Screening Hopes for Autism

Diagnosing ASD (autistic spectrum disorder) is often a challenge, especially in very young children, but early identification of those at risk could lead to better monitoring and earlier detection. Now, a team of international researchers are working on a blood test that they believe has the potential to be an autism screening tool. We spoke with Eric Courchesne, principle investigator and co-director of the Autism Center of Excellence at the University of California, San Diego, and co-author of the associated paper (1).

What motivated you to develop this test?

ASD is a highly heterogeneous disorder. Many genomic analyses have been conducted, which have led to the discovery of dozens of mutations. However, each mutation can only explain a small fraction of the cases, and the pathways involved in its development are unclear. This genetic complexity makes it difficult to conclusively diagnose the disorder before a child’s fourth year of life. In fact, the median age of diagnosis in the United States is 4 and a half years. This late identification is a serious concern – research shows early identification and treatment by age 2 or 3 years leads to much better clinical outcomes. At present, there is no successful early biomarker, leaving affected infants undetected and untreated.

How does it work?

We used weighted gene expression values to classify ASD vs typical and non-ASD developmentally delayed toddlers. The test uses over 700 genes in specific gene networks that we found to be abnormally expressed in blood leukocytes of 1- to 3-year-old toddlers with ASD, as compared with non-ASD toddlers. The expression of each gene is given a weight based on the gene’s aberrance in ASD and its contribution to dysregulation of the network.

So far, our test has 83 percent accuracy, 80 percent specificity and 85 percent sensitivity. This substantially outperforms all known biomarkers at very young ages in ASD, including genetic markers. A large number of genetic defects have been identified as possible risk factors in ASD, but each individual defect typically occurs in less than 1 percent of all patients with ASD. By some estimates, when combined these gene defects account for only 2–3 percent of variance. In my opinion this shows genetic markers are currently unsuitable for general pediatric screening. We are aiming to develop an accurate, cost-effective first or second tier screening test.

Practically, our test is simple: blood is drawn, then quickly processed so that leukocyte mRNA can be extracted and stabilized. We then use the Illumina HT-12 to obtain gene expression data and expression values for the genes are weighted, and the computed values are used to classify infants as either ASD or non-ASD.

Could this test be used clinically?

Our current test was a proof of concept. We are in the process of further developing and validating our classifier approach, and in the distant future, hopefully some clinical trials will be conducted. I believe that if a validated and robust test emerges, it could be used clinically. Many steps stand between successful proof of concept and actual clinical use – but a truly successful screening method for high risk of autism is very important to achieve, and someday someone will do it.

What are your next steps?

We are using RNA sequencing and new systems biology methods to develop even more accurate, specific and sensitive functional genomic biomarkers of early risk for ASD in infants and toddlers. We have quadrupled our sample size, and aim to double that again.