An Unexpected Takeaway From The Early Autism Diagnosis Study
Findings from a new autism study suggest that changes in the brain in early infancy may predict diagnosis at age 2 in children who have higher odds of autism because an older sibling is autistic. The study, published in Nature, doesn’t tell us anything about causation, but it does incidentally confirm that the MMR vaccine has nothing to do with autism.
The researchers imaged the brains of children at “high risk” for autism because of an older sibling’s diagnosis and the brains of children who had no family history of autism. They performed MRIs when the children were ages 6, 12 and 24 months and administered a test at age 24 months that is diagnostic of autism and another that assesses social skills.
Some of the measures they took with MRI included total brain volume, thickness of the cortex (the outer part of the brain) and surface area of the cortex. They then compared changes in these measures over time with outcomes like severity score on the autism diagnostic scale and scores on the social measures. Part of their analysis involved using an algorithm to take the imaging information and predict which children would later be diagnosed with autism.
The results suggest that rapid cortical surface growth from ages 6 to 12 months was a predecessor of a “dramatic” increase in brain volume at age 12 to 24 months in high-risk children diagnosed with autism at age 24 months. The machine learning algorithm predicted this diagnosis in 81% of the high-risk children who were ultimately diagnosed with autism. The machine also gave false positive result for four children.
The numbers in some of the analyses were 106 high-risk infants and 42 low-risk, although the original group consisted of 117 with no family autism history, 248 with an autistic sibling but not diagnosed with autism themselves and 70 with an autistic sibling who also were diagnosed with autism. The brain growth analysis included only 15 of the 70 high-risk children diagnosed with autism because they were the only participants who had MRI at all three timepoints (ages 6, 12 and 24 months).
Children who have an autistic sibling have increased odds of being autistic themselves, up to 1 in 5 compared to the current 1 in 68 (CDC) for the general population. Here, 70 out of 318 of the group with autistic siblings were diagnosed with autism whereas only three of the group with no family history met criteria under an older version of the DSM (Diagnostic and Statistical Manual of Mental Health Disorders). That’s 2.6% without a family history vs. 22% in the higher-odds group.
The study population was not large, especially considering the reductions in numbers for some analyses, and the authors caution that replication will be needed before anything along these lines could be used in practice for children with higher autism odds. Also, this work would be applicable only to children who are already known to have increased odds because of having an autistic sibling.