Highly sensitive detection of sepsis in burn patients via Raman spectroscopy of circulating extracellular vesicles (EVs)

Introduction: Sepsis following burn trauma is a global complication with high mortality, with estimates of more than 60% of burn patient deaths resulting from infectious complications. Diagnosing sepsis is often complicated by confounding clinical manifestations of the burn injury. Current investigations into sepsis biomarkers using cytokine/chemokine markers are ongoing, but lack the sensitivity and specificity required for prompt detection and treatment. There is a strong rationale to assess circulating EVs from patient liquid biopsy due to their primary release by pathogens from bacterial biofilms and their roles in subsequent inflammatory and immune response. Thus, EVs have promise to be potent biomarkers for sepsis in burn patients. This study applies label-free Raman spectroscopy of EVs isolated from patient plasma for rapid, sensitive, and specific detection of sepsis.

Materials and

Methods: Plasma-derived EVs from septic (n=8) and non-septic burn patients (n=6) were isolated from plasma by centrifugation then further purified via size-exclusion chromatography (SEC). They were then characterized via resistive pulse sensing (RPS), nanoparticle tracking analysis (NTA), hybrid immunofluorescence/interferometric imaging, cryo-electron microscopy (cryoEM), and Raman spectroscopy. For Raman spectral acquisition, 2µL of SEC-isolated EVs were dried on quartz substrates and analyzed using a custom-built inverted confocal Raman system. Five random spectra (90s each) were sampled from each patient sample (25mW laser power at sample, 785nm incident wavelength). Spectra were background corrected, smoothed, and normalized for downstream analysis by principal component analysis (PCA).

Results, Conclusions, and Discussions: EV characterization techniques validated the isolation of EVs from plasma as expected. PCA and quadratic discriminant analysis (QDA) effectively differentiated EV Raman spectral profiles of septic from non-septic burn patients. The first five principal components achieved 97.5% sensitivity, 90.0% specificity, and 94.3% accuracy in separating the two patient populations. Of the 70 patient spectra collected, just 1 was falsely predicted as non-septic and 3 were falsely predicted as septic, yielding a 2.5% false negative rate and a 10% false positive rate. Interestingly, the three false positive spectra were acquired from the same patient EV sample. The global average Raman spectrum of the difference between septic and non-septic burn patient EVs was compared with spectra of specific glycoconjugate biomarkers of four top bacterial strains associated with sepsis morbidity in burn patients and showed fitting with lipopolysaccharides specific to K. pneumoniae, E. coli, and P. aeruginosa, and lipoteichoic acid isolated from S. aureus. This demonstrates that label-free spectroscopy can pick up on bacterial EV signatures in peripheral circulation without the need to specifically isolate them, contributing to the growing field of bacterial EV characterization and understanding.
These proof-of-concept results suggest that Raman spectroscopy of circulating EVs from liquid biopsy can provide sensitive, specific, and label-free detection of sepsis through identification of bacterial EV markers for clinical use. Increasing patient samples analyzed and exploring additional bacterial EV biomarkers and underlying patient microbiome differences are follow-up work interests. Nevertheless, this work demonstrates the promising application of EVs as biomarkers in patient critical care.

Acknowledgements (Optional):