Five biomarkers separate a more invasive Klebsiella pneumoniae bloodstream infection subgroup
A five-marker PCR definition drew a sharp clinical line inside a routine bloodstream infection cohort in Japan. Among 207 episodes of Klebsiella pneumoniae bloodstream infection at a 907-bed tertiary hospital, 28 episodes met the hypervirulent definition built on simultaneous detection of rmpA, rmpA2, iucA, iroB, and peg-344. The remainder fell into the classical category, and the two groups behaved differently at the bedside. Abscess complications were far more common in the hypervirulent group, and treatment was longer in both hospital stay and antibiotic exposure.
The biologic logic fits the clinical pattern. Hypervirulent strains carry virulence determinants that favor siderophore production, mucoviscosity, and invasive spread, which helps explain why a bloodstream isolate can behave less like a simple bacteremia pathogen and more like a disseminating abscess-forming organism. In this cohort, abscess complications occurred in 61% of hypervirulent episodes versus 13% of classical episodes. In practice, that gap means a positive hypervirulent signature should raise concern for occult deep-seated infection even when the initial blood culture result looks similar to that of ordinary K. pneumoniae bacteremia.
The study used a retrospective single-center design at Kameda Medical Center in Kamogawa, Japan, and included consecutive patients with one blood culture set positive for the K. pneumoniae species complex from January 1, 2022, through December 31, 2024. Investigators excluded episodes involving K. variicola and K. quasipneumoniae after species-level PCR differentiation, leaving 207 K. pneumoniae isolates for analysis. Hypervirulence classification was assigned retrospectively and was not available to treating clinicians during routine care, which is important because the care team was not steering management with genomic knowledge in real time. That design makes the outcome differences more credible as a reflection of biology rather than clinician expectation.
Hospital course length reinforced the same signal. Median post-onset hospitalization was 28 days for hypervirulent episodes versus 14 days for classical episodes, and total antimicrobial treatment duration reached 43 days versus 14 days. Those figures are not abstract epidemiologic outputs; they reflect source control, imaging follow-up, and the prolonged treatment courses that abscess disease often demands. When a bloodstream infection seeds a liver or other deep focus, discharge usually waits for radiologic improvement, drainage decisions, and clinical stabilization rather than for clearance of fever alone.
A sharp lineage signal appeared within the hypervirulent subset. Most of the sequence type 23 hypervirulent episodes, 9 of 10, involved abscesses. ST23 has long been tied to invasive K. pneumoniae disease, and this cohort again placed that lineage in the center of the abscess phenotype. Genomic context mattered here because lineage was not just a label; it tracked with a more invasive clinical course.
Resistance convergence was present but still uncommon. Short-read whole-genome sequencing found CTX-M–type extended-spectrum β-lactamase genes in 3 of 28 hypervirulent isolates, representing 11% of the hypervirulent group and 1.4% of all 207 episodes. Even that small fraction matters clinically because it shows hypervirulence and antimicrobial resistance are no longer mutually exclusive categories. When virulence plasmids and β-lactam resistance genes begin to coexist in the same strain, the result is a pathogen that is harder to treat and more likely to spread into tissue beyond the bloodstream.
The sequencing work also showed why a simple phenotypic impression can miss the full picture. Investigators performed multilocus sequence typing, capsular locus typing, virulence gene profiling, and resistance determinant analysis on a subset enriched for hypervirulent and resistant isolates. They also evaluated the genomic context of ESBL-positive hypervirulent isolates to examine how virulence and resistance were organized on the bacterial genome. That approach is especially useful for surveillance because it distinguishes a true hypervirulent lineage from a more ordinary isolate that happens to share one or two virulence-associated markers.
The abscess endpoint was defined conservatively, based only on imaging-confirmed findings, which reduces the chance of overcalling disease severity. Metastatic infection included anatomically distinct secondary foci such as abscesses and nonsuppurative manifestations like endophthalmitis. Using those definitions, the five-biomarker assay separated a group with a clinically heavier burden of invasive disease. It also separated a group that needed longer antibiotic exposure, which has immediate implications for stewardship and for planning outpatient therapy.
The sample size was modest, but the clinical separation was large enough to be hard to ignore. Hypervirulent K. pneumoniae accounted for 14% of episodes in this Japanese tertiary hospital cohort, with a 95% confidence interval of 9%–19%. The prevalence, therefore, was not trivial, and the five-biomarker definition produced a cohort that looked epidemiologically and clinically distinct from classical bacteremia. The finding strengthens the case for genomic surveillance rather than relying only on conventional susceptibility patterns or crude clinical impressions.
What this study cannot do is settle whether every hospital should adopt immediate five-marker testing for all K. pneumoniae bacteremias. It does show that the genomic definition identifies a subgroup with more abscesses, longer hospitalization, and longer treatment, and it reveals that resistance genes are already entering that space. For clinicians, the practical consequence is straightforward: a bloodstream isolate carrying the hypervirulence signature should trigger a lower threshold for imaging, source evaluation, and close follow-up for metastatic infection.