The test supports clinician’s diagnostic process with the potential to lower the age of diagnosis leading to earlier treatment based on metabolites in blood samples able to predict with 88% accuracy. This work is a result from the larger emphasis on Alzheimer’s and neurodegenerative diseases at CBIS joining multiple approaches to develop better diagnostic tools and biomanufacturing new therapeutics.
According to the CDC 1.7% of all children are diagnosed with ASD which is characterized as a developmental disability caused by differences in the brain. Earlier diagnosis can lead to better outcomes with engaging in earlier intervention services at a possible 18-24 months of age, most diagnosis are based on clinical observation meaning most children are not diagnosed until after 4 years of age.
In 2017 success analyzing a group of 149 patients of which half had been diagnosed with ASD, data was obtained from 24 metabolites from each patient related on 2 cellular pathways: methionine cycle and transsulfuration pathway. Omitting one patient data from the group the rest were subjected to advanced analysis techniques that were used to generate a predictive algorithm which then made a prediction on the single remaining omitted patient data. Results were cross validated, swapping a different patient out of the group for another and repeating for all 149 participants. The method identified 96.1% of all typically developing participants and 97.6% of the ASD cohort correctly.
This new study applies the previous approach to an independent dataset. Existing datasets were used which included the metabolites analyzed in the original study. Appropriate data was identified from 3 different studies that included 154 autistic children; data included only 22 of the 24 metabolites used to create the original algorithm, but researchers were determined information available would be sufficient.
The original approach was used to create an algorithm using 22 metabolites from the original group of 149 participants then applied to the new group of 154 participants for testing purposes, when applied to each individual the algorithm was 88% accurate.
Difference in accuracy can likely be attributed to several factors such as most importantly 2 of the metabolites were not available in the second trial as each of the 2 had been strong indicators in previous work. According to the researchers overall the second study validates original results to provide insight into several variants on the approach. Most meaningfully were the results are of high degree accuracy using the approach on data collected years apart. Researchers are hopeful this will move forward into clinical trials to ultimately be used in a commercially available test.
