Genomic Factors Associated with Invasiveness in Klebsiella pneumoniae Isolates from Hospitalised Adults in Johannesburg, South Africa

Background Klebsiella pneumoniae invasive disease (KPn-ID) is a leading cause of healthcare-associated morbidity and mortality in low- and- middle- income countries. We report here on the genomic relatedness of invasive and colonising KPn isolates from adults hospitalised in Johannesburg, South Africa, and genetic factors associated with invasiveness. Methods A multicentre, prospective hospital-based study of adults with culture-confirmed KPn from blood and cerebrospinal fluid was undertaken from May 15th, 2023 to May 14th, 2024. Adults in hospital without evidence of KPn-ID were matched to cases (2:1) and assessed for KPn gastrointestinal colonisation. Whole genome sequencing was performed on invasive and colonising KPn isolates. We estimated the relative invasiveness of KPn isolates based on sequence types (STs), capsular loci (KL), lipopolysaccharide (O) antigen types, virulence factors, and antimicrobial resistance (AMR) factors by calculating crude case-carrier ratios (CCRs). Additionally, multiple multivariable logistic regression models were used to measure the association between host characteristics and KPn genomic characteristics, with invasiveness. Findings Whole genome sequencing was done on 508 invasive and 184 colonising KPn isolates. There was no single dominant clone amongst the invasive or colonising isolates on phylogenetic analysis. There was variability in the distribution of STs and K-loci between invasive (111 STs, 56 K-loci) and colonising isolates (87 STs, 57 K-loci). Several KPn genetic factors were associated with invasiveness in the crude CCR analyses: ST2497, ST39, KL2, KL64, O1αβ,2α, the virulence factors yersiniabactin, aerobactin, salmochelin and RmpADC, the genomic AMR markers for aminoglycosides, carbapenems, third generation cephalosporins, fluroquinolones and sulphonamides. Nevertheless, after adjusting STs for patient clinical characteristics and virulence factors in multiple logistic regression models, only ST2497 (aOR 3.81; 95% CI: 1.64- 8.86), ST39 (aOR 4.50; 95% CI: 1.62- 12.52), ST307 (aOR 7.50; 95% CI: 1.67- 33.77), and yersiniabactin (aOR 2.97; 95% CI: 1.83- 4.84) remained significantly associated with invasiveness. After further adjustment in a separate model for K-loci, O-antigen types, AMR factors and plasmids, the only genomic factor associated with invasiveness was carbapenem resistance (aOR 3.88; 95% CI: 1.28- 11.82). Similar results were obtained when using a cluster-adjusted model. Age over 65 years (aOR 2.65; 95% CI: 1.31- 5.38) was a significant clinical characteristic associated with enhanced invasive potential. Interpretation Age over 65 years, a host susceptibility factor, was associated with invasiveness. No genomic characteristics, other than the presence of genes encoding for carbapenem resistance, were associated with a higher potential for causing invasive disease. This poses a challenge for targeted preventive strategies against KPn-ID, including development of a vaccine, which would require multiple K- and/or O-antigens for adequate coverage.