Daily Sepsis Research Analysis
Three studies stand out today in sepsis research: a multi-omics/network-based platform repurposed acetaminophen and pyridoxal phosphate to protect the heart in septic cardiomyopathy; a large, international EMR-derived risk score accurately predicted harmful positive fluid balance after AKI diagnosis; and a MALDI-TOF MS plus machine learning approach rapidly identified carbapenem-resistant E. coli and K. pneumoniae directly from positive blood cultures.
Summary
Three studies stand out today in sepsis research: a multi-omics/network-based platform repurposed acetaminophen and pyridoxal phosphate to protect the heart in septic cardiomyopathy; a large, international EMR-derived risk score accurately predicted harmful positive fluid balance after AKI diagnosis; and a MALDI-TOF MS plus machine learning approach rapidly identified carbapenem-resistant E. coli and K. pneumoniae directly from positive blood cultures.
Research Themes
- Drug repurposing for septic cardiomyopathy via multi-omics/network biology
- Precision fluid stewardship in sepsis/AKI using validated risk prediction
- Rapid AMR diagnostics in bloodstream infection with MALDI-TOF plus ML
Selected Articles
1. Multi-Omics and Network-Based Drug Repurposing for Septic Cardiomyopathy.
An integrated UPLC-MS/MS and RNA-seq approach with network proximity analysis nominated 14 FDA-approved candidates for septic cardiomyopathy. Acetaminophen and pyridoxal phosphate improved EF/FS and reduced BNP and cTnI in LPS-SCM mice, mediated by reduced prostaglandin synthesis and restored amino acid balance.
Impact: Provides a scalable, translational platform and identifies two ICU-accessible agents with mechanistic plausibility to protect the septic heart, potentially expediting clinical testing.
Clinical Implications: Suggests repurposing acetaminophen and pyridoxal phosphate to mitigate septic cardiomyopathy, but clinical trials are needed before adoption given preclinical design and LPS model limitations.
Key Findings
- Network proximity across multi-omics identified 129 drugs; 14 prioritized for ICU suitability and safety.
- Acetaminophen and pyridoxal phosphate improved EF and FS and decreased BNP and cTnI in LPS-induced SCM mice.
- Mechanistically, acetaminophen reduced prostaglandin synthesis/inflammation; pyridoxal phosphate restored amino acid balance.
Methodological Strengths
- Integrated metabolomics and transcriptomics with network-based drug proximity screening
- In vivo validation in LPS-SCM mice and mechanistic assays in H9c2 cells
Limitations
- LPS-induced model may not capture polymicrobial/human SCM complexity
- Only two candidates were tested; no human data or safety endpoints in sepsis context
Future Directions: Prospective clinical trials to test dosing, safety, and efficacy in septic cardiomyopathy; expansion of the platform to polymicrobial and human tissue datasets.
2. Predicting a strongly positive fluid balance in critically ill patients with acute kidney injury: A multicentre, international study.
An EMR-derived AKI-FB score using routinely available variables predicted >2 L positive fluid balance within 72 hours of AKI diagnosis with AUC 0.805 (external AUC 0.761). Key predictors included sepsis/septic shock, creatinine, cumulative fluid input, vasopressor use/dose, lactate, transfusion, and nutrition.
Impact: Enables proactive fluid stewardship by identifying AKI patients at risk for harmful fluid accumulation, providing a ready-to-implement tool for ICU workflows and trial enrichment.
Clinical Implications: Supports early de-resuscitation strategies (diuretics, RRT, vasopressor-guided fluids) in high-risk AKI patients; can be embedded in EMRs to trigger alerts and guide fluid management protocols.
Key Findings
- Developed the AKI-FB risk score in 32,030 ICU patients; validated in 4,465 external patients.
- Threshold score of 32 predicted >2 L positive fluid balance at 72 h with 75% sensitivity and 72% specificity (AUC 0.805; external AUC 0.761).
- Sepsis/septic shock, highest creatinine, cumulative fluid balance, mechanical ventilation, noradrenaline use/dose, lactate ≥2 mmol/L, transfusion, and nutrition were key variables.
Methodological Strengths
- Very large development cohort with external validation across countries
- Uses routinely captured EMR variables facilitating implementation
Limitations
- Observational model without prospective interventional testing or outcome improvement demonstration
- Generalizability beyond the contributing health systems and calibration across diverse ICUs require further study
Future Directions: Prospective implementation trials to test clinical impact on fluid balance, AKI progression, and mortality; integration with decision support for fluid de-escalation protocols.
3. Rapid detection of carbapenem-resistant Escherichia coli and carbapenem-resistant Klebsiella pneumoniae in positive blood cultures via MALDI-TOF MS and tree-based machine learning models.
Tree-based ML models applied to MALDI-TOF spectra predicted carbapenem resistance directly from positive blood cultures with high AUROC for E. coli (up to 1.00) and strong performance for K. pneumoniae (up to 0.95), enabling earlier, targeted therapy.
Impact: Accelerates time-to-effective therapy for septic patients with CRE by leveraging existing MALDI platforms and deployable ML, with potential to reduce mortality and resistance spread.
Clinical Implications: Hospitals can integrate ML classifiers with MALDI-TOF to flag likely CREC/CRKP within hours, guiding escalation to effective agents and informing infection control.
Key Findings
- Analyzed 640 E. coli and 444 K. pneumoniae MALDI spectra; built DT, RF, GBM, XGBoost, and ERT models.
- For E. coli, AUROC reached 0.99–1.00 with accuracy up to 0.92 in 149 positive cultures.
- For K. pneumoniae, AUROC reached 0.90–0.95 with accuracy up to 0.86 in 127 positive cultures.
Methodological Strengths
- Leverages routinely acquired MALDI-TOF spectra with multiple ML algorithms and clear performance metrics
- Relatively large training datasets for E. coli and K. pneumoniae with separate prediction sets
Limitations
- Lack of external, multi-center validation; real-world clinical impact on time-to-therapy and outcomes not tested
- Lower and more variable performance for K. pneumoniae; potential spectrum/site-specific bias
Future Directions: Prospective multi-center implementation to test time-to-effective therapy, outcomes, and antibiotic stewardship impact; external validation and model calibration across MALDI platforms.