By Paul Nicolaus
February 20, 2019 | Autism spectrum disorder (ASD) could benefit from new and improved diagnostic methods, and the arrival of a newly commercialized blood test—along with other efforts headed toward a similar end goal—could be nudging the field toward earlier detection and intervention.
In November, NeuroPointDX, a division of Stemina Biomarker Discovery, announced the commercial launch of its NPDX AA test to help with the earlier diagnosis of ASD. The blood plasma-based test, which identifies amino acid metabotypes (subtypes based on metabolism) associated with ASD, can be used to screen children as young as 18 months.
The test is capable of identifying metabotypes in about 30% of children with ASD, based on data from the Children’s Autism Metabolome Project (CAMP), which is the largest clinical study to date focused on the metabolism of children with ASD. The subtypes identified by the CAMP study and published in Biological Psychiatry (doi: 10.1016/j.biopsych.2018.08.016) account for several of the 13 total metabotypes included in the new test panel.
Each of those subtypes is identified with greater than 95% specificity, Elizabeth Donley, CEO of Stemina Biomarker Discovery told Diagnostics World News. The NPDX AA test is intended for children who have a developmental delay and for younger siblings of children who have already been diagnosed, both of which indicate a higher risk for ASD.
The test, initially available on a limited basis, must be ordered by a physician and requires a fasting blood sample, which can be taken by a phlebotomy lab. The lab then ships the sample, following certain protocols, to NeuroPointDX’s laboratory. From there, a report is returned to the physician within two weeks to indicate whether or not the child is positive for any of the metabotypes. Children who receive a positive result are at an increased risk and should be prioritized for additional evaluation.
“It’s a big step,” Juergen Hahn of the Rensselaer Polytechnic Institute Department of Biomedical Engineering told Diagnostics World News, given that there was previously no test on the market. A notable limitation, though, is that the test is only able to flag a percentage of children. For the majority of those with ASD, the test results are inconclusive. “So it can be helpful for some,” he added.
More Tests on the Way?
Like NeuroPointDX, Hahn and his research team are focused on metabolism. The difference, he explained, is that NeuroPointDX—at least initially—took a broader approach whereas Hahn and colleagues have developed a blood test that focuses on just two cellular pathways believed to be linked to ASD. Their findings were published last year in Bioengineering & Translational Medicine (doi: 10.1002/btm2.10095).
Traditionally when researchers look at a biomarker they measure one component in the blood and then determine whether that component is high or low as compared to others, he explained, and if it’s extremely high or extremely low then sometimes that is an indicator of a certain condition. For autism, however, that doesn’t work.
“You’re not going to find one metabolite that tells you what’s going on,” he said, so his team applied machine learning approaches to metabolite data. They found that by measuring between five and seven metabolites and then analyzing them together, it is possible to distinguish between children who have autism and those who are typically developing. When the algorithm was applied to each individual, it correctly predicted autism with 88% accuracy.
Their latest study builds upon previous work published in 2017 in PLOS Computational Biology (doi: 10.1371/journal.pcbi.1005385) that analyzed data from a group of 149 people, roughly half of whom had been previously diagnosed with ASD. Hahn and colleagues gathered data on 24 metabolites for each member of the group and created an algorithm that correctly identified over 96% percent of all typically developing participants and over 97% of the ASD cohort.
The newer algorithm’s drop in accuracy can be attributed to several factors, Hahn pointed out. In the 2017 paper, the researchers measured 24 metabolites. Out of those 24, they found that using 7 together helped lead to a reliable prediction. The more recent study, however, used a dataset that included just 22 of the 24 metabolites used to create the original algorithm.
In addition, there was a “much more heterogeneous group of children in the second paper.” Part of the reason is that the data came from three separate studies, he said, “and whenever you put more variability into the system by taking data from different studies, you automatically expect that the accuracy will not be as good as if you basically run everything as a very well controlled study.”
According to Hahn, a pending patent application has been licensed to start-up company BioROSA Technologies, which is focused on commercializing the test. Looking ahead, the aim is to raise enough funding to conduct a study large enough to potentially garner FDA approval.
Meanwhile, a team led by Naila Rabbani, a researcher from the University of Warwick in England, has developed blood and urine tests to detect autism in children. Their findings, published in 2018 in Molecular Autism (doi: 10.1186/s13229-017-0183-3), found a link between ASD and damage to blood plasma proteins. “We have been working in this field specifically developing this method using mass spectrometry to measure a very trace amount of damaged amino acids,” she told Diagnostics World News.
The researchers examined blood and urine samples from 38 children between the ages of 5 and 12 with an ASD diagnosis as well as 31 healthy controls and discovered that there were chemical differences between the two groups. The tests they developed showed that children with ASD have higher levels of dityrosine and advanced glycation end products.
In the plasma protein, Rabbani said, “some of these markers were increased to a very high extent in children with ASD, which were not present in typically growing children. So that gave us the ability to use artificial intelligence to then discriminate between these two cohorts.” The data, used in conjunction with machine learning methods, led to a test performance with a sensitivity of 92% and specificity of 84%.
The discovery could lead to earlier diagnosis and intervention, according to Rabbani, and there is hope that additional testing may reveal new causes of ASD as well. The next step, which is currently in the planning stages, is to conduct larger studies in order to validate the test and develop it into a diagnostic kit. “We are acquiring funding and hoping by the middle of this year we should have something in place,” she added.
Researchers from State University of New York (SUNY) Upstate Medical University, Penn State College of Medicine, and Quadrant have their sights set on a test that instead relies on saliva. Their findings, published in Frontiers in Genetics (doi: 10.3389/fgene.2018.00534) in November 2018, reveal that a saliva-based biomarker panel and related algorithm could improve the ability to identify children with ASD in its earliest stages. In their study of over 450 children ages 18 months to 6 years, the researchers used a panel of 32 small RNAs to differentiate children with autism from children displaying typical development or non-ASD developmental delay with 85% accuracy.
As Diagnostics World News reported in September, Quadrant Biosciences has been awarded a $2 million Phase II Small Business Technology Transfer (STTR) grant from the National Institutes of Health (NIH) for the refinement and commercialization of an epigenetic ASD diagnostic test.
This Phase II study expands enrollment to five different academic medical center locations throughout the US and includes the recruitment of 750 additional children.
Why Earlier Diagnosis Matters
ASD is a developmental disability that can cause various social, communication, and behavioral challenges, such as speech disturbances, repetitive behaviors, limited interests, anxiety, and difficulty adapting to new environments. Although it can potentially be detected at 18 months or younger, current methods of diagnosis rely on behavioral assessments.
Because signs and symptoms vary from person to person and don’t necessarily appear right away, the reality is that many children do not receive a final diagnosis until after the age of 4. Research has revealed that early intervention services can allow children to learn important skills and improve their development during those early years. In other words, the sooner a child can be diagnosed, the faster they can receive crucial help and support.
“If we had a more objective biological marker that could be used during the first year of life or even between one and two, we could get children into effective intervention much, much earlier,” David Amaral, director of the University of California, Davis MIND Institute Autism Center of Excellence and principal investigator in the CAMP study, told Diagnostics World News.
The NPDX AA test panel is just the first of a series of tests expected to come out of the CAMP study. “This is not the end of the game,” added Amaral, who serves on the NeuroPointDX scientific advisory board. “This is the first step.”
A metabolomic study produces data on tens of thousands of small molecules. The difficult aspect is trying to decide where to go first and what to look at initially. The bioinformatics team at Stemina, based on their initial findings as well as findings in medical literature, decided to pursue what is recorded in the Biological Psychiatry paper.
“But we’re already working on other metabotypes, and over the next couple of years the vision is that there would be several others based on other metabolic irregularities or alterations,” he added. Over time, the goal is to build upon the utility of the test to identify a greater percentage of children with additional metabolic subtypes.
It’s clear that there is a sense of urgency about trying to find biological markers for autism and to do so in a way that can help families pinpoint a diagnosis so that they can get their kids into appropriate services, Connie Brooks, head of the psychology and diagnostics team at the University of Missouri’s Thompson Center for Autism & Neurodevelopmental Disorders told Diagnostics World News.
Because current methods rely on psychological, developmental, and autism-specific testing, it requires a lot of training and plenty of calibration among providers to make sure everyone is looking at symptoms the same way and interpreting them in the same way. In cases where it is less clear, it would be a huge help to offer families a clear yes or no answer from a simple, rigorous test. And yet, Brooks explained, coming up with that type of test is a huge challenge.
“I’m not sure if we’ll ever have one type of test that’s going to identify all kids—yes or no—as to whether they’re on the autism spectrum,” she added, which means we run the risk of using these emerging tools in a way that attempts to provide certainty when it cannot be guaranteed.
Paul Nicolaus is a freelance writer specializing in science, nature, and health. Learn more at www.nicolauswriting.com.