Monthly ReportNov 1–30, 2025
Sepsis, November 2025 edition
We read 84 papers and selected 5.
Summary
November 2025 sepsis research showed a clear shift toward precision diagnosis, biologically informed treatment stratification, and scalable implementation. Mechanistic studies identified new host-directed targets involving endothelial ALOX15, the Nur77–thrombomodulin pathway, immunometabolic PKM2 signaling, and gut–liver NETosis, while large clinical studies demonstrated that bundled care and structured neonatal examination can reduce infection-related harm and antibiotic exposure. Rapid host transcriptomics and explainable AI increasingly connected real-time biological or laboratory signals with antibiotic response, mortality risk, and potential treatment benefit. Multi-omics subgrouping further emphasized predictive enrichment as a strategy to overcome heterogeneous treatment effects and improve future sepsis trials.
Selected Articles
1. A rapid time-resolved host gene expression signature predicts responses to antibiotic treatment in neonatal bacterial sepsis.
Time-resolved transcriptomic analysis in microbiologically confirmed neonatal sepsis identified a host gene-expression signature that reversed within 24 hours of vancomycin initiation and tracked clinical improvement. The signature was reproducible across pediatric and adult sepsis cohorts, suggesting that host biology can provide an early readout of antibiotic response.
Impact: This study directly links a rapidly changing biological signal to antibiotic treatment response and addresses the major stewardship problem of identifying nonresponders or patients suitable for early de-escalation.
Clinical Implications: After prospective validation and adaptation to rapid clinical platforms, the signature could support antibiotic de-escalation, duration decisions, and early reassessment of treatment response. Its cross-age reproducibility increases translational value, although neonatal-specific performance and assay turnaround remain important implementation questions.
Key Findings
- A treatment-responsive host gene signature reversed within 24 hours of antibiotic initiation.
- Signature changes correlated with clinical improvement and included rapid adaptive immune pathway changes.
- The signature was reproducible across pediatric and adult sepsis cohorts.
2. A Multicomponent Intervention to Improve Maternal Infection Outcomes.
This cluster-randomized trial across 59 facilities in Malawi and Uganda evaluated the APT-Sepsis program, combining hand hygiene, evidence-based infection prevention and management, and the FAST-M bundle. Among 431,394 births, the intervention reduced the composite of infection-related maternal death, near-miss, or severe infection from 1.9% to 1.4%, with consistent effects across settings.
Impact: It provides high-quality randomized evidence that a scalable implementation package can reduce maternal infection-related harm at population scale in resource-limited settings.
Clinical Implications: Health systems serving similar populations should consider implementing WHO-aligned infection prevention, hand hygiene, and FAST-M pathways with fidelity monitoring, local adaptation, and cost-effectiveness evaluation.
Key Findings
- A cluster-randomized trial included 59 facilities and 431,394 births in Malawi and Uganda.
- The composite maternal infection outcome decreased from 1.9% to 1.4% with a risk ratio of 0.68.
- Benefits were consistent across countries, facility sizes, and follow-up periods.
3. Unexpected Protective Role of Thrombosis in Lung Injury via Endothelial Alox15.
Using LPS and cecal ligation and puncture models, this study found that mild pulmonary thrombosis reduced endothelial apoptosis, lung injury, and mortality through sustained endothelial ALOX15 expression. Endothelial-specific genetic manipulation and lipidomic rescue experiments supported a causal role for ALOX15-dependent protective lipids, whereas severe thrombosis or thrombocytopenia worsened outcomes.
Impact: The findings challenge the assumption that all pulmonary thrombosis is harmful in sepsis and provide a mechanistic explanation for why indiscriminate anticoagulation may fail in septic lung injury.
Clinical Implications: The study supports caution against blanket anticoagulation in sepsis-associated lung injury and motivates investigation of ALOX15-regulated lipid mediators. Translation requires confirmation in human tissue, large-animal models, and carefully selected clinical phenotypes.
Key Findings
- Mild pulmonary thrombosis reduced endothelial apoptosis, acute lung injury severity, and mortality in murine sepsis models.
- Endothelial-specific ALOX15 manipulation altered lung injury, and lipidomics identified protective ALOX15-dependent mediators.
- Severe thrombosis or thrombocytopenia worsened lung injury, helping explain failures of nonspecific anticoagulation strategies.
4. Deriving consensus sepsis clusters via goal-directed subgroup identification in multi-omics study.
A goal-directed subgroup identification framework integrated longitudinal transcriptomic, proteomic, metabolomic, and phenomic data from 1,327 patients across 43 hospitals. The resulting subgroups were optimized for differential treatment response and predicted survival differences associated with restrictive versus liberal fluid strategies and ulinastatin treatment, with external validation in international databases.
Impact: The study offers a methodological solution to sepsis heterogeneity by defining subgroups according to differential treatment benefit rather than clustering patients solely by baseline phenotype.
Clinical Implications: The framework could support predictive-enrichment designs for trials of fluids and immunomodulators, potentially reducing dilution of treatment effects. Prospective, assay-standardized validation is needed before clinical allocation of therapy.
Key Findings
- Goal-directed subgroup identification integrated longitudinal multi-omics data from 1,327 patients across 43 hospitals.
- Derived subgroups predicted differential survival associated with restrictive versus liberal fluid strategies.
- The framework also identified differential benefit signals for ulinastatin and was externally validated.
5. Explainable AI unravels sepsis heterogeneity via coagulation-inflammation profiles for prognosis and stratification.
SepsisFormer, an explainable transformer model, and SMART, a simpler routine-laboratory tool, were developed and validated in 12,408 patients. Seven coagulation-inflammatory laboratory variables plus age defined four risk tiers and two subphenotypes, while observational analyses suggested greater anticoagulant benefit in higher-risk or CIS2 patients.
Impact: The study combines high predictive performance with interpretability and implementation feasibility, allowing routine laboratory data to generate clinically testable hypotheses about treatment-effect heterogeneity.
Clinical Implications: SMART may support triage and monitoring using widely available laboratory data, while SepsisFormer can inform more complex risk prediction. The anticoagulation signals are hypothesis-generating and require prospective randomized evaluation.
Key Findings
- SepsisFormer achieved an AUC of 0.9301 in a multicenter cohort of 12,408 patients.
- SMART used seven coagulation-inflammatory laboratory variables and age to define four risk tiers and two subphenotypes.
- Observational analyses suggested greater anticoagulant benefit in moderate-to-severe risk groups and CIS2.