Genetic test to predict autism

13 September 2012

Researchers have developed a genetic test able to predict who will receive a diagnosis of autism with up to 85% accuracy.

The research published in the Nature journal Molecular Psychiatry looked at single nucleotide polymorphisms (SNPs) in the DNA of people who had been diagnosed with autism spectrum disorder (ASD) in comparison with non-ASD controls.

The ‘diagnostic classifier’ they arrived at consisted of 237 SNPs within 146 different genes. With this they were able to correctly predict whether an individual had received a diagnosis of ASD with 85% accuracy in two separate cohorts of people of central European (CEU) descent, with 71% accuracy in a further two further independent CEU-descended validation samples, and just over 56% in a cohort of Han Chinese origin.

The largest predictive effects were conferred by eight SNPs located within three genes – some of which increased risk, and others of which had a protective effect – but the accuracy of the test was diminished by any reduction from using all 237 SNPs.

The authors conclude that they have developed a diagnostic genetic test for ASD that is accurate within a genetically homogenous group. They consider that the test’s much lower predictive ability within the Han Chinese cohort is explained by a different SNP profile within this group, and that their method could likely be adapted to take account of this for different ethnicities.

Confirmation of these claims will require modification of the test and replication of the results within a range of other ethnic groups. Overall, their findings indicate that a SNP-based test could become a useful additional tool to aid early identification of ASD. One of the study’s co-authors Renee Testa said "Early identification of risk means we can provide interventions to improve overall functioning for those affected, including families".

More from us