Computational Method Uses Gene Expression to Judge Patient Sepsis Risk

A computational method that analyzes gene expression in blood to stratify patients with acute infections and predict their clinical outcomes is reported in Science Translational Medicine this week. Dysregulated host immune responses to infection can result in sepsis, which triggers severe organ dysfunction and accounts for around 11 million deaths globally each year. As a result, there is an urgent need for effective ways to identify patients with dysfunctional immune profiles that are amenable to intervention. To that end, a group led by University of Oxford scientists used whole-blood transcriptomics data derived from patients with sepsis and healthy individuals to create a multi-gene-based quantitative sepsis response signature score that reflects immune dysfunction and the trajectory of a patient's infection. The researchers then integrated the score into a machine learning framework, dubbed SepstratifieR, which they show can identify dysfunctional immune profiles and predict outcomes in patients with bacterial and viral sepsis, influenza, and COVID-19. "In combination with clinical biomarkers, SepstratifieR could improve risk estimation of immune dysfunction and clinical outcomes, as well as inform clinical trial design, bringing us closer to precision medicine for severe infection," they write.