Cross-Platform Metabolomic Profiling Uncovers Prognostic Serum Biomarkers of Sepsis Risk in Burn Patients

Authors:
Fadi Khalaf, Ana Stanciu, Diana Julia Tedesco, Philip Britz-Mckibbin, Marc Jeschke

Body of Abstract:
Background: Sepsis remains a major cause of mortality in burn patients, yet early diagnosis is hindered by its rapid progression, patient heterogeneity, and the confounding metabolic effects of burn injury. Existing clinical biomarkers are inconsistent and fail to reliably predict sepsis onset. To address this gap, we move beyond traditional markers and directly profile the metabolic disturbances underlying burn sepsis. Through comparative metabolomic and lipidomic analysis across multiple time points, we aim to identify novel circulating metabolites with strong diagnostic and prognostic potential, offering mechanistic insight and a more accurate framework for predicting sepsis in burn patients.

Methods: Serum samples from septic (n=39) and non-septic (n=37) burn patients were collected at 0–3, 4–10, 11–18, and >20 days post-injury (DPI). Untargeted analysis of serum metabolites and lipids was performed. A total of 91 metabolites and 84 lipids were annotated. Supervised orthogonal partial least squares-discriminant analysis (OPLS-DA) and receiver operating characteristic (ROC) curves were used to identify and classify discriminative metabolic signatures between septic and non-septic patients. Age, sex, DPI, and comorbidities were adjusted for.

Results: Established clinical markers—including creatinine, lactic acid, and CRP—showed no differences between septic and non-septic patients, underscoring the need for better biomarkers. Using all available time points, we identified several metabolites with strong diagnostic potential for burn sepsis, including the amino-acid–related metabolites S-methylcysteine, 3-methyl-2-oxovaleric acid, and N-methyllysine, as well as the bioactive lipids LPE 20:4 and PE 36:4 (p<0.001). The valerylcarnitine/crotonobetaine ratio demonstrated the strongest discriminative performance (AUC=0.83, p<0.0001). Using only early timepoints, we identified several metabolites with significant prognostic value for burn sepsis, including the amino acids arginine and isoleucine, and the methylated osmolytes/gut-derived metabolites betaine and crotonobetaine (all p< 0.001). The valerylcarnitine/crotonobetaine ratio demonstrated the strongest prognostic performance (AUC=0.88, p<0.0001). Lastly, enrichment analysis of the early time points identified methylhistidine metabolism as the most significantly altered metabolic pathway for distinguishing patients who later developed sepsis from those who did not. Conclusions: In summary, this is the first study to apply comprehensive serum metabolomics to burn sepsis, revealing diagnostic and prognostic biomarkers that outperform traditional clinical indicators and identifying early metabolic pathways linked to sepsis progression. These findings establish a foundation for timely, metabolically informed sepsis detection and have the potential to significantly improve clinical decision-making and outcomes in burn critical care.