Keywords
How to Cite
Abstract
Background: Carbapenem-resistant Klebsiella pneumoniae (CRKP) is a critical-priority pathogen, yet how carbapenemase mechanism, co-resistance to last-line agents, and the emergence of convergent resistance–virulence lineages jointly shape clinical outcomes remains incompletely resolved. We integrated genomic determinants with phenotypic resistance and clinical data to identify independent predictors of mortality.
Methods: We analysed 200 non-duplicate CRKP isolates from invasive and respiratory infections across three hospitals (2021–2024). Each isolate was characterised for sequence type (ST), clonal group (CG), capsular (KL) and O-antigen loci, carbapenemase class and allele, acquired resistance and porin determinants, and a defined virulence-locus panel (rmpA/rmpA2, aerobactin, salmochelin, yersiniabactin, colibactin, peg-344). A convergent genotype was defined as co-carriage of a carbapenemase with a hypervirulence-associated siderophore or regulator locus. Minimum inhibitory concentrations (MICs) were determined for seven agents. Associations with 30-day mortality were assessed by logistic and Cox proportional-hazards regression.
Results: Thirty-day mortality was 32.5% (65/200). KPC (33.5%) and OXA-48-like (30.5%) predominated, followed by metallo-β-lactamases (MBLs: NDM, VIM, IMP; 27.5%). All isolates were ceftazidime-resistant and 85.0% meropenem-resistant. Ceftazidime–avibactam resistance (30.5% overall) was almost entirely explained by MBL production (100% of MBL producers vs 4.1% of non-MBL isolates; p<0.001). Convergent resistance–virulence genotypes were present in 22.0% of isolates and carried higher 30-day mortality than non-convergent isolates (45.5% vs 28.8%; log-rank p=0.020). In multivariable analysis, independent predictors of 30-day mortality were patient age (adjusted OR [aOR] 1.03 per year, 95% CI 1.00–1.05), Pitt bacteraemia score (aOR 1.29, 1.07–1.56), ceftazidime–avibactam resistance (aOR 2.11, 1.06–4.20), and convergent genotype (aOR 2.28, 1.08–4.85). The convergent genotype remained independently associated with mortality in Cox analysis (HR 2.03, 1.17–3.54).
Conclusions: In this integrated genomic-epidemiological cohort, CRKP mortality was driven less by carbapenem resistance per se than by the loss of ceftazidime avibactam activity in MBL producers and by the convergence of carbapenem resistance with hypervirulence determinants. Genomic surveillance that reports carbapenemase class and virulence convergence not carbapenem resistance alone offers actionable prognostic and stewardship value.