Daily Sepsis Research Analysis
Three studies advance sepsis science across settings: a multicenter Ugandan proteomics analysis delineates an immunosuppressive host-response signature specific to malarial sepsis; a multi-omics mouse-to-human study proposes an early serum metabolite model (IC3) for sepsis-associated acute kidney injury; and a prehospital evaluation shows PRESEP as the only model outperforming unaided EMS assessment for identifying ED Sepsis-3 cases.
Summary
Three studies advance sepsis science across settings: a multicenter Ugandan proteomics analysis delineates an immunosuppressive host-response signature specific to malarial sepsis; a multi-omics mouse-to-human study proposes an early serum metabolite model (IC3) for sepsis-associated acute kidney injury; and a prehospital evaluation shows PRESEP as the only model outperforming unaided EMS assessment for identifying ED Sepsis-3 cases.
Research Themes
- Host-response proteomics and stratification in sepsis
- Early biomarkers and multi-omics for sepsis-associated acute kidney injury
- Prehospital sepsis prediction and EMS decision support
Selected Articles
1. Host Response Stratification in Malarial and Non-malarial Sepsis: A Prospective, Multicenter Analysis From Uganda.
Across two Ugandan cohorts, malarial and non-malarial sepsis shared a largely conserved host response, yet malarial sepsis exhibited a selective immunosuppressive signature (elevated IL-10, LILRB1, KIR3DL1, LAG3, TIM-4). A classifier comprising these proteins achieved AUCs of 0.73 and 0.72 with good calibration for identifying malarial sepsis.
Impact: This multicenter prospective analysis defines pathogen-associated immunologic heterogeneity in a high-burden setting and provides a validated protein signature for malarial sepsis stratification.
Clinical Implications: The immunosuppressive signature in malarial sepsis may guide future pathogen-stratified immunomodulatory trials and inform triage or risk enrichment strategies in endemic settings.
Key Findings
- Malarial sepsis accounted for 20% and 28% of cases in discovery and validation cohorts, respectively.
- ≤8% of proteins were differentially expressed between malarial and non-malarial sepsis after adjustment, indicating a largely conserved host response.
- Malarial sepsis showed higher levels of immunosuppressive/exhaustion markers (e.g., IL-10, LILRB1, KIR3DL1, LAG3, TIM-4).
- A proteomic classifier discriminated malarial sepsis with AUC 0.73 (0.65-0.81) and 0.72 (0.65-0.79), with Brier scores 0.14 and 0.18.
Methodological Strengths
- Prospective multicenter design with separate discovery and validation cohorts
- High-dimensional proteomic profiling with adjusted analyses for key confounders and external validation with calibration metrics
Limitations
- Generalizability beyond Ugandan public hospitals may be limited
- Observational design precludes causal inference and does not test targeted therapies
Future Directions: Test immunomodulatory strategies guided by the malarial immunosuppression signature and evaluate portability of the classifier across regions and assay platforms.
2. Metabolomics- and proteomics-based multi-omics integration reveals early metabolite alterations in sepsis-associated acute kidney injury.
Integrating renal proteomics and metabolomics in LPS-induced mice highlighted five core metabolites, with time-resolved serum changes detected as early as 8 hours. A three-metabolite serum model (IC3: inosine, creatine, 3-hydroxybutyrate) discriminated SA-AKI from sepsis without AKI in 56 patients with an AUC of 0.90.
Impact: This study bridges mechanistic mouse data and human serum metabolomics to deliver an early diagnostic panel for SA-AKI with high discriminative performance.
Clinical Implications: If validated in broader populations, the IC3 panel could enable early SA-AKI detection and timely renoprotective strategies in sepsis care.
Key Findings
- Identified 13 differential renal metabolites and 112 proteins in LPS-induced SA-AKI mice.
- Five core metabolites (3-hydroxybutyric acid, 3-hydroxymethylglutaric acid, creatine, myristic acid, inosine) were linked to renal function and proteomic changes.
- Serum myristic acid increased at 8 h; inosine decreased at 8 h; 3-hydroxybutyrate, 3-hydroxymethylglutaric acid, and creatine rose at 24 h in mice.
- In a 56-patient clinical cohort, the IC3 model (inosine, creatine, 3-hydroxybutyrate) achieved AUC 0.90 for early SA-AKI identification.
Methodological Strengths
- Multi-omics integration (renal proteomics and metabolomics) with time-resolved validation in vivo
- Translational validation in a human clinical cohort with a parsimonious logistic model
Limitations
- Clinical validation cohort was small (n=56) and from a single setting
- External validation and assessment against standard biomarkers (e.g., NGAL, KIM-1) were not reported
Future Directions: Prospective multicenter validation, head-to-head comparison with established AKI biomarkers, and clinical utility assessments for early intervention.
3. Performance Evaluation of Prehospital Sepsis Prediction Models.
In a nested case-control study across four EDs, only PRESEP outperformed unaided EMS infection assessment and qSOFA for predicting which ambulance-transported adults would meet ED Sepsis-3 criteria, with AUPRC 0.33 versus 0.17 for unaided EMS. PRESEP’s sensitivity was 60% with a positive predictive value of 20%.
Impact: This study provides the first broad, head-to-head prehospital comparison showing that most models underperform, highlighting PRESEP as a pragmatic choice and underscoring the need for better EMS-targeted tools.
Clinical Implications: EMS systems may consider implementing PRESEP while developing or validating improved models, with attention to modest PPV and operational integration.
Key Findings
- Among 21 prehospital sepsis models, only PRESEP outperformed unaided EMS assessment (AUPRC 0.33 vs 0.17; p < 0.001).
- PRESEP also outperformed qSOFA (AUPRC 0.13; p < 0.001).
- PRESEP showed 60% sensitivity and 20% positive predictive value among dichotomous predictors.
Methodological Strengths
- Nested case-control design with rigorous model discrimination using AUPRC appropriate for imbalanced outcomes
- Direct head-to-head evaluation of 21 models against real-world EMS assessment
Limitations
- Single-state (Utah) data may limit generalizability to other EMS systems
- Retrospective use of prospectively collected records; not an interventional deployment study
Future Directions: Prospective EMS implementation trials of PRESEP, model recalibration/augmentation with vital signs and biomarkers, and external validation across diverse EMS systems.