BLOOD-BASED AUTISM SPECTRUM DISORDER SIGNATURES FROM THE ITALIAN AUTISM NETWORK COLLECTION

Background A substantial genetic component accounts for Autism Spectrum Disorders (ASD) etiology. Some rare and, more recently, common genetic risk factors for ASD have been identified, thanks to recent large consortium efforts. The availability of blood samples from thoroughly characterized ASD families becomes an essential step to characterize genotype-phenotype correlations and identify novel candidate biomarkers. The Italian Autism Network holds a collection of about 250 families with extensive clinical information and a biobank of DNA, RNA and plasma derived from ASD patients and first-degree relatives (for a total of more than 800 subjects) enabling multidisciplinary research of ASD. Methods Here we present a study conducted on 76 sibling pairs from the Italian Autism Network (ITAN) collection selected from families formed by a proband and at least one unaffected sibling. RNA sequencing was conducted on PAX-gene blood samples and whole-genome genotype was derived by analysis on custom DNA arrays (PsychArray). To identify differential gene expression signature, a paired design model based on negative binomial distribution was used, correcting the model for batch effect, age, ethnicity and genetic similarity scores. In addition, based on the recent availability of GWAS data for ASD from a large meta-analysis, individual Polygenic Risk Score (PRS) for each subject was calculated and included in the gene expression model to highlight differences in gene expression between cases and controls independent from their burden of common risk factors. Results We have identified a gene expression signature for ASD which appears to reflect an unbalanced immune cell function. Using PRS as a covariate increased the number of genes above the statistical threshold. The gene set thus identified, being unrelated to the common genetic risk component, should more closely reflect disease status. Analysis of the PRS distribution confirms a higher score for affected versus unaffected siblings in spite of high similarity between related samples. Discussion As shown by investigations on ASD brain samples, dysregulation of gene networks both causally related (e.g. synaptic genes) and unrelated (e.g. immune function genes) to ASD etiology occur at transcriptome level. Additive effects of common variants likely playing a considerable role in ASD risk can be captured by PRS based on large GWAS datasets. By integrating transcriptome data with PRS it appears to be possible to tease apart the genetic and environmental factors affecting peripheral gene expression in ASD, suggesting novel candidate biomarkers for disease status.