By Sarah Anderson, PhD
In a featured article published today in Frontiers in Science, researchers from leading Canadian institutions describe how a bioinformatic systems immunology approach can transform the diagnosis and treatment of sepsis, signifying a major step toward greater pandemic preparedness.
Sepsis, a dysfunctional immune response to infection that results in multi-organ failure, is a widely underappreciated health crisis. It is the ultimate cause of almost 20 percent of deaths worldwide, including many formally attributed to the initial bacterial or viral trigger.
“Sepsis was the cause of death from the COVID-19 pandemic and will be the cause of death in the next pandemic,” said article lead author Robert Hancock, a professor at the Department of Microbiology and Immunology at the University of British Columbia. “If we're going to guard our present and our future, we really need to start to dig into this disease in great detail.”
Sepsis presents with non-specific early symptoms that evolve differently between individual patients and over time, making it difficult to understand and manage. To tease apart this heterogeneous disease, researchers have set out to collect vast amounts of data on information pathways within patients’ cells, from the genetic blueprint to protein production to small molecule signaling. Hancock and coauthors explain how systems immunology, a set of computational methods leveraging artificial intelligence, can process this data to track the biological signature that sepsis leaves on the body.
The authors used machine learning to analyze gene expression data in people with sepsis compared to a healthy cohort and found a set of six genes that could classify sepsis status with about 80 percent accuracy. As mortality rates skyrocket alongside delay in treatment, this genetic “Sepset” could provide a critical tool for early diagnosis and intervention.
Systems immunology also has the potential to take sepsis treatment from one-size-fits-all to personalized medicine. Researchers have used clustering algorithms to group the gene expression profiles of people with sepsis into five distinct categories known as endotypes. “Each endotype has a different underlying mechanism driving their sepsis, and this means that now you can start to think about using different therapies for each of those groups of patients,” Hancock said.
Other applications of systems immunology could enable temporal precision in sepsis treatment. RNA sequencing data analysis has revealed unique gene expression patterns corresponding to various phases of severe COVID-19 infections. Identifying these trends in sepsis could allow clinicians to follow a patient’s progression throughout the trajectory of the disease and tailor therapeutics accordingly.
The authors emphasize that computational methods alone can’t deliver definitive information, but rather provide data-driven hypotheses that require experimental validation. The more data that feeds into making these predictions, the more likely they are to be replicated at the lab bench.
That’s why Hancock urges for greater support for building the large-scale biological databases that sepsis systems immunology research depends on. “We need to have more concerted efforts by all of the stakeholders — whether it's government, funding, and professional agencies or researchers and the public — because only by collecting more data will we come up with better solutions,” he said.
Coauthors of the article are Andy An (Department of Microbiology and Immunology, University of British Columbia), Claudia C. dos Santos (Keenan Research Centre for Biomedical Science, Critical Care Medicine, St. Michael Hospital, University of Toronto) and Amy H. Y. Lee (Department of Molecular Biology and Biochemistry, Simon Fraser University). Research, authorship, and publication of the article was funded by the Canadian Institutes for Health Research.