Some autistic people seem to have unusually strong connections between different regions of their brain, while others have weakened links. These two connectivity patterns appear to be associated with different mechanisms within the brain, adding to the growing evidence that there are several kinds of autism that are at least somewhat distinct from each other.

“We were able to demonstrate that there are different dominant subtypes of autism that are associated with different biology,” says Alessandro Gozzi at the Italian Institute of Technology in Rovereto.

Autism is a form of neurodivergence, and it’s thought that about 780 individuals per 100,000 people are autistic. Common autistic traits include difficulties with social interactions, susceptibility to sensory overwhelm and restricted behaviours and interests. However, autistic people vary widely in how much they experience these traits.

For years, researchers have used brain-imaging methods, such as functional magnetic resonance imaging (fMRI), to look for a “signature” of autism in the brains of autistic people. “No such single signature has emerged,” says Gozzi. Some have found evidence of unusually strong connections between brain regions (“hyperconnectivity”), while others have found weaker links (“hypoconnectivity”), a mix of the two, or even a switch from one to the other during childhood.

Most of these studies neglected the diversity in autism, says Gozzi. To better understand what is really going on, his team first studied 20 strains of mice, each with a mutation in a different gene that is known to be associated with autism in people. fMRI scans revealed unusual patterns of connectivity: 11 of the mouse strains showed mostly hypoconnectivity, while the other nine showed mostly hyperconnectivity.

“These two different opposing signatures are associated with different mechanisms,” says Gozzi. The team showed this by mapping the proteins with which the mutated genes interact. In the mice with hypoconnectivity, the affected genes were interacting with proteins known to be involved with synapses, the connections between neurons. In the mice with hyperconnected brains, the relevant genes were interacting with a different group of proteins, some involved in gene regulation and others with the immune system.

Finally, the team studied fMRI data from 940 autistic people and 1036 age-matched controls. Among the autistic people, 24 per cent had hypoconnectivity, while 17 per cent had hyperconnectivity. “We have, at the very least, two dominant, biologically distinct subtypes of autism,” says Gozzi.

However, 59 per cent of the autistic people didn’t fit into either category, which may be related to the choice of genes for the study, so there is clearly much more variety to explore. “Our study does not claim that there are only these two subtypes”, says Gozzi, just that those were the two the researchers could detect and characterise.

Natalie Sauerwald at the Flatiron Institute in New York also thinks the number of autism subtypes remains unclear, but adds that this study helps to explain the heterogeneity of the condition and the biology behind it.

There are also issues in using animal models of autism. For one, people have hundreds of genes that each have a small effect on their likelihood of being autistic. Inevitably, the mice used in the study don’t “capture the full spectrum of autism”, says Sauerwald.

Some of the chosen genes are also linked to developmental delay, says Sauerwald. That means that such studies in mice are telling us about autistic people who also experience developmental delay, and can’t tell us about other autistic people.

In the long run, the challenge is to paint a complete picture of the diversity of autism, by linking genetics to brain connectivity, and then to behaviours, says Sauerwald.