Identifying Predictive Biomarkers for Sepsis in Trauma-Induced Systemic Inflammatory Response Syndrome Using Proteomics Data.

INTRODUCTION

Despite advances in surveillance, protocolized resuscitation, and critical care organ support, sepsis remains a common and fatal complication in trauma. Trauma-induced systemic inflammatory response syndrome (SIRS) can mask the occurrence of sepsis, complicating the diagnosis of sepsis. Thus, it is significant to identify high-risk patients of sepsis among trauma-induced SIRS for early prevention, early diagnosis, and early intervention of sepsis. This study seeks to identify predictive biomarkers for sepsis in patients with trauma-induced SIRS.

MATERIALS AND

METHODS

The existing plasma proteomics data from 62 severe trauma patients with (38 cases) and without (24 cases) SIRS were retrospectively analyzed. Among the 38 patients with SIRS, 15 SIRS patients who subsequently developed sepsis (SDS) were assigned to SDS group, while the remaining 23 SIRS patients who did not develop sepsis (SDDS) constituted SDDS group. By comparing the SDS and SDDS groups, predictive biomarkers for sepsis in trauma-induced SIRS were identified.

RESULTS

Fifteen differentially expressed proteins were identified between SDS and SDDS groups. We screened top 5 differentially expressed proteins as predictive biomarkers for sepsis in trauma-induced SIRS, including ANXA6 (area under the curve [AUC] = 0.881), FA12 (AUC = 0.873), TRY2 (AUC = 0.868), FIBG (AUC = 0.759), and DOPD (AUC = 0.745). The combined predictive performance of these biomarkers is better (AUC = 0.9913). To balance robust predictive performance with a minimal panel size, we developed a panel consisting of ANXA6 and FIBG, which demonstrated superior predictive performance (AUC = 0.9652).

CONCLUSIONS

This study identified predictive biomarkers for sepsis in trauma-induced SIRS. Notably, integrating coagulation and inflammatory markers enabled substantial reduction in the number of required modeling indicators while preserving superior diagnostic performance. These findings offer new insights into developing early predictive panels.