Machine Learning Identified Distinct THC and CBD Biomarker Signatures in Saliva of Children With Autism Receiving Medical Cannabis | Research

The takeaway

Machine learning analysis of saliva metabolomics from children with autism found that medical cannabis treatment shifted biomarker levels toward typically developing children, with THC and CBD each producing distinct metabolic signatures.

Autism researchers, pediatric medical cannabis clinicians, metabolomics researchers.

THC and CBD produce distinct metabolic biomarker signatures in saliva of autistic children

What the researchers found

Lysophosphatidylethanolamine distinguished ASD from typically developing groups. THC-associated and CBD-associated cannabis-responsive biomarkers formed two distinct groups, while CBG was associated with biomarkers from both. Novel phytochemicals beyond THC/CBD were identified as contributing to therapeutic effects through acetylcholinesterase inhibition. Medical cannabis treatment shifted biomarker levels in children with ASD toward typically developing levels.

Why it matters

This is the first application of machine learning to cannabis-responsive biomarkers in autism. Finding distinct THC and CBD metabolic signatures and that treatment shifts ASD biomarkers toward typical levels provides a potential framework for personalizing cannabis treatment and measuring response.

The numbers in context

Lysophosphatidylethanolamine identified as ASD-TD distinguishing biomarker. THC and CBD biomarker groups distinct. CBG overlaps both groups. Novel phytochemicals identified as acetylcholinesterase inhibitors.

How the study worked

Machine learning techniques applied to dynamic, high-resolution salivary metabolomics data from children with ASD before and after medical cannabis treatment and a typically developing control group.

What this study cannot tell us

Small sample size. Machine learning with limited data risks overfitting. Saliva metabolomics is an emerging field with limited validation. Cannot determine if biomarker changes cause clinical improvement. No placebo control.

How to read the evidence

Novel machine learning approach with limited sample, providing proof-of-concept requiring larger validation.

When this study was published

Published 2023.

The bigger picture

If validated, salivary biomarkers could provide an objective way to measure whether medical cannabis is working for a child with ASD, replacing subjective symptom ratings. The acetylcholinesterase inhibition finding also suggests a mechanism beyond just THC/CBD effects.

Questions still open

Can salivary biomarkers guide medical cannabis dosing for autism?
Do the novel phytochemicals beyond THC/CBD contribute meaningfully to clinical improvement?

Common questions

Can we measure if medical cannabis is working for autism?

This study found that salivary biomarkers shifted toward typical levels after medical cannabis treatment in children with ASD, potentially providing an objective way to assess response.

Do THC and CBD affect autism differently?

Machine learning analysis found THC and CBD each produced distinct patterns of metabolic biomarker changes, while CBG showed overlap with both, suggesting each cannabinoid has a unique biological signature.

Read the original research

A machine learning approach for understanding the metabolomics response of children with autism spectrum disorder to medical cannabis treatment.

Scientific reports, 13(1), 13022

View on PubMed ↗Publisher’s paper ↗

Citation

Quillet, Jean-Christophe; Siani-Rose, Michael; McKee, Robert; Goldstein, Bonni; Taylor, Myiesha; Kurek, Itzhak. (2023). A machine learning approach for understanding the metabolomics response of children with autism spectrum disorder to medical cannabis treatment.. Scientific reports, 13(1), 13022. https://doi.org/10.1038/s41598-023-40073-0