Survival of the Littlest: Navigating Sepsis Diagnosis beyond Inflammation in Preterm Neonates

Abstract

Sepsis diagnosis in preterm neonates is challenging due to symptom overlap with non-infectious inflammatory conditions, and slow, unreliable diagnostic practices. This case-control study aims to elucidate sepsis pathophysiology, and identify metabolic biomarkers for timely, accurate diagnosis, to prevent rapid health deterioration and unnecessary antibiotic use. Liquid chromatography-mass spectrometry was performed on 227 plasma samples, obtained from 94 preterm neonates, to measure 317 metabolites encompassing amines and signaling lipids. Linear mixed-effect modeling, LASSO and logistic regression models were calculated to assess metabolic alterations across control, systemic inflammation-no sepsis (SINS), and sepsis groups. Stratification by sex and pathogen type allowed identification of sex-specific responses and pathogen-driven variations in sepsis. Key findings include (i) shared metabolic changes in SINS and sepsis, (ii) progressive alterations from control to SINS to sepsis, and (iii) sepsis-specific markers. Males exhibited a pro-inflammatory phenotype while females showed an anti-inflammatory phenotype in response to sepsis. Gram-positive and gram-negative bacterial sepsis revealed distinct metabolic profiles. A diagnostic model comprising 5 metabolic features and IL-6 distinguished SINS from sepsis at clinical suspicion (AUC 0.79, sensitivity 0.85, specificity 0.82). These insights highlight the potential of metabolomics to revolutionize neonatal sepsis management with precision diagnostics and personalized treatment strategies.