(444g) Do Comorbid Medical Conditions Affect Autism Spectrum Disorder Diagnosis Via Big Data Analysis of Plasma Metabolites?
Despite these promising initial efforts towards the development of a blood test for ASD diagnosis, significant work remains in establishing the testâs robustness to heterogeneity in the clinical manifestations of ASD. Rather than being one homogenous condition, ASD is believed to be comprised of several distinct subtypes , which brings into question the validity of a one-size-fits-all approach to diagnosis. Comorbid medical conditions (CMCs), affecting up to 95% of children with ASD , are potentially significant sources of clinical heterogeneity in ASD that may interfere with metabolic processes  such as those in FOCM and TS. Understanding how CMCs contribute to variation in individualsâ FOCM/TS metabolite profiles is thus essential to the development of a viable blood test for diagnosing ASD.
To move towards this goal, FDA was used to first identify a model for differentiating 92 children with ASD from 82 TD children who were all participants in the Integrated Metabolic and Genomic Endeavor Study at Arkansas Childrenâs Research Institute . The FDA model with the best classification performance relies on a subset of five inputs of metabolite concentration measurements from the FOCM/TS pathways. Specifically, these are (1) the ratio of S-adenosylmethionine to S-adenosylhomocysteine, (2) glutamylcysteine, (3) oxidized glutathione, (4) the ratio of free cystine to free cysteine, and (5) the percentage of oxidized glutathione molecules. A new, independent cohort of 123 children with ASD was then evaluated with this FDA model to determine whether they would be classified as ASD or TD, and this data set contained information regarding childrenâs histories of CMC diagnoses in four categories (gastrointestinal problems, allergies, immune/metabolic dysfunction, and neurological disorders).
Overall, 91% (112/123) of the new ASD cohort was correctly identified as having ASD based upon the FDA model applied to the FOCM/TS metabolites. Within the subgroups of children currently diagnosed with and without a gastrointestinal problem, 92% (58/63) children and 90% (54/60) children, respectively, were correctly classified with ASD. For those currently with and without allergies, 93% (64/69) and 89% (48/54) of respective children were classified with ASD correctly. Children currently with and without immune/metabolic dysfunction were identified to have ASD with 90% (35/39) and 92% (77/84) accuracy, respectively. Finally, the subgroups of children with and without current neurological disorders were respectively classified with ASD with 93% (80/86) and 86% (32/37) accuracy.
The findings of this study offer several important conclusions. For one, 91% of a new cohort of children being correctly identified with ASD highlights the potential for supporting ASD diagnosis with FDA applied to plasma metabolites of the FOCM/TS pathways. Second, the accuracy of the classification model predictions was generally consistent regardless of whether a child was or was not diagnosed with a particular CMC. It is worth noting that accuracy tended to be higher in children with any given CMC as opposed to those without it, suggesting that the presence of a CMC may further contribute to metabolic abnormalities that aid with the identification of ASD. Future work with this data set should provide a more in-depth investigation of how the presence of multiple CMCs in an individual may interact to make that person more or less easily identified as having ASD. Further validation of the blood test on new cohorts of TD individuals, ideally with linked CMC diagnosis information, would also be valuable.
 S. Melnyk et al., âMetabolic imbalance associated with methylation dysregulation and oxidative damage in children with autism,â J Autism Dev Disord, vol. 42, no. 3, pp. 367â377, Mar. 2012.
 D. P. Howsmon, U. Kruger, S. Melnyk, S. J. James, and J. Hahn, âClassification and adaptive behavior prediction of children with autism spectrum disorder based upon multivariate data analysis of markers of oxidative stress and DNA methylation,â PLoS Comput. Biol., vol. 13, no. 3, p. e1005385, Mar. 2017.
 D. P. Howsmon et al., âMultivariate techniques enable a biochemical classification of children with autism spectrum disorder versus typically-developing peers: A comparison and validation study,â Bioengineering & Translational Medicine, vol. 3, no. 2, pp. 156â165, May 2018.
 T. Vargason et al., âComparison of three clinical trial treatments for autism spectrum disorder through multivariate analysis of changes in metabolic profiles and adaptive behavior,â Front. Cell. Neurosci., vol. 12, p. 503, Dec. 2018.
 O. Ousley and T. Cermak, âAutism spectrum disorder: Defining dimensions and subgroups,â Curr Dev Disord Rep, vol. 1, no. 1, pp. 20â28, Mar. 2014.
 G. N. Soke, M. J. Maenner, D. Christensen, M. Kurzius-Spencer, and L. A. Schieve, âPrevalence of co-occurring medical and behavioral conditions/symptoms among 4- and 8-year-old children with autism spectrum disorder in selected areas of the United States in 2010,â J Autism Dev Disord, vol. 48, no. 8, pp. 2663â2676, Aug. 2018.
 C. Tye, A. K. Runicles, A. J. O. Whitehouse, and G. A. Alvares, âCharacterizing the interplay between autism spectrum disorder and comorbid medical conditions: An integrative review,â Front. Psychiatry, vol. 9, 2019.
This paper has an Extended Abstract file available; you must purchase the conference proceedings to access it.
Do you already own this?
Log In for instructions on accessing this content.
|AIChE Graduate Student Members||Free|
|AIChE Undergraduate Student Members||Free|