Objectives Autism spectrum disorders and intellectual disability present a challenge for therapeutic and dietary management. We performed a re-analysis of plasma amino acid chromatography of children with autism spectrum disorders ( n = 22) or intellectual disability ( n = 29) to search for a metabolic signature that can distinguish individuals with these disorders from controls ( n = 30). Methods We performed univariate and multivariate analyses using different machine learning strategies, from the raw data of the amino acid chromatography. Finally, we analysed the metabolic pathways associated with discriminant biomarkers. Results Multivariate analysis revealed models to discriminate patients with autism spectrum disorders or intellectual disability and controls from plasma amino acid profiles ( P < 0.0003). Univariate analysis showed that autism spectrum disorder and intellectual disability patients shared similar differences relative to controls, including lower glutamate ( P < 0.0001 and P = 0.0002, respectively) and serine ( P = 0.002 for both) concentrations. The multivariate model ( P < 6.12.10−7) to discriminate between autism spectrum disorders and intellectual disability revealed the involvement of urea, 3-methyl-histidine and histidine metabolism. Biosigner analysis and univariate analysis confirmed the role of 3-methylhistidine ( P = 0.004), histidine ( P = 0.003), urea ( P = 0.0006) and lysine ( P = 0.002). Conclusions We revealed discriminant metabolic patterns between autism spectrum disorders, intellectual disability and controls. Amino acids known to play a role in neurotransmission were discriminant in the models comparing autism spectrum disorders or intellectual disability to controls, and histidine and b-alanine metabolism was specifically highlighted in the model.
Expression analyses of the mitochondrial complex I 75-kDa subunit in early onset schizophrenia and autism spectrum disorder: increased levels as a potential biomarker for early onset schizophrenia
