Machine Learning Maps Sepsis Immune Response to Revolutionize Patient Care

Researchers have mapped how the immune system changes over time in critically ill adults with sepsis, according to a study published in the journal Immunity. The work, led by King’s College London, aims to pave the way for targeted treatments that address specific parts of the immune system as the condition progresses.

Sepsis remains a life-threatening medical emergency triggered by a misfiring immune response to an infection. According to global health data cited in the research, the condition accounts for more than 160 million cases and roughly 21 million deaths worldwide each year. While standard medical care currently relies on antimicrobials to clear underlying infections alongside supportive therapies for failing organs, attempts to successfully regulate the misfiring immune system have largely fallen short.

Mapping the Immune Trajectory in Critical Care

To build a detailed picture of the body’s defenses during severe illness, investigators analyzed serial blood samples collected from critically ill patients at Guy’s and St Thomas’ NHS Foundation Trust. Samples were drawn across four distinct timepoints, spanning from admission to critical care through to discharge.

Researchers evaluated multiple layers of biological information within the blood, including changes in immune cells, gene expression, and protein expression. By applying machine learning approaches, the team combined these distinct data layers to generate a comprehensive immune profile for sepsis patients.

The analysis uncovered an underlying immune trajectory consisting of three distinct temporal immune states, termed STImS, as patients move from admission toward recovery. Each state features different immune cell activity and specific immune response programs.

Disconnect Between Clinical Stages and Immune States

A key finding of the study is that these identified sepsis immune states do not align with standard clinical stages. For instance, the earliest defined immune state, designated as STImS1, does not necessarily correspond to the early clinical stage of sepsis typically marked by a clinician’s initial diagnosis.

When people are admitted to hospital with sepsis, they are usually classed as having ‘early’ sepsis – but our findings show that this isn’t necessarily the case – their immune system may already be at stages of the immune response, said Dr. Matthew Fish, former PhD student at King’s College London and lead author of the paper. Fish carried out the research in the laboratory of Professor Manu Shankar-Hari under a National Institute of Academic Anesthesia BJA/RCoA fellowship, and now works as a postdoctoral researcher at Massachusetts General Hospital.

Senior author Professor Manu Shankar-Hari, Professor of Critical Care Medicine at King’s College London and honorary consultant at Guy’s and St Thomas’ NHS Foundation Trust, emphasized the limitations of traditional observational approaches. This research shows the importance of looking at the changing architecture of the immune system in sepsis over time, rather than just taking a snapshot view, Shankar-Hari stated in the findings.

Next Steps for Sepsis Therapeutics

The study brings together multidisciplinary teams, including Dr. Chad Swanson, Reader in the School of Immunology & Microbial Sciences from King’s College London and Professor Mervyn Singer, Professor of Intensive Care Medicine of University College London. The authors note that understanding these dynamic shifts in immune architecture could fundamentally alter how clinicians approach patient care, helping determine which therapies to administer and at what precise moment.

Future research efforts will focus on investigating what drives alterations in the immune system from the initial onset of an infection to the full development of sepsis. Pinpointing these upstream triggers could help identify novel treatment targets designed to halt disease progression or lessen illness severity for millions of patients globally.