SDF4 Is a Prognostic Factor for 28-Days Mortality in Patients With Sepsis via Negatively Regulating ER Stress

Sepsis is a heterogeneous syndrome induced by infection and results in high mortality. Even though more than 100 biomarkers for sepsis prognosis were evaluated, prediction of patient outcomes in sepsis continues to be driven by clinical signs because of unsatisfactory specificity and sensitivity of these biomarkers. This study aimed to elucidate the key candidate genes involved in sepsis response and explore their downstream effects based on weighted gene co-expression network analysis (WGCNA). The dataset GSE63042 with sepsis outcome information was obtained from the Gene Expression Omnibus (GEO) database and then consensus WGCNA was conducted. We identified the hub gene SDF4 (stromal cell derived factor 4) from the M6 module, which was significantly associated with mortality. Subsequently, two datasets (GSE54514 and E-MTAB-4421) and cohort validation (n=89) were performed. Logistic regression analysis was used to build a prediction model and the combined score resulting in a satisfactory prognosis value (area under the ROC curve=0.908). The model was subsequently tested by another sepsis cohort (n=70, ROC= 0.925). We next demonstrated that endoplasmic reticulum (ER) stress tended to be more severe in patients PBMCs with negative outcomes compared to those with positive outcomes and SDF4 was related to this phenomenon. In addition, our results indicated that adenovirus-mediated Sdf4 overexpression attenuated ER stress in cecal ligation and puncture (CLP) mice lung. In summary, our study indicates that incorporation of SDF4 can improve clinical parameters predictive value for the prognosis of sepsis, and decreased expression levels of SDF4 contributes to excessive ER stress, which is associated with worsened outcomes, whereas overexpression of SDF4 attenuated such activation.

Automated summary

This study identified the key candidate gene SDF4 through weighted gene co-expression network analysis (WGCNA) and built a prediction model for sepsis prognosis. Validation in multiple datasets showed the model's effectiveness, while also revealing the association between SDF4 and ER stress in patients with sepsis.