Children with autism spectrum disorder had higher urinary concentrations of multiple microbially derived metabolites than typically developing controls in a multisite study. Within the study cohort, classification based on one or more elevated metabolites yielded 90% sensitivity and 100% specificity. The researchers characterized the work as a pilot study and stated that replication in an independent cohort is needed.
The study, published in Molecular Psychiatry, evaluated urine samples from 52 children with autism spectrum disorder (ASD) and 47 typically developing children aged 2 to 11 years recruited from sites in Arizona, Texas, Tennessee, and Massachusetts. Researchers first performed semiquantitative liquid chromatography–mass spectrometry analysis, followed by targeted quantitative liquid chromatography–mass spectrometry analysis.
“[A]pproximately 8–9 out of 10 young children with ASD in this cohort had high concentrations of one or more microbially-related metabolites,” wrote lead study author Christina K. Flynn of the Biodesign Center for Health Through Microbiomes and the School for Engineering of Matter, Transport, and Energy at Arizona State University colleagues, and proposed a new phenotype termed “ASD associated with Microbially-Derived Metabolites [MDM].”
The researchers focused on microbially derived metabolites associated with phenylalanine, tyrosine, tryptophan, and yeast metabolism. The semiquantitative analysis identified more than 600 metabolites that differed between groups, including numerous microbial metabolites. Of 24 metabolites selected for further analysis, 23 were statistically significantly elevated in children with ASD compared with typically developing controls.
Among phenylalanine-related metabolites, p-cresol was 151% higher in the ASD group, phenylacetylglutamine was 64% higher, and p-cresol sulfate was 54% higher. Among tryptophan-related metabolites, methyl-3-indole acetate was 1,882% higher, beta-carboline was 1,537% higher, and 3-indolepropionic acid was 558% higher. Arabinitol, a yeast-associated metabolite, was 89% higher.
Researchers developed the MDM System, which assigns one point for each metabolite exceeding the highest value observed in the typically developing reference group. Children with ASD had an average MDM total score of 3.3, while all typically developing children scored zero.
Ninety percent of children with ASD had at least one markedly elevated metabolite, and 56% had elevations in both tryptophan- and phenylalanine-derived metabolites. Using a threshold of one or more elevated metabolites, the MDM System correctly classified 45 of 50 children with ASD, whereas all typically developing participants were correctly classified.
The targeted quantitative analysis generally supported the semiquantitative findings. P-cresol sulfate concentrations were 139% higher in children with ASD, while indoxyl sulfate concentrations were 171% higher. Additional metabolites with statistically significant elevations included p-cresol, hydroxybenzoic acid, benzoic acid, phenylacetylglutamine, hippuric acid, indole-3-acryloyl glycine, and arabinitol.
In the quantitative analysis, 57% of children with ASD had elevated phenylalanine-related metabolites, 42% had elevated tryptophan-related metabolites, and 16% had elevated yeast-related metabolites. Overall, 78% had at least one elevated microbially derived metabolite. The MDM System achieved 78% sensitivity and 100% specificity.
Fisher discriminant analysis produced combinations with area under the receiver operating characteristic curve values as high as 0.86 following leave-one-out cross-validation. Neural network and naïve Bayes approaches achieved accuracies of 83% and 82%, respectively.
The researchers reported that metabolite elevations were heterogeneous across patients, with different children exhibiting different combinations of abnormalities. No significant association was observed between MDM scores and age.
The study was described as a pilot study. Limitations included exclusion of participants with known single-gene disorders, lack of data on body mass index, diet, and medication use, and the absence of commercially available standards for some metabolites, which limited targeted quantitative analysis. The researchers stated that the findings require replication in another independent cohort.
The project received financial support from Arizona State University's Catalyst program, Zoowalk for Autism Research, Analutos, a private donor, and researcher James Adams. Additional support came through the Ira A. Fulton Schools of Engineering and the Fulton Entrepreneurial Professor Program at Arizona State University, the Integrated Mass Spectrometry Shared Resource supported by a National Cancer Institute grant of the National Institute of Health, the Robert Luft Foundation, and the Biodesign Center for Health Through Microbiomes.
Co-study authors James B. Adams, Christina K. Flynn, Rosa Krajmalnik-Brown, Juergen Hahn, Paul Whiteley, and Kevin Carr reported patents or patent applications related to ASD diagnosis. Co-study authors James B. Adams and Rosa Krajmalnik-Brown are co-founders of Autism Diagnostics LLC. Co-study author James B. Adams is founder of Autism Diagnostics Lab Inc. Co-study authors Paul Whiteley and Kevin Carr are co-owners of Analutos. Co-study authors James B. Adams and Rosa Krajmalnik-Brown are co-founders of Gut-Brain-Axis Therapeutics Inc.
Source: Molecular Psychiatry
