Autism spectrum disorder (ASD) encompasses several disabling neurodevelopmental conditions, characterized by impaired social and communication skills and repetitive and restrictive behaviors or interests. There are currently no approved biomarkers for ASD screening and diagnosis, which depends heavily on expert physician assessment and awareness of ASD symptoms within the affected families. A group from Hamad Bin Khalifa University (HBKU), Doha, Qatar undertook an extensive proteomic analysis and applied multiple computational methods in a study of ASD patients in order to explore common underlying dysfunctions between cases of ASD and identify new biomarkers. They measured the expression of 1196 serum proteins using 13 Olink® Target 96 panels, comparing 91 ASD cases and 30 healthy controls between 6 and 15 years of age.
The Olink analysis revealed 251 differentially expressed proteins between ASD and healthy controls, of which 237 proteins were significantly upregulated and 14 proteins were significantly downregulated. Machine learning analysis then identified 15 proteins (TNFSF14, EGF, LAP-TGFbeta1, JAM-A, CD40L, GP6, ARHGAP25, CLEC1B, EREG, ST1A1, ARSB, CASP2, LSP1, MANF & PTPN1) that could be biomarkers for ASD with an area under the curve (AUC) = 0.876 .
Pathway and biological function (GO) analysis of the top differentially expressed proteins revealed dysregulation of SNARE vesicular transport and ErbB pathways in ASD cases. Correlation analysis further showed that proteins from those pathways also correlate with ASD severity. Investigation of available data sources also showed that 18 of the differentially expressed proteins identified have their corresponding genes associated with ASD according to the Simons Foundation Autism Research Initiative (SFARI) genetic database. Given that the pathways identified were enriched in the ASD blood proteome profile despite the high heterogeneity of ASD, they might play an essential role as common underlying pathophysiological mechanisms in ASD.