This study introduces, for the first time, a non-invasive, accurate, and feasible model for predicting sepsis prognosis based on urinary protein markers. Traditional clinical diagnosis and prognosis assessment of sepsis necessitate multiple blood draws for marker detection, which not only risks iatrogenic blood loss but also offers limited diagnostic insights^12,13. Utilizing proteomics, we analyzed urine samples from both sepsis survivors and non-survivors, identifying three proteins with biomarker potential from a pool of 255 DEPs. Leveraging these three proteins, we developed a robust prognostic prediction model that demonstrated exceptional performance, with an ROC AUC reaching 0.902.

SLC25A24 plays a crucial role in sepsis by regulating mitochondrial calcium homeostasis, mitochondrial dysfunction, and inflammatory responses. As a mitochondrial calcium transporter, it may influence immune cell activation and NLRP3 inflammasome signaling, which drives excessive cytokine release during sepsis¹⁴. Dysregulation of SLC25A24 is associated with mitochondrial dysfunction, exacerbating oxidative stress and organ damage¹⁵. Additionally, it is highly expressed in metabolically active tissues such as the liver and heart, where it is closely related to cellular energy metabolism and may alleviate sepsis-induced organ damage by regulating mitochondrial function^16,17. Studies suggest that SLC25A24 is associated with the prognosis of sepsis¹⁸. Targeting SLC25A24 could offer a novel therapeutic strategy to counteract sepsis pathogenesis.

CREB3L3 is primarily involved in regulating lipid and glucose metabolism, as well as inflammatory responses, helping to mitigate the progression of sepsis through its effects on both metabolic and immune pathways. It plays a crucial role in balancing lipid and glucose metabolism, ensuring proper cholesterol and blood glucose levels, which in turn enhances immune cell functionality and supports the body in managing the metabolic stress induced by sepsis^19,20. Additionally, CREB3L3 influences the NF-κB signaling pathway, thereby modulating the release of key inflammatory factors like TNF-α and IL-6, which aids in controlling systemic inflammation and minimizing organ damage resulting from unchecked inflammation²¹. Moreover, CREB3L3 is instrumental in maintaining the integrity of the intestinal barrier, thereby reducing the impact of gut-derived infections and toxins on the entire body²².

UBQLN1 plays an essential role in sepsis by preserving protein homeostasis, modulating inflammatory responses, and safeguarding organ function²³. It manages the ubiquitin-proteasome system and autophagy pathways to facilitate the breakdown of misfolded proteins, thereby minimizing cellular damage. Furthermore, UBQLN1 influences the NF-κB signaling pathway, helping to suppress excessive inflammation and mitigate tissue injury²⁴. By enhancing immune cell function and reducing organ damage, UBQLN1 strengthens the body’s ability to combat infections, positioning itself as a promising protective factor and therapeutic target in sepsis management.

IGLV5-45/IGLJ2 and NART contribute significantly to sepsis progression via distinct mechanisms. IGLV5-45/IGLJ2 represents a specific immunoglobulin lambda light chain that enhances the specific recognition and clearance of pathogens in humoral immunity, which is vital for the anti-infective response during sepsis²⁴. In contrast, NART regulates key proteins through ADP-ribosylation, modulating immune response intensity and maintaining inflammation balance, thereby reducing excessive systemic inflammation and minimizing multi-organ damage²⁵. Together, they help maintain immune system balance, thereby alleviating the severity of sepsis and improving patient outcomes.

Sepsis triggers widespread inflammatory responses and immune dysregulation, leading to tissue damage and making organs more susceptible to Inflammation- and immune-mediated injury. This study gathered crucial evidence demonstrating the significant roles of SLC25A24, UBQLN1, and CREB3L3 in the pathogenesis of sepsis. Furthermore, we developed and validated a sepsis prognosis model comprising four biomarkers: SLC25A24, UBQLN1, and CREB3L3. The model exhibited accuracy that was comparable to or better than sepsis outcome models based on routine clinical hematology and biochemical parameters^25,26. Additionally, our study has the following limitations. The first limitation of this study derives from its single-center design and these proteins has not been confirmed by the enzyme-linked immunosorbent assay. In the further research, the confirmation of biomarkers derived from proteomics analyses was enhanced through additional verification by ELISA on a larger group, thereby increasing the reliability of the findings. Secondly, exploring the dynamic features of sepsis and the temporal changes in prognostic biomarkers should be priority in the further study, which will offer a deeper insight into the disease’s progression and the possibilities for preemptive treatment. Thirdly, due to potential bias from our limited sample size, multicenter prospective studies are required to confirm the generalizability and robustness of our results. Lastly, although we have identified three proteins linked to poor prognosis in sepsis, further study is needed to investigate the underlying mechanisms through which these proteins impact disease outcomes. It is particularly important to focus on their specific roles within the immune and pathological systems.