DEVELOPMENT OF A BLOOD-BASED MOLECULAR SIGNATURE FOR AUTISM

Melnick Auditorium

Objective. This study was conducted to validate and optimize a proposed gene expression signature for autism across two independent sets of patients, potentially providing the basis of a biological test for this disorder. Background. Autism is presently diagnosed using a variety of behavioral assessments. Significant gene expression differences in peripheral blood leukocytes found in 11 common genes (log-ratio > 1.5; p < 0.05; q2 0.05) from children with autism as compared with neurotypical controls were proposed as a potentially diagnostic “signature” for the disease by Gregg et al. Methods. Microarray datasets were downloaded from the Gene Expression Omnibus data archive established by the National Center for Biotechnology Information. The data was converted into a format analyzable using Microsoft Excel. Validation and optimization studies were performed using the previously described gene signature in the much larger Alter et al. dataset, in order to best differentiate the autistic subjects from controls across both datasets simultaneously. Results. 7 of 11 original genes in the signature were also significantly differentially expressed in the Alter et al. dataset. Through subset analyses, a group of 31 genes were developed as an optimized signature that was able to distinguish neurotypical from autistic subjects with a specificity of 0.96. Conclusion. The validation and optimization of this group of genes across two independent sets of patients suggests that there is a shared gene expression signature for autism that could potentially provide the basis of a biological test for this disorder. Grants. N/A