Introduction

Classical Klebsiella pneumoniae causes various infections, such as pneumonia, urinary tract infections, and bacteremia, commonly in hosts with comorbidities1,2. However, in recent decades, there have been several reports, primarily from Taiwan, of cases of community-acquired bloodstream infections (BSIs) with liver abscess caused by hypervirulent K. pneumoniae (hvKp) in healthy individuals1,2,3, and the spread of the bacterium is a major concern worldwide. Unlike classical K. pneumoniae, hvKp causes infections, including meningitis, necrotizing fasciitis, endophthalmitis, and metastatic infections in multiple organs4.

Four major virulence factors of K. pneumoniae (capsule, lipopolysaccharide, fimbriae, and siderophores) have been reported5. Capsules protect the bacterial cells from phagocytosis and antimicrobial peptides and suppress host immunological responses6,7,8,9. HvKp produces a hypercapsule, which consists of a mucoviscous extracellular polysaccharide; it envelopes the bacterial surface more robustly than a typical capsule. Specific capsular types such as K1 and K2 are associated with increased hvKp pathogenicity5.

Plasmid-borne regulator of mucoid phenotype A (rmpA) is a transcriptional regulator and enhances capsular polysaccharide synthesis and capsule production10,11. Previous studies have shown that the deletion of rmpA reduces colony mucoviscosity10,11, virulence in mice, and resistance to human serum10. Additionally, plasmid-borne rmpA has been found to be an accurate marker of hvKp with high sensitivity (0.98) and specificity (0.93)4. Therefore, in this study, we focused on K. pneumoniae harboring plasmid-borne rmpA and investigated the molecular epidemiology and clinical features of BSI caused by the bacterium in our university hospital located in western Japan.

Materials and methods

Study design

We retrospectively investigated K. pneumoniae isolated from blood samples at Nagasaki University Hospital from January 2010 to March 2021. Adult patients aged 20 years or older, from whose blood samples, K. pneumoniae was isolated, were listed from our clinical laboratory database. The first isolate was selected when the bacteria were repeatedly isolated from individual patients during the study12. Of the isolates listed, those that were available were included in this study. We collected clinical and microbiological information obtained through routine practice from the medical records and laboratory systems in our hospital. Among patients for whom the isolates were available, a matched case–control study in a 1:3 ratio was conducted to clarify the clinical features of BSI caused by rmpA-positive K. pneumoniae and the characteristics of the bacterium. Cases and controls were defined as patients from whom rmpA-positive and rmpA-negative K. pneumoniae were isolated, respectively. For case–control matching, age (± 5 years) and sex-matched patients to each case were listed among patients from whom rmpA-negative K. pneumoniae was isolated, and three patients per case were randomly selected as controls using Microsoft Excel (Microsoft Corporation). The clinical characteristics of the patients with BSI caused by K. pneumoniae were compared between the rmpA-positive and rmpA-negative groups. We evaluated the infection sites with bloodstream infection, such as pneumonia, biliary tract infection, and urinary tract infection, from which K. pneumoniae was isolated from each site. Other infection sites, including liver abscess, endophthalmitis, meningitis, and purulent spondylitis, were evaluated regardless of K. pneumoniae isolation from each site. The severity of BSI was assessed using the Pitt bacteremia score13,14. The study was performed in accordance with tenets of the Declaration of Helsinki and the Ethical Guidelines for Medical and Biological Research Involving Human Subjects. The study protocol including the waiver of consent was approved by the Institutional Review Board of Nagasaki University Hospital (approval number: 21071208).

Microbiological analysis

Hypermucoviscosity was assessed using the string test, which was considered positive if the viscous string was greater than 5 mm in length when the colony was stretched using a loop on an agar plate15. Bacterial DNA was extracted using the boiling method previously described12, with minor modifications. Three to five colonies were mixed with 100 µL Tris–EDTA buffer containing 250 U/mL achromopeptidase (Wako Pure Chemical Industries, Ltd.). After incubation at 40 °C for 15 min, 250 µL of 10% Chelex 100 Resin (Bio-Rad) was added, and the mixture was boiled at 99 °C for 5 min, cooled on ice for 1 min, and centrifuged at 12,000 rpm for 1 min. The supernatant was used for the subsequent analyses.

In this study, plasmid-borne rmpA, iucA, peg-344, and iroB, which have been reported to be accurate makers of hvKp4, as well as capsular types, including K1 (magA), K2, and K5, were evaluated using PCR. The PCR primers used were as follows: rmpA forward, 5′-ACTGGGCTACCTCTGCTTCA-3′; rmpA reverse, 5′-CTTGCATGAGCCATCTTTCA-3′16,17; iucA forward, 5′-AATCAATGGCTATTCCCGCTG-3′; iucA reverse, 5′-CGCTTCACTTCTTTCACTGACAGG-3′18; K1 (magA) forward, 5′-GGTGCTCTTTACATCATTGC-3′; K1 (magA) reverse, 5′-GCAATGGCCATTTGCGTTAG-3′15; K2 forward, 5′-GACCCGATATTCATACTTGACAGAG-3′; K2 reverse, 5′-CCTGAAGTAAAATCGTAAATAGATGGC-3′19. peg-344 forward, 5′-CTTGAAACTATCCCTCCAGTC-3′; peg-344 reverse, 5′-CCAGCGAAAGAATAACCCC-3′4; iroB forward, 5′-ATCTCATCATCTACCCTCCGCTC-3′; iroB reverse, 5′-GGTTCGCCGTCGTTTTCAA-3′4; K5 forward, 5′-TGGTAGTGATGCTCGCGA-3′; K5 reverse, 5′-CCTGAACCCACCCCAATC-3′19.

DNA was amplified under the following conditions: 5 min at 94 °C, 35 cycles of 30 s at 94 °C, 30 s at the annealing temperature [46 °C for rmpA, 50 °C for iucA, K1 (magA), and K2], and 1 min at 72 °C, and 7 min at 72 °C for the final extension; 10 min at 95 °C, 35 cycles of 30 s at 95 °C, 30 s at the annealing temperature (53 °C for peg-344 and 59 °C for iroB), and 40 s for peg-344 and 30 s for iroB at 72 °C, and 7 min at 72 °C for the final extension; for K5, 1 min at 94 °C, 30 cycles of 30 s at 94 °C, 45 s at 59 °C, and 90 s at 72 °C, and 6 min at 72 °C for the final extension.

Antimicrobial susceptibility was examined using BD Phoenix M50 (Becton Dickinson), according to the manufacturer’s instructions, and determined according to the Clinical and Laboratory Standards Institute (CLSI) M100-Ed33.

Multilocus sequence typing (MLST) was carried out for rmpA-positive isolates, based on the sequences of seven housekeeping genes (gapA, infB, mdh, pgi, phoE, rpoB, and tonB). The primers used have been described in the Klebsiella pneumoniae MLST database (https://bigsdb.pasteur.fr/klebsiella/primers-used/). Direct sequencing was performed as follows. DNA was amplified using primers for each housekeeping gene under the following conditions: 2 min at 94 °C, 35 cycles of 30 s at 94 °C, 1 min at 50 °C, and 30 s at 72 °C, and 5 min at 72 °C for the final extension. The products were purified using a QIA quick PCR purification kit (QIAGEN) or ExoSAP-IT (Applied Biosystems). Fluorescence-based cycle sequencing reactions were performed using the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems). After purification using the BigDye Xterminator Purification Kit (Applied Biosystems), the products were analyzed using the SeqStudio Genetic Analyzer (Applied Biosystems). Allele sequences and STs were determined according to the Klebsiella pneumoniae MLST database (https://bigsdb.pasteur.fr/klebsiella/).

Statistical analysis

Numerical variables are expressed as median (interquartile range) and compared using Wilcoxon rank-sum test between groups. Categorical variables were compared using Fisher’s exact test. In the multivariate analysis, variables with P < 0.2 in the univariate analysis were selected and adjusted using the conditional logistic regression model. Data were analyzed using JMP v16 (SAS Institute Inc.), and results with P < 0.05 were considered statistically significant.

Results

Microbiological characteristics of K. pneumoniae harboring rmpA

Of the 306 K. pneumoniae isolated from the blood of individual patients, 268 were available. Of these 268 isolates, rmpA was detected in 36 isolates (13.4%). Of the remaining 232 isolates without rmpA, 108 were matched as rmpA-negative controls based on the age (± 5 years) and sex of the patients (Fig. 1).

Figure 1
figure 1

Study design depicting adult patients with bloodstream infection (BSI) and bacterial isolates.

Full size image

Table 1 shows the characteristics of rmpA-positive and rmpA-negative isolates. Of the 36 rmpA-positive isolates, 31 (86.1%) harbored iucA and 35 (97.2%) each possessed peg-344 and iroB. Capsular types were identified as K1 in 9 (25.0%), K2 in 10 (27.8%), and K5 in 1 (2.8%) isolates, respectively. Among the 108 rmpA-negative isolates, 5 (4.6%) harbored iucA and 1 (0.9%) each possessed peg-344 and iroB; 2 (1.9%), 3 (2.8%), and 1 (0.9%) isolates had K1, K2, and K5 capsular types, respectively. Hyperviscosity was found in 30 rmpA-positive isolates (83.3%), which was higher than that in rmpA-negative isolates (four isolates, 3.7%).

Table 1 Microbiological characteristics of rmpA-positive and rmpA-negative K. pneumoniae isolates from patients with bloodstream infections.
Full size table

Table 2 shows the antimicrobial susceptibility of 36 rmpA-positive isolates. No carbapenem-resistant isolates and three (8.3%) extended-spectrum β-lactamase (ESBL) producers were identified. Two isolates (5.6%) were resistant to ciprofloxacin.

Table 2 Antimicrobial susceptibility of 36 rmpA-positive K. pneumoniae isolates from patients with bloodstream infections.
Full size table

Table 3 presents the relationship between MLST and capsular types of the 36 rmpA-positive isolates. ST23/K1 (eight isolates) was the most frequent ST/capsular type, followed by ST412/non-K1/K2 (seven isolates), ST86/K2 (five isolates), and ST268/non-K1/K2 (four isolates).

Table 3 Relationship between MLST and capsular type of 36 rmpA-positive K. pneumoniae isolates from patients with bloodstream infections.
Full size table

Clinical features of BSI caused by K. pneumoniae harboring rmpA

We investigated the baseline characteristics and clinical features of BSI caused by rmpA-positive K. pneumoniae, compared with those caused by rmpA-negative isolates (Table 4). Of the 144 patients analyzed, 91 (63.2%) developed K. pneumoniae BSI in the hospital, and the rates were similar between the rmpA-positive and rmpA-negative groups (63.9% and 63.0%, respectively). The use of anticancer drugs was significantly higher in the rmpA-negative group than in the rmpA-positive group. Similarly, the presence of malignancy tended to be higher in the rmpA-negative group than in the rmpA-positive group, but the difference was not significant. Other comorbidities and use of medical devices did not differ between the groups.

Table 4 Clinical characteristics of bloodstream infections caused by rmpA-positive and rmpA-negative K. pneumoniae.
Full size table

The biliary tract was the most frequent infection site with BSI (30 patients, 27.8%) in the rmpA-negative group, and the rate was higher than that in the rmpA-positive group (three patients, 8.3%). Conversely, liver abscess was a more frequent infection in the rmpA-positive group (eight patients, 22.2%) than in the rmpA-negative group (eight patients, 7.4%). Disease severity assessed using the Pitt bacteremia score was similar, and the mortality rates did not show significant differences between the groups.

Conditional regression analysis was performed to evaluate the correlation between rmpA-positive isolates and the clinical factors of patients with K. pneumoniae BSI. Variables with P < 0.2 in the univariate analysis (Table 4) were used for the analysis. The presence of liver abscess positively correlated with rmpA-positive isolates, whereas biliary tract infection and the use of anticancer drugs showed a negative correlation with rmpA-positive isolates in patients with K. pneumoniae BSI (Table 5).

Table 5 Correlation between rmpA-positive isolates and clinical factors in K. pneumoniae bloodstream infections.
Full size table

Discussion

Our study demonstrated the molecular epidemiology of K. pneumoniae harboring rmpA and the clinical features of BSI caused by the bacterium in our university hospital. Of the 268 K. pneumoniae isolates from blood, rmpA was detected in 13.4%. After case–control matching (rmpA-positive, 36 isolates; rmpA-negative, 108 isolates), the positive rates of iucA, peg-344, and iroB were remarkably higher in the rmpA-positive group (86.1%, 97.2%, and 97.2%, respectively) than in the rmpA-negative group (4.6%, 0.9%, and 0.9%, respectively). In addition to rmpA, iucA, peg-344, and iroB have been reported to be accurate markers of hvKp4. The high detection rates of these markers in rmpA-positive isolates support that rmpA is a useful marker of hvKp. Furthermore, K1 and K2 capsular types were identified in 25.0% and 27.8% of the isolates, respectively, in the rmpA-positive group, which were clearly higher than those in the rmpA-negative group (1.9% and 2.8%, respectively). The STs of K. pneumoniae from patients with BSI vary geographically20. Our results showed that ST23/K1 was the most prevalent (eight of 36 isolates) in rmpA-positive K. pneumoniae causing BSI, which is supported by the findings of a previous study on hvKp from Japan21. Additionally, we identified ST65/K2 and ST86/K2, similar to that in a previous study in Japan21.

This study showed the clinical characteristics of BSI caused by K. pneumoniae harboring rmpA. Liver abscess was recorded in 22.2% of the patients with BSI in the rmpA-positive group, three times more frequently than that in the rmpA-negative group (7.4%). The multivariate analysis showed that liver abscess significantly correlated with rmpA-positive isolates (odds ratio, 8.728). In addition, all eight rmpA-positive K. pneumoniae isolates causing liver abscess showed hyperviscosity and carried iucA. These results are supported by a recent report that rmpA, positive string test, and aerobactin are associated with K. pneumoniae causing liver abscess in patients with community-acquired BSI22. Furthermore, ST23/K1 (three isolates) and ST65/K2 (two isolates) were identified in five (62.5%) of the eight rmpA-positive K. pneumoniae isolates that caused liver abscess in our study, which is consistent with previous reports that they are the common ST/capsular types associated with liver abscess in East Asian countries23,24,25. The remaining types that caused liver abscess were ST412/non-K1/K2 (two isolates) and ST268/non-K1/K2 (one isolate).

Meanwhile, the use of anticancer drugs and the presence of biliary tract infection negatively correlated with rmpA-positive isolates. Classical K. pneumoniae is known to cause bacteremia especially in immunocompromised patients2. Therefore, the use of anticancer drugs may reflect the immunocompromised condition of the host. Additionally, biliary tract was a frequent infection site in the rmpA-negative group (27.8%) compared with that in the rmpA-positive group (8.3%) in this study. A recent study reported a similar result that biliary tract infection was observed more frequently in classical K. pneumoniae BSI26.

This study has a few limitations. First, as this was a retrospective study, some variables of clinical factors might not have been recorded by attending physicians. Second, the sample size was limited because this study was conducted in a single center, and some isolates were unavailable during the study period. Finally, because we focused on rmpA-positive isolates in this study, we could not analyze the microbiological characteristics of rmpA-negative isolates in detail.

In conclusion, our study revealed the molecular epidemiology of K. pneumoniae harboring rmpA, isolated from patients with BSI in our hospital. The presence of rmpA correlated with the clinical characteristics of K. pneumoniae BSI and can be used as a marker for understanding the pathophysiology of K. pneumoniae BSI.