Dr. Patricia Silveira

By Josh Kaiser 

For many years, scientists could only speculate that neuropsychiatric disorders such as attention deficit hyperactivity disorder, addiction, or schizophrenia were related to our genetics. A 2009 study showed that schizophrenia and bipolar disorder patients had hundreds of common genetic variants that each increased the risk of disease by a small amount. Over the past decade, these findings greatly influenced scientists as they performed similar whole genome studies on individuals with and without neuropsychiatric disorders, searching for significant common genetic variants. But what do the these variants tell us about the biological basis of these disorders? The answer is far from straightforward.

Now, researchers at McGill University’s Faculty of Medicine and the Douglas Hospital Research Centre, have developed a new method of examining the mechanisms associated with neuropsychiatric disorders. And it all starts with gene networks. Their findings were published recently in the journal EBioMedicine.

The current method of genome-wide association studies compares a pool of diagnosed and undiagnosed individuals in order to identify the most statistically significant gene variants (single-nucleotide polymorphisms) linked to the disease. “This is similar to trying to decipher a recipe to a cake by slicing the topmost layer and taste-testing it,” explains Shantala A. Hari Dass, lead author of the study and a postdoctoral fellow collaborating with Dr. Patricia Silveira, Assistant Professor in the Department of Psychiatry at McGill. “We tried to deconstruct the recipe by cutting the cake in “biologically coherent” slices and consider the ingredients independent of their position at the top or at the bottom of the slice.”

For this initial study, Dr. Silveira and her team decided to base their genetic variant search on a list of genes that interacted with the insulin receptor, as they are known to influence the development of neuropsychiatric traits like cognition and impulsivity. By then aggregating the common genetic variants in the insulin receptor gene network, the team was able to begin their research with a completely new approach.

To the researchers, though, it was important to not only start with a gene network, but also to be able to prove that their method can be applied to different brain regions. “Our study proposes the use of these networks of co-expressed genes as the relevant biological unit of influence for the development of specific characteristics or risk for disease later in life,” notes Dr. Silveira, senior author on the paper. “As genes can act very differently in diverse parts of the body, a gene network must be defined within a specific tissue.”

Because of this, the research was conducted in a brain region associated with addiction (the mesocorticolimbic region) and another region linked to dementia and Alzheimer’s (the hippocampus). In each tissue site the researchers used common genetic variants to create a score that captures physiological variation associated with the insulin receptor gene network.

After developing genetic scores in each region, the group analyzed whether the biologically-focused score could predict addiction or Alzheimer’s disease. For each corresponding brain site, the biologically-informed risk score was informative about the neuropsychiatric disorders diagnosis. Strikingly, the researchers showed that it could predict traits related to the adult diagnoses states in children, and with more sensitivity than the statistically driven genetic scores. This is important, as it highlights the mechanisms behind the development of these diseases can be in place even years before the conditions are established, with important consequences for a better understanding of risk factors and the development of preventive measures.

About the study

A biologically-informed polygenic score identifies endophenotypes and clinical conditions associated with the insulin receptor function on specific brain regions,” by Shantala Hari Dass, Patricia Silveira, et al, was published in the journal EBioMedicine online on March 26, 2019. https://doi.org/10.1016/j.ebiom.2019.03.051

April 18, 2019