Sepsis remains a leading cause of ICU mortality, with therapeutic progress limited to infection control and supportive care, as molecularly targeted treatments have failed in clinical trials due to the syndrome’s heterogeneity. Personalised medicine requires accurate biomarkers and mechanistic insights, and mass spectrometry–based plasma proteomics offers a promising approach. While few studies have applied this to sepsis, a landmark 2024 study demonstrated the potential to define molecular subtypes. However, the generalisability of phenotyping studies remains uncertain.
In a recent study, proteomics data from 333 patients across multiple centres were analysed at two early time points in sepsis. Cluster analysis revealed four plasma proteome subtypes and their temporal trajectories. The authors provided detailed molecular and clinical characterisation of these subtypes and developed a machine learning model to classify new patients using a reduced set of features, illustrating a path toward clinical implementation of proteome-guided sepsis stratification. Plasma samples from the patients at days 1 and 4 were analysed.
Plasma proteome analysis revealed four sepsis subtypes with distinct severity and immune profiles. Cluster 0 reflected terminal-stage sepsis with 100% mortality and rapid clinical deterioration, likely representing a stage of progression rather than a true molecular endotype. Clusters 1–3 showed differing immune responses: cluster 1 with strong adaptive immune activation and low albumin, cluster 2 with acute inflammation, neutrophil activation, and lowest Ig levels, and cluster 3 resembling a baseline subtype. Although SOFA scores decreased across clusters 1–3, short-term mortality did not differ, though cluster 1 may be linked to worse long-term outcomes.
A machine learning model using 10 proteins plus Ig levels successfully classified patients into clusters 1–3, with complement, coagulation, and albumin-related proteins among the strongest predictors. Many of these markers are supported by previous evidence linking them to sepsis severity and mortality.
Sepsis plasma proteome subtypes reveal distinct molecular and immune features, offering insights into disease mechanisms and potential therapeutic responses. They may enable predictive patient stratification in clinical trials and, with biomarker panels, could be applied in clinical practice. Overall, these findings represent a step forward toward precision medicine in sepsis management.
Source: Critical Care
Image Credit: iStock
