Genetic model predicts how immune system responds to sepsis and other infections

Genetic signatures have been used to predict how a patient's immune system might respond to sepsis, COVID-19, and influenza.

Sepsis is complicated, caused by the interaction of one or more infectious agents (eg bacteria or viruses) with an immune reaction involving multiple genes and their products. This led scientists at the Wellcome Sanger Institute in Cambridgeshire, in collaboration with the University of Oxford, Queen Mary University in London, and Imperial College London, to map the genetic landscape of patients with sepsis, which they believe is crucial to better understand the disease.

Dr Emma Davenport of the Wellcome Sanger Institute, one of the chief scientists involved in the study, said, 'Sepsis has long seemed an intractable problem because we simply didn't understand the disease as well as we needed to. Similarly, the early stages of the COVID-19 pandemic highlighted the stress doctors were under, trying to treat patients without having solid information to help them identify those most at risk. Our model provides a level of detail that finally allows us to start applying precision medicine techniques to sepsis and improve outcomes for patients.'

Sepsis is a condition in which the immune system responds to an infection in the body in an uncontrolled way, attacking healthy tissue. Sepsis is an acute, fast-moving condition, and kills around 11 million people worldwide every year. The causes underlying this condition are not well understood – sepsis is generally treated with antibiotics, however, as the condition is highly heterogeneous, patients respond differently to different treatments.

The scientists looked at genome sequences from 1655 sepsis patients and identified 19 genes, the expression profiles of which can be used to gauge immune system responses and clinical prognosis. They hope that using this panel of genes will aid in the administration of precision medicine, which could potentially target specific pathways and molecules based on a patient's individual genetic profile.

Publishing their results in Science Translational Medicine, the scientists used whole-blood transcriptomics, a method which evaluates the expression levels of genes in blood samples, and compares the results with clinical data. A quantitative signature scoring system was developed, using either a 7-gene or a 12-gene signature and a machine-learning framework was developed using this scoring system, trained on data from healthy individuals and patients with sepsis. This framework can predict the level of immune dysfunction, effectively separating patients according to their risk of severe infection.

They also applied this framework on SARS-CoV-2, the virus that causes COVID-19, and H1N1 influenza datasets, with promising results – high scores obtained using this system matched the severity of the illnesses, in terms of immune system dysregulation, and whether or not hospitalisation and intensive care were required.

The authors aim to use this model in clinical trials, to serve as a screening system for the inclusion of patients in trials for specific drugs, bringing precision medicine for sepsis one step closer to reality.