Molecular Epidemiology of Klebsiella pneumoniae Strains Causing Bloodstream Infections in Adults.

Molecular epidemiology of Klebsiella pneumoniae bacteremic strains allows for a better understanding of preventive and therapeutic strategies. Clinical and microbiological characteristics of 348 K. pneumoniae bacteremia cases (2007-2009) were retrospectively characterized by multilocus sequence typing and extended-spectrum beta-lactamases (ESBL) production. Overall, 223 (64.08%) cases were nosocomial (NA), 58 (16.67%) healthcare associated, and 67 (19.25%) community acquired. The main infection origins were urinary tract (16.6%, 50.0%, and 43.3%), biliary tract (10.8%, 24.2%, and 31.3%), and catheter-related infection (39.9%, 5.2%, and 0%). The 30-day mortality rate was around 20%. The rates of resistance were around 45% the highest being among NA cases, and ESBL production was detected in 7.2% of cases. A total of 161 different sequence types were grouped into 13 clonal sets by e-burst analysis. No relationship could be established between clonal sets and the origin of infection or the healthcare-related settings. The high genetic variability among the isolates suggests their intrapatient endogenous origin.


Introduction
K lebsiella pneumoniae is an important human pathogen in the community and in the hospital setting. K. pneumoniae is the second most common cause of Gramnegative bloodstream infections after Escherichia coli. This microorganism is also well known because of the increasing rates of antimicrobial resistance mainly associated with the hospital setting and with extended-spectrum beta-lactamase (ESBL) or carbapenemase production. 1,2 In recent years, the number of infections (pneumonia or urinary tract infections) caused by drug-resistant K. pneumoniae in hospitalized patients is increasing. Moreover, K. pneumoniae bloodstream infections are associated with patients with comorbidities such as diabetes mellitus, cancer, chronic liver disease, and biliary disease. 3 The molecular epidemiology of multidrug-resistant (MDR) K. pneumoniae strains and those associated with nosocomial (NA) outbreaks has been largely studied. 4,5 However, the molecular epidemiology of K. pneumoniae causing bacteremia outside the hospital is less known. 6 Recent studies have defined the major lineages of K. pneumoniae, including both NA and community-acquired isolates. 7 Bloodstream infections were detected in patients with comorbid conditions and complications, such as diabetes mellitus (related to community-acquired [CA] bacteremia), malignancy (mainly related to NA bacteremia or healthcare-associated [HCA] bacteremia) or cardiovascular diseases (mainly related to NA). 8,9 In our study, as in previous reports, urinary tract, biliary tract, catheter, and respiratory tract were the most common point of entry of bacteremia. 3,4,10 Nowadays this is a public health problem according to the clinical and epidemiological impact as well as to the economic and social costs. 11,12 In this study, we analyze the clinical and molecular epidemiology of K. pneumoniae strains isolated from blood cultures of adult patients attended over a 3-year period in a teaching hospital, in Barcelona, Spain. Moreover, NA, HCA, and CA bacteremia episodes caused by K. pneumoniae were analyzed. 30,000 admissions per year. It is a 900-bed university teaching hospital for adult patients located in the Barcelona metropolitan area, Spain, providing both medical and surgical care without pediatrics, obstetrics, and burn units. As hospital routine, blood cultures are performed on patients with symptoms of infection following standardized protocols.
K. pneumoniae isolates causing bloodstream infection from January 2007 to December 2009 were included in this study. Clinical data of patients, including patient age, sex, comorbidities, and infection source were prospectively collected and included in a computerized prospective.

Definitions
NA K. pneumoniae bacteremia was considered when the infection occurred after 48 hours of patient being admitted. HCA K. pneumoniae bacteremia was considered when the bloodstream infection sample was detected within 48 hours of the patient's admission, and it met any of the following criteria: a patient being with health system contact ( ‡2 hospitalization days in the 3 previous months to the recent admission, receiving any nursing care, undergoing hemodialysis in the previous month, or nursing home residents). Finally, CA K. pneumoniae bacteremia was considered when a positive bloodstream infection sample was detected in the first 48 hours of admission in a patient without any of the previous criteria. 13 Bacteremia was defined by the isolation of K. pneumoniae in blood cultures together with clinical symptoms of infection. Two episodes were considered different when they occurred at least 1 month apart.

Antimicrobial susceptibility testing
The minimum inhibitory concentration of antibiotics was tested by microdilution using the commercial method Mi-croScan (Beckman Coulter). The antibiotic susceptibility was interpreted according to the Clinical Laboratory Standards Institute (CLSI) criteria. 14 The multidrug resistance pattern and the presence of ESBL were assessed by the Kirby-Bauer disk diffusion method.
All multidrug-resistant strains were retested by disk diffusion method following the CLSI recommendations. The antibiotics tested were penicillins, third-generation cephalosporins, aztreonam, carbapenems, aminoglycosides, quinolones, and cotrimoxazole. To assess the ESBL phenotype, a double-disk diffusion test was performed to show the synergy of cefuroxime, cefotaxime, ceftazidime, and cefepime with amoxicillin/clavulanic acid.

Molecular typing
K. pneumoniae isolates were typed by pulsed-field gel electrophoresis (PFGE) after restriction with XbaI. Band patterns were visually compared using the criteria described by Tenover et al. 15 At least one isolate for each PFGE cluster was subjected to Multilocus Sequence Typing (MLST) following protocol 2, described at the MLST web site of the Institute Pasteur (Paris, France; http://bigsdb.web.pasteur.fr/ Klebsiella/Klebsiella.html). The allele number and Sequence Type (ST) were assigned using this MLST website.
Bacteremia data were analyzed using the goeBURST algorithm and an extension of the goeBURST rules to draw a full Minimum Spanning Tree (MST). The analysis of MLSTs was performed using Phyloviz software (https:// online.phyloviz.net/index). 16 The goeBURST Full MST algorithm represents STs in a combination of circles and lines separating the STs according to their allelic differences.
To differentiate between the recently described K. pneumoniae-related species (Klebsiella variicola, Klebsiella quasipneumoniae, and K. pneumoniae sensu stricto), we performed a MLSA (multilocus sequence alignment) as described previously and using the reference ST described for these three species. 17 Putative ESBL strains as deduced for the results of the double-disk diffusion assay were subjected to a multiplex polymerase chain reaction (PCR) assay confirming the detection of bla SHV , bla TEM , bla CTX-M , and bla OXA family genes. 18 For OXA-positive strains, the PCR amplicons were sequenced. Among CTX-M-positive strains, a PCR plus sequencing to differentiate families CTX-M-1 and CTX-M-9 groups was performed. Finally, because this PCR scheme did not differentiate ESBL or non-ESBL enzymes of TEM and SHV families, further PCR plus sequencing analysis was performed [19][20][21] (www.lahey.org/Studies). The PCR products were sequenced by Sanger methodology at Macrogen, Inc. (Seoul, Korea)

Statistical analyses
Statistical analyses were carried out using PASW Statistics 18. The odds ratios (ORs) and 95% confidence intervals were calculated, and Fisher's exact test or chi-square test were used for statistical analysis as appropriate. All probabilities were two-tailed, and a p-value <0.05 being considered statistically significant.

Results
Microbiological and clinical features of patients with NA, HCA, or CA bacteremia due to K. pneumoniae During the study period ( January 2007 to December 2009) 348 episodes of Klebsiella spp. bacteremia were detected in 335 patients. Eleven patients had two episodes and one patient had three episodes (defined as those occurred at least 1 month apart). Table 1 shows the characteristics of the 348 bacteremia episodes in the three groups analyzed. The overall mean age (-standard deviation) was 63.73 -15.81 years and 210/348 (60.35%) were males.
Comparative of molecular typing in NA, HCA, and CA bacteremia Molecular typing of bacteremias was performed by PFGE and MLST. Although we identified minor clusters among NA isolates, a high genetic diversity was observed in all three groups. In fact, the Simpson's diversity index was 0.983 in NA, 0.988 in HCA, and 0.996 in CA.
Overall, by PFGE/MLST analysis we obtained 161 different STs. By groups using Phyloviz software, we found 127 STs among 191 NA isolates, 42 STs among 54 HCA isolates, and 55 STs among 59 CA isolates. Fourteen STs were found in NA and HCA isolates, 14 STs in NA and CA, and 3 STs in HCA and CA. Finally, seven STs were found in all three groups: ST13, ST15, ST17, ST35, ST37, ST42, and ST45 (Supplementary Table S1 [Supplementary Data are available online at www.liebertpub.com/mdr] and Fig. 1). Seventeen STs (shared by three or more isolates) accounted for 38.1% of the isolates.
After e-burst analysis, there were 13 clonal groups (GC1-CG13; defined as an ST and single locus variant [SLV]), and 125 singletons. The 13 clonal sets accounted for 123 isolates; however, there were differences among them (Supplementary  Table S1). There was 28 singletons with more than one isolate (n = 87 isolates) and 97 singletons, which only presented one isolate. For instance, clonal complex 14 accounted for 28 isolates with 5 STs. Most of them had caused NA bacteremia. In fact, this CC14 was related to an outbreak of K. pneumoniae producing OXA-1 beta-lactamase in our hospital. 22 Other clonal groups accounting for more than 10 isolates were associated either with CA, NA, and HCA episodes. The same occurred among the 28 singletons accounting for more than one isolate. When the origin of bacteremia was analyzed, no relationship between ST and focus of infection could be established (Supplementary Table S2).

Discussion
Klebsiella pneumoniae is an important cause of Gramnegative bacteremia all over the world and has been largely analyzed in the hospital setting causing infections due to multiresistant strains. 9,23-26 In addition, K. pneumoniae is a frequent cause of other endogenous NA infections without bacteremia (urinary tract infection, surgical infection, etc.), although some geographical differences have been detected. 8,10 In the community, K. pneumoniae is the second cause of both urinary tract infections and Gram-negative CA bacteremia. [27][28][29] In this study, we analyzed clinical and molecular characteristics associated with K. pneumoniae bacteremia, including NA, HCA, and CA episodes. 13 Our study, in agreement with others, shows that K. pneumoniae bacteremia has different clinical features in the community or in the hospital. 9 Moreover, healthcarerelated episodes were similar to CA ones. There were differences regarding the foci of infection, the urinary and the bile tract being the most frequent origin in the CA and HCA, whereas catheter-related bacteremia was the most frequent among NA cases. In our series, as in others from Europe and North America, the bile tract is a main origin of K. pneumoniae bacteremia in outpatients, whereas in Asian or Oceania countries the origin related to liver abscess is more frequent. 3,9,29,30 In any case, K. pneumoniae bacteremia occurred mainly in elderly patients with underlying conditions. As described for other infections, diabetes mellitus was an important underlying disease for K. pneumoniae bacteremia in all the three groups. 6,26 Malignancy was mainly present among HCA and NA bacteremia episodes; in fact, this microorganism is a frequent cause of Gram-negative bacteremia in these patients. 31 Although the highest mortality rate was found among patients with NA bacteremia, these differences were not statistically significant and were similar to those found by other authors. 9,[32][33][34] In our study only 7.2% of the isolates produced an ESBL and these were mainly associated with CTX-M family. 2,35,36   Longer hospitalization, invasive procedures, and inappropriate antibiotic therapy or antibiotic use could raise the prevalence of resistant strains or ESBL-producing K. pneumoniae. 9,25,33 In the early 1990s and 2000s the ESBL-producing K. pneumoniae were associated with the hospital setting. However, the number of HCA and CA ESBL infections is currently increasing. 2,4,35,36 Nevertheless, community-acquired ESBL-K.pneumoniae infections were infrequent in our series, reflecting that hospital environment is a predominant site to disseminate these strains. 3,4,32,36 In this way, the higher antibiotic resistance rates were associated with NA cases. However, these rates were similar to CA or HCA isolates when the clonal outbreakrelated isolates were excluded. On the other hand, the rates of quinolone and cotrimoxazole resistance were up to 10% and 15% for CA and HCA isolates, which preclude their use as empirical therapy in this setting. Moreover, the presence of mobile elements with resistance determinants to fluoroquinolones and to cotrimoxazole has been described for other pathogens, such as E. coli. 2,37,38 The genetic diversity of K. pneumoniae isolates has not been extensively studied outside hospital outbreaks. In the present study, we analyzed this diversity and we found a high heterogeneity among the isolates. However, some clonal groups were found and were present in both the hospital and the community. Moreover, we found some antibiotic-susceptible strains with STs previously associated with ESBL or carbapenemase-producing strains. [39][40][41][42] For instance, all ST20 isolates in our series were antibiotic susceptible, whereas this ST has been associated with KPC-or NDM-producing isolates. However, all four isolates of this ST20 caused NA or HCA bacteremia, and this is probably a hospital-adapted clone with the ability to acquire resistance determinants.
Nevertheless, the high genetic variability among NA K. pneumoniae bacteremia isolates suggests that these infections had an endogenous origin. Despite the heterogeneity, the major cluster (CG1) included 16 patients with NA bacteremia episodes caused by ST14 strain harboring bla OXA-1 gene and resistant to quinolones and aminoglycosides. 22 We could not link the origin of infection with specific STs or clonal complex (CC), including the hypervirulent clones. For instance, ST23 commonly associated with patients with pyogenic liver abscess was also found causing catheterrelated bacteremia as described previously. 43 Similar results were observed in other studies, where no relationship was detected between pathogenicity of strains and specific MLST. 26 It seems that K. pneumoniae infections varied regarding clinical characteristics of patients, geographic distribution of serotypes, and the presence of epidemic strains. 29,44 The present study has two main limitations: first, it was performed at a single hospital; and second, the molecular typing was performed retrospectively, and some cases were missing. However, it is an important series which improves the knowledge of molecular features of K. pneumoniae causing bacteremia and its relationship with the origin of infection and the contact with the healthcare system.

Conclusions
This study shows a high genetic variability among K. pneumoniae isolates causing bacteremia and suggests they may be considered endogenous infections. The higher resistance rates observed in the hospital setting could be linked to an NA outbreak. Finally, high-risk clones associated to MDR could be found as susceptible in the healthcare-related settings.