A97-02 Evaluating the Relationship Between Pathogen Taxonomy and the Hyperinflammatory Phenotype on Critically Ill Patients With Sepsis From the Publicly Available MIMIC-IV Clinical Database

Abstract Rationale Distinct molecular phenotypes of sepsis, called Hyperinflammatory and Hypoinflammatory, are associated with markedly different outcomes. Emerging evidence from single center analyses suggest that pathogen taxonomy may influence phenotype classification, closely linking host response to pathogen exposure1. We sought to determine whether similar relationships between bloodstream isolates and molecular phenotype are observed in other critically ill cohorts. Methods Adult patients (age 18) with an index ICU admission for sepsis were identified from the MIMIC-IV (version 3. 1) clinical dataset2. Patients who met Sepsis-3 criteria with a suspected bloodstream infection were included in this analysis (PMID: 26903338). Clinical and laboratory data within 24 hours of ICU admission were used as inputs in a validated Clinical Classifier Model (CCM) to generate probability of belonging to the Hyperinflammatory phenotype (PMID: 32551817). Rigorous adjudication of microbiology results was conducted to identify bloodstream pathogens within 48 hours of CCM phenotyping and discard known commensals. Blood pathogens were organized based on organism characteristics, taxonomy, and frequency of pathogens identified. Kaplan-Meier estimator was used to compare 28-day mortality between phenotypes. Logistic regression was used to evaluate the relationship between blood pathogens and phenotypes for 1) pathogen characteristics, 2) most frequent pathogen species, and 3) pathogen taxonomy groupings. Results A total of 11, 695 unique index ICU admissions from 2008 to 2022 were included. Of these 1, 587 (13. 5%) were assigned to the Hyperinflammatory phenotype. 28-day mortality was significantly higher in Hyperinflammatory phenotype (50. 9%) compared to Hypoinflammatory phenotype (21. 8%; p 0. 01) Figure1a. Positive monomicrobial blood cultures were identified in 1, 453 participants, among these 317 (21. 8%) were Hyper and 1136 (78. 2%) Hypo, with higher 28-day mortality in the former (48. 9% vs. 19. 1%; p 0. 01). Gram-negative pathogens were predominant in the Hyperinflammatory phenotype (p 0. 01) Figure1b, with Escherichia coli (n = 99, 31. 2%) being the most common pathogen. Whereas gram-positive species were predominant in the Hypoinflammatory phenotype with Staphylococcus aureus (n = 246, 21. 7%) being the most common gram-positive pathogen. In a logistic regression model, Enterobacteriaceae was significantly associated with Hyperinflammatory phenotype classification after adjusting for age, gender, and Sequential Organ Failure Assessment (SOFA) score (OR 1. 05, 95% CI 1. 03-1. 08; p 0. 01) Figure1c. Conclusions These findings that implicate Enterobacteriaceae exposure, particularly Escherichia coli, with the Hyperinflammatory phenotype in critically ill septic patients align with prior studies. Further studies need to elucidate host and pathogen level factors that lead to increased susceptibility of manifesting the Hyperinflammatory phenotype. REFERENCES: 1. https: //doi. org/10. 1164/ajrccm-conference. 2024. 209. 1MeetingAbstracts. A4995; 2. https: //doi. org/10. 13026/kpb9-mt58 This abstract is funded by: R01HL173531, R35GM142992