Identification and Validation of an Endotoxin Tolerance–Based Prognostic Model with Therapeutic Insights in Sepsis

Yu Xie¹Yadong Su²Yin Qin²Qiuhong Zhang³Jie Liu²Yanglin Wu¹Ning Du^4*Yu Jiang^5*Gang Liu^6*

• ¹Department of Respiratory and Critical Medicine, University-Town Hospital of Chongqing Medical University, Chongqing, China
• ²Department of Emergency and Critical Care Medicine, University-Town Hospital of Chongqing Medical University, Chongqing Medical University, Chongqing, China
• ³Department of Emergency and Critical Care Medicine, University-Town Hospital of Chongqing Medical University,, Chongqing Medical University, Chongqing, China
• ⁴Center for Mental Health, University-Town Hospital of Chongqing Medical University, Chongqing, China
• ⁵Department of Respiratory and Critical Medicine, Chongqing Medical University, Chongqing, China
• ⁶Department of Emergency and Critical Care Medicine, THE AFFILIATED YONGCHUAN HOSPITAL OF CHONGQING MEDICAL UNIVERSITY,, Chongqing, China

Background Sepsis outcomes remain difficult to predict because immune trajectories are heterogeneous and can shift from early activation to immunosuppression. Endotoxin tolerance (ET) in circulating monocytes/macrophages is a central mechanism of sepsis-associated immunosuppression but has not been systematically leveraged for prognostication. We aimed to develop and clinically validate an ET-related gene (ETG) signature for short-term mortality risk stratification. Methods Public whole-blood transcriptomic datasets were intersected with curated ET gene sets to derive ETG candidates. An ensemble machine-learning framework screened 108 model/feature-selection combinations across 12 algorithms to build and rank prognostic models for 28-day mortality, yielding a parsimonious 10-gene signature (including IL4R, ATM, CX3CR1, FCGR1A) that generated an ETG risk score. A two-variable nomogram (age + ETG score) was then constructed. Performance was evaluated by bootstrapped calibration, time-dependent ROC/AUC at 7, 14 and 28 days, and Harrell’s C-index. Prospective validation used an ICU cohort (n=50; 13 survivors, 37 non-survivors). PBMCs underwent RT-qPCR and Western blotting; CD14⁺ monocytes were profiled by flow cytometry for FCGR1A/CD64, CX3CR1 and PD-1. Regulatory context (ceRNA network) and druggability (DGIdb query and molecular docking to CX3CR1, FCGR1A, TLR5) were explored. Results The 10-gene ETG signature stratified mortality risk with good discrimination and calibration (AUCs 0.76, 0.78 and 0.73 at 7, 14 and 28 days; Harrell’s C-index 0.782). The age-integrated nomogram showed close agreement between predicted and observed survival. Low-risk patients were enriched for immune-response pathways. In the clinical cohort, non-survivors had lower FCGR1A, TLR5 and CX3CR1 mRNA/protein in PBMCs and reduced proportions of FCGR1A⁺, CX3CR1⁺ and PD-1⁺ CD14⁺ monocytes (all p<0.01). ceRNA analysis highlighted a putative NEAT1/miR-1287-5p/CX3CR1 axis, and docking suggested ligandability of CX3CR1, FCGR1A and TLR5, nominating candidates such as valproic acid and CGP-52608. Conclusion An ET-anchored 10-gene signature captures a clinically relevant axis of sepsis-associated immunosuppression and enables short-term mortality risk stratification. Integration with age yields a simple nomogram with stable performance, and multilayer validation (transcript, protein, single-cell) supports biological plausibility and ET-guided immunomodulation as a potential therapeutic avenue.