By Laura Stefanik, MSc
Imagine if data from a brain image could be used to help guide treatment recommendations for people with a mental illness? And imagine if we could use that same image in combination with clinical and performance data to reliably predict who might experience difficulties in their social environment?
Right now we diagnose autism spectrum (ASD), schizophrenia spectrum (SSD) and bipolar disorder (BD) separately based on specific guidelines and a person’s individual experiences and behaviors. However, there is considerable overlap in clinical signs and symptoms across these disorders that make the diagnostic boundaries between them somewhat blurred.
Deficits in social communication, interaction and functioning are amongst these shared symptoms and are particularly problematic as they show little to no improvements with existing treatments in psychiatry. To further complicate the picture, there is a wide range of social performance and impairment within and across ASD, SSD, and BD and a lack of consensus on the brain wiring responsible for driving these deficits.
This overlap and variability raises the importance of adopting new techniques (outside the diagnostic approach) that focus on specific symptoms across people with a variety of diagnoses in order to understand what biological similarities and differences these individuals may have. If that question can be answered, then the potential to accelerate discovery and improve treatment success could be significant.
My MSc research at the Centre for Addiction and Mental Health (CAMH) through the Institute of Medical Science sought to answer this question. In late 2017 we published the first study using a data-driven grouping approach across disorders to better understand the brain mechanisms responsible for social performance (Neuropsychopharmacology). This study involved collecting and combining clinical, social cognitive, neurocognitive, and brain imaging data for 174 adolescents and young adults with ASD, SSD, BD, or no diagnosis using a computer algorithm created by our collaborators at SickKids called Similarity Network Fusion (SNF). This algorithm is powerful in that it finds new groups of people based on a comprehensive integration of all data types in order to identify subtypes of a given biological process or disease.
We discovered four specific groups of participants, each with distinct profiles of performance and brain wiring that were representative of real-word social functioning. All data-driven groups consisted of at least 3 diagnostic groups and showed considerably greater ability to differentiate deficits in brain wiring compared to diagnostic groups on their own and groups that were solely based on social performance. This was particularly interesting for groups that had similar levels of social difficulties but completely different underlying brain circuit impairments.
Our work addresses a gap in brain mapping and treatment innovation for adolescents and young adults with mental illness. It also highlights the importance of integrating data types and points to the potential that treatments targeting specific symptoms may help youth with a number of different diagnoses. Moreover, if researchers are able to pin down brain-based markers that reliably indicate or predict brain disorders (see some of our other work in Biological Psychiatry in March 2018 and December 2016), this could open up new opportunities for treatment and prevention.