Back

Abscess Complications and Prolonged Care in Five-Biomarker-Defined Hypervirulent Klebsiella pneumoniae Bloodstream Infection

Watanabe, N.; Watari, T.; Otsuka, Y.; Matsumiya, T.

2026-04-11 infectious diseases
10.64898/2026.04.10.26350004 medRxiv
Show abstract

BackgroundFive-biomarker-defined hypervirulent Klebsiella pneumoniae (hvKp) causes invasive infections, but its burden in bloodstream infections versus classical K. pneumoniae (cKp) is unclear. MethodsThis retrospective cohort study at a tertiary hospital in Japan included K. pneumoniae bloodstream infection episodes from January 2022-December 2024. hvKp was defined by the presence of all 5 genotypic biomarkers (rmpA, rmpA2, iucA, iroB, and peg-344). The primary outcome was abscess complications, and secondary outcomes were length of stay and antibiotic duration. Whole-genome sequencing was performed for 164 isolates. ResultsAmong the 207 episodes, 28 (14%) were of hvKp. Abscess complication occurred in 17 (61%) hvKp versus 23 (13%) cKp episodes (adjusted odds ratio 10.7; 95% CI, 4.36-26.2). Median length of stay in hvKp versus cKp was 28 versus 14 days (adjusted ratio 1.60; 95% CI, 1.18-2.16) and median antibiotic duration was 43 versus 14 days (adjusted ratio 2.13; 95% CI, 1.64-2.77). These associations were attenuated after adjusting for abscess-related complications. No significant difference in 30-day mortality was observed, although the study was underpowered. Multidrug resistance was less frequent in hvKp strains than in cKp strains (11% vs. 30%; P = .040). Among the sequenced hvKp episodes, abscess rates varied across lineages, from 9 of 10 in ST23 to 1 of 4 in ST412. ConclusionsFive biomarker-defined hvKp strains delineated a bloodstream infection subgroup with frequent abscess complications and prolonged care. hvKp and cKp present distinct clinical challenges; diagnostic tools distinguishing these subgroups may aid abscess evaluation and source control.

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.