In The Politics of Autism, I point out that there is currently no commercially-available blood test for autism. That situation may change someday.
Instead of looking for one single chemical to measure, the researchers — led by Prof. Juergen Hahn — used a big data approach and searched for patterns in metabolites.
In 2017, the researchers had their first success. They analyzed blood from 149 people with an ASD diagnosis, assessing each sample for levels of 24 metabolites. The chemicals were all related to two particular cellular pathways: the methionine cycle, and the transsulfuration pathway.
Having done this, they scientists were able to create a test that could correctly identify more than 96 percent of ASD cases within the group that they had recruited.
Recently, the same team set out to replicate its findings in a new dataset.
They assessed data from 154 children with ASD, taken by researchers from the Arkansas Children's Research Institute in Little Rock. This time, however, they only had access to information on 22 of the 24 metabolic markers they had used in the last trial.
Their results were published this month in the journal Bioengineering and Translational Medicine,and they are encouraging.
When they applied the algorithm, it predicted ASD correctly in 88 percent of cases.
While 88 percent is an impressive result, it is lower than the success rate from the previous studies. Prof. Hahn thinks that this is because the two missing metabolites were shown to be strong indicators in the last study. However, the results are still exciting.As always with autism, caution is in order. For decades, ASD people and their families have read of medical breakthroughs, only to meet with disappointment as more studies come in.