Daily Sepsis Research Analysis
Three impactful sepsis studies span mechanistic immunology, antibiotic stewardship, and AI-enabled diagnostics. A preclinical study identifies neutrophil-mediated trogocytosis of B-1a cells as a therapeutic target to attenuate sepsis-induced acute lung injury, while a large real-world cohort supports narrow-spectrum β-lactam plus gentamicin as safe empiric therapy. Separately, a machine-learning model using triage data accurately predicts bacteremia in febrile ED patients.
Summary
Three impactful sepsis studies span mechanistic immunology, antibiotic stewardship, and AI-enabled diagnostics. A preclinical study identifies neutrophil-mediated trogocytosis of B-1a cells as a therapeutic target to attenuate sepsis-induced acute lung injury, while a large real-world cohort supports narrow-spectrum β-lactam plus gentamicin as safe empiric therapy. Separately, a machine-learning model using triage data accurately predicts bacteremia in febrile ED patients.
Research Themes
- Immunomodulation and pathophysiology of sepsis-induced organ injury
- Antibiotic stewardship in empiric therapy for suspected sepsis
- AI/ML for early infection detection and risk stratification
Selected Articles
1. A novel molecule targeting neutrophil-mediated B-1a cell trogocytosis attenuates sepsis-induced acute lung injury.
Using a murine cecal ligation and puncture (CLP) model, the authors implicate neutrophil-mediated trogocytosis in the depletion of B-1a cells during sepsis and show that a novel molecule targeting this process preserves B-1a cells and attenuates acute lung injury. The work links Siglec-G–regulated B-1a biology to neutrophil interactions and identifies a potentially druggable pathway.
Impact: This provides a mechanistic framework connecting B-1a cell loss to lung injury in sepsis and proposes a novel therapeutic target by interrupting neutrophil-mediated trogocytosis.
Clinical Implications: Although preclinical, targeting trogocytosis to preserve protective B-1a cells could inspire immunomodulatory strategies to mitigate sepsis-induced acute lung injury and guide biomarker development.
Key Findings
- In a CLP mouse model of sepsis, neutrophils accumulated in lungs/serosa while B-1a cells declined, implicating neutrophil-mediated trogocytosis in B-1a depletion.
- A novel molecule that targets neutrophil-mediated trogocytosis preserved B-1a cells in vivo.
- Therapeutic targeting of trogocytosis attenuated sepsis-induced acute lung injury in treated mice.
Methodological Strengths
- Use of a well-established CLP model to recapitulate polymicrobial sepsis and lung injury
- Mechanism-driven intervention targeting a defined cellular process (trogocytosis)
Limitations
- Preclinical mouse study without human validation
- Incomplete mechanistic detail and safety/PK data for the novel molecule are not provided in the abstract
Future Directions: Validate the pathway in independent models and human samples; develop biomarkers of B-1a cell depletion/trogocytosis; and evaluate safety/efficacy in early-phase clinical studies.
Sepsis is a dysregulated immune response to infection. B-1a cells play a crucial role in maintaining immuno-physiologic homeostasis. Sialic acid-binding immunoglobulin-like lectin G (Siglec-G) regulates B-1a cell's behavior and function. Trogocytosis is the process by which one cell acquires portions of another cell's plasma membrane and cytoplasm through direct contact. During sepsis, neutrophils accumulate in the lungs and serosal cavities, while B-1a cells decrease. We hypothesized that neutrophil-mediated trogocytosis causes B-1a cell depletion in sepsis, and that targeting this process could preserve B-1a cells and attenuate sepsis-induced acute lung injury (ALI). Sepsis was induced in mice by cecal ligation and puncture (CLP). Twenty hours after CLP, B-1a cells (CD19
2. Empiric Antibiotic Therapy in Suspected Sepsis: Impact of Gentamicin-Based Regimens on Incident Renal Failure and Mortality.
In a 1,917-patient retrospective cohort of suspected sepsis, narrow-spectrum β-lactam plus gentamicin was not associated with higher AKI or mortality compared with broad-spectrum β-lactams. Broad-spectrum therapy was linked to worse ordinal outcomes (AKI stage/death; adjusted OR 1.61), and gentamicin cumulative dose did not correlate with peak creatinine.
Impact: Provides real-world evidence supporting stewardship-friendly empiric regimens without increased renal harm or mortality, potentially reducing broad-spectrum β-lactam use.
Clinical Implications: When local resistance patterns allow, consider narrow-spectrum β-lactam plus gentamicin for empiric coverage in suspected sepsis to limit broad-spectrum exposure while maintaining safety.
Key Findings
- Among 1,917 suspected sepsis patients, 33.1% received narrow-spectrum β-lactam/gentamicin and 66.9% received broad-spectrum β-lactams.
- Broad-spectrum β-lactams were associated with a higher posttreatment AKI stage or death (adjusted OR 1.61; 95% CI 1.27–2.04).
- No significant association was observed between cumulative gentamicin dose and peak creatinine; AKI cases on gentamicin normalized creatinine within 30 days.
Methodological Strengths
- Large real-world cohort with adjustment using an ordinal composite outcome from no AKI to death
- Direct comparison of stewardship-relevant regimens with clinically meaningful endpoints over 30 days
Limitations
- Retrospective, nonrandomized design with baseline imbalances and potential residual confounding
- Single health system; microbiological eradication and resistance emergence were not detailed
Future Directions: Prospective randomized trials to confirm safety/effectiveness; assess microbiologic outcomes and resistance; and evaluate subgroup effects (e.g., renal impairment).
BACKGROUND: The efficacy and safety of administering a narrow-spectrum β-lactam and gentamicin as empirical therapy for community-acquired sepsis has been questioned. We compared the efficacy and safety of this combination with that of broad-spectrum β-lactams (cefotaxime, piperacillin-tazobactam, or meropenem) in patients with suspected sepsis. METHODS: In this retrospective study, we included patients initiated on narrow-spectrum β-lactam/gentamicin or broad-spectrum β-lactams for suspected sepsis between January 2017 and December 2022. Patients without baseline creatinine and at least 1 subsequent creatinine measurement were excluded. We compared the impact of antibiotic regimens using a 5-level ordinal outcome scale ranging from no acute kidney injury (AKI) to all-cause death during 30-day follow-up. RESULTS: Among 1917 patients, 33.1% received narrow-spectrum β-lactam/gentamicin, and 66.9% received broad-spectrum β-lactams. Patients initiated on broad-spectrum β-lactams had more comorbidities, had lower estimated glomerular filtration rate on admission, and more frequently required treatment with noradrenaline, respiratory support, and admission to the intensive care and medical intermediate care units. Therapy with broad-spectrum β-lactams was associated with a higher posttreatment stage of AKI or death (adjusted odds ratio, 1.61 [95% confidence interval, 1.27-2.04]). We found no significant association between cumulative dose of gentamicin and peak creatinine value. For patients treated with gentamicin experiencing AKI, creatinine normalized during 30-day follow-up. CONCLUSIONS: In patients with suspected sepsis, empirical treatment with narrow-spectrum β-lactam/gentamicin was not associated with an increased risk of AKI or death. If local antimicrobial resistance patterns permit, narrow-spectrum β-lactam/gentamicin may reduce broad-spectrum β-lactam usage, addressing a key element of antibiotic stewardship.
3. Developing Machine-Learning Models to Predict Bacteremia in Febrile Adults Presenting to the Emergency Department: A Retrospective Cohort Study from a Large Center.
Using 80,201 febrile adult ED encounters with blood cultures, ML models trained on triage-available variables achieved strong performance for bacteremia prediction, with CatBoost AUC 0.844. The approach demonstrates feasibility for real-time risk stratification with minimal data burden.
Impact: The scale and performance suggest immediate translational potential for ED triage workflows, potentially improving timely diagnostics and antibiotic decisions.
Clinical Implications: Integration into ED triage could prioritize blood cultures, early antibiotics, and observation for high-risk patients, while reducing unnecessary workup in low-risk patients.
Key Findings
- Dataset included 80,201 febrile adult ED visits (2009–2018), with bacteremia prevalence ~12%.
- CatBoost achieved the highest AUC of 0.844 (95% CI 0.837–0.850); other gradient boosting models performed similarly.
- Models relied solely on triage-available features (demographics, symptoms, triage vitals/history), enabling potential real-time deployment.
Methodological Strengths
- Very large single-center cohort with predefined train/test split and K-fold validation
- Comparison of multiple supervised ML algorithms with confidence intervals for AUC
Limitations
- Single-center retrospective design without external prospective validation
- Outcome defined by blood culture positivity (may not capture all clinically significant sepsis); potential label noise
Future Directions: External validation and prospective impact studies; integration into EHR with clinician-in-the-loop; fairness and drift monitoring; calibration for different prevalence settings.
INTRODUCTION: Bacteremia, a common disease but difficult to diagnose early, may result in significant morbidity and mortality without prompt treatment. We aimed to develop machine-learning (ML) algorithms to predict patients with bacteremia from febrile patients presenting to the emergency department (ED) using data that is readily available at the triage. METHODS: We included all adult patients (≥18 years of age) who presented to the emergency department (ED) of National Taiwan University Hospital (NTUH), a tertiary teaching hospital in Taiwan, with the chief complaint of fever or measured body temperature more than 38°C, and who received at least one blood culture during the ED encounter. We extracted data from the Integrated Medical Database of NTUH from 2009-2018.The dataset included patient demographics, triage details, symptoms, and medical history. The positive blood culture result of at least one potential pathogen was defined as bacteremia and used as the binary classification label. We split the dataset into training/validation and testing sets (60-to-40 ratio) and trained five supervised ML models using K-fold cross-validation. The model performance was evaluated using the area under the receiver operating characteristic curve (AUC) in the testing set. RESULTS: We included 80,201 cases in this study. Of them, 48120 cases were assigned to the training/validation set and 32,081 to the testing set. Bacteremia was identified in 5,831 (12.1%) and 3,824 (11.9%) cases of the training/validation set and test set, respectively. All ML models performed well, with CatBoost achieving the highest AUC (.844, 95% confidence interval [CI] .837-.850), followed by extreme gradient boosting (.843, 95% CI .836-.849), gradient boosting (.842, 95% CI .836-.849), light gradient boosting machine (.841, 95% CI .834-.847), and random forest (.828, 95% CI .821-.834). CONCLUSION: Our machine-learning model has shown excellent discriminatory performance to predict bacteremia based only on clinical features at ED triage. It has the potential to improve care quality and save more lives if successfully implemented in the ED.