MRIs Could Help Predict Whether Infants at High Risk Will Develop Autism
May 12, 2017
Treating autism spectrum disorders is all about increasing function and skills. There are currently no drugs that treat the behavioral and communication symptoms of these disorders, so mental health professionals use language, sensory and occupational therapies to teach patients the skills they need to communicate and function.
Most of those who treat autism agree that these therapies work best when started very early in life, but recognizing autism is extremely tricky before symptoms first appear at two or three years old.
Now two new studies from UNC-Chapel Hill’s Carolina Institute for Developmental Disabilities describe the most accurate methods for predicting autism spectrum disorders to date. Using MRI scans of infants at high risk for developing autism spectrum disorders, the researchers were able to identify several anatomical markers that predict whether the infant will develop autism with 70 to 80 percent accuracy.
These findings would allow doctors to detect autism very early on. Children who are likely to develop autism could begin therapies like incorporating communication skills into play, or placing the child with typically developing children to learn appropriate social behavior before symptoms appear while the brain is still undergoing its most dramatic developments. By intervening earlier, children with autism spectrum disorders could develop more function and lead healthier lives.
With the prevalence of autism in the United States increasing 119 percent between 2000 and 2010, more effective interventions could improve the lives of millions of Americans.
In both studies, UNC School of Medicine psychiatry professor and senior author of both studies, Dr. Joe Piven, directed researchers throughout the United States and Canada. Researchers took MRI scans of hundreds of babies at six months, 12 months and 24 months old. Most of the babies were at higher risk for developing autism because they had an older sibling with autism, but there were also low-risk babies included as a control group.
The researchers were looking for structural and physiological changes in the babies’ brains, and after tracking which babies developed autism symptoms at 24 months, they correlated these changes to a likelihood of developing autism.
In the first study, published in the journal Nature, the researchers imaged 148 babies (106 at high risk and 42 at low risk). Among the high-risk infants, 15 developed autism at 24 months. Those 15 all showed an increase in brain surface area between six and 12 months and an increase in brain volume between 12 and 24 months compared to the other high risk infants.
The researchers then developed a computer algorithm, taking brain surface area, brain volume and the sex of the baby—boys are roughly four times as likely to develop autism than girls—and tried it out on a different set of babies. Using MRI data, the algorithm correctly predicted which infants would develop autism 81 percent of the time.
In the other study, published in the journal Biological Psychiatry, the researchers used MRIs to find out how much cerebrospinal fluid (CSF) the 221 high-risk and 122 low-risk babies had. CSF acts as a liquid cushion between the brain and the skull and performs immune functions in the brain. The brain manufactures it in cavities called the ventricles and it circulates out between the brain the skull, into the spinal cord and is eventually absorbed into the blood.
In the babies, the researchers found that those with autism had about 18 percent more fluid in their sub-arachnoid space—one of the layers between the brain and skull—than high-risk babies who did not develop autism. This effect scaled as well, as babies that developed the most severe autism symptoms had about 24 percent more CSF in the subarachnoid space. CSF measurements were able to predict with 69 percent accuracy whether a high-risk baby would develop autism or not, confirming the findings of a smaller study the group did in 2013.
The studies and their implications do have a few limitations. First, the researchers mainly tested how these MRI measurements predicted autism in babies that already had a sibling with autism, already a strong predictor. More research will be needed to determine whether babies with other risk factors, like having another family member with autism or certain other genetic markers, would exhibit the same physiological changes.
Second, it is possible that discovering these physical changes correlated with autism could provide another avenue for treatment. The studies, however, do not explain how the increases in brain size and CSF come about or even whether they cause the mental health symptoms associated with autism. Researchers will have to investigate further whether there are drugs or surgical treatments for these symptoms that can also help treat autism.
Third, these predictors are not 100 percent accurate. That could mean that some parents who would take early action to treat the autism their child was predicted to develop may not need to take that action. This might not be the most detrimental thing as the treatments do not involve dangerous drugs or surgeries, but when the situations are reversed, and parents do not take action because these tests predict their child will not develop autism, they could be missing out on crucial therapy time.
Finally, while MRI scans are among the safest imaging tests as they employ radio waves and magnets as opposed to potentially dangerous X-Rays, the scans take time and are not cheap, so getting access to these tests could be difficult for many parents.
Still, these two studies represent a major step forward in diagnosing and potentially treating one of the most common mental health disorders in the world.
Daniel Lane covers science, medicine, engineering and the environment in North Carolina.