Serum proteomic profiling of sepsis patients reveals a protein-based diagnostic model, with metabolomic insights into carbapenem-resistant Klebsiella pneumoniae infection

Introduction Sepsis caused by carbapenem-resistant Klebsiella pneumoniae (CRKP) is associated with high mortality. Current research is predominantly pathogen-centric, creating a knowledge gap regarding the host’s systemic molecular response, which is critical for understanding outcomes and developing diagnostic strategies. This study aimed to characterize the host serum proteomic and metabolomic landscape of CRKP sepsis and to develop a proteomics-derived biomarker panel for differential diagnosis, while integrating metabolomic data to gain mechanistic insights into host pathways. Methods Serum samples from sepsis patients, including culture-negative controls, and individuals infected with carbapenem-susceptible or carbapenem-resistant Klebsiella pneumoniae (CSKP and CRKP), underwent in-depth proteomic and metabolomic profiling. Differential expression, functional enrichment, and trend analyses were performed to characterize host molecular alterations associated with carbapenem resistance. Machine learning approaches were applied to construct diagnostic models based on host-derived molecular features. Candidate biomarker proteins were further validated using targeted parallel reaction monitoring (PRM) in an independent cohort. Results We identified 85 differentially expressed proteins (DEPs) distinguishing CRKP from CSKP, enriched in viral infection-related pathways, antigen presentation, and phagosome function. Trend analysis revealed a protein cluster (51 proteins) with stepwise increased abundance from controls to CSKP to CRKP, implicating coagulation, immune signaling, and metabolic dysfunction. Based on proteomic data, a four-protein biomarker panel (IGHV1-8, ITGA2, PKP1, IGFBP6) effectively differentiated CRKP from CSKP (test AUC = 0.920). Targeted proteomic validation in an independent cohort confirmed the differential expression of key model proteins, including IGFBP6 (CRKP vs. CSKP) and APOA2 (CSKP vs. controls). Metabolomics identified 128 differential metabolites in CRKP vs. CSKP, with enrichment in thermogenesis and amino acid/fatty acid degradation pathways. Integrated pathway analysis of proteomic and metabolomic data highlighted dysregulation in cysteine/methionine metabolism and the folate-mediated one-carbon pool, with MAT2B as a key connecting protein. Conclusion This first host-based multi-omics study of CRKP sepsis suggests distinct molecular signatures linked to resistance and disease severity. The independently validated biomarkers show high diagnostic potential, offering a preliminary foundation for early, non-invasive diagnosis and precision intervention strategies.