31602
Exposome-Wide Association Study in Autism Spectrum Disorder Using Targeted and Untargeted Metabolomics Platforms

Oral Presentation
Thursday, May 2, 2019: 3:06 PM
Room: 516ABC (Palais des congres de Montreal)
C. Hernandez-Ferrer1, D. I. Walker2, C. C. Johnson3, M. K. Chung4, I. H. Lee1, W. J. Alvarez1, C. J. Patel4, K. D. Pennell3, D. P. Jones5 and S. W. Kong1,6, (1)Computational Health Informatics Program, Boston Children's Hospital, Boston, MA, (2)Department of Environmental Medicine and Public Health, Icahn School of Medicine at Mount Sinai, New York, NY, (3)Civil and Environmental Engineering, Tufts University, Medford, MA, (4)Department of Biomedical Informatics, Harvard Medical School, Boston, MA, (5)Emory University, Atlanta, GA, (6)Department of Pediatrics, Harvard Medical School, Boston, MA
Background: Both genetic and environmental risk factors contribute to the liability for autism spectrum disorder (ASD). Epidemiologic studies show that multiple environmental factors could be associated with increased risks of ASD; however, direct measurement of environmental exposures during critical periods of brain development is challenging due to a lack of technology to monitor the exposome during the critical periods of brain development as well as limited availability of relevant biospecimens.

Objectives: To characterize the exposome - the totality of environmental exposures throughout development and lifespan - in patients with ASD and their family members compared to children without ASD using both targeted and untargeted metabolomics platforms.

Methods: Plasma samples (N=1,803) from patients with ASD, their family members and children without ASD were collected from the biorepositories of the Autism Speaks, Boston Children’s Hospital, and National Institute of Mental Health. Using gas chromatography - tandem mass spectrometry (GC-MS/MS), 80 chemicals including the neurotoxicants previously reported as environmental risk factors of ASD and the other toxicants were quantitively measured. For the same samples, we used liquid chromatography - high resolution mass spectrometry (LC-HRMS) to measure more than 22,000 features of both exogenous and endogenous origins. To prioritize exposure biomarkers of ASD, we fitted a generalized linear model to each of exogenous and endogenous chemicals after controlling for age, gender and shared environmental effect for each family. Furthermore, metabolic pathways perturbed in ASD were identified using annotated endogenous chemicals from LC-HRMS. Finally, the impact of exogenous chemicals on known biological pathways were analyzed by integrating GC-MS/MS measures with annotated chemicals in biological pathways from the Kyoto Encyclopedia of Genes and Genomes.

Results: GC-MS/MS analysis comparing cases with controls did not reveal any significant correlation between exogenous exposure biomarkers and ASD status after false discovery rate (FDR) correction. For 24 exogenous chemicals, the plasma concentrations were significantly correlated with blood gene expression levels of 822 transcripts at FDR < 0.05, which were enriched in focal adhesion, Fc epsilon RI signaling pathway, and proteoglycans in cancer. For LC-HRMS data, we found 191 features significantly associated with ASD compared to controls (FDR < 0.05). More than half of the significant features from LC-HRMS showed higher concentrations in ASD compared to controls including 35 annotated chemicals matching endogenous metabolites (N=13), dietary nutrients (N=8), pharmaceuticals (N=4), and environmental pollutants (N=2). Furthermore, concentrations of 10 exogenous exposures were significantly correlated with 736 features from LC-HRMS (FDR < 0.05) that were enriched in de novo fatty acid biosynthesis, C21-steroid hormone biosynthesis and metabolism, and vitamin D3 metabolism among others.

Conclusions: We performed the largest exposomic study in ASD and discovered endogenous and exogenous exposures significantly associated with ASD after controlling for shared environmental effect. By integrating GC-MS/MS data with gene expression and untargeted metabolomics profile, we highlighted physiological changes due to exogenous environmental exposures suggesting potential biological mechanism associated with environmental risk factors.