High Prevalence of Hypervirulent Klebsiella pneumoniae Infection in China: Geographic Distribution, Clinical Characteristics, and Antimicrobial Resistance

Abstract

Hypervirulent Klebsiella pneumoniae (hvKP) is traditionally defined by hypermucoviscosity, but data based on genetic background are limited. Antimicrobial-resistant hvKP has been increasingly reported but has not yet been systematically studied. K. pneumoniae isolates from bloodstream infections, hospital-acquired pneumonia, and intra-abdominal infections were collected from 10 cities in China during February to July 2013. Clinical data were collected from medical records. All K. pneumoniae isolates were investigated by antimicrobial susceptibility testing, string test, extended-spectrum β-lactamase (ESBL) gene detection, capsular serotypes, virulence gene profiles, and multilocus sequence typing. hvKP was defined by aerobactin detection. Of 230 K. pneumoniae isolates, 37.8% were hvKP. The prevalence of hvKP varied among different cities, with the highest rate in Wuhan (73.9%) and the lowest in Zhejiang (8.3%). Hypermucoviscosity and the presence of K1, K2, K20, and rmpA genes were strongly associated with hvKP (P < 0.001). A significantly higher incidence of liver abscess (P = 0.026), sepsis (P = 0.038), and invasive infections (P = 0.043) was caused by hvKP. Cancer (odds ratio [OR], 2.285) and diabetes mellitus (OR, 2.256) appeared to be independent variables associated with hvKP infections by multivariate analysis. Importantly, 12.6% of hvKP isolates produced ESBLs, and most of them carried bla_CTX-M genes. Patients with neutropenia (37.5% versus 5.6%; P = 0.020), history of systemic steroid therapy (37.5% versus 5.6%; P = 0.020), and combination therapy (62.5% versus 16.7%; P = 0.009) were more likely to be infected with ESBL-producing hvKP. The prevalence of hvKP is high in China and has a varied geographic distribution. ESBL-producing hvKP is emerging, suggesting an urgent need to enhance clinical awareness, especially for immunocompromised patients receiving combination therapy.

INTRODUCTION

Over the past few decades, increasing rates of hypervirulent Klebsiella pneumoniae (hvKP) infection have been reported worldwide (1–4). Such strains are notorious for their capacity to cause serious and metastatic infections in young and healthy individuals, such as pyogenic liver abscesses and endophthalmitis (5). Hypermucoviscosity is an important in vitro parameter for identification of hvKP (6, 7). However, several controversies have arisen regarding the association of hypermucoviscosity phenotype and virulence (8, 9). Hypermucoviscosity-negative strains are more prone to cause severe infections and have a higher mortality rate in diabetic mice than hypermucoviscous K. pneumoniae (8). Our previous study also demonstrated that one of the five hypermucoviscous K. pneumoniae isolates showed high virulence in both in vitro and in vivo assays (9). Therefore, it is apparent that hvKP cannot be defined by string test alone (10). Aerobactin accounts for increased siderophore production and is a major virulence determinant and new defining trait for hvKP based on genetic background (11). Some advances have been made recently in the epidemiology of hvKP isolates in China (6, 7, 12), although most of the studies defined hvKP by hypermucoviscosity phenotype rather than genotype, which may have led to biased results. In the present study, hvKP strains were defined on the basis of aerobactin detection.

Although hypervirulent and antimicrobial-resistant populations of K. pneumoniae were largely nonoverlapping (13), some isolates with combined virulence and resistance were detected (9, 12, 14–16). More importantly, hypervirulent carbapenem-resistant K. pneumoniae isolates have emerged in clinical settings (9, 12, 16). The confluence of multidrug resistance and enhanced virulence has the potential to be the next clinical crisis (5). So far, research on antimicrobial-resistant hvKP strains has been limited and mainly based on case reports (9, 12). A systematic study on the molecular and clinical characteristics of antimicrobial-resistant hvKP infections is still lacking.

To reduce the bias caused by subjective identification methods and to investigate the recent pattern of hvKP in China, we conducted a nationwide study and systematically analyzed the clinical and molecular characteristics and antimicrobial resistance of hvKP isolates.

MATERIALS AND METHODS

Clinical K. pneumoniae isolates.

A total of 230 consecutive cases of K. pneumoniae infection were collected from a national prospective surveillance program (February to July 2013 from the Chinese Antimicrobial Resistance Surveillance of Nosocomial Infections, including Beijing [n = 45], Tianjin [n = 22], Shenyang [n = 22], Jinan [n = 14], Xi'an [n = 15], Wuhan [n = 23], Changsha [n = 17], Zhejiang [n = 12], Shanghai [n = 21], and Guangzhou [n = 39]). Each center was asked to enroll the cases and collect strains isolated from hospital-acquired pneumonia, intra-abdominal infections, and bloodstream infections. Duplicate isolates from the same individual were excluded. All isolates were stored at −80°C until use. Confirmation of the K. pneumoniae isolates was performed by 16S rRNA sequencing.

Clinical data collection.

Clinical data and patient information were obtained from medical records and included demographic characteristics, underlying medical conditions, clinical presentations, antimicrobial therapy administration, and outcomes. The study was approved by the Research Ethics Board at Peking University People's Hospital (Beijing, China).

Antimicrobial susceptibility testing and phenotypic confirmation of ESBL.

Susceptibility testing was determined by the agar dilution method according to Clinical and Laboratory Standards Institute (CLSI) guidelines (17). The following antimicrobial agents were tested: cefoxitin, cefotaxime, ceftriaxone, ceftazidime, cefepime, piperacillin-tazobactam, imipenem, meropenem, amikacin, ciprofloxacin, minocycline, and tigecycline. The results were interpreted according to CLSI breakpoints (18). The interpretive criteria for tigecycline were based on the Food and Drug Administration (FDA) breakpoints (≤2 mg/liter was susceptible, 4 mg/liter was intermediate, and ≥8 mg/liter was resistant). Pseudomonas aeruginosa ATCC 27853 and Escherichia coli ATCC 25922 were used as controls for antimicrobial susceptibility testing. Extended-spectrum β-lactamase (ESBL) production was confirmed by agar dilution test using cefotaxime and ceftazidime along with clavulanate, in accordance with CLSI guidelines (17).

Detection of virulence-associated features and cps genotyping.

Hypermucoviscosity was identified by the string test as previously described (5). The formation of a viscous string of >5 mm was considered positive. We conducted cps genotyping of K serotype-specific alleles for K1, K2, K5, K20, K54, and K57 by PCR as previously described (19). Virulence-associated factors, such as the exopolysaccharide synthesis regulator gene (rmpA) and an iron uptake system (aerobactin), were determined by PCR (20, 21). Previously reported primers used for PCR were the following: aerobactin forward, 5′-GCATAGGCGGATACGAACAT-3′; aerobactin reverse, 5′-CACAGGGCAATTGCTTACCT-3′ (21). The reaction mixture was kept at 95°C for 5 min, followed by 30 cycles of 95°C for 1 min, 50°C for 1 min, 72°C for 1 min, and 72°C for 10 min. The PCR products were visualized and analyzed by agarose gel electrophoresis and sequencing.

β-Lactamase gene identification.

PCR was used to determine the prevalence of β-lactamase genes, including CTX-M, SHV, and TEM β-lactamase, as previously described (22). The bla_CTX-M gene was first identified with the PAN-CTX-M primer pair as previously described (forward, 5′-TTTGCGATGTGCAGTACCAGTAA-3′; reverse, 5′-CGATATCGTTGGTGGTGCCATA-3′), followed by the group-specific primers (22). The products were submitted for sequencing.

MLST.

Multilocus sequence typing (MLST) was performed as described on the Pasteur Institute MLST website (http://bigsdb.pasteur.fr/klebsiella/klebsiella.html), including DNA sequencing analysis of the seven housekeeping genes. New alleles and STs were submitted to the MLST website and were approved.

Statistical analysis.

Statistical analysis was performed using IBM SPSS Statistics version 20.0 and Graphpad Prism version 5. We used the χ² or Fisher's exact test for categorical variables. P < 0.05 was considered statistically significant. Logistic regression was used to identify variables associated with hvKP and ESBL-producing hvKP infections. All variables with P values of <0.1 were included in the multivariate model.

RESULTS

Virulence-associated features of K. pneumoniae isolates.

Aerobactin-positive strains were designated hvKP, which was determined by PCR. Eighty-seven of 230 (37.8%) isolates were hvKP. The prevalence of hvKP infection ranged from 8.33% to 73.9% in China, with the highest rate in Wuhan and the lowest rate in Zhejiang. The geographic distribution in China is shown in Fig. 1. Among all these strains, 74.7% (65/87) were hypermucoviscous, while six hypermucoviscous isolates were detected in the aerobactin-negative group. Hypermucoviscosity was strongly associated with hvKP (74.7% versus 4.2%; P < 0.001). A total of 71 (71/230; 30.9%) isolates were positive for K1, K2, K5, K20, and K57 serotypes, and the K54 serotype was not found in this study. K1, K2, and K20 were more likely to be identified in hvKP isolates (Table 1). The rmpA gene, a regulator of exopolysaccharide synthesis, had a strong relation to aerobactin (97.7% versus 3.5%; P < 0.001) (Table 1).

FIG 1.

Geographic distribution of hvKP infection cases in China. The prevalence of hvKP infections ranged from 8.33% to 73.91%, with a varied geographic distribution. Virulence-associated ST types (including ST23, ST65, and ST86) and resistance-associated ST types (including ST11 and ST437) demonstrated different distributions among K. pneumoniae isolates from different cities.

TABLE 1.

Microbiological and clinical characteristics of hvKP isolates

Characteristic	Value (no. [%]; no. positive/total no.) fora:		P value
	hvKP (n = 87)	cKP (n = 143)
Microbiological features
Hypermucoviscosity	65 (74.7)	6 (4.2)	<0.001
K serotype
K1	29 (33.3)	8 (5.6)	<0.001
K2	15 (17.2)	4 (2.8)	<0.001
K5	2 (2.3)	0	0.142
K20	6 (6.9)	0	0.003
K57	4 (4.6)	3 (2.1)	0.431
rmpA	85 (97.7)	5 (3.5)	<0.001
Antimicrobial susceptibility
Cefoxitin	78 (89.7)	103 (72.0)	0.002
Cefotaxime	75 (86.2)	70 (49.0)	<0.001
Ceftazidime	78 (89.7)	88 (61.5)	<0.001
Cefepime	82 (94.3)	91 (63.6)	<0.001
Piperacillin-tazobactam	86 (98.9)	111 (77.6)	<0.001
Imipenem	85 (97.7)	129 (90.2)	0.033
Meropenem	86 (98.9)	129 (90.2)	0.011
Amikacin	81 (93.1)	130 (90.9)	0.629
Ciprofloxacin	83 (95.4)	89 (62.2)	<0.001
ESBL production	11 (12.6)	61 (42.7)	<0.001
Clinical characteristics
Male sex	62 (77.5; 62/80)	79 (67.5; 79/117)	0.149
Age (yr) (means ± SD)	55.9 ± 1.5	51.8 ± 1.7	0.230
Coexisting conditions
Pulmonary disease	12 (15; 12/80)	11 (9.4; 11/117)	0.263
Cancer	27 (33.8; 27/80)	22 (18.8; 22/117)	0.020
Liver disease	8 (10; 8/80)	14 (12.0; 14/117)	0.819
Neurological disease	10 (12.5; 10/80)	21 (17.9; 21/117)	0.327
Diabetes mellitus	22 (27.5; 22/80)	17 (14.7; 17/116)	0.030
Neutropeniab	7 (8.8; 8/80)	13 (11.1; 13/117)	0.640
Splenectomy	0	2 (1.1; 2/117)	0.515
Surgery within 1 mo	22 (27.5; 22/80)	39 (33.3; 39/117)	0.434
Systemic steroid therapy	7 (8.8; 7/80)	16 (13.7; 16/117)	0.369
Use of immunosuppressant	7 (8.8; 7/80)	10 (8.6; 10/116)	1.000
Hospitalization within last 90 days	20 (25; 20/80)	53 (45.3; 53/117)	0.004
Transferred from another hospital	12 (15; 12/80)	17 (14.5; 17/117)	1.000
Receipt of antibiotics 30 days before infection	25 (31.3; 25/80)	52 (45.2; 52/115)	0.054
Use of invasive devices
Artery cannula	4 (8.5; 4/47)	3 (3.6; 3/83)	0.253
Presence of central venous catheters	9 (19.1; 9/47)	12 (14.6; 12/82)	0.621
Tracheal cannula	9 (19.1; 9/47)	11 (13.3; 11/83)	0.450
Tracheotomy	3 (6.4; 3/47)	4 (4.8; 4/83)	0.703
Presence of Foley catheter	15 (31.9; 15/47)	23 (27.7; 23/83)	0.689
Clinical presentation
Bacteremia	24 (30; 24/80)	35 (30.2; 35/116)	1.000
Abdominal infection	16 (20; 16/80)	27 (23.3; 27/116)	0.604
Hospital-acquired pneumonia	36 (45; 36/80)	43 (37.1; 43/116)	0.301
Liver abscess	10 (11.5; 10/87)	5 (3.5; 5/143)	0.026
Septic shock	6 (6.9; 6/87)	13 (9.1; 13/143)	0.629
Sepsis	43 (49.4; 43/87)	50 (35.0; 50/143)	0.038
Invasive infectionc	11 (12.6; 11/87)	7 (4.9; 7/143)	0.043
Infection occurred in ICU	11 (13.8; 11/80)	31 (26.9; 31/115)	0.033
Surgical treatment	13 (18.3; 13/71)	24 (21.6; 24/111)	0.706
Outcomes
Clinical improvement without modification of initial treatment	50 (70.4; 50/71)	86 (76.1; 86/113)	0.395
Change of initial antibiotics due to clinical worsening	17 (23.9; 17/71)	17 (15.0; 17/113)	0.172
Relapse	2 (2.8; 2/71)	1 (0.9; 1/113)	0.560
In-hospital mortality	2 (2.8; 2/71)	9 (8.0; 9/113)	0.208
Bacterial clearance after 72-h treatment	64 (83.1; 64/77)	95 (81.9; 95/116)	1.000
Persistent infection after 72-h treatment	7 (9.1; 7/77)	18 (15.5; 18/116)	0.274

^aK. pneumoniae strains carrying aerobactin were designated hvKP, while cKP was defined by aerobactin negativity. Values in boldface are statistically significant.

^bAbsolute neutrophil count of <500 cells/μl.

^cInvasive infections include liver abscess and meningitis.

Clinical characteristics of hvKP infection.

Table 1 shows the baseline characteristics of patients with hvKP infections. Compared with the classic K. pneumoniae (cKP) group, more patients with hvKP infections had cancer (33.8% versus 18.8%; P = 0.02) and diabetes mellitus (27.5% versus 14.7%; P = 0.03) as their underlying diseases. A significantly higher number of patients presenting with liver abscess (11.5% versus 3.5%; P = 0.026), sepsis (49.4% versus 35%; P = 0.038), and invasive infections (12.6% versus 4.9%; P = 0.043) were infected by hvKP. It was also noted that patients with a history of hospitalization within the previous 90 days (25% versus 45.3%; P = 0.004) and intensive care unit (ICU) stay (13.8% versus 26.9%; P = 0.033) were more likely to have a cKP infection.

Variables associated with hvKP infection.

Univariate regression analysis revealed that cancer (odds ratio [OR], 1.963) and diabetes mellitus (OR, 2.086) were statistically significant variables associated with hvKP infections (Table 2). In addition, cancer (OR, 2.285) and diabetes mellitus (OR, 2.256) were independent variables associated with hvKP infections, while hospitalization within the previous 90 days (OR, 0.345) and infection in the ICU (OR, 0.412) were independent protective factors for hvKP by multivariate analysis.

TABLE 2.

Regression analysis of variables associated with hvKP infections

Variable	Analysis result
	Univariate		Multivariate
	OR (95% CIa)	P value	OR (95% CI)	P value
Cancer	1.963 (1.012–3.806)	0.046	2.285 (1.102–4.738)	0.026
Diabetes mellitus	2.086 (1.034–4.210)	0.040	2.256 (1.050–4.846)	0.037
Hospitalization within last 90 days	0.403 (0.216–0.751)	0.004	0.345 (0.171–0.695)	0.003
Receipt of antibiotics 30 days before infection	0.551 (0.303–1.002)	0.051	0.549 (0.279–1.078)	0.081
Infection occurred in ICU	0.432 (0.202–0.922)	0.030	0.412 (0.183–0.928)	0.032

^aCI, confidence interval.

Antimicrobial susceptibility and prevalence of ESBL genes among K. pneumoniae isolates.

The antimicrobial resistance rate of cKP was significantly higher than that of hvKP strains (Table 1). ESBL was identified in 72 isolates (72/230; 31.3%) and was less common in the hvKP group (12.6% versus 42.7%; P < 0.001). Among 87 hvKP isolates, 12.6% (11/87) produced ESBLs, and one of them exhibited resistance to carbapenem.

The designation of the 11 ESBL-producing hvKP strains was by patient number. The demographics, virulence-associated features, and clinical characteristics of the 11 patients are summarized in Table 3. A total of six patients developed septicemia or bacteremia, and most of the strains were isolated from blood samples. All of these isolates harbored CTX-M, SHV, and TEM β-lactamase, and most of them had one or two genes that encoded ESBLs.

TABLE 3.

Clinical and microbiological characteristics of ESBL-producing hvKP isolates

Clinical characteristic	Value for patient isolate:
	1	2	3	4	5	6	7	8	9	10	11
Age (yr)	76	76	49	43	60	53	20	53	62	48	1 mo
Gender	Male	Male	Male	Female	Female	Male	Female	Male	Female	Female	Male
City	Shanghai	Guangzhou	Guangzhou	Shenyang	Shenyang	Shenyang	Tianjin	Xi'an	Beijing	Jinan	Wuhan
Specimen type	Blood	Sputum	Sputum	Bile	Bile	Blood	Blood	Blood	Blood	Blood	Blood
Date of specimen collection (yr/mo/day)	2013/4/13	2013/4/2	2013/3/6	2013/1/31	2013/2/18	2013/1/29	2013/3/17	2013/2/25	2013/7/23	2013/7/26	2013/3/26
Infection type	Septicemia	Pneumonia	Pneumonia	Liver abscess	Abdominal infection	Liver abscess	Septicemia	Bacteremia	Septicemia	Bacteremia	Septicemia
Clinical outcome	Survived	Survived	Survived	Unknown	Unknown	Unknown	Survived	Survived	Survived	Survived	Survived
β-Lactamase(s)	SHV-75, CTX-M-55	SHV-11, TEM-1, CTX-M-55	SHV-11, TEM-1, CTX-M-55	SHV-11, TEM-1, CTX-M-like	SHV-11, TEM-1, CTX-M-like	SHV-11, TEM-1, CTX-M-like	SHV-148	CTX-M-14	SHV-11, CTX-M-14	CTX-M-14	SHV-11, TEM-53
Virulence-associated features
String test	+	−	+	+	+	+	+	+	+	−	+
cps genotyping	K20	K1	K1	K1	K20	K20	K2	K1	K nontypeable	K nontypeable	K2
rmpA gene	+	+	+	+	+	+	+	+	+	+	+
MLST	ST420	ST23	ST23	ST268	ST268	ST268	ST1658	ST367	ST17	ST35	ST65

Patient characteristics and variables associated with hvKP isolates with and without ESBL production.

Patients with neutropenia (37.5% versus 5.6%; P = 0.020), history of systemic steroid therapy (37.5% versus 5.6%; P = 0.020), and combination therapy (use of multiple antimicrobials) (62.5% versus 16.7%; P = 0.009) were more likely to be infected with ESBL-producing hvKP.

MLST genotyping.

MLST identified 100 STs among 230 K. pneumoniae isolates. The most prevalent ST in K. pneumoniae isolates was ST23 (10.9%), followed by ST11 (5.2%). ST23, ST268, ST375, ST412, and ST660 were strongly associated with hvKP, while ST11, ST15, and ST37 were more common in the cKP group. ST17, ST23, ST35, ST65, ST268, ST367, ST420, and ST1658 were identified in hvKP isolates with ESBL production. In addition, STs showed different distributions among different cities (Fig. 1). In Wuhan, ST23 was the most dominant, with a prevalence of 21.7%, while ST11 was the most common in Zhejiang, with a prevalence of 25%.

DISCUSSION

To our knowledge, this is the first systematic multicenter study on hvKP in China, and it provides the most comprehensive understanding of hvKP defined by genetic background. According to previous designation criteria for hvKP, 30.9% of K. pneumoniae isolates were identified as string test positive in our study. The proportion was lower than that reported in previous studies conducted in a single center in China, with a prevalence of 33% in Beijing (7). However, it remains unclear whether all hvKP isolates had this phenotype. Serotypes, STs, and other genomic background data should be taken into consideration when defining hvKP. Therefore, the prevalence of hvKP might be underestimated due to the lack of objective diagnostic methods.

Aerobactin-positive isolates are associated with more serious infections (11, 23). Recently, aerobactin was considered a defining genetic trait and potential antivirulence target for hvKP (24). In our study, 37.8% of K. pneumoniae isolates were defined as hvKP by aerobactin. The prevalence showed a varied geographic distribution, with the highest rate in Wuhan and the lowest in Zhejiang. In Wuhan, 34.8% of the isolates belonged to ST23, ST65, and ST86, which are closely related to high virulence (9, 25). However, resistance-associated STs, including ST11 and ST437, were prevalent in Zhejiang, with a prevalence of 50%, and no isolate belonged to ST23 (26).

The prevalence of resistance in hvKP was still lower than that in cKP in the present study, particularly with regard to ESBLs. Contrary to the previous view that the prevalence of antimicrobial-resistant hvKP is low, a high percentage (12.6%) of ESBL production was found among hvKP in our study. Most of them carried bla_CTX-M genes, indicating the compatibility of plasmids containing bla_CTX-M genes with hvKP strains. Even more disturbing is that CTX-M-type ESBL is the most common genotype in China (27). The widespread dissemination of the CTX-M plasmid into hypervirulent variants from different STs is a frightening prospect in the future. Such isolates could produce various broad β-lactamases and ESBLs with a high level of resistance to multiple antimicrobials. Recently, carbapenemase-producing hvKP isolates have been identified in clinical settings, causing fatal infections (12, 16). K. pneumoniae carbapenemase-producing plasmid could be transferred into hvKP strains successfully (28). The situation calls for an immediate response to multidrug-resistant hvKP infections.

hvKP is clinically important because it can cause severe infections, and it should be given priority in clinical settings. Our results also demonstrate that hvKP is more likely to cause liver abscess, sepsis, and invasive infections than cKP, especially for patients in non-ICU departments and with no hospitalization prior to infection. The risk of ESBL-producing hvKP infection increased with neutropenia, systemic steroid therapy, and combination therapy. Therefore, compromised immunity is an important factor for antimicrobial-resistant hvKP infections.

There were some limitations to our study. Genotypic identification did not necessarily imply the expression of aerobactin; thus, further studies are needed to better define the hvKP strains. In addition, K. pneumoniae isolates were collected only in China, and whole-genome framework was not investigated. Therefore, little is known about the lineage diversification in our isolates, and it is unclear that our results can be generalized to hvKP outside China (29, 30). A study that includes more isolates from other regions, especially for antimicrobial-resistant strains, is needed.

In conclusion, the prevalence of hvKP was high in China, with a varied geographic distribution. Such strains were more likely to cause serious infections, such as liver abscess and sepsis in clinical settings. Our study highlights the urgent need to enhance clinical awareness and management of hvKP infections. Given the high prevalence of CTX-M ESBLs in China, the propensity for the acquisition of resistance genes among hvKP isolates from different STs, especially for immunocompromised patients with multiple antimicrobials treatment, should increase our concern.