Impacts of Hypervirulence Determinants on Clinical Features and Outcomes of Bacteremia Caused by Extended-Spectrum β-Lactamase-Producing Klebsiella pneumoniae

Abstract

We investigated the implications of hypervirulence determinants on clinical features of 48 adult patients with bacteremia caused by extended-spectrum β-lactamase-producing Klebsiella pneumoniae. Isolates in the hypervirulence group included any of the following virulence determinants: K1/K2 capsule serotypes, hypermucoviscosity phenotype, rmpA gene, or rmpA2 gene. Nonhypervirulence group isolates were negative for all of the above virulence factors. In this study, all isolates used were non-K1/K2 strains. Statistically significant differences were observed in clinical features of patients between the two groups. The hypervirulent isolates (n = 19), including 11 isolates with the hypermucoviscosity phenotype, 15 with the rmpA gene, and 16 with the rmpA2 gene, were more commonly recovered from diabetic patients and mainly manifested as secondary bacteremia (such as pneumonia, urinary tract infections, or other localized infections). The nonhypervirulent isolates (n = 29) were more commonly recovered from patients after prolonged hospital stays (>30 days) and mostly manifested as primary bacteremia. The overall in-hospital mortality was 56.3%. Hazard ratio (HR) analysis revealed the following positive predictors for mortality: nosocomial infection, stay in an intensive care unit, no removal of the central venous catheter, Charlson comorbidity score, and APACHE II score (≧15). The negative predictors were initial appropriate antibiotic therapy (HR 0.42) and urinary tract infection (HR 0.19). Charlson score was an independent confounder based on multivariate analysis (HR 1.43, 95% confidence interval 1.04–1.99). In conclusion, hypervirulence determinants played a role in causing secondary infections in diabetic patients; however, the presence of morbidity cofactors could themselves influence mortality, despite the absence of hypervirulence determinants.

Introduction

K lebsiella pneumoniae (KP) is a major cause of community-acquired infections, including septicemia, pneumonia, urinary tract infections, tissue abscesses, and, in particular, liver abscesses.^1,2 Infection-related mortality occurs in 34.5% of intensive care unit patients with community-onset KP bacteremia.³ Extended-spectrum β-lactamase-producing K. pneumoniae (ESBL-KP) are rarely isolated from liver abscesses.^4,5 However, the mortality rate for patients with ESBL-KP bacteremia is between 55.8% and 63.2%, which is higher than the rate for patients with non-ESBL-KP bacteremia.^6,7

Several hypervirulence determinants in KP have been discovered, of which K1 or K2 capsular serotype, hypermucoviscosity phenotype, and the presence of the plasmid-borne regulator of mucoid phenotype A (rmpA) gene or the transcriptional activator (rmpA2) gene are associated with invasive syndrome and pyogenic abscess formation.^8–15 KP isolates with the rmpA gene and hypermucoviscosity phenotype are significantly correlated with high serum resistance and tissue abscess formation,^1,8,13 virulent enough to cause high mouse lethality.¹¹ They are also a frequent cause of community-acquired bacteremic pneumonia in Taiwan.¹⁶

In one small study in China, ESBL-KP isolates were identified in 5 (17.2%) of 29 clinical isolates of hypermucoviscous KP.¹⁷ In addition, we recently reported that ESBL-KP isolates are less hypermucoviscous and less virulent than non-ESBL-KP isolates, 22.7 versus 54.0% in blood isolates, respectively. This is mostly due to the concurrently lower carriage and higher mutation rates of the rmpA and rmpA2 genes in ESBL-KP.¹⁸ Nonetheless, the underlying reason for higher mortality rates in patients with ESBL-KP bacteremia than in patients with non-ESBL-KP bacteremia is not clear^6,7 given that the latter is considered more virulent, but less antibiotic resistant than the former.¹⁸ With the exception of the antibiotic resistance factor, the effects of the virulence factors on clinical features and outcomes in the patients with ESBL-KP bacteremia are not known. Therefore, the purpose of our investigation was to compare the clinical characteristics of adult patients with ESBL-KP bacteremia with or without hypermucoviscosity-associated virulence determinants. We hypothesized that virulent determinants do not affect clinical features and outcomes in ESBL-KP bacteremia patients.

Materials and Methods

Study design

This study was conducted at an academic tertiary medical center with a 1,300-bed capacity in southern Taiwan. Adult patients with at least one ESBL-KP blood culture were retrospectively enrolled for an 18-month period from January 2009 through June 2010. Patients younger than 18 years old or with polymicrobial bacteremia were excluded. KP isolates were stored at −80°C until analyzed. The patients received management and therapy as per usual practice by attending physicians without study-directed intervention during hospitalization. After discharge, the electronic records for the patients were retrospectively reviewed for demographic data (age, sex, comorbid disease, and Charlson score¹⁹), source of infection, radiologic images and reports, laboratory data, acute respiratory distress syndrome, insertion of central venous catheter, days of hospitalization, stay in the intensive care unit, susceptibility antibiogram, empirical antibiotic treatment, acute physiology and chronic health evaluation (APACHE) II score,²⁰ in-hospital death, and number of days between onset of bacteremia and outcome.

Definitions

Pneumonia diagnoses were based on confirmatory findings from clinical examination, radiographic pulmonary infiltrates, and bacterial cultures of sputum or endotracheal aspirates yielding ESBL-KP with the same antibiogram as the bacteremic isolate. Urinary tract infections were diagnosed on the basis of observations from concurrent ESBL-KP bacteremia and bacteriuria with the same antibiogram in the presence of pyuria (≥10 white blood cells per high-power field of urine sample). Other localized infections were defined as ESBL-KP pyogenic infections other than pneumonia or urinary tract infections.

Primary bacteremia was defined as cryptogenic infection with no obvious focal source identifiable from clinical specimen cultures and image studies. Central venous catheter-related bloodstream infection was classified as primary bacteremia and was defined as concurrent ESBL-KP growth in blood and catheter tip cultures in the absence of other infectious sources accounting for the bacteremia. Nosocomial infection was defined as an infection that was identified at least 48 hr following hospital admission. Healthcare-associated infections were grouped as nosocomial infections. Community-acquired infections excluded episodes of residence in a nursing home or long-term care facility. Acute respiratory distress syndrome (ARDS) was defined according to the American–European Consensus definition of the syndrome.²¹ The APACHE II score was calculated based on data obtained within 24 hr after the onset of ESBL-KP bacteremia. The appropriateness of initial antimicrobial therapy was defined as treatment with an active antibiotic initiated at the onset of the bacteremia and continued for at least 3 days.

The hypervirulence group in our study consisted of patients in whom isolates that were positive for any virulence determinant of hypermucoviscosity phenotype, K1/K2 capsule serotypes, or plasmid-borne rmpA or rmpA2 genes were identified. The nonhypervirulence group consisted of patients in whom all isolates identified were negative for all of the aforementioned virulence determinants.

Microbiological characteristics

The KP was identified using a Phoenix system (Becton, Dickinson and Company, Baltimore, MD) and API 20E system (bioMerieux, Marcy l'Etoile, France). Strains were considered positive for a hypermucoviscosity phenotype based on the observation of a viscous string of >5 mm from the bacterial colony cultured on a blood agar plate.^1,10 Antimicrobial susceptibility tests were performed using the standard Kirby–Bauer disc diffusion method.²² Antibiotics such as gentamicin, amikacin, ciprofloxacin, ampicillin, piperacillin–tazobactam, cefazolin, cefuroxime, ceftriaxone, ceftazidime, flomoxef, ertapenem, and imipenem were tested. ESBL production was screened for and identified by the double disc test using cefotaxime and ceftazidime along with an amoxicillin–clavulanate disc, in accordance with Clinical and Laboratory Standards Institute (CLSI) standards.²² The control strains were Escherichia coli American Type Culture Collection (ATCC) 25922, K. pneumoniae ATCC 700603, and Pseudomonas aeruginosa ATCC 27853.

DNA manipulation

Genomic DNA was extracted using a QIAamp DNA Mini kit (Qiagen, Hilden, Germany) for screening for K capsule serotypes. Plasmid DNA was extracted using a QIAprep Spin Miniprep kit (Qiagen) in preparation for detection of rmpA and rmpA2 genes.

Polymerase chain reaction amplification of virulence genes

The genome of the KP strain, NTUH-K2044 (http://genome.nhri.org.tw/KP/), carries three different copies of rmpA. Two of the open reading frames (ORFs), KPP020 (rmpA) and KPP302 (rmpA2), are on the plasmid pK2044, and the other (ORF KP3619) is on the chromosome.^9,23 Since the well-known plasmid-borne rmpA and rmpA2 genes are the most important virulent determinants,^1,10–15 we focused on analyses of these two genes for the current study.

The rmpA, rmpA2, and genes specific for K capsule serotypes were identified by polymerase chain reaction (PCR) using specific primers as previously described and PCR products were confirmed by DNA sequencing. The rmpA gene was amplified using the following primers: forward, 5′-ACT GGG CTA CCT CTG CTT CA-3′; and reverse, 5′-CTT GCA TGA GCC ATC TTT CA-3′.²¹ The rmpA2-specific primers were as follows: forward, 5′-TGT GCA ATA AGG ATG TTA CAT TAG T-3′; and reverse, 5′-TTT GAT GTG CAC CAT TTT TCA-3′. The primers for magA (forward, 5′-GGT GCT CTT TAC ATC ATT GC-3′; and reverse, 5′-GCA ATG GCC ATT TGC GTT AG-3′) were used to amplify wzy_KpK1, a capsule serotype K1-antigen-specific polymerase gene.^8,24 The primers for k₂A (forward, 5′-CAA CCA TGG TGG TCG ATT AG-3′; and reverse, 5′-TGG TAG CCA TAT CCC TTT GG-3′) were used to specifically identify isolates with a K2 capsule serotype.^24,25

DNA sequencing

All amplicons were purified with PCR clean-up kits (Roche Diagnostics, GmbH, Penzberg, Germany) and sequenced on an ABI PRISM 3730 sequencer analyzer (Applied Biosystems, Foster City, CA). Sequences were analyzed using BLAST online through the National Center for Biotechnology Information database.²⁶

Statistical analyses

Fisher's exact test was used to analyze the significance of categorical variables. Continuous variables were analyzed using nonparametric Kruskal–Wallis rank tests or Mann–Whitney U tests. Exact logistic regression was used to identify significant variables to estimate the risk ratio of the virulence group. Mortality hazard ratios were assessed using Cox proportional hazards regression model analyses. The Kaplan–Meier survival curves for patients with bacteremia after onset were compared between the groups of isolates with and without virulence determinants. A two-tailed p < 0.05 was considered statistically significant. All statistics were performed using Stata version 12.1 (Stata Press, College Station, TX).

Results

Demographic data and clinical syndromes

Individual patients (n = 55) with ESBL-KP bacteremia were enrolled in this study. Of these, 7 patients (infant, n = 1; polymicrobial bacteremia, n = 6) were excluded, leaving 48 adult patients for inclusion in the investigation. Primary (cryptogenic) bacteremia from unknown sources and secondary bacteremia due to focal infections were identified in 23 (47.9%) and 25 (52.1%) patients, respectively. The focus of secondary bacteremia included pneumonia (n = 9), urinary tract infections (n = 13), and three others (perianal abscess, n = 1; pancreatic abscess, n = 1; liver abscess, n = 1).

The hypervirulence and nonhypervirulence groups included 19 (39.6%) and 29 (60.4%) patients, respectively. Primary bacteremia was significantly associated with nonhypervirulence (p = 0.001), liver cirrhosis (p = 0.046), higher Charlson score (p = 0.027), central venous catheter insertion (p = 0.008), thrombocytopenia (p = 0.005), and in-hospital mortality (p = 0.018), whereas, secondary bacteremia was significantly associated with hypervirulence (p = 0.001) and diabetes mellitus (p = 0.017) (Table 1).

Table 1.

Comparison of Bacterial Virulence, Demographic Data, and Clinical Characteristics of Patients with ESBL-KP Bacteremia with Primary (or Cryptogenic) and Secondary (or Focal) Infections

Variables	Primary infection (n = 23)	Secondary infection (n = 25)	p
With virulence determinants (19, 39.6%)	3 (13.0)	16 (64.0)	0.001^a
Female (13, 27.1%)	7 (30.4)	6 (24.0)	0.616
Age (median, Q1–Q3)	73 (53–77)	68 (59–76)	0.928
Community-acquired (12, 25.0%)	4 (17.4)	8 (32.0)	0.324
Diabetes mellitus (17, 35.4%)	4 (17.4)	13 (52.0)	0.017^a
End-stage renal disease (4, 8.3%)	2 (8.7)	2 (8.0)	1.000
Liver cirrhosis (4, 8.3%)	4 (17.3)	0 (0.0)	0.046^a
Chronic obstructive pulmonary disorder (4, 8.3%)	1 (4.3)	3 (12.0)	0.610
Coronary artery disease (16, 33.3%)	10 (43.5)	6 (24.0)	0.222
Old cerebral vascular accident (14, 29.1%)	6 (26.1)	8 (32.0)	0.497
Cancer (12, 25.0%)	8 (34.8)	4 (16.0)	0.187
Charlson score (median, Q1–Q3)	5 (3–7)	3 (2–4)	0.027^a
APACHE II score ≥15 (20, 41.7%)	11	9	0.559
Stay of hospitalization (median, Q1–Q3)	26 (17–38)	23 (12–41)	0.516
Stay in intensive care unit (23, 47.9%)	13 (56.5)	10 (40.0)	0.301
Central venous catheter insertion (19, 39.6%)	14 (60.9)	5 (20.0)	0.008^a
C-reactive protein (median, Q1–Q3)	97.8 (72.3–178.8)	87.3 (70.4–144.0)	0.347
White blood cell count (median, Q1–Q3)	9,200 (3,700–13,600)	11,800 (5,900–16,500)	0.219
Platelet count (median, Q1–Q3)	74,000 (25,000–131,000)	192,000 (79,000–273,000)	0.005^a
Initial antibiotic therapy
Inappropriate (25, 52.1%)	13 (56.5)	12 (48.0)	0.555
Acute respiratory distress syndrome (9, 18.8%)	3 (13.0)	6 (24.0)	0.466
Mortality	17 (73.9)	10 (40.0)	0.018^a
Days of outcome from onset (median, Q1–Q3)	9 (2–16)	10 (3–23)	0.490

p < 0.05.

ESBL-KP, extended-spectrum β-lactamase-producing Klebsiella pneumoniae.

Microbiological characteristics in hypervirulence and nonhypervirulence groups

The genes for wzy_KpK1 and wzy_KpK2, specific to isolates with K1 and K2 capsular serotypes, respectively, were not identified in any ESBL-KP isolates in our study. The hypervirulence group (n = 19) included 11 isolates (22.9%) with the hypermucoviscosity phenotype, 15 isolates (31.3%) with the rmpA gene, and 16 isolates (33.3%) with the rmpA2 gene.

Initial comparisons indicated that the hypervirulence group exhibited a significantly higher proportion of patients with diabetes mellitus (p = 0.013) and lower proportion of those with primary bacteremia (p = 0.001) than the nonhypervirulence group. Charlson scores, APACHE II scores, antibiotic susceptibility patterns, antimicrobial therapy administration, and clinical outcomes were not significantly different between the two groups (Table 2). Univariate analysis of the exact logistic regression model revealed significantly higher proportions of pneumonia, urinary tract infection, and diabetic comorbidity in the hypervirulence group. Furthermore, multivariate analysis revealed that pneumonia and other localized infections were significantly associated with the hypervirulence group, whereas prolonged hospital stay (> 30 days) was significantly associated with the nonhypervirulence group (Table 3).

Table 2.

Comparison of Demographic and Clinical Data from Patients with ESBL-KP Bacteremia With and Without Virulent Determinants

Variables	With virulence (n = 19)	Without virulence (n = 29)	p
Sex (n,%)
Female (13, 27.1%)	5 (26.3%)	8 (27.6%)	1.000^a
Age (median, Q1–Q3)	68 (57–76)	72 (61–77)	0.527^b
Community-acquired (12, 25%)	6 (31.6%)	6 (20.7%)	0.501^a
Diabetes mellitus (17, 35.4%)	11 (57.9%)	6 (20.7%)	0.013^a,c
End-stage renal disease (4, 8.3%)	1 (5.3%)	3 (10.3%)	1.000^a
Liver cirrhosis (4, 8.3%)	1 (5.3%)	3 (10.4%)	1.000^a
Chronic obstructive pulmonary disorder (4, 8.3%)	2 (10.5%)	2 (6.9%)	1.000^a
Coronary artery disease (16, 33.3%)	5 (26.3%)	11 (37.9%)	0.535^a
Old cerebral vascular accident (14, 29.1%)	5 (26.3%)	9 (31.0%)	0.927^a
Cancer (12, 25.0%)	4 (21.1%)	8 (27.6%)	0.739^a
Primary infection (23, 47.9%)	3 (15.8%)	20 (69.0%)	0.001^a,c
Stay of hospitalization (median, Q1–Q3)	17 (10–28)	27 (17–54)	0.050^b
Stay in intensive care unit (23, 47.9%)	10 (52.6%)	13 (44.8%)	0.142^a
Insertion of central venous catheter (19, 39.6%)	5 (26.3%)	14 (48.3%)	0.147^a
C-reactive protein (median, Q1–Q3)	87.2 (73.8–144)	109.8 (60.7–147.5)	0.791^b
White blood cell count (median, Q1–Q3)	13,200 (3,700–17,800)	9,900 (5,900–13,000)	0.776^b
Platelet count (median, Q1–Q3)	137,000 (61,000–231,000)	99,000 (31,000–231,000)	0.736^b
Charlson score (median, Q1–Q3)	4 (2–5)	4 (2–7)	0.376^b
APACHE II score ≥15	8 (42.1%)	12 (44.8%)	0.960^a
Susceptibility antibiogram^d (n, %)			0.280^a
Type 2 (10, 20.8%)	7 (41.2%)	3 (16.7%)
Type 3 (19, 39.6%)	7 (41.2%)	12 (66.7%)
Type 5 (6, 12.5%)	3 (17.7%)	3 (16.7%)
Initial antibiotic therapy
Inappropriate (25, 52.1%)	10 (52.6%)	15 (51.7%)	1.000^a
Acute respiratory distress syndrome (9, 18.8%)	5 (26.3%)	4 (13.8%)	0.451^a
Mortality (27, 56.3%)	10 (52.6%)	17 (58.6%)	0.770^a
Days of outcome from onset (median, Q1–Q3)	10 (2–17)	9 (3–17)	0.727^b

Fisher^'s exact test for categorical variable.

Mann–Whitney U test for continue variable.

p < 0.05.

Types of susceptibility antibiograms as follows: (1) ertapenem, imipenem (n = 1); (2) ertapenem, imipenem, flomoxef (n = 10); (3) amikacin, ertapenem, imipenem, flomoxef (n = 19); (4) ertapenem, imipenem, flomoxef, piperacillin–tazobactam (n = 1); (5) amikacin, ertapenem, imipenem, flomoxef, piperacillin–tazobactam (n = 6); (6) amikacin, ertapenem, imipenem, flomoxef, piperacillin–tazobactam, ciprofloxacin (n = 3); (7) amikacin, ertapenem, imipenem (n = 4); (8) ertapenem, imipenem, ciprofloxacin (n = 1); (9) ertapenem, imipenem, ciprofloxacin, piperacillin–tazobactam (n = 1); (10) amikacin, gentamicin, imipenem, piperacillin–tazobactam (n = 1); and (11) amikacin, gentamicin, ertapenem, imipenem, flomoxef, piperacillin–tazobactam (n = 1).

Q1–Q3, interquartile range between 25% and 75%.

Table 3.

Exact Logistic Regression Analysis of Isolates With Versus Without Virulence Determinants from Patients with Bacteremia

	Univariate		Multivariate
Variables	Odds ratio (95% CI)	p	Odds ratio (95% CI)	p
Infection classification
Primary (cryptogenic) infection	1.000 (reference)	—	1.000 (reference)	—
Pneumonia	19.973 (2.443–289.616)	0.002^a	10.983 (1.474–81.844)	0.019^a
Urinary tract infection	7.254 (1.208–57.858)	0.027^a	4.704 (0.712–31.056)	0.108
Other localized infections	11.334 (0.466–823.977)	0.169	26.820 (1.101–653.235)	0.043^a
Stay of hospitalization
≤30 days	1.000 (reference)	—	1.000 (reference)	—
>30 days	0.238 (0.036–1.103)	0.072	0.138 (0.020–0.938)	0.043^a
Central venous line-related
No central line insertion	1.000 (reference)	—	1.000 (reference)	—
No growth in tip culture	0.145 (0.000–1.125)	0.067	0.853 (0.029–25.550)	0.927
Growth of K. pneumoniae in catheter tip culture	0.277 (0.005–3.265)	0.502	0.528 (0.055–5.061)	0.580
No removal of catheter	1.069 (0.164–6.965)	1.000	1.892 (0.294–12.161)	0.502
Diabetes mellitus
No	1.000 (Ref.)	—	1.000 (Ref.)	—
Yes	5.066 (1.245–23.143)	0.020^a	3.526 (0.765–16.251)	0.106

p < 0.05.

CI, confidence interval.

Outcome and risk factors for mortality

The overall in-hospital mortality rate for the patients with ESBL-KP bacteremia was 56.3%. Kaplan–Meier plots depicting survival of patients after onset of ESBL-KP bacteremia were not significantly different between the hypervirulence and nonhypervirulence groups (Fig. 1). Univariate analysis of hazard risk ratios indicated that nosocomial infection, stay in an intensive care unit, no removal of central venous catheter, Charlson scores, and APACHE II score (≧15) were significant risk factors for mortality. On the other hand, urinary tract infection and initial appropriate antibiotic therapy were negatively associated with mortality. Multivariate analysis of the above risk factors revealed Charlson score as an independent predictor of mortality (Table 4).

FIG. 1.

Kaplan–Meier survival plots for patients with ESBL-KP bacteremia were not significantly different between the hypervirulence group (Group 1) and nonhypervirulence group (Group 2).

Table 4.

Hazard Risk Ratios for Mortality of Patients with ESBL-KP Bacteremia

	Univariate		Multivariate
Variable	HR	95% CI	HR	95% CI
Without virulence	0.93	0.42–2.07
Male	0.76	0.33–1.79
Age	0.99	0.97–1.02
Nosocomial	4.54^a	1.06–19.36	2.29	0.41–12.80
Diabetes mellitus	1.10	0.49–2.46
End-stage renal disease	1.15	0.34–3.86
Alcoholism	1.38	0.18–10.37
Liver cirrhosis	1.10	0.25–4.76
Chronic obstructive pulmonary disorder	0.97	0.23–4.13
Coronary artery disease	1.62	0.73–3.62
Old cerebral vascular accident	0.95	0.39–2.34
Cancer	1.59	0.66–3.82
Pneumonia versus primary infection	1.03	0.40–2.66
Urinary tract infection versus primary infection	0.19^a	0.04–0.84	0.92	0.15–5.64
Other localized infections versus primary infection	0.65	0.14–2.95
Stay in intensive care unit	3.58^a	1.07–12.02	0.99	0.22–4.45
No growth in CVP tip culture versus no CVP line	1.91	0.59–6.19
ESBL-KP in CVP tip culture versus no CVP line	2.63	0.88–7.86
No removal of CVP versus no CVP line	3.50^a	1.24–9.82	0.83	0.21–3.21
Acute respiratory distress syndrome	1.91	0.79–4.61
C-reactive protein	1.00	0.99–1.00
White blood cell count	1.00	0.99–1.00
Platelet count	1.00	0.99–1.00
Initial appropriate antibiotic therapy	0.42^a	0.18–0.95	0.88	0.28–2.77
Recurrent bacteremia	0.24	0.05–1.03
Charlson score	1.44^a	1.20–1.72	1.43^a	1.04–1.99
APACHE II score ≥15	4.76^a	1.96–11.55	2.55	0.82–7.88

p < 0.05.

CVP, central venous catheter; HR, hazard ratio.

Discussion

We compared the clinical features and outcomes of 48 patients with bacteremia caused by ESBL-KP with or without hypervirulence-associated determinants. All bacterial isolates were negative for genes associated with serotypes K1 and K2. These observations are similar to those of a study by Lee et al., in which K1 and K2 serotypes were not detected in 24 ESBL-KP isolates.²⁷

Overall, 11 (22.9%), 15 (31.3%), and 16 (33.3%) isolates were positive for the hypermucoviscosity phenotype, rmpA gene, and rmpA2 gene, respectively. Patients in the hypervirulence group showed higher proportions of diabetic comorbidity and secondary focal infections than those in the nonhypervirulence group. In contrast to a previous study, in which a high prevalence of liver abscesses was seen in patients with community-acquired KP bacteremia,²⁸ we found only one case with liver abscess associated with ESBL isolates. The capability of an ESBL-KP strain to cause an invasive focal infection might be enhanced by bacterial virulence and deficient phagocytosis in hosts with diabetes mellitus.^{4,5,8,9,11,29,30} This hypothesis is consistent with the observations of the hypervirulence group in our study.

Lee et al. reported one (4.2%) ESBL strain expressing a hypermucoviscosity phenotype and five (20.8%) strains harboring rmpA or rmpA2 genes, which suggests a negative correlation between virulence and ESBL production.²⁷ However, in the current study, 39.6% of ESBL strains expressed a hypermucoviscosity phenotype, rmpA gene, and/or rmpA2 gene, implying higher virulence in our isolates. Furthermore, an increasing trend of up to 17% of ESBL strains in the hypermucoviscous KP isolates has been observed in China.¹⁷ Therefore, continuous monitoring of hypervirulence in the ESBL-KP isolates is needed.

Based on previous reports, APACHE II scores, disease severity, Charlson scores, inappropriate antimicrobial therapy, and primary bacteremia are risk factors for mortality in patients with KP liver abscess or bacteremia.^5,28 In a small series, clinical severity predicted mortality following ESBL-KP bacteremia.³¹ With regard to virulence, K1 and K2 isolates do not appear to confer a poor outcome in patients with community-onset KP pneumonia.¹⁶ Hypervirulent KP-infected patients have lower mortality rates than nonhypervirulent KP-infected patients (4.5% vs. 16.7%).³² All the aforementioned studies indicate that disease severity and/or comorbidity, rather than bacterial virulence, correlate with mortality in KP-infected patients. The primary limitation of these studies is their inclusion of only a few virulent ESBL-KP isolates.

Hence, we studied a higher number of virulent ESBL-KP strains than those in previous studies. However, hypervirulence did not have a statistically significant effect on in-hospital mortality. Kaplan–Meier survival plots for patients with bacteremia after onset were not significantly different between the two groups. The in-hospital mortality rate of 56.3% is similar to that observed in previous studies.^6,7

Patients with primary bacteremia had a significantly higher rate of mortality compared with those with secondary bacteremia. Multiple risk factors affecting mortality were identified in our study, including nosocomial infection, prolonged stay in an intensive care unit, no removal of central venous catheter, Charlson score, and APACHE II score. Initial administration of appropriate antibiotic therapy and urinary tract infections were negatively correlated with mortality. However, a multivariate analysis revealed Charlson score as the only independent predictor of mortality. We acknowledged that unidentified variables might have contributed to the high mortality rate. Early diagnosis and adequate source control at the infection focus would likely have lowered the mortality rate in diabetic patients with secondary bacteremia. This might also have negatively affected the role of virulence determinants on mortality due to ESBL-KP bacteremia. Therefore, it is important that physicians identify the infection focus of ESBL-KP isolates, particularly those that are hypervirulent.

As this was a retrospective design, the use of appropriate antibiotic therapies was seen to include diverse regimens, including ertapenem, imipenem, meropenem, flomoxef, or piperacillin–tazobactam alone, or in combination with fosfomycin, colistin, or tigecycline. Thus, we could not analyze the effects of specific antibiotic regimens on the observed outcomes. However, a previous case study did report a minimal role of appropriate antibiotic treatment on mortality due to bacteremia caused by ESBL-producing bacteria.³³ In addition, a recent study reported that piperacillin–tazobactam is less effective than carbapenems for the treatment of ESBL bacteremia.³⁴

Moreover, the study had a few other limitations. First, a small study population may have contributed to an imbalance in the proportion of diabetes and infection foci within the two groups. However, in addition to basic association tests, we conducted exact logistic regression analysis, which is used to model binary outcome variables (e.g., hypervirulent or nonhypervirulent) when sample sizes are too small for standard logistic regression.

Second, it is contentious that eight rmpA/rmpA2-positive isolates without a hypermucoviscosity phenotype were classified into the hypervirulence group. The observation of nonhypermucoviscous hypervirulent KP strains^35,36 may be explained by a potential mutation of rmpA and/or rmpA2 genes.¹⁸ In addition, other determinants simultaneously located on the rmpA-encoded plasmid, such as siderophores, can also contribute to virulence.^10,11,13

Third, our study did not include ESBL-KP isolates with K1/K2 serotypes, but these appear to be extremely rare.²⁷ Last, the virulence of the isolates was not confirmed using animal studies.¹⁸ Thus, we cannot exclude the possibility of inconsistencies present within the virulence experiments conducted between groups.

Conclusion

Hypervirulence had a significant effect on clinical features, but not mortality. Hypervirulent ESBL-KP isolates were significantly correlated with secondary bacteremia and diabetes mellitus. Nonhypervirulence was correlated with primary bacteremia, especially in patients with prolonged hospital stay. The overall mortality rate was substantially increased (56.3%) by multiple risk factors, but not by virulence classification. The impact of multiple risk factors was overshadowed by the Charlson score, which was identified as an independent predictor of mortality. Future efforts in identifying hypervirulence determinants of ESBL-KP might help physicians better understand the clinical implications of ESBL-KP pathogens and prompt them to search for infection focuses for targeted treatment, especially in diabetic patients.

Footnotes

Acknowledgments

The authors thank Chia-Yu Wang and Li-Yin Liou for the laboratory work. This work was supported by grants from the Chi Mei Medical Center Research Foundation (CMFHT10202, CMFHR10208, and CMFHR10508). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Ethics Statement

This was a retrospective study and an exemption from the requirement for informed consent was granted by the Institutional Review Board (IRB) of the Chi Mei Medical Center, Tainan, Taiwan (IRB Approval No. 10301-014).

Disclosure Statement

No competing financial interests exist.

References

W.L.

, Ko

W.C.

, Cheng

K.C.

, Lee

H.C.

, Ke

D.S.

, Lee

C.C.

, Fung

C.P.

, and Chuang

Y.C.

. 2006. Association between rmpA and magA genes and clinical syndromes caused by Klebsiella pneumoniae in Taiwan. Clin. Infect. Dis., 42:1351–1358.

Y.H.

, Chuang

Y.C.

, and Yu

W.L.

. 2008. Clinical spectrum and molecular characteristics of Klebsiella pneumoniae causing community-acquired extrahepatic abscess. J. Microbiol. Immunol. Infect., 41:311–317.

Tseng

C.P.

, Wu

H.S.

, Wu

T.H.

, Lin

Y.T.

, and Fung

C.P.

. 2013. Clinical characteristics and outcome of patients with community-onset Klebsiella pneumoniae bacteremia requiring intensive care. J. Microbiol. Immunol. Infect., 46:217–223.

S.C.

, Siu

L.K.

, Ma

, Yeh

K.M.

, Fung

C.P.

, Lin

J.C.

, and Chang

F.Y.

. 2008. Community-acquired liver abscess caused by serotype K1 Klebsiella pneumoniae with CTX-M-15-type extended-spectrum β-lactamase. Antimicrob. Agents Chemother., 52:804–805.

Yoon

J.H.

, Kim

Y.J.

, Jun

Y.H.

, Kim

S.I.

, Kang

J.Y.

, Suk

K.T.

, and Kim

D.J.

. 2014. Liver abscess due to Klebsiella pneumoniae: risk factors for metastatic infection. Scand. J. Infect. Dis., 46:21–26.

Lee

N.Y.

, Lee

C.C.

, Huang

W.H.

, Tsui

K.C.

, Hsueh

P.R.

, and Ko

W.C.

. 2012. Carbapenem therapy for bacteremia due to extended-spectrum-β-lactamase-producing Escherichia coli or Klebsiella pneumoniae: implications of ertapenem susceptibility. Antimicrob. Agents Chemother., 56:2888–2893.

Yang

C.C.

, Li

S.H.

, Chuang

F.R.

, Chen

C.H.

, Lee

C.H.

, Chen

J.B.

, Wu

C.H.

, and Lee

C.T.

. 2012. Discrepancy between effects of carbapenems and flomoxef in treating nosocomial hemodialysis access-related bacteremia secondary to extended spectrum β-lactamase producing Klebsiella pneumoniae in patients on maintenance hemodialysis. BMC Infect. Dis., 12:206.

Fang

C.T.

, Chuang

Y.P.

, Shun

C.T.

, Chang

S.C.

, and Wang

J.T.

. 2004. A novel virulence gene in Klebsiella pneumoniae strains causing primary liver abscess and septic metastatic complications. J. Exp. Med., 199:697–705.

Fang

C.T

, Lai

S.Y.

, Yi

W.C.

, Hsueh

P.R.

, Liu

K.L.

, and Chang

S.C.

. 2007. Klebsiella pneumoniae genotype K1: an emerging pathogen that causes septic ocular or central nervous system complications from pyogenic liver abscess. Clin. Infect. Dis., 45:284–293.

10.

V.L.

, Hansen

D.S.

, Ko

W.C.

, Sagnimeni

, Klugman

K.P.

, von Gottberg

, Goossens

, Wagener

M.M.

, V.J. and Benedi and International Klebseilla Study Group. 2007. Virulence characteristics of Klebsiella and clinical manifestations of K. pneumoniae bloodstream infections. Emerg. Infect. Dis., 13:986–993.

11.

W.L.

, Ko

W.C.

, Cheng

K.C.

, Lee

C.C.

, Lai

C.C.

, and Chuang

Y.C.

. 2008. Comparison of prevalence of virulence factors for Klebsiella pneumoniae liver abscesses between isolates with capsular K1/K2 and non-K1/K2 serotypes. Diagn. Microbiol. Infect. Dis., 62:1–6.

12.

Cheng

H.Y.

, Chen

Y.S.

, Wu

C.Y.

, Chang

H.Y.

, Lai

Y.C.

, and Peng

H.L.

. 2010. RmpA regulation of capsular polysaccharide biosynthesis in Klebsiella pneumoniae CG43. J. Bacteriol., 192:3144–3158.

13.

Hsu

C.R.

, Lin

T.L.

, Chen

Y.C.

, Chou

H.C.

, and Wang

J.T.

. 2011. The role of Klebsiella pneumoniae rmpA in capsular polysaccharide synthesis and virulence revisited. Microbiology., 157:3446–3457.

14.

Wacharotayankun

, Arakawa

, Ohta

, Tanaka

, Akashi

, Mori

, and Kato

. 1993. Enhancement of extracapsular polysaccharide synthesis in Klebsiella pneumoniae by RmpA2, which shows homology to NtrC and FixJ. Infect. Immun., 61:3164–3174.

15.

Lai

Y.C.

, Peng

H.L.

, and Chang

H.Y.

. 2003. RmpA2, an activator of capsule biosynthesis in Klebsiella pneumoniae CG43, regulates K2 cps gene expression at the transcriptional level. J. Bacteriol., 185:788–800.

16.

Lin

Y.T.

, Jeng

Y.Y.

, Chen

T.L.

, and Fung

C.P.

. 2010. Bacteremic community-acquired pneumonia due to Klebsiella pneumoniae: clinical and microbiological characteristics in Taiwan, 2001–2008. BMC Infect. Dis., 10:307.

17.

, Sun

, Yu

, Li

, Chen

, Jin

, Jiao

, and Wu

. 2014. Increasing occurrence of antimicrobial-resistant hypervirulent (hypermucoviscous) Klebsiella pneumoniae isolates in China. Clin. Infect. Dis., 58:225–232.

18.

W.L.

, Lee

M.F.

, Tang

H.J.

, Chang

M.C.

, and Chuang YC

Y.C.

. 2015. Low prevalence of rmpA and high tendency of rmpA mutation correspond to low virulence of extended spectrum β-lactamase-producing Klebsiella pneumoniae isolates. Virulence. 6:162–172.

19.

Charlson

M.E.

, Pompei

, Ales

K.L.

, and MacKenzie

C.R.

. 1987. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J. Chronic Dis., 40:373–383.

20.

Knaus

W.A.

, Draper

E.A.

, Wagner

D.P.

, and Zimmerman

J.E.

. 1985. APACHE II: a severity of disease classification system. Crit. Care Med., 13:818–829.

21.

ARDS Definition Task Force, Ranieri

V.M.

, Rubenfeld

G.D.

, Thompson

B.T.

, Ferguson

N.D.

, Caldwell

, Fan

, Camporota

, and Slutsky

A.S.

. 2012. Acute respiratory distress syndrome: the Berlin Definition. JAMA., 307:2526–2533.

22.

Clinical and Laboratory Standards Institute. 2010. Performance Standards for Antimicrobial Susceptibility Testing: CLSI M100-S20. Clinical and Laboratory Standards Institute Press, Wayne, PA.

23.

Cheng

N.C.

, Yu

Y.C.

, Tai

H.C.

, Hsueh

P.R.

, Chang

S.C.

, Lai

S.Y.

, Yi

W.C.

, and Fang

C.T.

. 2012. Recent trend of necrotizing fasciitis in Taiwan: focus on monomicrobial Klebsiella pneumoniae necrotizing fasciitis. Clin. Infect. Dis., 55:930–939.

24.

Yeh

K.M.

, Lin

J.C.

, Yin

F.Y.

, Fung

C.P.

, Hung

H.C.

, Siu

L.K.

, and Chang

F.Y.

. 2010. Revisiting the importance of virulence determinant magA and its surrounding genes in Klebsiella pneumoniae causing pyogenic liver abscesses: exact role in serotype K1 capsule formation. J. Infect. Dis., 201:1259–1267.

25.

W.L.

, Fung

C.P.

, Ko

W.C.

, Cheng

K.C.

, Lee

C.C.

, and Chuang

Y.C.

. 2007. Polymerase chain reaction analysis for detecting capsule serotypes K1 and K2 of Klebsiella pneumoniae causing abscesses of the liver and other sites. J. Infect. Dis., 195:1235–1236.

26.

Boratyn

G.M.

, Camacho

, Cooper

P.S.

, Coulouris

, Fong

, Ma

, Madden

T.L.

, Matten

W.T.

, McGinnis

S.D.

, Merezhuk

, Raytselis

, Sayers

E.W.

, Tao

, Ye

, and Zaretskaya

. 2013. BLAST: a more efficient report with usability improvements. Nucleic Acids Res., 41(Web Server issue):W29–W33.

27.

Lee

C.H.

, Liu

J.W.

, Su

L.H.

, Chien

C.C.

, Li

C.C.

, and Yang

K.D.

. 2010. Hypermucoviscosity associated with Klebsiella pneumoniae-mediated invasive syndrome: a prospective cross-sectional study in Taiwan. Int. J. Infect. Dis., 14:e688–e692.

28.

H.S.

, Wang

F.D.

, Tseng

C.P.

, Wu

T.H.

, Lin

Y.T.

, and Fung

C.P.

. 2012. Characteristics of healthcare-associated and community-acquired Klebsiella pneumoniae bacteremia in Taiwan. J. Infect., 64:162–168.

29.

Lin

J.C.

, Chang

F.Y.

, Fung

C.P.

, Xu

J.Z.

, Cheng

H.P.

, Wang

J.J.

, Huang

L.Y.

, and Siu

L. K.

. 2004. High prevalence of phagocytic-resistant capsular serotypes of Klebsiella pneumoniae in liver abscess. Microbes Infect., 6:1191–1198.

30.

W.L.

, Chan

K.S.

, Ko

W.C.

, Lee

C.C.

, and Chuang

Y.C.

. 2007. Lower prevalence of diabetes mellitus in patients with Klebsiella pneumoniae primary liver abscess caused by isolates of K1/K2 than with non-K1/K2 capsular serotypes. Clin. Infect. Dis., 45:1529–1530.

31.

Lee

C.H.

, Su

L.H.

, Tang

Y.F.

, and Liu

J.W.

. 2006. Treatment of ESBL-producing Klebsiella pneumoniae bacteraemia with carbapenems or flomoxef: a retrospective study and laboratory analysis of the isolates. J. Antimicrob. Chemother., 58:1074–1077.

32.

Liu

Y.M.

, Li

B.B.

, Zhang

Y.Y.

, Zhang

, Shen

, Li

, and Cao

. 2014. Clinical and molecular characteristics of emerging hypervirulent Klebsiella pneumoniae bloodstream infections in mainland China. Antimicrob. Agents Chemother., 58:5379–5385.

33.

De Rosa

F.G.

, Pagani

, Fossati

, Raviolo

, Cometto

, Cavallerio

, Parlato

, Guglielmi

, Serra

, and Di Perri

. 2011. The effect of inappropriate therapy on bacteremia by ESBL-producing bacteria. Infection., 39:555–561.

34.

Tamma

P.D.

, Han

J.H.

, Rock

, Harris

A.D.

, Lautenbach

, Hsu

A.J.

, Avdic

, Cosgrove

S.E.

, and Antibacterial Resistance Leadership Group. 2015. Carbapenem therapy is associated with improved survival compared with piperacillin-tazobactam for patients with extended-spectrum β-lactamase bacteremia. Clin. Infect. Dis., 60:1319–1325.

35.

Shon

A.S.

, Bajwa

R.P.

, and Russo

T.A.

. 2013. Hypervirulent (hypermucoviscous) Klebsiella pneumoniae: a new and dangerous breed. Virulence., 4:107–118.

36.

, Zhou

, Jiang

, Shi

K.R.

, Li

, Shen

, Wei

Z.Q.

, and Yu

Y.S.

. 2015. Clinical and microbiological characteristics of Klebsiella pneumoniae liver abscess in East China. BMC Infect. Dis., 15:161.