Urinary Untargeted Metabolic Profile Differentiates Children with Autism from Their Unaffected Siblings

Abstract

Autism Spectrum Disorder (ASD) encompasses a clinical spectrum of neurodevelopmental conditions that display significant heterogeneity in etiology, symptomatology, and severity. We previously compared 30 young children with idiopathic ASD and 30 unrelated typically-developing controls, detecting an imbalance in several compounds belonging mainly to the metabolism of purines, tryptophan and other amino acids, as well as compounds derived from the intestinal flora, and reduced levels of vitamins B6, B12 and folic acid. The present study describes significant urinary metabolomic differences within 14 pairs, including one child with idiopathic ASD and his/her typically-developing sibling, tightly matched by sex and age to minimize confounding factors, allowing a more reliable identification of the metabolic fingerprint related to ASD. By using a highly sensitive, accurate and unbiased approach, suitable for ensuring broad metabolite detection coverage on human urine, and by applying multivariate statistical analysis, we largely replicate our previous results, demonstrating a significant perturbation of the purine and tryptophan pathways, and further highlight abnormalities in the “phenylalanine, tyrosine and tryptophan” pathway, essentially involving increased phenylalanine and decreased tyrosine levels, as well as enhanced concentrations of bacterial degradation products, including phenylpyruvic acid, phenylacetic acid and 4-ethylphenyl-sulfate. The outcome of these within-family contrasts consolidates and extends our previous results obtained from unrelated individuals, adding further evidence that these metabolic imbalances may be linked to ASD rather than to environmental differences between cases and controls. It further underscores the excess of some gut microbiota-derived compounds in ASD, which could have diagnostic value in a network model differentiating the metabolome of autistic and unaffected siblings. Finally, it points toward the existence of a “metabolic autism spectrum” distributed as an endophenotype, with unaffected siblings possibly displaying a metabolic profile intermediate between their autistic siblings and unrelated typically-developing controls.