A year after the researchers published their work on the physiological test for signs of autism, a subsequent study confirms its exceptional success in assessing whether a child has an autistic spectrum disorder.

A physiological test that supports the diagnostic process of a clinician may reduce the age of autism diagnosis, leading to earlier treatment. The results of a study that uses an algorithm to predict whether a child has an autism spectrum disorder (ASD) based on metabolites in a blood sample are published in the June issue of Bioengineering & amp; Translational Medicine.

"We looked at groups of children with ASD regardless of our previous research and got a similar result. We can predict with children's autism with 88 percent accuracy, "said Jurgen Khan, lead author, system biologist, professor, head of Rensselaer, Polytechnic Institute of the Department of Biomedical Engineering and member of the Center for Biotechnology and Interdisciplinary Research. Rensselaer (CBIS). "This is very promising."

An estimated 1.7% of all children diagnosed with ASD are diagnosed as "developmental disabilities caused by differences in the brain," according to the Centers for Disease Control and Prevention. An earlier diagnosis is usually considered the best result, because children receive early care in the treatment of autism. The diagnosis of ASD is possible at the age of 18-24 months, however, since the diagnosis depends solely on clinical observations, most children do not diagnose ASD for up to 4 years. Instead of searching for a single ASD indicator, the developed approach uses large data methods to find the patterns of metabolites related to two related cellular pathways (a series of interactions between molecules that control cell function) with suspicious connections to ASD.

"Jürgen's work on the development of a physiological test for signs of autism is an example of how the interdisciplinary life science and technology interface brings new perspectives and solutions to improve human health," said Deepak Vashisht, CBIS Director. "This is an excellent result because of the greater emphasis on Alzheimer's and neurodegenerative diseases in CBIS, where our work combines several approaches to developing better diagnostic tools and biomanipulating new therapeutic drugs."

Initial success in 2017 analyzed data from a group of 149 people, about half, of which, previously were diagnosed with ASD. For each member of the group, Khan received data on 24 metabolites associated with two cellular pathways-the methionine cycle and the trans-sulfation route. By deliberately omitting data from one person in the group, Khan transferred the remaining set of data to advanced analysis methods and used the results to generate the prediction algorithm. Then the algorithm made a prediction about the data from the missed individual. Khan cross-validated the results, replacing another person from the group and repeating the process for all 149 participants. His method was correctly identified by 96.1 percent of all typically developing participants and 97.6 percent of ASD cohorts.

The new study applies Khan's approach to an independent set of data. To avoid a lengthy process of collecting new data through clinical trials, Khan and his team searched for existing data sets that included the metabolites that he analyzed in the initial study. Researchers have identified relevant data from three different studies, in which there were a total of 154 autistic children, conducted by researchers at the Arkansas Children's Research Institute. The data included only 22 of the 24 metabolites that he used to create the original prediction algorithm, but Khan determined that the available information would be sufficient for the test.

The team used its approach to recreate the prediction algorithm, this time using data from 22 metabolites from the original group of 149 children. Then the algorithm was applied to a new group of 154 children for testing purposes. When the prediction algorithm was applied to each person, he correctly predicted autism with 88 percent accuracy.

Khan said that the difference between the initial accuracy rate and the speed of the new study may be related to several factors, the most important being that two of the metabolites were not available in the second set of data. Each of the two metabolites was a strong indicator in the previous study.

In general, the second study confirms the initial results and gives an idea of ​​several approaches.

"The most significant result is the high degree of accuracy that we can obtain using this approach on data collected over the years, in addition to the initial set," Khan said. "This is an approach that we would like to see in clinical trials and, ultimately, in commercially available tests."

Khan joined the research of doctoral students: Troy Vargason, Daniel P. Hausmon, Robert A. Rubin from Whittier College; Leanna Delhi, Marie Tippett, Shannon Rose and Sirish K. Bennuri of the Children's Research Institute in Arkansas and the Arkansas University of Medical Sciences; John Slater, Stepan Melnik and S. Jill James of the University of Arkansas for Medical Sciences; and Richard E. Fry of Phoenix Children's Hospital. The study was partially funded by the National Institutes of Health.


Materials provided by the Rensselaer Polytechnic Institute.


Daniel P. Hausmon, Troy Vargason, Robert A. Rubin, Leanna Delhi, Marie Tippet, Shannon Rose, Sirish K. Bennuri, John Slater, Stepan Melnik, S. Jill James, Richard E. Fry, Jurgen Khan. Multivariate methods allow for a biochemical classification of children with autism spectrum disorder compared to typically developing peers: comparison and validation. Bioengineering and translational medicine, 2018; DOI: 10.1002 / btm2.10095

Source: www.sciencedaily.com