Valerie W. Hu, PhD;* Yinglei Lai, PhD

Autism spectrum disorders (ASD) are neurodevelopmental disorders which are currently diagnosed solely on the basis of abnormal stereotyped behavior as well as observable deficits in communication and social functioning. Although a variety of candidate genes have been identified on the basis of genetic analyses and up to 20% of ASD cases can be collectively associated with a genetic abnormality, no single gene or genetic variant is applicable to more than 1-2 percent of the general ASD population. In this report, we apply class prediction algorithms to gene expression profiles of lymphoblastoid cell lines (LCL) from several phenotypic subgroups of idiopathic autism defined by cluster analyses of behavioral severity scores on the Autism Diagnostic Interview-Revised diagnostic instrument for ASD. We further demonstrate that individuals from these ASD subgroups can be distinguished from nonautistic controls on the basis of limited sets of differentially expressed genes with a predicted classification accuracy of up to 94% and sensitivities and specificities of ~90% or better, based on support vector machine analyses with leave-one-out validation. Validation of a subset of the “classifier” genes by high-throughput quantitative nuclease protection assays with a new set of LCL samples derived from individuals in one of the phenotypic subgroups and from a new set of controls resulted in an overall class prediction accuracy of ~82%, with ~90% sensitivity and 75% specificity. Although additional validation with a larger cohort is needed, and effective clinical translation must include confirmation of the differentially expressed genes in primary cells from cases earlier in development, we suggest that such panels of genes, based on expression analyses of phenotypically more homogeneous subgroups of individuals with ASD, may be useful biomarkers for diagnosis of subtypes of idiopathic autism.

Key Words: Autism, subphenotypes, gene expression, class prediction, blood biomarkers

Valerie W. Hu, PhD;^1* Yinglei Lai, PhD^2
1Department of Biochemistry and Molecular Medicine, George Washington University School of Medicine and Health Sciences, DC
²Department of Statistics, George Washington University, DC
*Corresponding Author: Department of Biochemistry and Molecular Medicine, George Washington University School of Medicine and Health Sciences, Washington, DC 20037. Tel: 202-994-8431.
(Email: valhu@gwu.edu)

CONFLICT OF INTEREST
None.

ACKNOWLEDGEMENTS
VWH thanks Ms. Mara Steinberg for assistance in cell culturing and RNA preparation from the new set of LCL for the qNPA analysis. This study was supported by a supplement to NIH grant # R21 MH073393 (VWH). The funding agency had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. Intellectual property (GWU and VWH) arising from this study has been licensed by SynapDx Corp. (Boston, MA) which played no role in any part of this study nor in the decision to publish this manuscript.

FUNDING
This study was supported by a supplement to NIH grant # R21 MH073393 (VWH).