MONTREAL, November 17, 2020 – While some diseases have known root causes, many are characterized by multiple abnormal interactions whose specific origin is not known.

Recently, a research team from CHU Sainte-Justine, Harvard University, Stanford University and the University of California, San Francisco used computational biology to analyze a large volume of data and deconstruct and reconstruct the networks of interactions leading to a disease, thus identifying new phenotypes that can guide clinicians toward optimized therapeutic strategies for each patient.

The results of the study are presented in the journal Patterns published by Cell Press.

A "toolbox" for diagnostic assistance

Complex disorders, due to their multifactorial etiology, require new tools capable of leveraging large quantitative data sets to sketch out the equivalent of a "food web" in ecology, i.e. a network of interactions between various cells and proteins.

Constructing this network allows us to study the differences between healthy and sick individuals and to identify, on one hand, the phenotypes at work in the disease and, on the other, potential therapeutic strategies.

"Here, we applied these quantitative approaches to cyclic thrombocytopenia (CT), a very rare and complex blood disorder," explains Morgan Craig, a researcher at CHU Sainte-Justine and professor at Université de Montréal.

"Our study consists of three subjects with biomarkers for CT, in which multiple cells and proteins undergo abnormal oscillations, leading to a deficient platelet concentration in the blood, which corresponds to the clinical picture of the disease," she continues.

"However, one of the subjects had atypical biomarkers.”

"We then applied a collection of mathematical techniques (empirical dynamics), which, based on longitudinal data, allowed us to dissect the dynamics of the abnormal physiological interactions in the disease in each patient," says Madison Ski Krieger, first author of the study.

The development of this diagnostic toolbox has helped us validate a plausible therapeutic intervention. "

"What's really new here is that our approach allows us to distinguish the multiple phenotypes of the disease by examining its pathophysiological mechanisms, regardless of whether they have the same clinical manifestation," adds Professor Craig.

"In other words, although the symptoms of a disease may be the same, the mechanisms responsible for that disease may differ."

This insight would explain why some drugs do not work in certain patients.

Looking ahead

"What's exciting is that this quantitative approach has the potential to be applied to multiple complex disorders and populations to improve our understanding of a given disease and guide clinicians toward personalized therapeutic targets," concludes Professor Craig.

The team is continuing its research in order to study drug resistance in cancer and to better understand the mechanisms underlying COVID-19. 

About the study

The article "A blueprint for identifying phenotypes and drug targets in complex disorders with empirical dynamics" was published in the journal Patterns by Cell Press in November 2020. The first author is Madison S. Krieger, postdoctoral fellow in the Department of Organic and Evolutionary Biology at Harvard University, where he was co-supervised by Morgan Craig. The principal author is Morgan Craig, PhD, researcher at CHU Sainte-Justine and professor in the Department of Mathematics and Statistics at Université de Montréal. The study was funded by a Discovery Grant and a Discovery Accelerator Supplement from the Natural Sciences and Engineering Research Council of Canada (NSERC), an NSERC Postdoctoral Fellowship), a grant from the National Institutes of Health (NIH) and a Long-Term Research Fellowship from the Human Frontiers Science Program.