A transcriptome-based risk model in sepsis enables prognostic prediction and drug repositioning

Septic management presented a tremendous challenge due to heterogeneous host responses. We aimed to develop a risk model for early septic stratification based on transcriptomic signature. Here, we combined genes OLAH, LY96, HPGD, and ABLIM1 into a prognostic risk score model, which demonstrated exceptional performance in septic diagnosis (AUC = 0.99–1.00) and prognosis (AUC = 0.61–0.70), outperforming that of Mars and SRS endotypes. Also, the model unveiled immunosuppressive status in high-risk patients and a poor response to hydrocortisone in low-risk individuals. Single-cell transcriptome analysis further elucidated expression patterns and effects of the four genes across immune cell types, illustrating integrated host responses reflected by this model. Upon distinct transcriptional profiles of risk subgroups, we identified fenretinide and meloxicam as therapeutic agents, which significantly improved survival in septic mice models. Our study introduced a risk model that optimized risk stratification and drug repurposing of sepsis, thereby offering a comprehensive management approach. [Display omitted] •A risk model for septic prognosis was construed by LY96, HPGD, OLAH, and ABLIM1•The high-risk subgroup was associated with an immunosuppressive profile•The low-risk subgroup had a poor response to hydrocortisone•The model screened fenretinide and meloxicam for septic treatment Health sciences; Intensive care medicine; Internal medicine; Medical microbiology; Medical specialty; Medicine; Pharmacology