A pilot study found that elevated gut microbe-derived metabolites could identify many children with autism, raising the possibility of a simple urine-based screening tool and a newly proposed ASD subtype tied to microbiome dysfunction.
Study: Elevated microbially-derived metabolites in autism: a possible diagnostic screening test for a distinct ASD phenotype. Image credit: Prostock-studio/Shutterstock.com
A recent study published in Molecular Psychiatry investigated whether urinary concentrations of microbially derived metabolites (MDMs) can objectively distinguish children with autism spectrum disorder (ASD) from typically developing children.
ASD: Clinical complexity, microbiota associations, and unresolved mechanisms
ASD is a complex neurodevelopmental condition characterized by early-onset challenges in social communication, restricted interests, and repetitive behaviors. Symptom presentation and severity vary widely, leading to a spectrum ranging from individuals with profound support needs to those with relatively mild difficulties who can function independently. ASD arises from complex genetic-environmental interactions.
Over recent decades, ASD prevalence in the United States has increased dramatically, placing growing pressure on families, healthcare systems, and support services. Although around 10% of cases are associated with identifiable genetic syndromes, the underlying causes of most cases remain unknown. Given this heterogeneity, researchers increasingly view identifying biologically distinct ASD subtypes as a key step toward developing targeted therapeutic strategies.
Although early behavioral intervention is most effective within the first two years of life, diagnoses typically occur much later. This delay highlights the urgent need for non-invasive early screening tools to enable timely intervention and reduce long-term clinical and economic impacts.
A substantial subset of individuals with ASD experience chronic gastrointestinal (GI) symptoms, which often parallel ASD severity and emerge within the first three years of life. Consistent evidence demonstrates gut dysbiosis in ASD, with distinct microbial profiles compared to those of neurotypical individuals. This dysbiosis alters metabolic and immune pathways, including the production of short-chain fatty acids (SCFA), cytokines, and neurotransmitters, supporting a possible mechanistic link between gut microbiota and neurodevelopment via the gut-brain axis.
Microbial metabolites like p-cresol and indoxyl sulfate are found at higher levels in people with ASD. These compounds may harm gut health, immune function, and brain signaling, especially when present in large amounts early in life. Therefore, it is necessary to identify specific microbial and metabolic markers that could help distinguish different types of ASD and support earlier diagnosis or more targeted treatments.
Analyzing urinary metabolites to screen for ASD
The current study developed a biomedical screening test for ASD by measuring MDMs in the urine of children with ASD and typically developing (TD) children. A total of 52 children with ASD and 47 TD children between the ages of 2 and 11 were recruited from four U.S. sites. ASD diagnoses were confirmed by expert assessors using the Childhood Autism Rating Scale and the Social Responsiveness Scale-2 (SRS-2; score >68). Urine samples were collected.
Metabolite extraction and liquid chromatography–mass spectrometry (LC-MS) analysis were performed according to standard protocols, and metabolites were annotated and quantified relative to urinary creatinine. An initial untargeted LC-MS approach was used for discovery, followed by a targeted quantitative LC-MS follow-up analysis.
A novel multivariate analysis, the Microbially-Derived Metabolite System™ (MDM System™), was created to identify children with ASD who have intestinal dysbiosis. Each participant’s MDM concentration was compared to the TD range; scores reflected the number of elevated MDMs.
Distinct microbial metabolite profiles differentiate ASD from TD children
Both the TD and ASD groups were age-matched, and the TD group was purposely balanced for gender. No significant gender differences were found in metabolite analysis. Analysis focused on microbially produced metabolites, grouped as phenylalanine-derived, tryptophan-derived, or yeast/other.
Six phenylalanine-derived and eight tryptophan-derived metabolites were significantly elevated in ASD, with increases ranging from 29% to 1882%. Many ASD participants had metabolite levels above those of all TD cases. Arabinitol, a yeast metabolite, was also 51% higher in ASD, while N-formyl methionine was 70% lower. These findings highlight a distinct metabolic profile in ASD.
Most ASD participants had very high levels of tryptophan- or phenylalanine-derived metabolites, or both, compared to TD children. Elevated arabinitol and decreased N-formyl methionine often co-occurred in a subgroup of ASD participants. Except for N-formyl methionine, metabolite levels were generally higher in ASD.
The MDM System™ Total Score, which reflects the number of highly elevated metabolites per participant, averaged 3.3 in ASD and 0 in TD. Using a threshold of one elevated metabolite, the semiquantitative analysis achieved 90% sensitivity and 100% specificity for ASD.
There was no significant correlation between MDM Total Score and age in ASD. Multivariate models, including Fisher Discriminant Analysis (FDA), Neural Networks, and Naïve Bayes, consistently achieved high diagnostic accuracy, with area under the receiver operating characteristic curve values up to 0.86.
Univariate analysis identified ten metabolites, mostly phenylalanine- and tryptophan-related, as significantly higher in ASD. Some, like p-cresol, were elevated only in a subset of ASD cases, and detection limits influenced results for certain compounds.
Metabolites like p-cresol and indole-3-propionic acid provided strong group separation. While substantial numbers of ASD participants had elevated tryptophan- or phenylalanine-related metabolites, yeast metabolite increases were less frequent. Highly correlated metabolites, such as indole propionic acid and beta-carboline, likely reflect microbial dysbiosis rather than external exposure. Metabolites with many undetectable samples contributed less to the MDM System™ algorithm.
In the targeted quantitative analysis, the MDM System™ had 100% specificity and 78% sensitivity, demonstrating reproducibility of the overall methodology, although performance was lower than in the initial semiquantitative analysis. FDA also showed that the most effective metabolite combinations had area under the curve values above 0.7, with minimal additional gain from adding more metabolites.
The findings also led the researchers to propose a hypothetical ASD subtype termed “ASD associated with Microbially-Derived Metabolites” (ASD-MDM). Based on the study data, the authors suggest that approximately 80–90% of children with ASD in their cohort may belong to this metabolically distinct subgroup, although the proposed classification requires independent validation before it can be considered an established ASD phenotype.
Conclusions
The current study highlighted the significance of MDMs in a substantial subset of children with ASD. The development of the MDM System™ offers a promising proof-of-concept approach for future early screening and identification of children at increased likelihood of ASD.
However, the findings are based on a relatively small pilot cohort, and the authors emphasize that independent validation in larger cohorts is still needed before the test can be considered clinically established. Continued research, including validation in independent cohorts and the exploration of microbiome-based therapies, is essential to fully realize the potential of these advances for improving outcomes in children with ASD.
The paper also notes that several authors hold patents, patent applications, or commercial interests related to ASD diagnostics and the MDM System™, underscoring the importance of independent replication of the findings.
Download your PDF copy by clicking here.
Shared brain connectivity patterns found across major neurodevelopmental disorders