Abstract
Hypervirulent Klebsiella pneumoniae (hvKP) can cause severe invasive infections in healthy and immunocompromised individuals. It is an important clinical pathogen, with the recent identification of highly invasive community-acquired K. pneumoniae that produce purulent liver abscess. Herein, the frequency, clinical characteristics, and molecular epidemiology of K. pneumoniae isolates obtained from January to October 2016 in Wenzhou, China, were examined. In the present study, 33 isolates (68.8%, 33/48) had a positive string test and were identified as hvKP. Both age and sex were associated with a positive string test (p < 0.05). Univariate analysis revealed that liver abscess (odds ratio = 10.154) was a significant risk factor for hvKP. Among the hvKP isolates, K1 was the most common capsular serotype, followed by K2 (p < 0.05). The prevalence of K1 and K2 was significantly higher in hvKP than non-hvKP isolates. The rates of virulence-associated genes, rmpA, iroB, fib, and hib, were significantly higher for hvKP than for non-hvKP (p < 0.05). In this study, kfuB, ybtA, and wcaG were associated with K1 isolates. ST23 was the predominant hvKP type, which belonged to serotype K1. Pulsed-field gel electrophoresis showed that two clusters had >75% similarity and each accounted for >3 isolates. The homology of the 48 clinical isolates was diverse. In conclusion, hvKP isolates had a high frequency of virulence factors and a wide variety of homologs. These results suggest that, because of the toxicity associated with K. pneumoniae, increased clinical understanding of the disease is important to prevent larger, severe outbreaks.
Introduction
Klebsiella pneumoniae, a major opportunistic Gram-negative bacteria, is responsible for hospital- and community-acquired infections among immunocompromised hosts.1,2 Over the past two decades, hypervirulent K. pneumoniae (hvKP) with hypermucoviscosity has emerged, causing serious invasive infections. As a clinically significant pathogen, it is unlike the “classic” K. pneumoniae (cKP) found in both healthy and immunocompromised individuals. 3
The hvKP associated with pyogenic liver abscesses has the ability to cause serious community-acquired infections. In Taiwan, 88.8% of K. pneumoniae isolates from patients with pyogenic liver abscesses had the hypermucoviscosity phenotype. 4 Life-threatening infections caused by hvKP were subsequently reported in many southeast Asian countries, including Korea,5,6 Hong Kong,7,8 and Singapore, 9 with an increasing number of cases reported from America, Europe, Africa, and the Caribbean, indicating that this unique syndrome has become a globally emerging disease. 10
These hvKP strains efficiently overproduce a polysaccharide capsule, which is hypermucoviscous and can be detected by the string test. A positive string test is defined as a viscous string of >5 mm in length produced when a bacterial colony is touched with an inoculation loop. 11
Unlike hvKP strains, most non-hvKP strains are rarely resistant to commonly used antimicrobial agents, with the exception of ampicillin. 12 Recently, Gu et al. reported that a fatal outbreak of ventilator-associated pneumonia was caused by a prevalent ST11 type carbapenem-resistant hvKP. 13 In China, the prevalence of hvKP, carbapenem-resistant K. pneumoniae, and carbapenem-resistant hvKP is largely unknown. 14
In this study, the prevalence and molecular characteristics of community-acquired hvKP were investigated using isolates from Wenzhou, China.
Materials and Methods
Collection and identification of K. pneumoniae
A total of 48 K. pneumoniae were isolated consecutively from the blood of outpatients diagnosed with K. pneumoniae bloodstream infections at the first Affiliated Hospital of Wenzhou Medical University located in Wenzhou, China, from January to October 2016. K. pneumoniae identification was performed by colony morphology, Gram-staining, and using a VITEK-2 automated platform (bioMérieux, Marcy l'Etoile, France). Escherichia coli ATCC25922 and Pseudomonas aeruginosa ATCC87253 were used as control strains for the identification of bacterial clinical isolates. Patient information was extracted from medical records and informed consent was obtained from patients participating in the current study. The study was approved by the Ethics Committee of the First Affiliated Hospital of Wenzhou Medical University.
Antimicrobial susceptibility testing
Susceptibility of the isolates to antimicrobial agents was examined using the VITEK-2 automated platform (bioMérieux) in accordance with the manufacturer's instructions. E. coli ATCC25922 was used for antimicrobial susceptibility quality control.
Hypermucoviscosity phenotype test
The string test was used to detect the hypermucoviscosity phenotype, as described previously. 11 K. pneumoniae isolates were subcultured on Columbia blood agar overnight at 37°C. Single colonies of K. pneumoniae isolates were subjected to the string test. This involved placing an inoculation loop (BIO-KONT, Wenzhou, China) into the suspension to observe whether a string was formed when the loop was drawn slowly away from the suspension. Formation of a mucoviscous string >5 mm was defined as a positive test.
Capsular polysaccharide serotype genotyping and detection of virulence-associated features
Capsular serotypes were identified by polymerase chain reaction (PCR), as described previously. 15 The presence of genes encoding virulence factors (rmpA, wabG, entB, kfuBC, wcaG, ybtA, uge, iutA, ironB, urea, ycf, alls, Aer, and magA) was assessed by PCR using documented primers. 16 The first strain of the virulence-related gene identified by PCR and identical to the published sequence was selected as a positive control for subsequent PCR analysis.
Multilocus sequence typing and pulsed-field gel electrophoresis
Multilocus sequence typing (MLST) was performed for the K. pneumoniae isolates by amplifying and sequencing seven housekeeping genes (gapA, infB, mdh, pgi, phoE, rpoB, and tonB) according to the protocol described on the K. pneumoniae MLST website. 17 Genomic DNA from all clinical isolates was digested with the restriction enzyme, XbaI, and subjected to electrophoresis at 14°C for 20 hr, at an angle of 120°, with switching times of 6 and 36 sec at 6 V/cm, using a Bio-Rad CHEF III system (CHEF MAPPER XA; Bio-Rad). Pulsed-field gel electrophoresis (PFGE) patterns were compared with Bionumerics software (AppliedMaths, Sint-Martens-Latem, Belgium) using the Dice Similarity coefficient. The threshold for clustering PFGE patterns was >75% similarity.
Detection of extended-spectrum beta-lactamases and K. pneumoniae carbapenemase enzymes
Extended-spectrum beta-lactamases (ESBLs) and carbapenemases were screened for and detected according to the recommendations of the Clinical and Laboratory Standards Institute (CLSI). 18 A suspension of E. coli ATCC 25922 (indicator bacteria) was adjusted to 0.5 McFarland standard with sterile saline, then diluted 10-fold with saline. The diluted bacterial suspension was evenly spread on a Mueller-Hinton agar plate. The plate was allowed to dry for 3–10 min. An ertapenem or meropenem tablet (10 μg/tablet) was placed in the center of the plate. Using a 1 μL inoculation loop to pick up colonies, the bacteria or quality control strains were cultured on blood agar plates overnight. This involved inoculating the plates in a straight line out from the edge; the streak was ≥20–25 mm in length. Each plate was incubated at 35°C for 16–20 hr.
Statistical analysis
Statistical analysis was performed using GraphPad Prism 6 software (version 6.00, La Jolla, CA). We used chi-square or Fisher's exact tests for categorical variables. p Values <0.05 (two-tailed) were considered statistically significant. Multivariate logistic regression was used to identify variables associated with hvKP infections. All variables with p < 0.1 in the univariate analyses were entered into the multivariate model. Liver abscess, cancer, and hospitalization within the last 90 days were used for the logistic regression model.
Results
Clinical characterization of hvKP bloodstream isolates
From January to October 2016, a total of 48 clinical isolates of K. pneumoniae were obtained from blood samples. The characteristics of patients with hvKP and non-hvKP infections are shown in Table 1. The proportions of male and female patients were 72.9% (35/48) and 27.1% (13/48), respectively. Among the 48 clinical isolates, 33 (68.8%) had positive string test results and were identified as hvKP. Both age and sex were associated with positive string tests (both p < 0.05). Among the 48 isolates of community-acquired K. pneumoniae, the incidence of liver abscess in patients with hvKP infections was 33.3% (11/33). In patients with cancer, the incidence of hvKP infections was 6.1% (2/33).
Microbiological and Clinical Characteristics of Hypervirulent Klebsiella pneumoniae Isolates
hvKP, hypervirulent Klebsiella pneumoniae.
Risk factors for hvKP and non-hvKP
Univariate analysis demonstrated liver abscess (odds ratio [OR] = 0.154) to be a statistically significant risk factor for hvKP infection. Cancer (OR = 0.078) and hospitalization within the last 90 days (OR = 0.692) appeared to be protective factors for hvKP. Multivariate regression analysis showed that liver abscess, cancer, and hospitalization within the last 90 days were not independent factors (Table 2).
Variables Associated with Hypervirulent Klebsiella pneumoniae Infections in Regression Analyses
CI, confidence interval; OR, odds ratio.
Antimicrobial resistance among hvKP isolates
Table 3 summarizes the antimicrobial resistance results for hvKP and non-hvKP isolates. For the tested antimicrobial agents, the resistance rate was significantly lower among hvKP isolates compared with non-hvKP isolates (p < 0.05), including for piperacillin/tazobactam, cefotetan, cefoperazone, imipenem, tobramycin, and amikacin. All hvKP and non-hvKP isolates were resistant to ampicillin. Of the 48 isolates of K. pneumoniae, 1 produced carbapenemase and 6 produced ESBL, comprising 1 hvKP isolate and 5 non-hvKP isolates. The carbapenemase-producing isolates, which contained the NDM-1 resistance gene, was resistant to all antibacterial agents, except aztreonam, amikacin, and nitrofurantoin. Production of ESBL by hvKP isolates (6.1%, 2/33) was significantly lower than among non-hvKP isolates (33.3%, 5/15), (p < 0.05).
Antimicrobial Resistance Profiles of 33 Hypervirulent Klebsiella pneumoniae and 15 Non-Hypervirulent Klebsiella pneumoniae Isolates
ESBL, extended-spectrum beta-lactamase.
Molecular characteristics and virulence-associated genes of hvKP and non-hvKP isolates
MLST analysis of the 48 community-acquired K. pneumoniae isolates identified ST23 (29%, 14/48), and ST11, ST29, ST65, ST25, ST86, and ST828 (4%, 2/48 for each). The remaining isolates had unknown STs (Table 4). ST23 was a major epidemic strain among the hvKP isolates (39.4%, 13/33) compared with with non-hvKP isolates (6.7%, 1/15) (p < 0.05). Capsular serotype K1 (31.3%, 15/48), K2 (16.7%, 8/48), wzi64 (8.3%, 4/48), wzi323 (4.2%, 2/48), wzi5 (4.2%, 2/48), wzi50 (4.2%, 2/48), wzi115 (4.2%, 2/48), and other unknown genotypes (27%, 2/48) were identified. K1 and K2 were significantly more frequent among hvKP than among non-hvKP isolates, whereas non-K1 non-K2 serotype was the opposite (p < 0.05) (Table 4). PFGE DNA fingerprinting showed that only two PFGE clusters with >75% similarity accounted for more than three isolates each (Fig. 1). Cluster A accounted for five isolates comprising two ST23-K1, one ST29-K2, one ST2715-wzi323, and one K-non-typable isolate. Cluster B accounted for four isolates comprising two ST23-K1, one ST23-K1, and one ST65-K2. Among the 48 clinical isolates tested, the positivity rates for virulence-associated genes were >90%, including for wabs (97.9%), ebtb (97.9%), uge (91.7%), ureA (100%), and ycf (100%). The positivity rates of virulence-associated genes among hvKP isolates were as follows: rmpA (90.9%, 30/48), iroB (72.7%, 24/48), FIB (87.9%, 29/48), and HIB (66.7%, 22/48), which were significantly higher than for non-hvKP isolates (p < 0.05).

PFGE results for 48 community-acquired Klebsiella pneumoniae isolates. (
Microbiological Characteristics of 33 Hypervirulent Klebsiella pneumoniae and 15 Non-Hypervirulent Klebsiella pneumoniae Isolates
iutArmpA*iutArmpA2, containing iutArmpA or iutArmpA2.
Discussion
K. pneumoniae is a conditionally pathogenic, Gram-negative bacillus that is a major pathogen associated with both hospital- and community-acquired disease. It infects various parts of the human body, with lung infections being especially prevalent. Compared with classic K. pneumoniae bloodstream infections, hvKP bloodstream infections are mainly community acquired, demonstrating that hvKP isolates play an important role in community-acquired disease. Previous studies demonstrated K. pneumoniae liver abscesses to be associated with community-acquired infections. 11 A significant factor in the K. pneumoniae liver abscess invasion syndrome is the hypermucoviscous phenotype.19,20 Neither age nor sex was associated with hvKP. 21 However, our study showed that patients infected with hvkp were younger and more male than those infected with non-hvkp. Patients with hvKP infections were more likely to develop liver abscess (33.3% vs. 6.7%, p = 0.002), whereas non-hvKP isolates were associated with cancer (46.7% vs. 6.1%, p = 0.002). Univariate analysis indicated that liver abscess (OR = 10.154) was positively associated with hvKP infection (p < 0.05). Univariate analysis also indicated that cancer (OR = 0.078) was negatively associated with hvKP infection (p < 0.05). However, there was no significant association in the multivariate analysis.
Carbapenems, such as imipenem, are a class of highly efficient and broad-spectrum beta-lactam antibiotics, which are considered the most effective drugs for the treatment of Gram-negative bacilli infections. Hence, these antibiotics have become important antimicrobial agents in the treatment of multidrug-resistant K. pneumoniae infections. In recent years, with the emergence of carbapenem-resistant K. pneumoniae, it has become increasingly difficult to control these infections. 22 In the present study, of six beta-lactamase-producing isolates, only one was highly virulent. We speculate that beta-lactamase-resistant carbapenems were more prevalent in non-hvKP strains. In this study, only one of the seven multidrug-resistant clinical isolates was an hvKP isolate, which is consistent with the conclusion that high virulence and antibiotic resistance in K. pneumoniae do not largely overlap. 23
Serotype K1 and K2 strains have been identified as the most virulent and have been associated with pyogenic liver abscesses, among known K. pneumoniae serotypes. 24 This may be due to the robust resistance to phagocytosis exhibited by the highly virulent serotypes K1 and K2, providing favorable conditions for bacterial colonization. 25 In this study, the prevalence of hvKP in 23 isolates of serotype K1 and K2 was 87% (20/23). The study also showed that K1 was the most common capsule serotype among hvKP isolates (42.4%, 14/33), followed by K2 (32%, 8/33), which is consistent with the literature. 26 Yu et al. reported that serotype K1 has strong associations with kfuB and alls, 27 and we found that wcaG, kfuB, and ybtA were associated with K1 isolates. The wcaG gene increases bacterial escape from macrophage phagocytosis by fucose inclusion in envelope synthesis. 28 Kim et al. identified aerobactin and kfu as essential for bacterial virulence. Their protein products are associated with iron acquisition. 29 In this study, the prevalence of kfu and aerobactin in hvKP isolates was only 39% (11/33) and 24% (8/33), respectively. We speculate that this may be due to the small sample size studied. A previous study reported that the protein-encoding product of ybt, which had a high prevalence in serotype K1 strain, was also a virulence factor in K. pneumoniae pulmonary infection. 30 This is consistent with the high prevalence of ybt among serotype K1 isolates (75.8%, 25/33) in this study. The rmpA gene was a plasmid gene that regulated the synthesis of extracellular polysaccharides and associated with a high-viscosity phenotype. 31 In this study, the rmpA gene was found in 25/33 hypermucoviscous isolates (75.8%), which was considerably higher than in a report from Beijing, China, 32 in which 33.3% (5/15) of isolates were nonhypermucoviscous. These results suggest that other regulatory mechanisms for the hypermucoviscous phenotype may exist.
Previous studies have shown that st23 is the main prevalent strain of hvkp isolates in China, and has strong correlation with serotype K1 and liver abscess.11,26 In this study, all hypervirulent K1 isolates belonged to ST23, which was included in the clonal complex (CC) CC23K1 described by Brisse. 33 CCs are defined as groups of two or more independent isolates that share identical alleles at six or more loci. 34 Previous research has shown that ST57 and ST82 are related to the K1 serotype and to pyogenic liver abscesses, 35 but these were not found in this study. Serotype K2 isolates differed from serotype K1 isolates and belonged to ST25, ST29, ST65, ST86, and ST375, reflecting high genetic diversity. Almost ST25 and ST65 were identified as virulence clones of serotype K2. 16 In recent years, ST25, ST65, and ST86 have been recognized as hypervirulent strains together with K1-ST23. 36 In this study, almost all serotypes, K1 and K2, belonged to these STs, indicating that serotypes K1 and K2 have hypervirulence in the community. Prevention of their widespread prevalence in the community is an important objective.
In conclusion, this study reports on the prevalence of community-acquired bloodstream K. pneumoniae infections in the Wenzhou area. HvKP isolates from blood infections were more frequently associated with concurrent bacteremia than isolates without the hypermucoviscosity phenotype. In community-acquired K. pneumoniae infections, the frequency of high-virulence isolates was higher, and the types and quantities of drug-resistant isolates were considerably higher, than in previous studies. This requires implementation of effective monitoring and strict infection control strategies to prevent the spread of multidrug-resistant hvKP.
Footnotes
Acknowledgment
This study was supported by grants from the National Natural Science Foundation of China (81672078).
Disclosure Statement
No competing financial interests exist.
