receptor ligand, and it crosses the blood-brain barrier.
The human data are consistent and improving. IPA is reduced in obesity and type 2 diabetes and in nonalcoholic fatty liver disease. A 2025 study in Science Advances found lower serum IPA in patients with mild cognitive impairment and Alzheimer disease than in controls, with serum IPA correlating with MoCA and MMSE scores, and showed in transgenic mice that blocking the neuronal pregnane X receptor abolished the cognitive benefit of intermittent fasting. In migraine, serum IPA correlated inversely with attack frequency. Low IPA on a metabolomics panel is one of the most actionable findings in functional medicine, because it usually reflects a fiber and polyphenol deficit rather than a fixed trait.
The barrier, and why it matters upstream of everything else
The intestinal barrier deserves its own note, because it is the mechanism by which a gut problem becomes a brain problem. A single layer of epithelial cells, sealed by tight junctions and covered in mucus, separates roughly a hundred trillion microbes from your bloodstream. When that seal loosens, bacterial cell wall components cross into circulation and the immune system responds. That response does not stay local. Circulating inflammatory signals activate microglia, the brain’s resident immune cells, and shift neurotransmitter metabolism in ways that show up as low mood, fatigue, and impaired concentration long before anything appears on a conventional test. Every mechanism described earlier in this article operates faster when the barrier is compromised.
No blood panel measures intestinal permeability directly, and this one is no exception. Zonulin, LPS-binding protein, and the lactulose-mannitol ratio remain separate tests. What iollo offers instead is a view of the conditions that maintain the barrier, and of the consequences when they fail. Indole-3-propionic acid is the most informative of these. Beyond its antioxidant and cognitive associations, IPA acts on the pregnane X receptor in the intestinal lining, where it suppresses inflammatory signaling and increases production of the tight junction proteins that hold the barrier together. Animals unable to make it develop a demonstrably leaky gut and higher circulating endotoxin, and levels are reduced in inflammatory bowel disease. Of major significance is the finding of elevated brain levels of endotoxin in Alzheimer’s diseased brains – up to 42x the level of healthy brains! Because IPA production depends on fermentable fiber, a low value describes a gut that is not being given the raw material to maintain its own barrier.
Three other markers fill in the picture. Citrulline is produced almost exclusively by small intestinal cells, which makes it the closest thing here to a measure of functioning absorptive tissue; low levels suggest there is less healthy intestinal lining to work with. Glutamine is the primary fuel those cells burn, and a low value means the barrier is being asked to maintain itself on a thin energy budget. And the kynurenine-to-tryptophan ratio sits at the other end of the sequence, because the inflammatory signaling that follows barrier failure is precisely what diverts tryptophan away from serotonin production. Read together, low IPA with low citrulline and glutamine and a rising kynurenine ratio describes a coherent story: a barrier without the substrate to maintain itself, fewer cells to do the maintaining, and the inflammatory consequence already measurable in circulation.
p-Cresol sulfate and indoxyl sulfate
These are the mirror image of IPA. Both are protein putrefaction products: p-cresol sulfate derives from bacterial fermentation of tyrosine and phenylalanine, indoxyl sulfate from the indole branch of tryptophan metabolism, and both are sulfated in the liver before entering circulation. They are classified as protein-bound uremic toxins, and their neurological relevance is not limited to kidney disease. Indoxyl sulfate promotes oxidative stress and endothelial dysfunction, which matters for cerebral small vessel integrity, and p-cresol has been repeatedly implicated in animal models of autism-like behavior through effects on dopamine metabolism and the gut barrier.
Clinically, high p-cresol sulfate and indoxyl sulfate with low indole-3-propionic acid is a recognizable pattern: a proteolytic, low-fiber fermentation profile. It typically responds to reducing excess animal protein, increasing resistant starch and diverse plant fiber, and adding polyphenol-rich foods that shift bacterial metabolism away from putrefaction. Renal function must be interpreted alongside these markers, since impaired clearance elevates both independently of gut ecology.
Secondary bile acids
Bile acids are made by the liver and then extensively modified by gut bacteria, which makes the primary-to-secondary ratio a direct readout of microbial enzymatic activity. The Alzheimer Disease Neuroimaging Initiative measured bile acid profiles in 1,464 subjects and found lower cholic acid and higher deoxycholic acid, including its glycine and taurine conjugates, in Alzheimer disease compared with cognitively normal older adults. The deoxycholic acid to cholic acid ratio, which reflects bacterial 7-alpha-dehydroxylation, associated strongly with cognitive decline, and the finding replicated in serum and postmortem brain samples from the Rush Religious Orders and Memory and Aging Project.
The panel measures deoxycholic acid along with cholic acid and a wide set of glycine- and taurine-conjugated bile acids, so this ratio can be tracked, though cholic acid currently sits in the raw-data tier rather than the processed report. Elevated deoxycholic acid also correlates with prolonged transit time and increased colorectal cancer risk, which makes it a useful dual-purpose marker.
TMAO, choline, and where I stay skeptical
Trimethylamine N-oxide is the marker most often cited in gut-brain marketing and the one where I would counsel the most restraint. Mechanistic work in mice is compelling, with TMAO administration producing cognitive impairment, hippocampal cell senescence, and neuroinflammation. But in the Rotterdam Study, with more than 3,100 participants for cognition and over 2,500 followed for incident dementia, plasma TMAO showed no association with cognition, neuroimaging markers, or dementia risk. It predicted dementia only in the subgroup with impaired renal function. A plausible mechanism and a validated biomarker are not the same thing.
What did associate with poor cognition, brain atrophy, and white matter hyperintensity volume was plasma choline, a TMAO precursor. Carnitine, another precursor and arguably the more efficient one given how readily an omnivore-adapted microbiota converts it, went the opposite way and tracked with lower white matter hyperintensity volume. That divergence argues that whatever choline is marking is not bacterial trimethylamine production. Three hypotheses are worth holding loosely. Choline is the head group of the phospholipids that dominate neuronal membranes, so a rise may be downstream of membrane breakdown rather than a cause of it. Alternatively, free choline may accumulate when its two consuming routes, phospholipid synthesis and oxidation to betaine for methyl donation, are functioning poorly, which would fit the white matter finding given the established link between homocysteine and small vessel disease. Confounding by renal, hepatic, and cardiometabolic status remains a third possibility. None of this is established.
The practical consequence is that an isolated elevated choline is not a reason to restrict dietary choline, and here the chemical form matters more than the amount. In a randomized trial, choline bitartrate tablets raised both fasting TMAO and platelet reactivity, while an equivalent quantity of total choline delivered as four large eggs daily raised neither. Phosphatidylcholine capsules, also a supplement, behaved like the eggs and raised neither. The dividing line is not whole food versus supplement but free versus esterified choline, and the dosing makes the mechanism visible: four eggs supplied roughly 467 mg of total choline that was almost entirely esterified, with negligible free choline. Free choline arrives at the colon as substrate for bacterial choline TMA-lyase, and esterified choline largely does not. Eggs are not the problem the TMAO literature is sometimes taken to indict, and if a patient is taking a choline supplement, the form on the label is the question worth asking. I read choline against betaine, and both against B12, folate, and B6 status, before drawing any conclusion.
The HPA axis, and kynurenine as an inflammation proxy
The stress axis is the third channel in the gut-brain model, and the panel reads it better than most people realize. Cortisol and cortisone together describe not just adrenal output but the activity of 11-beta-hydroxysteroid dehydrogenase, the enzyme pair that interconverts the active and inactive forms in tissue. A high cortisone relative to cortisol suggests active inactivation, and the reverse suggests local amplification, which is a different clinical situation than either value alone conveys. DHEA-S is measured as well, and the cortisol to DHEA-S relationship is the more informative reading. The two adrenal steroids act in opposition, cortisol catabolic and immunosuppressive, DHEA-S anabolic and immune-supporting, and their ratio has been studied as an index of catabolic-anabolic balance, chronic stress load, and the immunological drift of aging. A patient with normal cortisol and depressed DHEA-S is not a patient with a normal stress axis.
This connects directly to the tryptophan section, and the connection is the reason I read those markers together rather than separately. Tryptophan is diverted down the kynurenine pathway by two different enzymes with two different triggers. Indoleamine 2,3-dioxygenase is induced by inflammatory cytokines, interferon-gamma most prominently. Tryptophan 2,3-dioxygenase is induced by cortisol. So a rising kynurenine-to-tryptophan ratio is a downstream readout of both immune activation and HPA drive, arriving through separate doors.
That is genuinely useful on a panel with no cytokine measurements, because the ratio functions as an indirect index of inflammatory tone. It is also the limitation, and I want to be clear about it: an elevated ratio does not tell you which enzyme is responsible. Reading it alongside cortisol, cortisone, and DHEA-S is what begins to separate the two. A high ratio with elevated cortisol and suppressed DHEA-S points toward stress-driven diversion. A high ratio with an unremarkable steroid picture points toward immune activation, and that is when I would order actual inflammatory markers rather than continuing to infer them.
The supporting cast
Several other iollo markers inform the same axis without being microbial products themselves. Glutamine is the primary fuel of the enterocyte, and low levels suggest a compromised intestinal barrier, which is the upstream permission slip for immune-mediated neuroinflammation. Glycine and taurine are both bile acid conjugation partners and inhibitory neuromodulators, and taurine is separately tracked as a longevity marker. DHA speaks to membrane fluidity and the resolution of neuroinflammation. Spermidine, partly bacterial in origin, ties into autophagy and cognitive aging, and the ceramides and sphingomyelins on the panel connect to the lipid alterations reported in Alzheimer disease cohorts. Acylcarnitines report on mitochondrial fatty acid oxidation, and neurons are unforgiving of energy shortfalls.
Acetylcarnitine
One acylcarnitine deserves separate treatment. Acetylcarnitine is the acetylated form that crosses the blood-brain barrier readily, and it is the marker on this panel with the most direct evidence in depression. Plasma acetylcarnitine is reduced in major depressive disorder, and the deficit is largest in the patients who are hardest to treat: those with treatment-resistant illness, earlier age of onset, greater severity, and a history of childhood adversity. The proposed mechanism is epigenetic rather than purely bioenergetic, since acetylcarnitine supplies acetyl groups for histone acetylation and has been shown to modulate expression of the metabotropic glutamate receptor mGlu2 in animal models.
I would not present this as a diagnostic finding. The human work is observational, sample sizes are modest, and the supplementation trials in depression are heterogeneous and largely small. But when a patient presents with a low mood that has not responded to conventional approaches, a depressed acetylcarnitine sits at an interesting intersection of mitochondrial function, one-carbon metabolism, and the glutamatergic system, and it is one of the few findings on this panel with a plausible intervention attached to it.
Reading the pattern, not the marker
No single metabolite on this panel diagnoses anything. What is informative is the constellation. Low indole-3-propionic acid with elevated p-cresol sulfate and indoxyl sulfate, a rising kynurenine-to-tryptophan ratio, low glutamine, and an elevated deoxycholic acid to cholic acid ratio describe a coherent physiology: a fiber-poor, protein-heavy fermentation pattern producing neurotoxic metabolites, an inflamed barrier, and inflammation-driven diversion of tryptophan away from serotonin. That is a picture worth acting on, and every element of it is modifiable.
The intervention set is unglamorous and effective: substantially increased fiber diversity rather than fiber quantity alone, polyphenol-rich plant foods, moderation of excess animal protein, fermented foods, correction of digestive insufficiency and transit abnormalities, targeted probiotics where indicated, and attention to sleep and stress load given the cortisol-kynurenine link. Retesting in three to six months tells you whether the physiology moved, which is a more honest endpoint than whether the patient feels better on any given week.
The honest limitations
iollo is a wellness test, not a diagnostic one, and it is not a substitute for clinical evaluation. One practical note before the scientific caveats: the panel is tiered. Some metabolites appear in the processed report, while others are measured but currently released only as raw, unprocessed data, with reports rolling out over time. Several markers discussed here, including TMAO, cholic acid, glycine, valine, and homocysteine, sit in that second group at present, so what arrives in a given report may not match what the panel measures.
A related caution for anyone reading their own results. Serotonin and GABA both appear on this panel, and neither is a window into the brain. Circulating serotonin is overwhelmingly enterochromaffin in origin and does not cross the blood-brain barrier, and plasma GABA faces the same problem. They say something about gut and peripheral physiology and almost nothing about central neurotransmission, which is one of the most common misreadings in this field.
Most of the neuropsychiatric literature cited here is observational, which means association rather than causation, and several of the strongest findings come from disease cohorts rather than from healthy people being tracked forward. Renal function confounds the uremic toxins and TMAO. Single-timepoint metabolomics captures a moment, and diet in the preceding days moves several of these markers meaningfully. Interpreted with those caveats, though, this panel gives something functional medicine has wanted for a long time: a direct look at what the microbiome is actually sending to the brain.