Contribution of prognostic ferroptosis-related subtypes classification and hub genes of sepsis

Background: Sepsis in patients is a great threat to human health due to its high incidence rate, its rapid and unpredictable progression, as well as it is difficult to treat, and it has poor prognosis. Ferroptosis is a newly discovered type of cell death characterized by the iron-dependent peroxide aggregation. Furthermore, ferroptosis is different from other forms of cell death, namely apoptosis, necrosis, pyroptosis and autophagy. Our study investigated the role of ferroptosis-related genes in sepsis.

Methods: The GSE65682 dataset from the Gene Expression Omnibus (GEO) database was used to screen ferroptosis-related genes associated with sepsis, and the GSE134347 dataset for the external validation of selected hub genes. The univariate Cox regression analysis, Kaplan-Meier (K-M) survival analysis and weighted gene co-expression network analysis (WGCNA) were used to identify hub genes. Evaluation of the immune cell infiltration in sepsis was used to explain the immune heterogeneity among the cell subtypes. Gene set variation analysis (GSVA) and transcriptional regulatory analysis of selected hub genes further elucidated the probable mechanism of ferroptosis-related genes associated with prognosis in sepsis. Finally, we constructed a competing endogenous RNA (ceRNA) network model.

Results: A total of 479 RNA-seq data points were used for analysis, including 365 samples from patients who survived sepsis and 114 samples from patients who succumbed to sepsis from the available GSE65682 dataset. Consequently, the univariate Cox regression analysis and consensus clustering analysis divide all 479 sepsis samples into two clusters of "survivals" vs. "non-survivals". Following complex analysis were identified as the most important ferroptosis-related genes. Indeed, the WGCNA and K-M analyses associated the expression patterns of NEDD4L and SIAH2 hub genes as the best prognosis for the survival of sepsis (p < 0.05). The expression trend was also consistent with the survival trend of the NEDD4L and SIAH2 hub genes by the external validation of GSE134347 (p < 0.05). Immune cell infiltration analysis indicated that the types and numbers of different immune cells vary among different subtypes and NEDD4L and SIAH2 hub genes. For example, NEDD4L and SIAH2 gene expression had a positive correlation with M0 macrophages and a negative correlation with neutrophils (p > 0.05). Finally, analysis of two hub genes and transcription factors (TFs) showed that 71 TFs were predicted to be related to NEDD4L while 64 TFs to SIAH2 by the Cistrome DB online database.

Conclusion: We suggest that NEDD4L and SIAH2 hub genes are involved in the ferroptosis-associated sepsis. The pattern of NEDD4L and SIAH2 expression in patients undergoing sepsis may have prognostic potential for the severity of sepsis and eventually for patients' survival.