Written in Blood: A Promising Autism Prediction Model using Expression Profiling
Autism Spectrum Disorder (ASD) encompasses a broad range of social and developmental delays with onset before three years of age. Characteristics of children with ASD include restricted and repetitive behavior, communication deficits, and social impairments. ASD is the result of neurodevelopmental disorders, but genetics cannot fully explain the onset. Although some cases have been linked to particular DNA mutations and most have been shown to be highly heritable, epigenetics may also be a contributing factor. The diagnostic procedures that are currently used for ASD are limited to certain behavioral milestones and therefore postpone the diagnosis of ASD. However, early diagnosis and treatment have been linked to improved outcomes, and as the prevalence of ASD continues to increase, the establishment of a reliable evaluation technique becomes more urgent. For this reason, researchers have turned to the blood transcriptome in search of signature patterns that are able to distinguish children with ASD from controls.
Scientists have recently published data that suggest that gene expression signatures obtained from blood samples may indicate genetic changes that point to ASD. The study mainly focused on two independently collected data sets of blood samples (P1 and P2) from males between four and 13 years of age who had either been diagnosed with ASD or did not exhibit any signs of chronic diseases or neurological disorders. Separate microarray profiling was done on the P1 and P2 blood samples and hundreds of transcripts (489 in P1 and 610 in P2) were found to be differentially expressed in each group, reflecting disturbances in multiple signaling pathways. The scientists then analyzed the 489 genes from P1 with the EASE (Expression Analysis Systematic Explorer) exact test to determine which processes were overrepresented relative to normal levels. 22 pathways were found to be overrepresented; most notably, the neurotrophin signaling pathway, which is critical to neural and brain development. These results led the scientists to generate peripheral blood gene expression profiles that could be used as molecular parameters for identifying ASD. Using the P1 ASD and control data sets, Kong S.W., et. al. developed a prediction model based off of combinations of differential gene expression levels. By testing the independently acquired P2 data sets against the prediction model, the scientists were able to correctly diagnose 73% of the patients with a 95% confidence interval. Additional testing for potential inconsistencies found that while the age of the patient at the time of blood draw significantly influenced levels of gene expression, all but two of the core genes used in the prediction model were unaffected.
The prediction model based off of gene expression signatures described here is promising for distinguishing children with ASD from controls yet remains relatively noninvasive. Unfortunately, it is still short of perfect. While the current model can correctly identify 73% of ASD cases in males between the ages of four and 13 years, further studies are necessary to produce models that can more accurately assess the disorder in both genders. Also, at what age can the gene expression signature be detected? Is the signature present at birth? And if so, what are the tell-tale signs that would subject a baby to testing for ASD?
Kong SW, Collins CD, Shimizu-Motohashi Y, Holm IA, Campbell MG, Lee IH, Brewster SJ, Hanson E, Harris HK, Lowe KR, Saada A, Mora A, Madison K, Hundley R, Egan J, McCarthy J, Eran A, Galdzicki M, Rappaport L, Kunkel LM, & Kohane IS (2012). Characteristics and predictive value of blood transcriptome signature in males with autism spectrum disorders. PloS one, 7 (12) PMID: 23227143