Molecular Epidemiology, Risk Factors and Outcomes of Carbapenem-Resistant Klebsiella Pneumoniae Infection in a Tertiary Hospital of Eastern China: For a Retrospective Study Over 4 Years

Carbapenem-resistant Klebsiella pneumoniae (CRKP) have undergone extensive dissemination in worldwide resulting in increased mortality. We performed a retrospective analysis of epidemiology and risk factors for the outcome of CRKP infection in a general teaching hospital in China. Methods A molecular and clinical study were conducted for 98 CRKP in a tertiary hospital from January 2013 to December 2016. Carbapenemase gene detection, pulsed-eld gel electrophoresis (PFGE) and multilocus sequence typing (MLST) were performed. Logistic regression was also used to identify the risk factors associating with the 30-day mortality.

Background Carbapenem-resistant Klebsiella pneumoniae (CRKP) has emerged as a major nosocomial pathogen and are being increasing reported worldwide [1][2][3], which has been associated with a high mortality rate range from approximately 16-70% [4,5]. Carbapenem resistance in K. pneumoniae is reported to be caused either by epidemic clones or by the horizontal dissemination of mobile elements. Additionally, the predominant production of Klebsiella pneumoniae carbapenemase (KPC) contributed to the most important mechanism of carbapenem resistance in K. pneumoniae [6]. Of epidemiological signi cance, the international spread of KPC-producing K. pneumoniae is primarily associated with a single multilocus sequence type (ST) or its related variants. In Europe and America, KPC-producing ST258 K. pneumoniae is regarded as one of the most successful multidrug-resistant nosocomial pathogens [3]. However, the international high-risk clone of K. pneumoniae ST11 is frequently reported as a successful pathogen at infections in Asia [6][7][8]. Of note, Wang et al [9] reported that KPC-positive ST11 K. pneumoniae with a highly resistant to meropenem had a signi cant difference on the mortality.
Indeed, data regarding the distribution of these resistant determinants among CRKP isolates collected during a longtime period is still limited in China, and the relationship between molecular characteristics and the clinical outcome is also sparing. In our study, we aimed to describe the epidemiology of CRKP, focus on the molecular characteristics of circulating strains and to evaluate which proportion of strains was susceptible to the clinical outcome.

Study population
A retrospective study for hospitalized specimens which were isolated with CRKP bacteria from January 2013 to December 2016 was conducted in a hospital with 2800 beds. 327 strains of CRKP bacteria (no repeated isolates) were screened out, including 98 strains of nosocomial CRKP pathogenic bacteria.
Medical charts were reviewed and demographic and clinical information was collected. This study was approved by the Ethics Committee of the rst a liated hospital of Anhui Medical University.

Bacterial isolates and genotypic investigation of resistance
Non-duplicate K. pneumoniae isolates which developed 48 h following hospital admission were collected. These isolates were subsequently characterized as hospital-acquired infections with resistant to imipenem, meropenem or ertapenem. Identi cation and antimicrobial drug susceptibility testing were performed using automated systems from VITEK®2 (bioMérieux, France). Organisms were considered resistant to carbapenems according to the CLSI guidelines (2016) [10]. Polymerase chain reaction (PCR) for the common bla KPC gene was carried out in order to identify the carbapemenase gene of all CRKP.

Molecular typing
To track and characterize the clonal relationships of the K.pneumoniae isolates, pulsed-field gel electrophoresis (PFGE) of the XbaI-digested genomic DNA was conducted using a CHEF-DRIII system (Bio-Rad, United Kingdom). Strain relatedness analysis was performed on the BioNumerics software version based on the Dice coefficient (SD) for genetic relatedness. SD values re ect the degree of similarity between PFGE images of different strains, ranging from 0 to 1, where 0 represents completely different and 1 represents completely identical. Different bands are judged to be different types.
According to the similarity coe cient between each two images, the unweighted pair group average method (UPGMA) was used for clustering to construct the cluster tree, which is described as the reference [11]. Concomitantly, multi locus sequence typing (MLST) was the most common technique, which based on genetic variation in seven housekeeping genes (rpoB, gapA, mdh, pgi, phoE, infB and tonB) that together provided a relative genetic pro le. Then, CRKP isolates were assigned a ST number according to the allelic pro les available in the Institute Pasteur database. These ST data can be further de ned by eBURST.

Clinical characterization and de nitions
Clinical data was collected using electronic medical records which extract patient information, including demographic characteristics, chronic underlying diseases, invasive procedures, disease status, antimicrobial therapy, clinical outcomes with 30-day mortality. 1). Infection: all patients were evaluated according to CDC criteria [12] to assess whether the infection was due to CRKP. 2). Emergency sate was de ned as life-threatening organ dysfunction, which was evaluated by an acute change in total Sequential Organ Failure Assessment (SOFA) score of ≥ 2 [13]. 3). Empirical and de nitive therapy: treatment administered before the susceptibility testing results were available was characterized as empirical, whereas treatment given after the results were available (at least one active drug for ≥48 h) was considered as de nitive therapy. The former is divided into appropriate treatment and inappropriate treatment in our study. Appropriate antibiotic treatment was de ned as appropriate antimicrobial therapy with at least one active agent initiated within 48 h after infection onset [14]. De nitive treatment regimens were classi ed as monotherapy (treatment with one in vitro active agent) and combination therapy (treatment with two or more in vitro active agents), and the latter included associated with Tigecycline or with others. 4). Clinical outcome: Infection-related mortality was de ned as the proportion of patients who died only as a direct consequence of the CRKP infection, and with no other plausible explanation in the opinion of the local investigator. Therefore, 30-day infection-related mortality was considered as death for patients who died within 30 days of the onset of the CRKP infection.

Statistical analysis
Statistical analysis was performed using SPSS version 21 software (IBM Corp., Armonk, NY). Univariate analysis was performed to identify factors related to mortality and the ST11 of CRKP. The chi-square test was used for categorical variables and Student's t-test for continuous variables. A logistic regression model was used to identify factors independently associated with 30-day-mortality. Sets of variables that had P ≤ 0.1 in the univariate analysis on mortality were entered into the model. P-values were interpreted together with 95% con dence interval (CI) for the logistic regression model. All tests were two tailed and a P-value of < 0.05 was considered statistically signi cant.

Isolates
A total of 98 patients with CRKP infection were identi ed during the study period. Of these, the median age was 62 years, with nearly three-quarters were male, and 42 patients (42.9%) were admission to intensive care unit (ICU with clone type of PT09 or PT16 were just discovered from ICU admission in 2016 which corresponding to ST11 (Figure 1).

Outcome and ST11
During the study period, 87 patients with CRKP infection were included in the analysis of clinical outcome, while the remaining ones were excluded due to incomplete electronic information or for the inability to follow up with the study.  (Table 3).

Discussion
Nosocomial infections due to CRKP are associated with substantial mortality, yet the clinical signi cance of isolating CRKP from the hospital is unknown. To our knowledge, this study is the rst to describe the epidemiology and 30-day mortality of patients with CRKP infection from the long-term retrospective observation. Through this extensive inquiry, a clear picture has emerged which gradually reveal the particularly noteworthy ndings. First of all, due to the trend of the KPC-positive CRKP strains belonging to the dominant ST11 lineage were observed, thus, our investigation took further insight on the epidemiology. The outcome stressed the diversity in CRKP strains with 24 clones of ST and 59 clusters of PT, however, the strains of predominant ST11 CRKP corresponding to PT10 and PT15 remained in vogue. Importantly, it is worthy noting that the new advantage cluster of PT09 or PT16 were just discovered from ICU admission in 2016 which corresponding to ST11. Secondly, the 30-day mortality of the patients due to CRKP infection was associated with the pathogens of ST11 CRKP. Summary together, our ndings provided a new insight into the risk factors of the clinical outcome and further demonstrated the importance of epidemiology.
In our study, the investigation was conducted on CRKP isolates in 2013-2016. The molecular epidemiological methods had been used for tracking and characterization of CRKP isolates by PFGE and MLST which were accepted to differentiate the species-both of which indicated that CRKP ST11 was identi ed as the predominant sequence type strain in our study. The epidemiology of CRKP is continually evolving, and now the advantage strains of bla KPC -producing K. pneumoniae ST11 have been reported, which is the most frequently detected KPC-producing K. pneumoniae clone in China [15][16][17]. Interestingly, ST11 also occupied predominant isolates according to the prepotent PFGE pattern. As well, the dominant clone PT10 and PT15 corresponded to the advantage type of ST11 in ICU where took a important role in acquisition of carbapenem resistant isolates [18]. It is worthy noting that the only two strains of CRKP with clone type of PT09 or PT16, only second to the dominant cluster PT, were just discovered from ICU admission in 2016 which corresponding to ST11. In this study, it stated that ST11 CRKP payed a dominant role in the process, but aslo emphasized that the homology of these strains was polymorphic and the advantage clusters were changed by evolution. Naturally, an e cient measure could be established to monitor the variation of the prevalence of these CRKP strains.
In this report, the overall 30-day mortality rate of patients infected with CRKP was 36.78% (32/87), which was similar to that reported by previous studies for CRKP infections [19,20]. Several previous studies [21][22][23] had showed that the long-term hospitalization and empirical treatment were independent predictors of death. Through our retrospective study, the length of hospitalization and appropriate empirical treatment both were independently associated with 30-day mortality. The key point of this study was to evaluate the correlation of ST11 CRKP infection with the clinical outcomes, which was rst reported, as we know so far [7][8][9].
Next, this focused investigation on the ST11 CRKP would be taken into consideration. Sequence types of globally prevalent carbapenem-resistant Enterobacteriaceae, including ST11 of K. pneumoniae clonal complex 258, were considered as a high-risk clones [24]. Thus, we believe that the analysis of type ST11 and the association with clinical baseline factor were concerning. Though the underlying mechanisms behind the emergence of CRKP ST11 clones, as well as their associations with the outcomes are yet to be elucidated, we used our acquisition of knowledge by reviewing the relevant literature and attempted to provide some perspectives on the issue. Closely related to ST258, ST11's factors that contribute to the epidemiologic success remain unknown. However, several articles report that chromosomal or plasmid factors, beyond antibiotic resistance, may increase the strain's tness and provide an advantage that underlies its prevalence [25,26]. Additionally, a recent report [27] showed that the KPC-producing K. pneumoniae ST11 clone was resistant to serum killing, which may also be proof for the prevalence. In addition, a review [28] explained that ST11 genome strains carried distinct capsular polysaccharide (cps) regions, and the cps locus was one of the primary determinants of antigenicity associated with K. pneumoniae. Capsule switching was a species-speci c mechanism used by the microbe to escape the host immune response. Croucher, et al [29] inferred that DNA exchange in-and-around the cps regions may be an important mechanism used by K. pneumoniae to rapidly diversify and evolve. Therefore, the success prevalence of CRKP ST11 may be due largely to chromosomal recombination instead of antibiotic resistance. Brisse S, et al [30] found that strains owning some serotypes usually carried a relevant virulence factor content and had been associated with serious human infections, which underscored the potential threat represented by the emergence and diffusion of carbapenem-resistance clones with increased virulence potential [31,32]. Meanwhile, CRKP associated with plasmid-encoded carbapenemases, which could acquire multiclass antibiotic resistance to produce distinct clinical challenges and result in invasive infections with high mortality. Undoubtedly, careful monitoring of carbapenem susceptibilities and rapid identi cation of epidemiological lineages was quite necessary for implementation of infection-control measures to prevent the endemicity of CRKP.
Our study has several limitations. This paper does not unify the the patients of CRKP infection, difference from the other observational studies, which were based on available data that include only patients with bacteremia or bacteriuria [33][34][35]. Additionally, for the fewer strains acquired from the saved isolates in 2014, the study appeared to be incomplete. And, our further research should take insight into the virulent CRKP ST11 strains. Together this research will allow us to expand our knowledge on such strains.

Conclusion
ST11 CRKP payed a dominant role in the process, but the homology of these strains was polymorphic and the advantage clusters would be changed by evolution. Additionally, besides appropriate empirical treatment and hospital stays, ST11 CRKP was independently associated with the 30-day mortality. Therefore, efforts are imminently needed to improve the understanding to the knowledge of the epidemiological status to prevent its spread.