Answers are grounded in this paper and cite their sources.
Comparative glycomics and machine learning classify survivors and non-survivors of septic shock, indicating potential for early diagnosis and treatment.
Septic shock, the excessive immune response to pathogen infection, accounts globally for ∼20% of all deaths. Current methods to establish disease severity are unacceptably slow, unspecific and insensitive, hindering timely and effective treatment. Aiming to establish easy-to-measure glyco-signatures that may identify the most critically unwell patients, we applied comparative glycomics and glycoproteomics to sera longitudinally collected from septic shock survivors (n=29) and non-survivors (n=8). Glycomics of all 134 serum samples (sampled daily until recovery/death) revealed significant N-glycome dynamics across both patient groups. Unsupervised clustering of the serum N-glycome upon ICU admission (day 1) indicated survivorship-specific glyco-signatures. We therefore employed machine learning to train a random forest model using the serum N-glycome data. The model accurately classified survivorship outcomes of 35 of 37 patients (accuracy 94.6%) and correctly predicted 29 of 29 survivors (specificity 100%) and 6 of 8 non-survivors (sensitivity 75%). Interrogation of the serum N-glycome data revealed that Lewis x (Lex)-type N-glycans are elevated in non-survivors relative to survivors at ICU admission, a finding recapitulated by glycoproteomics. Amongst the 58 other Lex-containing serum glycoproteins that were strongly associated with acute phase response and stress pathways, alpha-1-acid-glycoprotein (AGP-1) was identified as a principal carrier of Lex glycoepitopes with a potential to stratify septic shock survivors from non-survivors (AUC 0.90). This study lays a foundation for risk stratification of septic shock patients by uncovering easy-to-assay AGP-1-Lex glycoforms that identify individuals experiencing poor survival outcomes already upon ICU admission, with the potential to translate to early individualised clinical care at the bedside.
No takes yet. Share an insight, caveat, or question.
Chau et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: