Screening biomarkers for autism spectrum disorder using plasma proteomics combined with machine learning methods

Area of Science:

• Biochemistry
• Proteomics
• Neurodevelopmental Disorders

Background:

• Autism spectrum disorder (ASD) is a prevalent neurodevelopmental disorder in children.
• Early intervention significantly improves outcomes for individuals with ASD.
• Identifying reliable biomarkers is crucial for early ASD detection and timely intervention.

Purpose of the Study:

• To investigate novel blood-based protein biomarkers for the early detection of autism spectrum disorder (ASD).
• To identify and validate a panel of proteins in plasma that can differentiate ASD cases from controls.

Main Methods:

• Utilized Sequential Window Acquisition of All Theoretical Spectra-Mass Spectrometry (SWATH-MS) for initial proteomic profiling.
• Employed machine learning algorithms to screen for candidate biomarkers from differentially expressed proteins.
• Validated potential biomarkers using targeted proteomics Multiple Reaction Monitoring (MRM) in an independent cohort.

Main Results:

• Identified 51 differentially expressed proteins (DEPs) associated with immune response and metabolic pathways.
• Machine learning identified 10 candidate protein biomarkers.
• Four proteins (PPBP, APCS, FGG, PF4V1) were validated as significant discriminators, achieving an Area Under the Curve (AUC) of 0.8087 for ASD screening.

Conclusions:

• A specific combination of four plasma proteins demonstrates significant potential as a screening tool for autism spectrum disorder (ASD).
• These findings support the development of blood-based assays for early ASD identification.
• Further research is warranted to confirm the clinical utility of these protein biomarkers in diverse populations.