Saliva Testing Accurately Identifies Children With Autism

By Will Boggs MD

December 17, 2018

NEW YORK (Reuters Health) - A salivary test based on multiple RNA features can accurately identify children with autism spectrum disorder (ASD), researchers report.

"Though additional validation of these results is needed (and is currently being funded by the National Institutes of Health), our findings suggest that a biologic test for autism may be part of physicians' toolkits in the near future," Dr. Steven D. Hicks from Penn State College of Medicine, in Hershey, told Reuters Health by email.

Previous ASD biomarker studies have been hampered by insufficient sample sizes, focus on single molecules, a lack of separate training and test sets, and poor generalizability and validity.

Dr. Hicks and colleagues evaluated levels of human and microbial salivary RNAs to train and then test a biomarker classification tool in 456 children between the ages of 19 and 83 months.

The selection algorithm identified a panel of 32 diagnostic RNA features, including 12 microbial taxa, seven mature microRNAs, four precursor microRNAs, eight piwi-interacting RNAs and one small nucleolar RNA.

In the training-set validation, the algorithm identified ASD status with 80% sensitivity, 78% specificity and 87% accuracy (AUC), the researchers report in Frontiers in Genetics, online November 9.

In the test set, the algorithm accurately predicted ASD status in 41/50 ASD children, 18/21 children with typical development, and 12/13 children with developmental delay, which translates into 82% sensitivity, 88% specificity, 88% accuracy and a positive predictive value of 91%.

One-fourth of misclassified children with ASD had above-average social scores.

Evaluation of putative microRNA targets revealed significant enrichment of several potentially relevant pathways, including axon guidance, neurotrophin signaling, and circadian entrainment, as well as such pathway interactions as xenobiotic metabolism and NF-kappa B signaling.

"Based on the make-up of our study cohort (about 50% autism, 25% non-autism developmental delay, 25% children with typical development), such a test would need to be employed as a diagnostic adjunct (not a broad screening tool)," Dr. Hicks said. "For example, general pediatricians might apply it in patients with a positive score on the modified checklist for autism in toddlers (MCHAT) to improve the specificity of referrals to developmental specialists."

"Alternatively, developmental specialists might employ this type of technology to provide an additional level of diagnostic evidence (e.g. in children with borderline behavioral assessments, or in cases where parents were skeptical of initial diagnoses)," he said. "I envision the results of such a test would not definitively identify a child as having (or not having) autism, but would estimate the extent to which a child's RNA profile was consistent with autism."

Dr. Hicks and two coauthors are co-inventors of patent applications for RNA biomarkers in autism spectrum disorder that are licensed to Quadrant Biosciences Inc.; Dr. Hicks has financial ties to the company.


Front Genet 2018.