Abstract
Sepsis, a pressing global health challenge, is characterized by immune dysregulation and high mortality. This study, in response to this urgent need, integrated Single-cell RNA sequencing data from 53 samples and bulk transcriptomic data from 479 sepsis patients to analyze immune microenvironment heterogeneity and develop a prognostic signature. Using the scVI algorithm, eight immune cell types were identified. B, T, and NK cells were reduced in sepsis, while neutrophils were elevated. "BayesPrism" analysis linked higher macrophage abundance to improved survival. A pro-inflammatory FCGR3A + macrophage (Macro_1_FCGR3A) subset exhibited high M1 scores, enhanced antigen presentation, and activation of the PI3K-AKT-mTOR and TNF-NFκB pathways. Pseudotime analysis showed terminal differentiation with elevated HLA-DRB1 and CXCL10. A prognostic model using "StepCox + Ridge" selected 13 key genes, including CX3CR1, achieving AUCs (Area Under the Curve) of 0.77 (training) and 0.69 (validation). CX3CR1 showed strong diagnostic potential (AUC = 0.982). These findings reveal the pivotal role of FCGR3A + macrophages in sepsis prognosis.