Autism can be accurately diagnosed in blood by a new algorithm assessing blood metabolites, according to a new study by Rensselaer Polytechnic Institute scientists.
The algorithm uses big data to find two cellular pathways that indicate a person on the spectrum with roughly 97 percent accuracy – meaning earlier diagnosis and better outcomes, according to the paper, in the journal PLOS Computational Biology.
The test would be the first physiological test for autism, which is currently diagnosed by a multidisciplinary team in each case that involves pediatricians, psychologists and others.
Their method was found using a more complex look at the physiological mechanics of the blood, said Juergen Hahn, the lead author, a system biologist.
“Instead of looking at individual metabolites, we investigated patterns of several metabolites and found significant differences between metabolites of children with ASD and those that are neurotypical,” Hahn said in a school statement on the work. “By measuring 24 metabolites from a blood sample, this algorithm can tell whether or not an individual is on the Autism spectrum, and even to some degree where on the spectrum they land.”
The team used a big data analytical method known as the Fisher Discriminant Analysis to look into the markers in the blood of 149 people, half of whom were on the spectrum.
The technique took the full run of data – and then removed a single participant at a time, in order to predict their missing markers. He repeated the process for all 149 people.
The data predicted 97.6 percent of the autistic individuals, and 96.1 percent of those who had typical neurological systems, they report.
The two key pathways are the methionine cycle and the transulfuration pathway. The former is linked to DNA methylation and epigenetic factors, and the latter produces an antioxidant glutathione, which is crucial in reducing oxidative stress in the body.
Together, with the two dozen metabolites, the best biological marker yet has been identified, they conclude.
“A lot of studies have looked at one biomarker, one metabolite, one gene, and have found some differences, but most of the time those differences weren’t statistically significant or the results could not be reliably replicated,” added Hahn. “Our contribution is using big-data techniques that are able to look at a suite of metabolites that have been correlated with ASD and make (statistically) a much-stronger case.”
The latest CDC estimates find that one in 68 children of schooling age are diagnosed with autism in the United States – though some debate exists as to whether the spectrum is more prevalent, or if detection is better. Currently about 3 million people of all ages live with autism in the U.S.
The early diagnosis of ASD has been seen as a crucial milestone. Last month a team from the Children's Hospital of Philadelphia and the University of North Carolina - Chapel Hill reported that brain measurements allowed them to predict autism by age 2 with 90 percent accuracy.