We hypothesized metabolomic profiling could be utilized to identify children who scored poorly on the communication component of the Ages and Stages Questionnaire (ASQ); which assesses development in childhood, and to provide candidate biomarkers for autism spectrum disorders (ASD). In a population of three-year-old children, 15 plasma metabolites, were significantly (p < 0.05) different between children who were categorized as having communication skills that were “on schedule” (n = 365 (90.6%)) as compared to those “requiring further monitoring/evaluation” (n = 38 (9.4%)) according to multivariable regression models. Five of these metabolites, including three endocannabinoids, were also dysregulated at age one (n = 204 “on schedule”, n = 24 “further monitoring/evaluation”) in the same children. Stool metabolomic profiling identified 11 significant metabolites. Both the plasma and stool results implicated a role for tryptophan and tyrosine metabolism; in particular, higher levels of N-formylanthranilic acid were associated with an improved communication score in both biosample types. A model based on the significant plasma metabolites demonstrated high sensitivity (88.9%) and specificity (84.5%) for the prediction of autism by age 8. These results provide evidence that ASQ communication score and metabolomic profiling of plasma and/or stool may provide alternative approaches for early diagnosis of ASD, as well as insights into the pathobiology of these conditions.
