Autism spectrum disorders (ASD) are a group of neurodevelopmental disorders resulting from multiple factors. Diagnosis is based on behavioural and developmental signs detected before 3 years of age, and there is no reliable biological marker. The purpose of this study was to evaluate the value of gas chromatography combined with mass spectroscopy (GC-MS) associated with multivariate statistical modeling to capture the global biochemical signature of autistic individuals. GC-MS urinary metabolic profiles of 26 autistic and 24 healthy children were obtained by liq/liq extraction, and were or were not subjected to an oximation step, and then were subjected to a persilylation step. These metabolic profiles were then processed by multivariate analysis, in particular orthogonal partial least-squares discriminant analysis (OPLS-DA, R(2)Y(cum) = 0.97, Q(2)(cum) = 0.88). Discriminating metabolites were identified. The relative concentrations of the succinate and glycolate were higher for autistic than healthy children, whereas those of hippurate, 3-hydroxyphenylacetate, vanillylhydracrylate, 3-hydroxyhippurate, 4-hydroxyphenyl-2-hydroxyacetate, 1H-indole-3-acetate, phosphate, palmitate, stearate, and 3-methyladipate were lower. Eight other metabolites, which were not identified but characterized by a retention time plus a quantifier and its qualifier ion masses, were found to differ between the two groups. Comparison of statistical models leads to the conclusion that the combination of data obtained from both derivatization techniques leads to the model best discriminating between autistic and healthy groups of children.

Autism spectrum disorders (ASD) are neurodevelopmental diseases with complex genetic and environmental etiological factors. Although genetic causes play a significant part in the etiology of ASD, metabolic disturbances may also play a causal role or modulate the clinical features of ASD. The number of ASD studies involving metabolomics is increasing, and sometime with conflicting findings. We assessed the metabolomics profiling of urine samples to determine a comprehensive biochemical signature of ASD. Furthermore, to date no study has combined metabolic profiles obtained from different analytical techniques to distinguish patient with ASD from healthy individuals. We obtained (1)H-NMR spectra and 2D (1)H-(13)C HSQC NMR spectra from urine samples of patients with ASD or healthy controls. We analyzed these spectra by multivariate statistical data analysis. The OPLS-DA model obtained from (1)H NMR spectra showed a good discrimination between ASD samples and non-ASD samples (R(2)Y(cum) = 0.70 and Q(2) = 0.51). Combining the (1)H NMR spectra and the 2D (1)H-(13)C HSQC NMR spectra increased the overall quality and predictive value of the OPLS-DA model (R(2)Y(cum) = 0.84 and Q(2) = 0.71), leading to a better sensitivity and specificity. Urinary excretion of succinate, glutamate and 3-methyl-histidine differed significantly between ASD and non-ASD samples. Urinary screening of children with neurodevelopmental disorders by combining NMR spectroscopies (1D and 2D) in multivariate analysis is a better sensitive and a straightforward method that could help the diagnosis ASD.