Genotypic Analysis of Klebsiella pneumoniae Isolates in a Beijing Hospital Reveals High Genetic Diversity and Clonal Population Structure of Drug-Resistant Isolates

Background The genetic diversity and the clinical relevance of the drug-resistant Klebsiella pneumoniae isolates from hospital settings are largely unknown. We thus conducted this prospective study to analyze the molecular epidemiology of K. pneumoniae isolates from patients being treated in the 306 Hospital in Beijing, China for the period of November 1, 2010–October 31, 2011. Methodology/Principal Findings Antibiotic susceptibility testing, PCR amplification and sequencing of the drug resistance-associated genes, and multilocus sequence typing (MLST) were conducted. A total of 163 isolates were analyzed. The percentage of MDR, XDR and PDR isolates were 63.8% (104), 20.9 (34), and 1.8% (3), respectively. MLST results showed that 60 sequence types (STs) were identified, which were further separated by eBURST into 13 clonal complexes and 18 singletons. The most dominant ST was ST15 (10.4%). Seven new alleles and 24 new STs were first identified in this study. Multiple logistic regression analysis revealed that certain clinical characteristics were associated with those prevalent STs such as: from ICU, from medical ward, from community acquired infection, from patients without heart disease, from patients with treatment success, susceptible to extended spectrum cephalosporin, susceptible to cephamycins, susceptible to fluoroquinolones, and with MDR. Conclusions/Significance Our data indicate that certain drug-resistant K. pneumoniae clones are highly prevalent and are associated with certain clinical characteristics in hospital settings. Our study provides evidence demonstrating that intensive nosocomial infection control measures are urgently needed.

Rapid and discriminative genotyping methods are useful for determining the clonality of the isolates in nosocomial or household outbreaks [15,16]. Multilocus sequencing typing (MLST) is a nucleotide sequence-based approach for characterizing bacterial isolates (http://www.mlst.net/), with the advantage over traditional pulsed-field gel electrophoresis (PFGE) of ease of manipulation and convenient comparison [17,18]. In our previous study, we observed alarmingly high rates of MDR, XDR and PDR strains among K. pneumoniae isolates from a tertiary care hospital in Beijing, China [19]. In addition, data from that study indicate that many of the drug resistance genes were transmissible [19]. Since the genetic diversity, transmission patterns and the clinical relevance of the drug-resistant K. pneumoniae isolates from hospital settings are largely unknown, we thus further conducted this MLST genotyping analysis for K. pneumoniae isolates from the 306 Hospital, a tertiary care hospital in Beijing, China for the period of November 1, 2010-October 31, 2011 with an aim to assess the molecular epidemiology as well as clinical characteristics associated with prevalent K. pneumoniae clones.

Ethics statement
All of the investigation protocols in this study were approved by the institutional ethics committee of the 306 Hospital, Beijing, China. Written informed consent for K. pneumoniae isolates to be collected as well as for their information to be stored in the hospital database for research purposes was provided by participants. Written informed consent was obtained from the next of kin, caretakers, or guardians on the behalf of the minors/children participants involved in this study. Permission for using the information in the medical records of the patients for research purposes was obtained from the 306 Hospital. The Institute ethics committee of the 306 Hospital reviewed that relevant ethical issues in this study were considered.

Study population, bacterial isolate identification, and drug susceptibility testing
The 306 Hospital in Beijing, China is a tertiary care hospital, with 1,100 beds and approximately 25,000 hospital admissions per year. Consecutive non-repetitive K. pneumoniae isolates were collected from patients being treated in the 306 Hospital for the period of November 1, 2010-October 31, 2011. Isolates with ambiguous sequence data for one or more alleles were excluded from the analysis. All isolates were cultured in Luria-Bertani (LB) medium. A total of 175 isolates were confirmed as K. pneumoniae by 16S rDNA sequencing. Drug susceptibility testing (DST) for the K. pneumoniae isolates was performed using the bioMérieux VITEK2 system following manufacturer's instructions. The following 18 drugs were tested: ampicillin (AMP), piperacillin/tazobactam (TZP), ampicillin/sulbactam (SAM), cefazolin (CFZ), ceftriaxone (CRO), ceftazidime (CAZ), cefepime (FEP), cefotetan (CTT), ertapenem (ETP), imipenem (IM), aztreonam (ATM), ciprofloxacin (CIP), levofloxacin (LVX), gentamicin (GM), tobramycin (TOB), amikacin (AMK), trimethoprim-sulfamethoxazole (SXT), and nitrofurantoin (FD). The ESBLs were detected by the bioMérieux VITEK-2 AST-GN13 test. In some cases, the ESBL positivity was further confirmed by the double disk diffusion method according to standard protocols by the Clinical Laboratory Standard Institute (CLSI) [20]. Escherichia coli strains ATCC 25922 and ATCC 35218, Klebsiella pneumoniae strain ATCC 700603 and Pseudomonas aeruginosa strain ATCC 27853 were used as quality control strains for the DST. Clinical records of patients from whom the K. pneumoniae isolates were obtained were reviewed retrospectively.

Genotyping of K. pneumoniae isolates by MLST analysis
Genotyping was determined by MLST analysis. MLST with seven genes (gapA, infB, mdh, pgi, phoE, rpoB and tonB) was performed on isolates according to the protocol described on the K. pneumoniae MLST website (www.pasteur.fr/mlst) [18]. Alleles and sequence types (STs) were assigned by using the MLST database (www.pasteur.fr/mlst/Kpneumoniae.html). Alleles and STs that had not been previously described were submitted to the curator of the database and were assigned new designations.

Assignment to clonal complexes
The program eBURST v 3.0 was used to identify the different clonal complexes [21]. Clonal complexes were defined as groups of two or more independent isolates that shared identical alleles at six or more loci; each complex was named after the putative founder ST. Data from additional 1,380 isolates of K. pneumoniae were obtained from the MLST isolate database deposited at the Pasteur Institute (http://www.pasteur.fr/cgi-bin/genopole/PF8/ mlstdbnet.pl?file = klebs_isolates.xml) [18].

Sequence analysis
The proportions of nucleotide alterations that led to a change in the amino acid sequence (non-synonymous substitution, dn) and the proportions of nucleotide alterations that did not lead to a change in the amino acid sequence (synonymous substitution, ds) were calculated with START2 [22]. Phylogenetic analysis was performed using ClonalFrame algorithm with the software package ClonalFrame version 1.1 [23], using 50,000 burn-in cycles and 100,000 further iterations. Maximum likelihood tree was constructed using PhyML 3.0 under a GTR+I+G model based on the alignment of concatenated sequence from the seven MLST gene loci with 1000 bootstrap replicates [24].

Measurement of clonality
The standardized index of association (I A S ) for the seven loci was calculated using START2 software and 1,000 iterations [22]. I A S was also estimated using 764 unique STs including the new STs in this study and existing STs in the entire MLST isolate database as a whole.

Definitions
Every patient was counted only once during his hospital stay regardless of the number of positive cultures. Each case was differentiated between CA infection and HA infection based on a temporal definition. A CA case was defined as a case with known carriage of K. pneumoniae on admission or with the first culture positive for these bacteria within 48 h of admission [6,25]. A HA case was defined by the first culture positive obtained more than 48 h after admission. MDR was defined as acquired nonsusceptibility to at least one agent in three or more antimicrobial categories, XDR was defined as non-susceptibility to at least one agent in all but two or fewer antimicrobial categories (i.e. bacterial isolates remain susceptible to only one or two categories) and PDR was defined as non-susceptibility to all agents in all antimicrobial categories [7].

Statistical analyses
Data were entered and analysed using the statistical package SPSS for windows (version 15). Two people were independently cross-checked each entry to ensure the quality of data entered into the computer. For categorical data, different groups were compared using the Chi-square test. Univariate and multivariate analyses were used to determine the factors associated with prevalent K. pnuemoniae clones. The factors examined were shown in Table S5. All Potentially associated factors were included in a logistic regression model for multivariate analysis, and they were eliminated using a backward stepwise selection method using a P value threshold of 0.1 for the variables to be remained in the model. Mantel-Haenszel odds ratios (ORs), 95% confidence intervals (CIs) and corresponding P values were reported. P value of ,0.05 was considered to be statistically significant.

Demographic and clinical characteristics of the patients
From November 1, 2010 to October 31, 2011, a total of 175 non-repetitive hospitalized patients who had K. pneumoniae isolates available were subjected to DST at the 306 Hospital. Twelve (6.9%) patients were excluded from the study as a result of lowquality sequencing results for one or more of the 7 house-keeping genes of their isolates. Among the remaining 163 patients, 126 (77.3%) were HA cases and 37 (22.7%) were CA cases. Eighty (49.1%) isolates were ESBL positive and 83 (50.9%) isolates were ESBL negative. The proportion of the male and female patients were 73.6% (120/163) and 26.4% (43/163), respectively. Fiftyfour (33.1%) of the patients were Beijing residents and the rest were from other provinces of China (non-Beijing residents). The median (6SD) age of the patients was 74.50619.16 years (range 1-98 years). The majority of the patients were from medical ward (75/163, 46.0%) and intensive care unit (ICU) (49/163, 30.1%). The main source of the specimens was sputum (121/163, 74.2%). The proportion of MDR, XDR, PDR, and other types of K. pneumoniae isolates were 63.8% (104/163), 20.9% (34/163), 1.8% (3/163) and 13.5% (22/163) respectively. Notably, the rates of resistance to most drugs were much higher among ESBL positive isolates than ESBL negative isolates. The epidemiology of CA and HA cases based on the admission date is shown in Fig. 1. More detailed information on relevant demographic and clinical characteristics of the study population is summarized in Table 1.

Measurement of clonality and selection pressure
Analysis of the data set of 163 isolates from patients yielded an I A S value of 0.1251. This was decreased to 0.0841 when only one representative of each sequence type was included. Significant linkage disequilibrium was detected in both analyses. It remained significant when only one representative isolate for each ST in the  entire isolate database were considered (I A S = 0.1133), thus the observed linkage disequilibrium is not due to sampling bias. The values of I A S in these analyses were low suggesting the weakly clonal population. The dn/ds ratios for all loci were significantly less than 1 (Table S2), indicating there was no strong positive selective pressure on the genes.

Phylogenetic analysis
The maximum likelihood tree show a phylogenetically distinct cluster of related STs (ST877, ST886, ST894, ST856 and ST526) is formed with strong bootstrap support. Four of them (ST877, ST886, ST894 and ST856) are new STs in this study, and 2 of these (ST886 and ST894) differed from recognized STs at three loci (Fig. 4). The drug resistance profiles and epidemiological information of those clones belonging to phylogenetically distinct cluster of related STs are shown in Table S3.

Drug resistance profiles of K. pneumoniae isolates grouped by CCs and STs
The detailed information on drug resistance profiles of K. pneumoniae isolates grouped by CCs is shown in Table 3. All CCs showed high proportion of resistance to penicillin. We did not observe significant differences between ESBL blaCTX-M and blaTEM/SHV groups ( Table 2 and Table 3). We also compared the drug resistance profiles, the corresponding drug resistanceassociated genes, as well as the clinical characteristics of the prevalent K. pneumoniae clones (Table S4). We observed that some isolates with the same STs were from patients who were hospitalized in the same period and who shared the same wards, especially in the first ward of the department of respiration, neurointensive care unit (NICU), and ICU. In addition, we detected a large variety of ESBL genes (such as bla SHV and bla CTX-M , and bla TEM ), as well as genes associated with resistance to fluoroquinolones (such as qnrA, qnrB, qnrC, qnrD, qnrS, aac(69)-Ib-cr, and qepA) and aminoglycosides (such as aacC2, addA1, and aacA4) in the Factors associated with prevalent K. pneumoniae clones Factors associated with prevalent STs compared with nonprevalent STs upon univariate and multivariate analysis are shown in Table S5 and Table 4, respectively. Multiple logistic regression analysis revealed that isolates from ICU (OR, 13.802), from medical ward (OR, 5.154), from community acquired infection (OR, 3.106), from patients without heart disease (OR, 3.446), from patients with treatment success (OR, 6.691), susceptible to extended spectrum cephalosporins (OR, 8.633), susceptible to cephamycins (OR, 3.430), susceptible to fluoroquinolones (OR, 6.247), with MDR (OR, 3.111) were significantly associated with the prevalent STs.

Discussion
The present study describes the genetic diversity of drugresistant K. pneumoniae isolates in a tertiary hospital in Beijing. Twenty-four new STs were detected, demonstrating that the MLST database is still novel and continuously growing. The isolates originated from both CA and HA infections. It is noteworthy that in contrast to the observations from a study from Germany [1], the levels of resistance were equally high among both HA and CA K. pneumoniae isolates in our study, indicating a very large reservoir of resistance in the community around Beijing. We further analyzed the drug resistance profiles, the corresponding drug resistance-associated genes, as well as the clinical characteristics of those isolates with the same STs. We observed that the rates of resistance to most drugs were much higher among ESBL positive isolates than ESBL negative isolates. But we did not observe significant differences between ESBL blaCTX-M and blaTEM/blaSHV groups in their distribution of STs or CCs. We noticed that some prevalent isolates with the same STs were from patients who were hospitalized in the same period and who shared the same wards, especially in the first ward of the department of respiration, NICU, and ICU, this observation suggests that those clonal isolates were transmitted in the hospital, causing infections among immunocompromised patients in those wards. In addition, many patients with this ST shared the same wards such as the first ward of the department of respiration and cardiac care unit, suggesting there is currently an ongoing transmission of isolates of this ST in those wards of the hospital.
Notably, among the three isolates of PDR K. pneumoniae, two isolates belonged to ST15, which is the most prevalent ST in this study. The clinical and molecular epidemiological data suggest that the majority of those ST15 isolates were from patients with severe underlying diseases such as pulmonary infection, renal dysfunction, heart failure, and chronic obstructive pulmonary disease, etc. A study from Spain reported that VIM-1 producing K. pneumoniae ST15 clone has a high capacity to spread among ICU patients with severe underlying conditions [2]. ST15 is also widespread in other countries such as Denmark, Hungary, Korea, Malaysia, Singapore and Taiwan [26][27][28]. Interestingly, one of the PDR K. pneumoniae isolate (TZSKP-82) possesses the bla OXA-48 gene. OXA-48 carbapenemases were first isolated from K. pneumoniae in Turkey in 2008 [10,29]. To the best of our knowledge, this is the first documented case of OXA-48-producing K. pneumoniae in China.
ST23 was another prevalent ST in this study. ST23 was the primary founder of CC23, and isolates sharing this ST were found in other countries. Previous studies showed that ST23 isolates were closely related to liver abscess [30,31]. However, the ST23 isolates in this study were diagnosed with different kinds of illnesses including cerebral infarction, renal dysfunction, and liver abscess, etc.
Another frequently identified ST is ST11, which is a single locus variant of ST258. ST258 is a well known lineage of K. pneumoniae which plays an important role in the global spread of carbapenemases. ST258 was not found in our study. There are only a few nucleotide differences between ST11 and ST258 in their tonB alleles. ST258 was proposed to be probably arisen from ST11 by acquisition of the tonB-79 allele, followed by acquisition of carbapenem-resistance genes on mobile elements [32]. A recent study was conducted to analyze carbapenem-resistant K. pneumoniae isolates from 13 hospitals in nine cities covering five provinces in China, and they found that ST11 was the most dominant clone among the 95 carbapenem-resistant K. pneumoniae isolates in China [33]. Although ST11 is not the most dominant one in our study, it is among one of the prevalent clones and all those ST11 isolates harbored ESBL genes. In addition, the genetic relatedness of ST11 with ST258 is of great concern.
From the maxiumum likelihood tree, a phylogenetically distinct cluster of related STs (ST877, ST886, ST894, ST856 and ST526) is formed. Four of them (ST877, ST886, ST894 and ST856) are new STs in this study, and 2 of these (ST886 and ST894) differed from recognized STs at three loci. These results suggest a recent clone is emerging locally. Further identification of the drug resistance profiles and epidemiological information of the 4 isolates with new STs imply that they are relatively susceptible (with resistance to only a few drugs and no resistance genes detected).
One of the challenges for infection control is to discern the prevalent clones as well as their clinical relevance, especially the treatment outcome, of those isolates, so as to provide information for better management measures. We thus further examined the association between certain prevalent K. pneumoniae isolates (based on the frequency of the STs) and the demographic and clinical features as well as mortality of the patients from whom the isolates were obtained. We noticed that some results (For example, the results for the association with the intensive care unit) from univariant and multivariant analysis differ greatly. We think that the results from the multivariant analysis should be more reliable and those variables with significant P values in the multivariant analysis were identified to be independently associated with prevalent K. pneumoniae clones after excluding some less significant variables and taking into consideration of the confounding factors during the multivariate analysis. Data from multiple logistic regression analysis revealed that isolates from ICU, from medical ward, from community acquired infection, and with MDR were significantly associated with those prevalent clones. In addition, we noticed that those prevalent clones were more frequently associated with patients without heart disease, who were susceptible to extended spectrum cephalosporins, cephamycins and fluoroquinolones, and who had better treatment outcome. A recent study from Taiwan reported that the ESBL positive E. coli ST131, which has emerged in bloodstream infections in Taiwan, is not related to more health-care-associated risk factors, and the E. coli bacteremia caused by this clone did not exhibited a higher mortality rate [34]. Thus the prevalent clones are not always the Figure 3. 50% majority-rule consensus phylogenetic tree derived from ClonalFrame for the 7 housekeeping loci in K. pneumoniae, displaying the clonal relationship between the STs and drug resistance of the K. pneumoniae population. doi:10.1371/journal.pone.0057091.g003        most virulent ones or those associated with more severe clinical features or outcome.
Since the prevalent clones have a great potential of transmission among patients, the observation that those clones were significantly associated with MDR, HA infection, as well as nosocomial infections in the crowded ICU, together with the identification of a large variety of drug resistance-associated genes, particularly those ESBL genes, as well as genes associated with resistance to fluoroquinolones and aminoglycosides [35][36][37][38][39][40][41], in those prevalent clones suggest that although those isolates are associated with less severe clinical features and outcome, they could be a dangerous reservoir for transmission of drug resistance genes, thus warrant a high degree of awareness and monitoring of those drug resistance determinants in clinical isolates. In addition, these isolates from China had different ESBL genotypes, implying multiple acquisition events and the presence of multiple circulating variants of the clone.
In conclusion, the diversity of the genotypes and the complexity of the resistance phenotypes and determinants found, as well as the potential for widespread dissemination of those prevalent isolates detected in our study suggest that certain possibly less virulent (based on the clinical manifestations of the patients) but highly transmissible drug-resistant clones of K. pneumoniae isolates are currently prevalent among patients in hospital settings in Beijing, emphasizing the continuous hospital-wide surveillance of phenotypic and genotypic drug resistance data, as well clinical characteristics and treatment outcome for the prevalent K. pneumoniae clones is necessary to understand the spread of those successful clones, so as to make better infection control measure against nosocomial infection caused by K. pneumoniae and Enterobacteriaceae, which are closely related to K. pneumoniae and interchange resistance determinants frequently with them. Further in-depth investigation of other important population genetic forces, such as gene flow, natural selection, etc., with more extensive sampling, would validate the interesting observation of an inverse relationship between prevalence and virulence in a statistically robust fashion, as well as to provide more insights into the spatial and temporal population dynamics of drug-resistant K. pneumoniae isolates.      (17), ST562 (14), ST23 (12), ST716 (10), ST11 (9), ST147 (8). b All variables included in the univariate analysis (shown in Table S5) were included in the logistic regression model for multivariate analysis, and they were eliminated using a backward stepwise selection method using a P value threshold of 0.1 for the variables to remain in the model. P values,0.05 were considered to be statistically significant. doi:10.1371/journal.pone.0057091.t004