Machine learning of brain-behavior dimensions reveals four subtypes of autism spectrum disorder linked to distinct molecular pathways. Here, the 3D prism cube represents the machine learning of the three brain-behavior dimensions, etched onto the prism’s glass. White light or “data” passes into the prism or “machine learning algorithm,” splitting into four colored light paths that represent the spectrum of autistic people in the four autism subtypes. The painted background of a sequencing array represents the molecular associations of the autism subtypes. CREDIT Weill Cornell Medicine; Dr. Amanda Buch

People with autism can be classified into four distinct subtypes based on their brain activity and behavior, according to a study from Weill Cornell Medicine investigators.

The study, published March 9 in Nature Neuroscience, leveraged machine learning to analyze newly available neuroimaging data from 299 people with autism and 907 neurotypical people. They found patterns of brain connections linked with behavioral traits in people with autism, such as verbal ability, social affect, and repetitive or stereotypic behaviors. They confirmed that the four autism subgroups could also be 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 different types of difficulties with social interaction, communication and repetitive behaviors. Scientists believe there are probably many different types of autism spectrum disorder that might require different treatments, but there is no consensus on how to define them,” said co-senior author Dr. Conor Liston, an associate professor of psychiatry and of neuroscience in the Feil Family Brain and Mind Research Institute at Weill Cornell Medicine. “Our work highlights a new approach to discovering subtypes of autism that might one day lead to new approaches for diagnosis and treatment.”

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

“If you put people with depression in the right group, you can assign them the best therapy,” said lead author Dr. Amanda Buch, a postdoctoral associate of neuroscience in psychiatry at Weill Cornell Medicine.

Building on that success, the team set out to determine if similar subgroups exist among individuals with autism, and whether different gene pathways underlie them. She explained that autism is a highly heritable condition associated with hundreds of genes that has diverse presentation and limited therapeutic options. To investigate this, Dr. Buch pioneered new 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 in the autism subgroups.

“One of the barriers to developing therapies for autism is that the diagnostic criteria are broad, and thus apply to a large and phenotypically diverse group of people with different underlying biological mechanisms,” Dr. Buch said. “To personalize therapies for individuals with autism, it will be important to understand and target this biological diversity. It is hard to identify the optimal therapy when everyone is treated as being the same, when they are each unique.”

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

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

Those advances allowed the team to identify four clinically distinct groups of people 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 other had more repetitive behaviors and less social impairment. The connections between the parts of the brain that process visual information and help the brain identify the most salient incoming information were hyperactive in the subgroup with more social impairment. These same connections were weak in the group with more repetitive behaviors.

“It was interesting on a brain circuit level that there were similar brain networks implicated in both of these subtypes, but the connections in these same networks were atypical in opposite directions,” said Dr. Buch, who completed 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 other two groups had severe social impairments and repetitive behaviors but had verbal abilities at the opposite 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 better understand what was causing the differences and found many were genes previously linked with autism. They also analyzed network interactions between proteins associated with the atypical brain connections, and looked for proteins that might serve as a hub. Oxytocin, a protein previously linked with positive social interactions, was a hub protein in the subgroup of individuals with more social impairment but relatively limited repetitive behaviors. Studies have looked at the use of intranasal oxytocin as a therapy for people with autism with mixed results, Dr. Buch said. She said it would be interesting to test whether oxytocin therapy is more effective in this subgroup.

“You could have treatment that is working in a subgroup of people with autism, but that benefit washes out in the larger trial because you are not paying attention to subgroups,” Dr. Grosenick said.

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

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

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

In 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 could have been helpful for him,” Dr. Buch said.  

Researchers have demonstrated that starting intervention coaching parents of autistic toddlers as early as 18 months leads to better gains in language, social communication, and daily living skills.

While prior studies provided strong evidence for the benefits of early intervention in autism, many are correlation studies rather than randomized controlled studies that can provide more conclusive results. Additionally, prior research has not demonstrated an ideal age to begin interventions.

“Many of us in the autism community say earlier is better, but we do not have enough high-quality evidence for that, so a randomized controlled trial like this one helps address that issue,” said lead author Whitney Guthrie, PhD, a clinical psychologist.

Researchers used the Early Social Interaction (ESI) model, a parent-implemented intervention for toddlers diagnosed with autism that provides a framework to support a child’s development in social communication and active engagement. It also is a framework for parents as they receive group education and individualized coaching to incorporate evidence-based strategies into everyday activities in the family’s natural environments.

In this study, families received both group and individual conditions but were randomized as to whether the Individual-ESI intervention coaching began when the toddlers were 18 months old or 27 months old. A total of 82 toddlers and their parents participated in the study, conducted at Florida State University and the University of Michigan.

The study found that children who received individualized ESI coaching earlier showed greater gains in use and understanding of language, social use of communication skills, and self-help skills. Additionally, the findings were specific to the intensive and individualized parent-coaching model, compared with the group-education treatment.

“This study provides research findings that show starting just nine months earlier, at 18 months versus 27 months, makes a difference in the child’s progress,” said Amy Wetherby, PhD, director of the Autism Institute at Florida State University. “And that makes sense because this is when a child’s brain is developing the most rapidly, in the first two years of life. That doesn’t mean they won’t continue to learn later, but by providing these interventions even earlier, they have the greatest chance of having the most impact.”

“This program is also not something that is likely to cause harm, and it could help many children, with or without autism,” UCLA Professor Catherine Lord, PhD, said. “The intent is to support parents and other caregivers in their everyday interactions, not to replace other possible sources of intervention, including preschool programs and direct therapies.”

The research team included Guthrie and co-principal investigators Wetherby and Lord. Guthrie worked with Lord during her undergraduate career at Michigan, then worked with Wetherby during her doctoral program in clinical psychology at FSU. The findings were part of Guthrie’s ambitious doctoral dissertation.

This study highlights the importance of the American Academy of Pediatrics recommendation that all children are first screened for autism at 18 months, which is critical to get started earlier with intervention, Wetherby said.

Lord hopes the study results “will stimulate discussions among governmental agencies about how to get children into services faster, perhaps by not waiting for formal diagnoses but requiring an eventual diagnosis at some point, and also by providing more and better reimbursed diagnostic services.”

Lord also wants frontline workers who might be the first to see the children —like pediatricians, primary care physicians, speech-language pathologists and occupational therapists— to be able to admit children into programs.

Guthrie said that while it is important to expand future research to include more children from a wider and more diverse community, the results of this study provide measurable evidence supporting earlier intervention.

“Our findings underscore the importance of a swift referral for evaluation if a child screens positive for autism, or there is any professional or caregiver concern for autism,” Guthrie said. “We want to make sure that toddlers don’t miss this important window for earlier intervention.”

Pregnant women whose household tap water had higher levels of lithium had a moderately higher risk of their offspring being diagnosed with autism , according to a new study led by a UCLA Health researcher.

“Any drinking water contaminants that may affect the developing human brain deserve intense scrutiny,” said lead study author Beate Ritz, MD, PhD, professor of neurology in the David Geffen School of Medicine at UCLA and professor of epidemiology and environmental health at the UCLA Fielding School of Public Health. “In the future, anthropogenic sources of lithium in water may become more widespread because of lithium battery use and disposal in landfills with the potential for groundwater contamination. The results of our study are based on high-quality Danish data but need to be replicated in other populations and areas of the world.”

Because of lithium’s mood-stabilizing effects, some lithium compounds have long been used to treat depression and bipolar disorders. However, there has been debate about whether mothers can safely take lithium during pregnancy amid increasing evidence it is associated with a higher risk of miscarriage and cardiac anomalies or defects in newborns.

Ritz, whose research focuses on how environmental exposures influence neurodevelopmental disorders and neurodegenerative diseases, said she decided to examine the possible association between lithium and autism risk after finding little research in humans about how lithium affects brain growth and development. Still, she found that some experimental research indicated lithium, which is among several naturally occurring metals often found in water, could affect an important molecular pathway involved in neurodevelopment and autism.

Zeyan Liew, PhD, MPH, first author of the study and an Assistant Professor of Epidemiology at the Yale University School of Public Health, added that this study was important because prior research findings from Denmark using high-quality medical registry data have already shown that ingestion of chronic and low-dose lithium from drinking can influence the occurrence of adult onset neuropsychiatric disorders. However, no study has been performed to assess whether lithium from drinking water consumed by pregnant women affects their child’s neurodevelopment.

Ritz and Liew worked with Danish researchers who analyzed lithium levels in 151 public waterworks in Denmark, representing the water supply for about half of the country’s population. To identify which waterworks supplied mothers’ homes at the time of their pregnancy, the researchers used address information from Denmark’s comprehensive civil registry system. Using a nationwide database of patients with psychiatric disorders, the researchers identified children who were born in 1997-2013, and compared 12,799 diagnosed with autism against 63,681 children who did not have an autism diagnosis. The researchers alsocontrolled for maternal characteristics, some socioeconomic factors and air pollution exposures, all of which have been linked to increased risk of autism in children.

As lithium levels increased, so did the risk of an autism diagnosis, the researchers reported. Compared to the lowest quartile of recorded lithium levels – in other words, those in the 25th percentile – lithium levels in the second and third quartiles were associated with a 24-26% higher risk of autism. In the highest quartile, the risk was 46% higher compared to the lowest quartile.

The researchers found a similar relationship between increased lithium levels and higher risk of autism diagnosis when the data were broken down by subtypes of the disorder. They also found the association between lithium levels and autism risk was slightly stronger for those living in urban areas compared to smaller towns and rural areas.

In addition to Denmark’s comprehensive civil databases that have proven to be valuable resources for public health researchers, several other factors made Denmark ideal for this study. Denmark’s consumption of bottled water ranks among the lowest in Europe, meaning Danes largely rely on tap water. The country also has a robust system for measuring trace metals and other contaminants in their water supply. Ritz said lithium levels in Denmark’s water, when compared to other countries, are likelyin the low to moderate range.