Serum Metabolome Profile in GI Symptomatic ASD Children: A Pilot Study

Background: We have previously shown that children with ASD and gastrointestinal (GI) inflammation display a unique gene expression profile both in GI biopsy tissue and peripheral blood, when compared to non-ASD GI-symptomatic children without GI inflammation. The differential gene expression patterns identified in peripheral blood can provide clues as to the pathobiology that underlies the inflammation and/or ASD. To explore this hypothesis further we have assayed serum from the same individuals to determine if the metabolite profiles: (1) also differ significantly between the groups and, (2) if they provide additional mechanistic insight.

Objectives: The goal of this study was to measure serum metabolite profiles in GI-symptomatic children with ASD and ileocolonic inflammation (ASD^IC+) and from GI-symptomatic typically developing children (without evidence of ileocolonic inflammation; TD^IC-), in order to compare metabolite levels between these groups.

Methods:  Global metabolite profiles were determined in serum samples derived from 40 individuals as follows: (a) the “control” group (TD^IC-) consisting of 14 males and 5 females and, (b) the “case” group (ASD^IC+) consisting of 17 males and 4 females. Sample processing involved metabolite extraction, followed by ultrahigh performance liquid chromatography-tandem mass spectroscopy (UPLC-MS/MS). Raw data were extracted, peak-identified and QC processed. Compounds were identified by comparison to library entries of purified standards or recurrent unidentified entities. Following log transformation and imputation of missing values, if any, with the minimum observed value for each compound, Welch’s two-sample t-test was used to identify biochemicals that achieved statistical significance (p≤0.05), as well as those approaching significance (0.05<p<0.01).

Results: The present dataset comprises a total of 612 compounds of known identity (named biochemicals) measured in 40 individual serum samples. A comparison between the 2 groups showed that 292 metabolites (48%; 72 higher and 200 lower in ASD^IC+ compared to controls) reached a level of statistical significance and another 36 metabolites (12 higher/24 lower in ASD^IC+) approached significance. Many of these metabolite level differences between ASD and TD samples have been reported in other studies, while some of the individual findings here are either in disagreement, or are altogether new, compared to published reports. Overall, the data showed that the two groups were distinguishable (i.e. significantly separated) by principal component analysis and the ASD^IC+ profiles, as a group, appeared to be more heterogeneous. Using the primary groupings of “case” and “control”, random forest analysis resulted in a predictive accuracy of 95% for the serum samples.

Conclusions: Comparison of metabolome profiles between ASD^IC+ and TD^IC- showed that levels of nearly half of the serum metabolites are statistically significantly different between the two groups and that these metabolome profiles are largely phenotype-specific. Moreover, many of the specific metabolite level differences have previously been reported in metabolome studies comparing ASD and control samples, however in some cases the changes we found are either discordant or novel compared to published reports. This is not surprising since our pilot study examined differences in serum from children with ASD and gastrointestinal inflammation to serum from children with neither condition.