Individuals with autism spectrum disorder may be classified into 4 distinct subtypes based on their brain activity and behavior, in keeping with a study from Weill Cornell Medicine investigators.

The study, published March 9 in Nature Neuroscience, leveraged machine learning to investigate newly available neuroimaging data from 299 individuals with autism and 907 neurotypical people. They found patterns of brain connections linked with behavioral traits in individuals with autism, resembling verbal ability, social affect, and repetitive or stereotypic behaviors. They confirmed that the 4 autism subgroups is also replicated in a separate dataset and showed that differences in regional gene expression and protein-protein interactions explain the brain and behavioral differences.

Like many neuropsychiatric diagnoses, individuals with autism spectrum disorder experience many differing types of difficulties with social interaction, communication and repetitive behaviors. Scientists consider there are probably many differing types of autism spectrum disorder that may require different treatments, but there isn’t any consensus on find out how to define them. Our work highlights a recent approach to discovering subtypes of autism that may sooner or later result in recent approaches for diagnosis and treatment.”

Dr. Conor Liston, co-senior writer, associate professor of psychiatry and of neuroscience within the Feil Family Brain and Mind Research Institute at Weill Cornell Medicine

A previous study published by Dr. Liston and colleagues in Nature Medicine in 2017 used similar machine-learning methods to discover 4 biologically distinct subtypes of depression, and subsequent work has shown that those subgroups respond otherwise to varied depression therapies.

“Should you put individuals with depression in the appropriate group, you possibly can assign them one of the best therapy,” said lead writer Dr. Amanda Buch, a postdoctoral associate of neuroscience in psychiatry at Weill Cornell Medicine.

Constructing on that success, the team set out to find out if similar subgroups exist amongst individuals with autism, and whether different gene pathways underlie them. She explained that autism is a highly heritable condition related to a whole bunch of genes that has diverse presentation and limited therapeutic options. To research this, Dr. Buch pioneered recent analyses for integrating neuroimaging data with gene expression data and proteomics, introducing them to the lab and enabling testing and developing hypotheses about how risk variants interact within the autism subgroups.

“One among the barriers to developing therapies for autism is that the diagnostic criteria are broad, and thus apply to a big and phenotypically diverse group of individuals with different underlying biological mechanisms,” Dr. Buch said. “To personalize therapies for people with autism, it would be necessary to know and goal this biological diversity. It is tough to discover the optimal therapy when everyone seems to be treated as being the identical, once they are each unique.”

Until recently, there have been not large enough collections of functional magnetic resonance imaging data of individuals with autism to conduct large-scale machine learning studies, Dr. Buch noted. But a big dataset created and shared by Dr. Adriana Di Martino, research director of the Autism Center on the Child Mind Institute, in addition to other colleagues across the country, provided the big dataset needed for the study.

“Latest methods of machine learning that may cope with hundreds of genes, brain activity differences and multiple behavioral variations made the study possible,” said co-senior writer Dr. Logan Grosenick, an assistant professor of neuroscience in psychiatry at Weill Cornell Medicine, who pioneered machine-learning techniques used for biological subtyping within the autism and depression studies.

Those advances allowed the team to discover 4 clinically distinct groups of individuals with autism. Two of the groups had above-average verbal intelligence. One group also had severe deficits in social communication but less repetitive behaviors, while the opposite had more repetitive behaviors and fewer social impairment. The connections between the parts of the brain that process visual information and help the brain discover probably the most salient incoming information were hyperactive within the subgroup with more social impairment. These same connections were weak within the group with more repetitive behaviors.

“It was interesting on a brain circuit level that there have been similar brain networks implicated in each of those subtypes, however the connections in these same networks were atypical in opposite directions,” said Dr. Buch, who accomplished her doctorate from Weill Cornell Graduate School of Medical Sciences in Dr. Liston’s lab and is now working in Dr. Grosenick’s lab.

The opposite two groups had severe social impairments and repetitive behaviors but had verbal abilities at the alternative ends of the spectrum. Despite some behavioral similarities, the investigators discovered completely distinct brain connection patterns in these two subgroups.

The team analyzed gene expression that explained the atypical brain connections present in each subgroup to raised understand what was making the differences and located many were genes previously linked with autism. In addition they analyzed network interactions between proteins related to the atypical brain connections, and searched for proteins that may function a hub. Oxytocin, a protein previously linked with positive social interactions, was a hub protein within the subgroup of people with more social impairment but relatively limited repetitive behaviors. Studies have checked out using intranasal oxytocin as a therapy for individuals with autism with mixed results, Dr. Buch said. She said it might be interesting to check whether oxytocin therapy is simpler on this subgroup.

“You would have treatment that’s working in a subgroup of individuals with autism, but that profit washes out within the larger trial since you aren’t listening to subgroups,” Dr. Grosenick said.

The team confirmed their results on a second human dataset, finding the identical 4 subgroups. As a final verification of the team’s results, Dr. Buch conducted an unbiased text-mining evaluation she developed of biomedical literature that showed other studies had independently connected the autism-linked genes with the identical behavioral traits related to the subgroups.

The team will next study these subgroups and potential subgroup-targeted treatments in mice. Collaborations with several other research teams which have large human datasets are also underway. The team can be working to refine their machine-learning techniques further.

“We try to make our machine learning more cluster-aware,” Dr. Grosenick said.

Within the meantime, Dr. Buch said they’ve received encouraging feedback from individuals with autism about their work. One neuroscientist with autism spoke to Dr. Buch after a presentation and said his diagnosis was confusing because his autism was so different than others but that her data helped explain his experience.

“Being diagnosed with a subtype of autism might have been helpful for him,” Dr. Buch said.

Source:

Journal reference:

Buch, A. M., et al. (2023). Molecular and network-level mechanisms explaining individual differences in autism spectrum disorder. Nature Neuroscience. doi.org/10.1038/s41593-023-01259-x.