Decoding sepsis and the secrets of the immune system

Sepsis remains one of the most devastating and elusive conditions in modern medicine. It begins with an infection – even from something as ordinary as an infected cut, yet within hours, it can trigger a dysregulated immune response that leads to organ failure and death. What makes sepsis especially dangerous is how easily it hides in plain sight – there is no single test to confirm it, and often by the time it is recognised, it is too late.

Despite decades of research and clinical trials aimed at controlling the immune response, results have been inconsistent, reflecting the reality that not all patients labelled with sepsis are experiencing the same disease. At the Wellcome Sanger Institute, Dr Emma Davenport’s group aims to uncover what is really happening beneath the surface. Emma and her team are decoding the molecular signals in the blood to understand each person’s unique immune response, and ultimately, find ways to identify and treat sepsis before it becomes fatal.

The word sepsis is derived from the Greek word σήψη meaning ‘putrefaction’ and was first recorded in a medical sense by Homer over 2,500 years ago.¹ Since this time, our understanding of the causes and features of sepsis has significantly evolved. Sepsis is considered a response to infection that causes organ dysfunction due to a dysregulated immune response.¹ The definitions of sepsis have changed over the years, but there is still not a single test that is able to determine whether someone absolutely has or does not have sepsis.

It was estimated in 2020 that there were 48.9 million cases and 11 million sepsis-related deaths worldwide – accounting for 20 per cent of deaths globally.² Almost half of these cases were found in children under five years of age.² Sepsis affects people worldwide, but its burden is significantly higher in low- and middle-income countries.² Alongside its impact on mortality and morbidity, sepsis also has a significant economic burden. In the UK, estimated costs of sepsis each year are £7.76 billion.³

Decades of clinical trials have attempted to understand sepsis by trying to dampen the immune response. However, using the same drugs in different cohorts has generated inconsistent results. For some patients these drugs were beneficial, for others they seemed not to do anything and in others they even seemed to be harmful. This is because sepsis is not one disease but many overlapping biological states and each patient can be very different.

At the Sanger Institute, Dr Katie Burnham, Senior Staff Scientist, and her colleagues in Dr Emma Davenport’s group have had a long-standing collaboration with Professor Julian Knight’s group at the University of Oxford as part of the UK Critical Care Genomics Group. Together they are attempting to uncover the molecular mechanisms behind sepsis and identify opportunities for a precision medicine approach to manage it.

The immune system is very strongly regulated through gene expression. During an infection and the development of sepsis, large-scale changes in gene expression are triggered – thousands of genes alter their activity by turning on or off. However, given the complexity of the immune system, numerous immune pathways might be altered in any given individual. Knowing what type of immune dysregulation a person has can enable individuals to be separated out into different subgroups. This in turn can enable more targeted treatment to be delivered faster.⁴

As part of their work, Katie and her team have been using whole blood transcriptomics – an approach that measures gene expression in blood – to capture a snapshot of a person’s immune system and the molecular mechanisms that are driving this. In 2016, the team published a paper outlining two sepsis response signature (SRS) subgroups in the UK Genomic Advances in Sepsis (GAinS) study.⁵ Using white blood cells of adult patients admitted to intensive care with sepsis, they performed genome-wide gene expression profiling and identified two distinct SRS subgroups.⁵ The SRS1 subgroup was more immunosuppressed and was associated with higher mortality, while the SRS2 group was more immunocompetent and was likely to have better outcomes.⁵ Importantly, SRS status was found to be associated with different responses to steroid treatment in the VANISH (VAsopressin versus Noradrenaline as Initial therapy in Septic sHock) clinical trial.

The team are now interested in identifying appropriate treatments for each of the subgroups but to do this they need a better understanding of the molecular mechanisms driving the differences between the groups. In one study, they used single cell RNA-sequencing on whole blood from sepsis patients to show that the dysregulated SRS1 response was strongly driven by immature neutrophil subsets.⁷ This revealed key insights into what was going wrong in the immune system and pointed to possible targets for treatment. These immature neutrophils dampen normal immune responses, and their increased production is linked to worse outcomes in SRS1 patients, suggesting that they may play a key role in driving harmful immune dysfunction. Additionally, in a different paper, it was demonstrated that a patient’s genetic variation can affect their immune gene expression response to sepsis.⁸ The team is now working on integrating additional omic datasets, such as proteomics (study of all proteins), together with clinical data to further refine the subgroups and understand potential drivers for future drug targets.⁹

People presenting to intensive care with sepsis are typically very unwell and therefore, in order to advance to personalised medicine, we would need a test that could turn results around very quickly. To make patient stratification more accessible, Katie alongside Dr Kiki Cano-Gamez and their team at the University of Oxford developed a machine learning framework called SepstratifieR.¹⁰ Starting with genome-wide gene expression data, the team used machine learning methods to go from all 20,000 genes in the genome to approximately 10 genes that could help categorise patients into these SRS subgroups.

The idea is that a blood sample would be taken and added into a device with pre-loaded probes for these 10 genes. The returned gene expression measurements would then be aligned to their existing reference dataset and the individual could be categorised into one of the subgroups. In 2023, the University of Oxford partnered with the life sciences firm Danaher to build on the team’s work in this space. The aim is to develop a test that would pinpoint the different subgroups and enable the development of novel personalised care pathways.¹¹

Although these developments are still early, the team is interested to see how these sorts of tools will be used alongside clinician judgement as we gain more information. Hopefully, with advanced technologies, the team and their collaborators will be able to sample and measure more densely with real-time analysis to reveal how the immune response dynamically evolves over the course of an infection.

Moving forward, Katie and her colleagues hope to do more work monitoring patients over time. They have several datasets with samples collected over multiple days, which suggest that different immune response trajectories may be associated with different outcomes. In some cases, individuals show more stable immune profiles over time, while in others the immune response appears to evolve, with these differing patterns potentially relating to prognosis. While these are early results, the team thinks that these insights could inform clinical trial design, emphasising the need to take gene expression measurements before and after treatment to understand if the treatment is modulating the response in the way you would expect.

Work to date from the team has focussed on adult patients admitted to intensive care with sepsis. Having identified these subgroups, the team is now exploring their applicability in other settings. For example, the Bioresource for Adult Infectious Disease (BioAID) has been used to validate the findings in less severely ill patients with infection at presentation to the emergency department. Going forward, the team will be assessing whether SRS subgroups are relevant in other critical care syndromes, such as traumatic brain injury (TBI) through a collaboration with TBI Reporter. Finally, a collaboration with the DIAMONDS* (Diagnosis and Management of Febrile Illness using RNA Personalised Molecular Signature Diagnosis) consortium is providing a very exciting new opportunity to assess the utility of the SRS subgroups across the age spectrum from newborns through to the elderly. Emma’s group will be generating whole genome sequencing and transcriptomic data for thousands of patients, which will be integrated with detailed clinical information performed by the DIAMONDS researchers.

Sepsis moves fast, often becoming deadly before it is even recognised. By decoding the body’s molecular response to infection, we are starting to understand why it affects people so differently – insights that hopefully in the future will be integrated into clinical settings, helping healthcare professionals identify risk earlier and guide the right treatment to the right patient more quickly.