Elevated microbially-derived metabolites in autism

 

 

 

A new study reports that many children with autism have elevated levels of microbially-derived metabolites (MDMs) in their urine. The authors propose that this pattern is so common that it defines a distinct subtype of autism, which they call ASD-MDM (Autism Spectrum Disorder associated with Microbially-Derived Metabolites).

The authors claim that approximately 90% of autistic children have ASD-MDM and also suggest that ASD-MDM is a distinct subtype of autism. But that would mean almost all autism is ASD-MDM, so it would not really be a focused sub-type. 

It is striking that there are 22 authors listed, but only 52 ASD children studied. There are some familiar names among the 22.

Elevated microbially-derived metabolites in autism: a possible diagnostic screening test for a distinct ASD phenotype

The study is interesting and deserves attention. However, like many autism studies, it raises as many questions as it answers.

 

What did the researchers find?

The researchers measured a range of metabolites produced by gut bacteria and yeasts in the urine of 52 children with autism and 47 typically developing controls.

The metabolites fell into three broad categories:

• Phenylalanine and tyrosine-derived metabolites such as p-cresol and p-cresol sulfate
• Tryptophan-derived metabolites such as indoxyl sulfate and various indole compounds
• Yeast-associated metabolites such as arabinitol

Many of these compounds were significantly elevated in the autism group.

The most convincing findings involved p-cresol, p-cresol sulfate, phenylacetylglutamine and indoxyl sulfate. These metabolites have been reported repeatedly in previous autism studies and are among the best-replicated metabolic findings in the autism literature.

Using a scoring system based on the number of metabolites exceeding the range seen in any control child, the authors reported that approximately 78–90% of children with autism had at least one markedly elevated microbial metabolite.

 

What is new?

The most important contribution of this study is not the individual metabolites. We have known about elevated p-cresol for many years. It has been covered extensively in previous posts and in Stephen’s comments.

The novelty lies in combining multiple microbial metabolites into a single framework and proposing that they collectively define a biological subtype of autism.

This is an attractive idea.

Autism is clearly not a single disorder. Two people can receive the same diagnosis while having entirely different underlying biology. One person may have a monogenic disorder, another a mitochondrial dysfunction, another a channelopathy, and another an immune-mediated condition.

The notion that a substantial subgroup of autistic children may have a characteristic pattern of microbial metabolites is therefore entirely plausible.

 

Reasons for caution

The authors make some ambitious claims regarding diagnosis and screening. Several limitations should be kept in mind.

First, the study was very small, involving just under one hundred participants. This is adequate for a pilot study, but much too small to establish a diagnostic test with confidence.

Second, the control group was unusual. The autism group was predominantly male, which is expected, but the control group contained more females than males. This creates the possibility that some of the observed differences may be influenced by sex-related differences rather than autism alone. Comparing autistic boys with very restricted diets to typical girls with rich varied diets, springs to mind.

Sex Distribution of Study Participants

	ASD Group	Typically Developing (TD) Controls
Male	41	20
Female	11	27
Total	52	47
Male (%)	79%	43%
Female (%)	21%	57%
Male:Female Ratio	3.7:1	0.74:1

Third, the study collected no information on diet or medication use. This is a major limitation. Many autistic children have restricted diets, gastrointestinal problems, food selectivity, supplements or medications that can influence both the microbiome and the metabolome. Without these data, it is difficult to determine how much of the observed metabolic profile is attributable to autism itself.

Diet is one of the strongest known determinants of microbial metabolism. Many young autistic children, particularly those with more severe autism, consume highly restricted diets consisting of a small number of preferred "safe foods", often ultra-processed foods and very little dietary fiber. Such eating patterns can profoundly alter both the composition of the gut microbiome and the metabolites it produces. A boy whose diet consists largely of chicken nuggets, fries, white bread and sweetened drinks may be expected to have a very different microbiome from a girl consuming a varied diet rich in fruit, vegetables and fiber, regardless of whether either child has autism.

Super Size Me was a 2004 documentary by Morgan Spurlock in which he ate only food from McDonald's for 30 days.

The rules included:

Every meal had to come from McDonald's.

If asked whether he wanted to "super size" a meal, he had to accept.

He tried to eat three meals a day.

He reduced his exercise to match the average American activity level.

By the end of the month he reported:

·        Weight gain of about 11 kg (24 lb)

·        Increased cholesterol

·        Abnormal liver function tests

·        Reduced energy

·        Mood changes

If it had been 2026, they would have analyzed changes to his microbiome and looked at his urine metabolites. You can imagine the results.

The film became very influential and helped draw attention to the health effects of fast food.

 

Fourth, the study did not directly examine the gut microbiome. Instead, it measured microbial metabolites excreted in urine. Elevated urinary metabolites may reflect altered microbial activity, but can also be influenced by intestinal permeability, liver metabolism, sulfation capacity and kidney function. The study therefore provides direct evidence of altered metabolite profiles, but only indirect evidence of gut dysbiosis.

Finally, this was largely a study of classic childhood autism rather than the full autism spectrum. The participants were predominantly male and had an average CARS score of 41, consistent with substantial autistic symptoms  (A CARS score above about 37 is generally considered severe autism). The findings therefore cannot automatically be generalized to those with Level 1 or 2 autism, or those diagnosed later in life. It remains possible that elevated microbial metabolites are particularly common in children with more severe autism and gastrointestinal dysfunction.

 

Only urine was tested

An important limitation of this study is that the researchers did not directly examine the gut microbiome itself. They analyzed urine samples and measured concentrations of metabolites thought to be produced by gut bacteria or yeasts, such as p-cresol sulfate, p-cresol and indoxyl sulfate. This approach was chosen because these metabolites may provide a functional readout of microbial activity and can be measured using a simple, non-invasive urine test. However, elevated urinary metabolites do not necessarily prove the presence of gut dysbiosis, since their levels can also be influenced by diet, intestinal permeability, liver metabolism, sulfation capacity and kidney excretion.

A stronger study would have combined urinary metabolomics with stool microbiome sequencing, dietary assessments, medication histories and measurements of gastrointestinal symptoms. Such an integrated approach would have helped determine whether the abnormal metabolites truly arose from altered microbial populations and whether specific bacteria or fungi were responsible. Therefore, while the study provides convincing evidence that many autistic children have abnormal patterns of microbial metabolites, it provides only indirect evidence that gut dysbiosis itself is the underlying cause, and its conclusions should be interpreted accordingly.

  

Cause or consequence?

This is perhaps the most important question.

The paper often implies the following sequence:

Gut dysbiosis → microbial metabolites → autism

But the reverse sequence is also possible:

Autism → altered diet, gut motility and gastrointestinal function → microbial metabolites

The study cannot distinguish between these possibilities.

To demonstrate causation, researchers would need to identify elevated metabolites before autism symptoms emerge and show that those metabolites predict later diagnosis.

That would be a much stronger result.

 

Why this matters

Despite the limitations, this study fits remarkably well with a growing body of evidence suggesting that gut-derived metabolites can influence brain function.

P-cresol is particularly noteworthy because it has been associated with mitochondrial dysfunction, immune activation, impaired intestinal barrier function and behavioural abnormalities in animal models.

The repeated appearance of p-cresol and related compounds across many studies suggests that these findings should not be dismissed.

What remains unclear is whether these metabolites are merely biomarkers or whether they actively contribute to symptoms.

 

The broader perspective

Readers of this blog will know that I have never viewed autism as a single condition with a single treatment. Instead, I view autism as a behavioural diagnosis that sits on top of multiple underlying biological disorders.

Some people may have:

• Mitochondrial dysfunction
• Ion channel dysfunction
• Folate pathway abnormalities
• Neuroinflammation
• Gastrointestinal dysfunction
• Microbial metabolite abnormalities

and often several of these at the same time.

The goal should not be to debate whether autism is genetic or environmental, neurological or gastrointestinal.

The goal should be to identify the specific abnormalities present in each individual and address them where possible.

This study adds weight to the argument that microbial metabolism deserves investigation as part of that process.

 

The clinically important question

The most interesting question is not whether microbial metabolites can help diagnose autism.

The most important question is whether reducing abnormal metabolites improves symptoms.

If a child has markedly elevated p-cresol sulfate or indoxyl sulfate, can we normalize those levels?

If we do, does language improve? Does anxiety improve? Do gastrointestinal symptoms improve? Does adaptive functioning improve?

Those are the questions that matter to families.

The authors point to previous studies of microbiota transfer therapy that reported reductions in p-cresol sulfate accompanied by improvements in gastrointestinal and autism-related symptoms. Whether those findings can be replicated in larger controlled studies remains to be seen.

 

A look at the detailed results

 

Looking more closely at Tables 4 and 5

The paper presents two sets of metabolite data that are easy to confuse. Table 4 contains results from the initial semi-quantitative (untargeted) metabolomics analysis, while Table 5 contains results from the subsequent quantitative (targeted) analysis using authentic chemical standards.

Readers should focus primarily on Table 5, because it represents the validation phase of the study. Table 4 was designed to identify potentially interesting metabolites, but untargeted metabolomics is prone to both measurement error and occasional metabolite misidentification. In contrast, the metabolites in Table 5 were measured directly against known standards, allowing both their identity and concentration to be determined with much greater confidence.

In simple terms, Table 4 generated the hypotheses, while Table 5 tested them.

One of the most interesting aspects of the paper is that some dramatic findings from Table 4 became much less impressive in Table 5. Several tryptophan-derived metabolites appeared to increase by more than 1000% in the discovery phase, but these effects were greatly reduced or no longer statistically significant when measured using quantitative methods. This is not unusual and illustrates why validation studies are so important.

On the other hand, some findings survived the transition from discovery to validation. Most notably, p-cresol, p-cresol sulfate, phenylacetylglutamine and indoxyl sulfate remained significantly elevated in the autism group. These are therefore the metabolites that deserve the greatest attention.

The overall picture from comparing Tables 4 and 5 is that the evidence for widespread abnormalities in microbial metabolism remains convincing, but the evidence for some individual metabolites is weaker than the headline figures from the discovery phase might suggest. The quantitative data in Table 5 provide the most reliable basis for interpreting the study and assessing its clinical relevance.

 

Useful observations from Tables 4 and 5

1. p-Cresol survives both discovery and validation

The strongest finding is not a new metabolite but an old one.

In Table 4:

• p-Cresol increased by 151%
• p-Cresol sulfate increased by 54%

In Table 5:

• p-Cresol increased by 76%
• p-Cresol sulfate increased by 139%

Many findings became weaker during quantitative validation, but p-cresol and p-cresol sulfate remained significant. This strengthens the case that elevated p-cresol metabolism is a genuine feature of a subgroup of autistic children.

 

2. Phenylacetylglutamine may deserve more attention

Phenylacetylglutamine (PAGln) is increasingly recognized as a microbiome-derived metabolite with important biological effects.

In Table 4:

• 64% increase
• 32% of ASD children exceeded the highest control value

In Table 5:

• 80% increase
• Highly significant (p = 0.002)

Compared with p-cresol, PAGln receives relatively little attention in autism research but may prove to be an important marker of altered aromatic amino acid metabolism.

 

3. Tryptophan metabolism appears abnormal in many children

Although the individual metabolites differed between the two analyses, the overall signal remained.

The study reports:

• 64% of ASD children with elevated tryptophan metabolites in Table 4
• 42% in Table 5

This suggests that altered microbial metabolism of tryptophan may be common in autism. This is particularly interesting because tryptophan is the precursor of serotonin, melatonin and kynurenine pathway metabolites.

 

4. Indoxyl sulfate deserves attention

Indoxyl sulfate is another well-known microbial metabolite.

In Table 5:

• 171% increase
• Statistically significant (p = 0.03)

Like p-cresol sulfate, it has been linked to inflammation, oxidative stress and mitochondrial dysfunction. It may be one of the more biologically important findings in the study.

 

5. The abnormalities are highly heterogeneous

Perhaps the most important finding is that no individual metabolite identified most autistic children.

For example:

• p-Cresol sulfate: 21% above the control range
• p-Cresol: 19%
• Hydroxybenzoic acid: 17%
• Indole-3-acryloyl glycine: 17%
• Arabinitol: 10%

Different children had different abnormalities. This strongly supports the view that autism consists of multiple biological subtypes rather than a single disorder with a single biochemical signature.

 

6. The yeast findings are relatively weak

The paper devotes considerable attention to yeast metabolites, but the quantitative data are less convincing.

Only arabinitol remained significant in Table 5. Other proposed yeast markers, including citramalic acid, tartaric acid and tricarballylic acid, were not statistically significant.

The results support the existence of a yeast-associated subgroup, but not a major role for yeast in most autistic children.

 

7. The quantitative data support a lower prevalence than the headline claim

The paper's headline message is that approximately 90% of autistic children have elevated microbial metabolites.

However, the quantitative data suggest:

• 57% with elevated phenylalanine-related metabolites
• 42% with elevated tryptophan-related metabolites
• 16% with elevated yeast metabolites
• 78% with at least one elevated microbial metabolite

The validated figure is therefore closer to 78% than 90%.

 

8. A possible aromatic amino acid subtype of autism

Taken together, the clearest pattern involves metabolites derived from phenylalanine, tyrosine and tryptophan.

These amino acids are precursors for important neurotransmitters including:

• Dopamine
• Noradrenaline
• Serotonin
• Melatonin

The study therefore suggests that a substantial subgroup of autistic children may have altered microbial metabolism of aromatic amino acids. This broader observation may ultimately prove more important than any individual metabolite measured in the study.

What matters clinically?

The most important question raised by these findings is not whether they can be used to diagnose autism. The more important question is whether these metabolites are merely biomarkers or whether they contribute directly to symptoms.

If elevated p-cresol sulfate, p-cresol, phenylacetylglutamine or indoxyl sulfate prove to be biologically active drivers of symptoms, then they become potential treatment targets. This would fit with a growing body of evidence suggesting that at least some forms of autism involve treatable metabolic and physiological abnormalities.

From a personalized medicine perspective, the most valuable contribution of this study is not the proposed diagnostic test but the identification of potentially actionable metabolic pathways that may define a distinct subgroup of autistic individuals.

 

Conclusion

This study provides further evidence that abnormal microbial metabolites are common in autism and may define a biologically meaningful subtype.

The findings are intriguing and broadly consistent with decades of research on p-cresol and other gut-derived compounds.

However, the study does not prove that gut dysbiosis causes autism, nor does it establish a clinically validated screening test.

What it does provide is another reminder that autism is heterogeneous and that meaningful progress is likely to come from identifying and treating specific biological abnormalities rather than assuming that all autistic people share the same underlying pathology.

For those interested in personalized medicine, that is perhaps the most important message of all.

 

How might altered microbial metabolism of aromatic amino acids be treated?

The study suggests that a substantial subgroup of autistic children have abnormal microbial metabolism of the aromatic amino acids phenylalanine, tyrosine and tryptophan, leading to elevated levels of compounds such as p-cresol, p-cresol sulfate, phenylacetylglutamine and indoxyl sulfate. While no proven treatment exists specifically for this metabolic pattern, several approaches could potentially be helpful.

The most obvious strategy is to modify the gut microbiome itself through dietary changes, prebiotics, probiotics, synbiotics or, in selected cases, Microbiota Transfer Therapy (MTT). The goal would be to reduce production of potentially harmful metabolites and encourage a healthier microbial ecosystem.

Another approach is to increase populations of beneficial bacteria that preferentially ferment dietary fiber into short-chain fatty acids such as butyrate rather than producing aromatic metabolites.

Improving intestinal barrier function may also reduce absorption of microbial metabolites into the bloodstream. Compounds such as butyrate and some probiotics have been proposed for this purpose.

Since several of the metabolites identified in the study are sulfate conjugates, supporting sulfation and glutathione pathways through interventions such as NAC or taurine may also deserve further investigation.

Because p-cresol and related compounds have been linked to oxidative stress and mitochondrial dysfunction, mitochondrial support therapies may help reduce downstream effects even if they do not address the underlying source of the metabolites.

Finally, gastrointestinal motility should not be overlooked. Chronic constipation increases the time available for bacterial fermentation of amino acids and may contribute to the production of p-cresol and related compounds. Treating constipation and other gastrointestinal problems may therefore be an important part of the solution.

At present, the evidence is strongest for identifying these metabolites as biomarkers rather than proven treatment targets. The key question for future research is whether reducing elevated microbial metabolites leads to meaningful improvements in autism symptoms, gastrointestinal function and quality of life.