Grant: Project Grant

Sepsis, systemic organ dysfunction triggered by a dysregulated host response to infection, incurs enormous human and health care burden. Globally it is estimated to affect more than 30 million people each year, ~40% of whom die. The sequelae include physical, cognitive, and psychological disability; profoundly impacting quality of life of survivors. The biggest problem is that we have no specific treatments. Patients receive fluid resuscitation, antibiotics and supportive care, but an under appreciation for patient heterogeneity and inadequate characterization of the complexity of the host response to infection has made it difficult to design clinical trials that show benefit. The issue is that, despite development of robust therapeutic candidates, we lack the tools to accurately categorize sepsis (prognostic enrichment) and identify those patients that are more likely to respond to therapy (predictive enrichment). In the cancer literature, personalizing treatment on the basis of the molecular characteristics of a tumor is simply 'good practice'- there is an urgent need to do the same for septic patients. Here we propose to use an approach that has never been used for sepsis before; to identify DNA methylation changes that classify septic vs non-septic critically ill patients (often it is very difficult to differentiate these patients on the basis of clinical tools alone) and determine the association between methylation changes with time-to-death and/or response to therapy. We will capitalize on banked DNA samples collected in the context of RCTs where patient heterogeneity was minimized through randomization and bias was reduced by blinding. We will use predictive modeling to integrate clinical and molecular data to identify a set of informative methylation marks - and validate their ability to risk stratify patients using independent cohorts. Successful completion of this project is expected to significantly advance care.