Scientists have a new tool in the race to improve the diagnosis and prognosis of sepsis

Sepsis is a serious condition in which the body fails to respond properly to an infection. This can progress to septic shock, which can affect the lungs, kidneys, liver and other organs. When the injuries are severe, they can lead to death, with an estimated 11 million sepsis-related deaths worldwide each year. A new combined approach using a blood test and artificial intelligence (AI) could catch the condition earlier and save lives, according to experts who combined the unique molecular signature of sepsis with AI tools to estimate the risk of organ failure and death of a person. The findings of the new research will be presented at the European Congress of Clinical Microbiology and Infectious Diseases in Barcelona next month.

Given the challenges associated with timely diagnosis and the fact that sepsis kills millions of people worldwide each year, there is an urgent demand for alternative approaches.

Now, Swedish researchers have identified unique molecular signatures of sepsis and used artificial intelligence (AI) to improve diagnosis and identify patients most likely to develop severe symptoms and suffer poor outcomes.

A group of researchers from Lund University in Sweden has identified distinct molecular signatures associated with clinical signs of sepsis that could provide a more accurate diagnosis and prognosis, and help target specific therapies to patients who would benefit the most. more, according to new research that will be presented at this year’s European Congress of Clinical Microbiology and Infectious Diseases (ECCMID 2024), which will take place in Barcelona, ​​Spain, between April 27-30.

“A simple blood test, when combined with a personalized risk model, has the potential to save lives by providing a more accurate diagnosis of sepsis and determining who is likely to develop more severe clinical manifestations,” says co-senior author Dr. Lisa Mellhammar of Lund University, in a statement.

According to the researcher, it is vital that patients with suspected sepsis are identified before organ failure occurs.

Sepsis is a life-threatening organ dysfunction triggered by a severe infection. Severe sepsis and septic shock are progressive stages of sepsis associated with multiorgan failure and death.

According to the WHO, in 2017 there were an estimated 49 million cases of sepsis and 11 million potentially preventable deaths – accounting for almost 20% of all deaths worldwide [1].

One of the persistent challenges in treating sepsis is the lack of timely diagnosis, as there is no single diagnostic test that reliably detects the presence of generalized infection.

Current practices rely on broad-spectrum biomarkers such as PCR (C-reactive protein, a marker of inflammation), PCT (procalcitonin, a pro-hormone) and lactate to detect sepsis.

In addition, sepsis is a highly variable condition that can arise from a multitude of causes and, despite hundreds of clinical trials, there are no specific treatments and physicians currently rely on the use of a broad spectrum of antibiotic, antiviral, and antifungal.

“It is difficult to predict who will develop sepsis, who will recover and who will have poor outcomes,” says co-lead author Dr. Adam Linder of Lund University.

“We urgently need better ways to understand sepsis at the molecular level so that we can classify patients with suspected sepsis based on the clinical manifestations of their disease and identify high-risk patients and develop more effective treatments,” he says. .

In the new study, the researchers aimed to investigate the distinct proteomic signatures (unique patterns of proteins that are associated with the immune response in sepsis patients) associated with different clinical symptoms and outcomes, such as different organ dysfunctions and infections.

They included 1,364 plasma samples from randomly selected adult patients with suspected sepsis admitted to the emergency department of Skåne University Hospital between 1 September 2016 and 31 March 2023. In total, 1,073/1,364 patients had an infection , and of these 913 had sepsis.

The researchers used mass spectrometry to analyze the plasma samples and generate comprehensive molecular maps to better understand protein patterns that were predictive of septic shock.

Each group of proteins was then combined into a molecular signature to train a machine learning model, allowing the researchers to predict which patients would develop septic shock with high accuracy.

Patients were then classified according to low, medium or high likelihood of developing septic shock, and the model was able to show how increased risk was associated with higher mortality.

The researchers also identified clusters of proteins that were predictive of six different types of organ dysfunction (cardio, central nervous system, coagulation, liver, kidney and respiratory) and infection. They analyzed the biological processes associated with each group to show how their unique proteomic signatures influence sepsis.

Patients were then classified into five risk categories based on the likelihood of organ dysfunction and infection and the risk of death.

A rapid test that would provide a more accurate diagnosis of sepsis and could also estimate who is at greater risk of poorer outcomes now seems a real possibility, Swedish researchers say.

Any research of this kind needs clinical validation, and many hurdles must be overcome before these biomarkers can be used in the clinic.

“We envision this tool, which could be implemented worldwide, as the future of early detection of sepsis,” the authors state.

Despite the promising results, they point out some limitations of the study, including that because sepsis is a highly variable condition, it is important that the results are validated in diverse settings and cohorts.

Furthermore, because sepsis is a dynamic syndrome that progresses over the course of the disease, it is necessary to collect repeated samples from patients with sepsis to study how the progression from earlier to later states affects the proteome. The authors also note that these data could benefit from further analysis using, for example, transcriptomics.

[1] Sepsis, World Health Organization (WHO)