Abstract
BACKGROUND: Invasive Klebsiella pneumoniae (KP) infections complicated by sepsis are often associated with substantial mortality, which may partly result from dysregulated host immune-inflammatory responses to bacterial components, including lipopolysaccharide. KP O-antigen types have gained increasing attention in epidemiological surveillance and vaccine development, yet their relationship with sepsis risk in invasive KP infection remains insufficiently evaluated, particularly given potential confounding by clonal background and antimicrobial resistance. Here, we integrated clinical data with pathogen genomic and phenotypic characteristics to identify factors associated with sepsis in invasive KP infection and to characterize the underlying clonal and resistance context. METHODS: This retrospective cohort study screened all consecutive patients with culture-confirmed KP infection at Qingdao Municipal Hospital (November 2016-October 2023) and identified invasive cases based on isolation from normally sterile sites; only the first invasive episode per patient was included. Sepsis was defined according to Sepsis-3 criteria. Antimicrobial susceptibility testing was performed routinely, and resistance phenotypes (CRKP/ESBL/MDR) were derived from susceptibility results. All isolates underwent whole-genome sequencing (paired-end 150 bp) on the MGI Tech MGISEQ-2000 platform; serotypes, sequence types, and virulence determinants were inferred using Kleborate. Core-genome SNP phylogenies were reconstructed using Snippy with recombination filtered by Gubbins and maximum-likelihood inference by IQ-TREE. Multivariable logistic regression was used to identify factors associated with sepsis. RESULTS: Among 127 patients with invasive KP infection, the incidence of sepsis was 55.1% (70/127). In prespecified multivariable logistic regression models, bloodstream infection, higher PCT levels, and lymphocyte counts <1×10^9/L were associated with higher odds of sepsis after multivariable adjustment. Regarding pathogen-related factors, a "replacement exposure" strategy was applied by entering O2a, K64, and the ST11-K64 clonal background separately; O2a (OR 6.777, 95% CI 1.118-41.070), K64 (OR 16.674, 95% CI 1.588-175.022), and ST11-K64 (OR 19.525, 95% CI 1.991-191.461) were each significantly associated with sepsis, and Firth's penalized regression yielded directionally consistent results, supporting robustness under sparse-data/separation concerns. Cross-tabulation analyses further demonstrated strong aggregation between O2a and K64 as well as the ST11-K64 background (K64 present in 83.3% and ST11-K64 in 77.8% of O2a-positive isolates, versus 0.9% and 0% among O2a-negative isolates, respectively), indicating that O2a and K64 predominantly arose from a high-risk ST11-K64 clonal subgroup. Phylogenetic analysis showed that this ST11-K64 lineage largely overlapped with carbapenem-resistant Klebsiella pneumoniae (CRKP) and harbored selected virulence determinants (e.g., aerobactin and rmpA2), supporting its designation as a high-risk lineage with combined resistance and virulence features. CONCLUSION: In this retrospective cohort of invasive KP infections, O2a was associated with a higher likelihood of sepsis. Given the strong clustering of O2a with the ST11-K64 clonal background and carbapenem-resistant phenotypes, O2a is more likely to serve as a clinically observable adjunct marker of a high-risk clonal/resistance lineage. These findings suggest the potential incremental value of O2a for early risk stratification, which warrants further evaluation for informing sepsis surveillance and assessment of resistance risk.