Two metabolic biomarkers accurately predict autistic spectrum disorders (ASDs) in children, new research shows.
A multivariate statistical analysis conducted by investigators at Rensselaer Polytechnic Institute in Troy, New York, "provides unprecedented quantitative classification results" that accurately identify children with ASD and neurotypical children.
For the study, investigator Juergen Hahn, PhD, professor and head, Department of Biomedical Engineering, Rensselaer Polytechnic Institute, and colleagues measured blood concentrations of two metabolites — the folate-dependent one-carbon mechanism (FOCM) and transulfuration (TS) pathways — in 83 children with ASD, 47 siblings, and 76 age-matched neurotypical peers. The children ranged in age from 3 to 10 years.
The researchers used two advanced modeling and statistical analysis tools to analyze the metabolic data. Fisher Discriminant Analysis enabled multivariate classification of participants, resulting in the correct classification of 97.6% of the children with autism and 96.1% of the neurotypical children.
A second statistical method, called kernel partial least squares (KPLS), was used to predict adaptive behavior in children with ASD, as measured by the Vineland Adaptive Behavior Composite score. The researchers measured five metabolites of the pathways, which accurately predicted the Vineland score with an R 2 of 0.45 after cross validation.
"Autism is a complex condition that requires investigation of multiple variables simultaneously," Dr Hahn told Medscape Medical News. "Looking at multiple metabolites together using multivariate statistical techniques provides far greater accuracy and more meaningful results than looking only at one metabolite, or even at ratios of two."
The study was published online March 16 issue of PLOS Computational Biology.
Important Diagnostic Tool
The diagnosis of ASD is currently reached through the assessment by a multidisciplinary team of medical and mental health professionals using psychometric tools. Identifying biomarkers that might support this diagnostic process is "ongoing," according to the study authors.
Previous research has demonstrated that FOCM and TS contribute to the genetic and environmental predisposition to ASD. FOCM contributes to epigenetic gene expression through DNA methylation, and TS contributes to intracellular redox status. These organic toxins induce oxidative stress. Moreover, heavy metals disrupt transsulfuration by binding glutathione, which is the major contributor to intracellular redox homeostasis.
"Investigators have struggled with identifying a single, predictive measurement of these pathways that separates individuals with ASD from neurotypical controls or that correlates well with ASD severity," the authors note, stating that multivariate methods are necessary to accomplish these objectives.
To enable the discovery of important multivariate interactions that could improve classification, the researchers utilized latent variable techniques that assess the importance of individual variables.
The authors explain that Fisher Discriminant Analysis, which has been used extensively in problems of biological classification, "achieves optimal linear separability using a typically small set of latent variables that are linear combinations of the original variable set." Latent variable regression techniques also include KPLS.
The investigators applied these statistical techniques to a comparison of FOCM/TS pathways in neurotypical participants as well as children with ASD and their siblings.
Cross-validatory Fisher Discriminant Analysis was applied to measurements of FOCM/TS metabolites, generating a linear classifier for ASD and neurotypical participants. The analysis yielded accurate classification in 97.6% of the children with ASD and in 96.1% of neurotypical children.
The performance of this classification was then evaluated in the sibling cohort, which the study authors describe as "a more challenging classification problem, due to partially shared genetic and environmental effects with the ASD cohort."
They found that the probability distribution function (PDF) of the siblings showed significantly more overlap with the PDF of the neurotypical children than with the PDF of the ASD cohort, supporting the hypothesis that siblings of those with ASD have FOCM/TS metabolite profiles significantly more similar to neurotypical peers than to ASD siblings.
Five metabolites in the FOCM/TS pathway (GSSG, tGSH/GSSG, nitrotyrosine, tyrosine, and f-cysteine) were used to assess the prediction of adaptive behavior in children with ASD, using the Vineland Adaptive Behavior Composite score. That scale evaluates adaptive skills in communication, socialization, daily living, and motor domains.
The investigators found that the KPLS accurately predicted the Vineland score with an R 2 of 0.45 after cross validation.
Dr Hahn stated that these techniques can be a helpful adjunct to conventional diagnostic approaches to ASD.
"If you are evaluating someone to determine if he or she is on the autistic spectrum, you want to make sure your diagnosis is correct. Analyzing metabolites using these statistical tools does not replace but can support the diagnostic process."
He added, "Our study suggests that DNA methylation plays key role in epigenetics and that glutathione plays a key role in oxidative stress, which has been linked to autism."
Enabling Early Diagnosis
"This study is an impressive and important contribution to our understanding of the origins of autism, confirming in a convincing manner that oxidative stress and impaired methylation are the defining features of autism," commented Richard C. Deth, PhD, professor of pharmacology in the Department of Pharmaceutical Sciences, Nova Southeastern University, Fort Lauderdale, Florida.
"Methylation of DNA and histone proteins is the backbone of epigenetic regulation during neurodevelopment, so it makes mechanistic sense that impaired methylation would cause autism," Dr Deth told Medscape Medical News.
He emphasized that "the metabolic factors, which this study shows to be associated with autism, are potentially reversible and amenable to treatment, particularly if they are recognized early.
"Testing for these specific biochemical factors may therefore allow early diagnosis, with a better chance to correct the neurodevelopmental trajectory," he said.
The study authors note that psychometric instruments currently used to diagnose ASD are rarely able to make the diagnosis in children younger than 2 years, because these instruments are based solely on behavioral assessment. Testing metabolites can potentially enable earlier intervention, they add.
The potential applicability of the study findings should generate much interest on the part of researchers and clinicians, according to Dr Hahn.
"Given that these metabolites and pathways so closely allow classification of whether an individual is or is not on the ASD spectrum, researchers and practitioners should be very interested in looking further into this approach," he said.
The authors have disclosed no relevant financial relationships.
PLoS Comput Biol. Published online March 16, 2017. Full text
Medscape Medical News © 2017 WebMD, LLC
Send comments and news tips to news@medscape.net.
Cite this: Blood Test May Lead to Earlier, More Accurate Autism Diagnosis - Medscape - Mar 17, 2017.
Comments