Using a novel multi-modal approach to data analysis, scientists have identified a specific subtype of autism linked with a gene cluster known to affect both neurodevelopment and lipid metabolism. The research suggests this subtype could account for nearly seven percent of autism cases and future study will explore whether cholesterol-lowering drugs such as statins could be an effective treatment.
The collaborative research project – spanning Harvard Medical School, Massachusetts Institute of Technology and Northwestern University – ostensibly used autism as a test case to explore the efficacy of a novel data analysis approach. The goal is to identify distinct disease subtypes using large volumes of clinical, genomic and transcriptomic data.
The method combines disparate datapoints, including healthcare claims, electronic health records, familial whole-exome sequences and neurodevelopmental gene expression patterns, to better identify specific disease subtypes. The ultimate plan is to establish a multi-modal diagnostic approach allowing for more precise treatment outcomes.
“Think of a Google map and how it overlays various types of information on top of one another – cities, streets, parcels, land use, electrical grids, elevations – for a more detailed representation,” says Yuan Luo, co-lead on the new research. “This is what we did with our data to get a complete view of genes that have multiple regulatory functions and are implicated in autism.”
The research began by looking at gene expression patterns, with a focus on genes known to play a role in prenatal and postnatal brain development. Progressively zooming in on novel mutations more frequently appearing in subjects with autism, the study ultimately hit on a previously unrecognized cluster of exons (the parts of genes that code for amino acis). The cluster identified seemed to regulate both lipid metabolism and neurodevelopment.
Dyslipidemia is a condition where individuals present with abnormally high levels of lipids, such as triglycerides or cholesterol, in their blood. The next step in the new study was to conduct a massive analysis of patient data records to find out whether the correlation between autism and dyslipidemia appeared in real-life conditions.
Scanning millions of patient health records, encompassing over 100,000 subjects with autism, the researchers found 6.5 percent of those diagnosed with autism also displayed abnormal lipid levels. This was nearly double the rate of abnormal lipid levels identified in subjects without autism.
This is not the first study to suggest a link between dyslipidemia and autism, but it is the first to effectively quantify the commonality of the association and point to a possible molecular explanation. The researchers do suggest the findings raise a number of questions that will need to be explored in future studies. It is unclear exactly how lipid abnormalities could be influencing neurodevelopment, and it is also not known whether treating lipid abnormalities could improve autism outcomes.
Senior investigator on the study Isaac Kohane, from the Blavatnik Institute at Harvard Medical School, says the findings affirm the deeply diverse nature of autism. It is suggested the suspected variety of autism subtypes is analogous to cancer. Just like different types of cancer require fundamentally different types of treatments, autism may require a similarly heterogeneous approach.
“Our results are a striking illustration of the complexity of autism and the fact that autism encompasses many different conditions that each arise from different causes – genetic, environmental or both,” says Kohane. “Identifying the roots of dysfunction in each subtype is critical to designing both treatments and screening tools for correct and timely diagnosis – that is the essence of precision medicine.”
Beyond autism, this novel multi-modal approach should help break down a variety of other genetically complex diseases into distinct subtypes. Alzheimer’s disease, for example, has proved notoriously difficult to treat and some researchers are beginning to suggest it needs to be broken down into sub-types and not considered a homogenous neurodegenerative condition.
“Conceptually, this is the same framework that we can apply in complex inherited neurodevelopmental disorders like autism and beyond,” says Kohane. “Our multimodal approach combining multiple types of data demonstrates that this is not only possible but imminent.”
The new study was published in the journal Nature Medicine.
Source: Harvard Medical School
