Molecular Epidemiology of and Risk Factors for Extensively Drug-Resistant Klebsiella pneumoniae Infections in Southwestern China: A Retrospective Study

Background: The increasing prevalence of extensively drug-resistant Klebsiella pneumoniae (XDR-KP) poses a serious threat to clinical anti-infective treatment. This retrospective study assessed the molecular epidemiology of and risk factors for infections with XDR-KP to investigate the mechanism of drug resistance and the epidemiological characteristics. Methods: A retrospective 1:2 case-control study was conducted at Chongqing Renji Affiliated Hospital of the Chinese Academy of Sciences University from January 2015 to December 2017. A total of 69 non-repetitive XDR-KP strains were collected. Patients infected with XDR-KP comprised the case group, and 138 matched patients with non-XDR-KP infection at the same site comprised the control group. The chi-square test and logistic regression were performed to evaluate the related risk factors. Molecular typing was performed by multilocus sequence typing (MLST). Potential resistance genes were detected by polymerase chain reaction (PCR) and sequencing. Predictors of 28-day mortality in patients with XDR-KP infection were also identified in our study. Results: Only tigecycline and polymyxin B showed favorable in vitro drug sensitivity tests. These XDR-KP strains had a high prevalence rate (n = 66, 95.7%) of carbapenemase-related drug resistance genes. Among them, KPC-2 was the most frequently detected gene (n = 52, 75.4%). Particularly, all of the isolates harbored multiple drug resistance genes. Epidemiological analysis showed that fifty-eight XDR-KP isolates were resistant strains with the ST-11 genotype. Multivariate logistic regression analysis showed that ICU admission (OR: 3.28, 95% CI: 1.66–6.49, P < 0.001), tracheal cannula (OR: 3.16, 95% CI: 1.48–6.76, P = 0.003), and carbapenem exposure (OR: 3.16, 95% CI: 1.25–7.98, P = 0.015) were independent risk factors for XDR-KP infection. Solid tumors (OR: 7.22, 95% CI: 1.84–28.34, P = 0.005) and septic shock (OR: 9.46, 95% CI: 2.00–44.72, P = 0.005) were independent risk factors for 28-day mortality from XDR-KP infection. Conclusion: This study showed that XDR-KP isolates were highly resistant and exhibited clonal transmission. ST11 was the predominant epidemic type of XDR-KP producing KPC-2 in Southwestern China. Physicians should be aware of these high-risk patients with notable predictive factors for XDR-KP infection. These findings may provide some recommendation for the diagnosis and treatment of patients infected with XDR-KP strains in Southwestern China.


InTRODUCTIOn
Klebsiella pneumoniae (KP) is one of the most common pathogens of nosocomial infections that often cause severe or fatal infections (Blin et al., 2017;. Carbapenems are currently the most effective beta-lactam antibiotics for the treatment of such severe bacterial infections. With the overuse of carbapenems in clinical treatment over the past decade, the emergence of carbapenem-resistant K. pneumoniae (CRKP) has become the main cause of the failure of clinical antiinfective treatment (Arabaghian et al., 2019;Barnes et al., 2019;Meng et al., 2019). It is worth noting that extensively drug-resistant K. pneumoniae (XDR-KP), which has a broader spectrum of drug resistance than CRKP, should be paid more attention. Because XDR-KP is resistant to most of the antibiotics currently used in the clinic and is only sensitive to tigecycline and polymyxin B, it further reduces the types of drugs available in the clinic, which poses a serious challenge to clinical anti-infective treatment (Du et al., 2016;Liu et al., 2017;Moradigaravand et al., 2017).
High morbidity and mortality are closely related to infection with XDR-KP, with a mortality rate of approximately 40-50% (Hoxha et al., 2016;Zheng et al., 2017). Currently, colistin is one of the few effective antibiotics in the treatment of XDR-KP infection, and the therapeutic options for infections are extremely limited. However, with the emergence of mcr-1positive Klebsiella pneumoniae, which may lead to resistance to colistin, the treatment of XDR-KP infection is increasingly difficult (Rojas et al., 2017;Petrosillo et al., 2019). In addition, because carbapenem genes and other resistance genes can be transmitted frequently through mobile DNA elements, the increasing incidence of XDR-KP infection reported worldwide reflects a serious threat to public health and has even caused fatal outcomes in vulnerable patients. Therefore, it is necessary to explore the molecular epidemiology of and risk factors for infections with XDR-KP (de Campos et al., 2018;Li et al., 2018;.
At present, there have been no reports on the molecular epidemiology and risk factors for XDR-KP infection in Southwestern China. In this retrospective study, we attempted to analyze the clinical characteristics, molecular epidemiology and risk factors for XDR-KP infection collected from Chongqing Renji Affiliated Hospital of the Chinese Academy of Sciences University to provide a theoretical basis for the clinical diagnosis, treatment and control of the spread of XDR-KP infection. Moreover, predictors for mortality in patients with XDR-KP infection were also identified in our study.

Study Design and Setting
This retrospective study was performed from January 2015 to December 2017 in the Chongqing Renji Affiliated Hospital of the Chinese Academy of Sciences University, a tertiary university hospital with a 1,800-bed academic medical centre located in Chongqing, Southwest China. Patients with XDR-KP infection were only sensitive to tigecycline or polymyxin B and resistant to all other drugs in this study. To assess for risk factors for XDR-KP infection, a retrospective 1:2 case-control study was performed. The case-control study groups in our analysis were defined as follows: the case group comprised patients with XDR-KP infection during hospitalization, and patients with non-XDR-KP infection at the same site comprised the control group. Patients admitted for <48 h and those with duplicate isolates were excluded.

Data Collection and Definitions
After patients in both the case and control groups had been confirmed, relevant demographics and clinical data of the enrolled patients were collected from the electronic medical records and clinical microbiology laboratory databases. Survivor and nonsurvivor subgroups were compared to identify predictors of 28-day mortality.

Variables
Thirty-six clinical variables were applied in the analysis of the risk factors. The following parameters were used: (1) demographics (age, sex, hospital transfer, intensive care unit admission, and 28-day mortality); (2) baseline diseases and acquired infection (hypertension, diabetes, severe anemia, hypoproteinemia, hypokalemia, septic shock and solid tumors, respiratory, liver, gastrointestinal, cardiovascular, renal, and endocrine system diseases, as well as pulmonary, intra-abdominal, and urinary tract infection); (3) invasive procedures during the hospital stay (parenteral nutrition, mechanical ventilation, urinary catheter, tracheal cannula, drainage tube, previous surgery in the past 6 months, central venous catheter, and arterial catheter); and (4) antibiotic exposure within 3 months (cephalosporins, carbapenem, aminoglycosides, quinolones, tetracyclines, macrolides, metronidazole, and glycopeptides). These variables were analyzed to identify risk factors for XDR-KP infection.

Definitions
The following terms were defined prior to data analysis. Patients >65 years old were defined as elderly. Severe anemia was defined as a hemoglobin level <60 g/L. Hypoproteinemia was defined as serum total protein level <60 g/L or albumin level <25 g/L. Hypokalemia was defined as a serum potassium level <3.5 mmol/L. Septic shock was defined as sepsis associated with organ dysfunction and accompanied by persistent hypotension after volume replacement (Plante 2016;Howell and Davis, 2017). The source of infection was determined to be pneumonia, urinary tract infection (UTI), surgical site infection, intra-abdominal infection, or line-related infection using definitions from the US Centers for Disease Control and Prevention (CDC) (Kang et al., 2005).

Sample Size Calculations and Statistical Analysis
Based on the report of the CHINET Antimicrobial Resistance Surveillance Program in 2015 (Fupin et al., 2016), we assumed that XDR-KP will comprise 2.2% of the cases and non-XDR-KP controls will comprise 14.1%. To determine a difference at the 0.05 significance level with 80% power, we estimated that we would need at least 66 XDR-KP vs. 132 non-XDR-KP control cases (EpiInfo, version 3.3.2).
All statistical analyses were performed using SPSS version 23.0 for Windows (SPSS, Inc., Chicago, IL, USA). Categorical variables were presented as frequencies and percentages. In the analysis of risk factors for XDR-KP infection and mortality, the univariate analysis was used for each variable. To identify the independent risk factors, variables with P < 0.10 in the univariate analysis were included in the logistic regression model for the multivariate analysis and analyzed using backward stepwise regression. The odds ratio (OR) and 95% confidence interval (CI) were also calculated to evaluate the strength of any association that emerged. For all statistical analyses, P < 0.05 indicated statistical significance.

Antibiotic Susceptibility Test
As shown in Table 1, all 69 XDR-KP isolates were resistant to common clinical antibacterial drugs including carbapenems, by in vitro drug sensitivity tests but were only sensitive to tigecycline and polymyxin B (sensitivity rate 100%). The resistance rate of strains in the XDR-KP infection group was significantly higher than that in the control group, and the difference was statistically significant (P < 0.05).
Drug resistance genes can be transmitted horizontally among bacteria through mobile DNA elements, such as

Risk Factor Analysis of XDR-KP Infection
Univariate analysis of the demographics, baseline diseases, acquired infection, invasive procedures, and antibiotic exposure was performed in the case-control groups. The following factors were most frequently associated with the development of XDR-KP infection: ICU admission, transferring from another hospital, respiratory diseases, pulmonary infection, tracheal cannula, and exposure to carbapenem and glycopeptide antibiotics ( Table 3).

DISCUSSIOn
At present, the emergence and dissemination of the XDR-KP strain is a major worldwide source and shuttle for antibiotic resistance. The increasing prevalence and global spread of these clinically extensively drug-resistant strains gravely threatens public health (Liu et al., 2017). According to the global epidemic trend of XDR-KP strains, the detection rate of XDR-KP strains in the northeastern United States, Italy, Greece, Israel and other European countries was relatively high, and there were even reports of local epidemic outbreaks in some areas (Zhou et al., 2015;Gu et al., 2018). However, molecular epidemiological characteristics differ from region to region (Garg et al., 2019;Kim et al., 2019). To our knowledge, there have been no reports on the epidemiology of and risk factors for XDR-KP infection in Southwestern China. This retrospective study was the first to systematically analyze the clinical characteristics, molecular epidemiology, risk factors of XDR-KP infection and the predictors of mortality. For the drug sensitivity of XDR-KP strains, only tigecycline and polymyxin B still showed favorable in vitro drug sensitivity tests, and they were the most sensitive antibiotics for XDR-KP infection in the present study (sensitivity rate 100%). Currently, tigecycline and polymyxin B are considered to be the last line of defense against XDR-KP infection (Pragasam et al., 2016;Braun et al., 2018;Alves et al., 2019). However, increasing resistance is reported worldwide, in particular due to the plasmid-encoded protein mcr-1 present in pathogens such as Escherichia coli and Klebsiella pneumoniae, which further increases the difficulty of anti-infective treatment (Zhao et al., 2017;Shankar et al., 2019). Focusing on the molecular detection and genetic characterization of XDR-KP strains, we found that all the XDR-KP isolates contained multiple drug resistance genes, suggesting that the wide resistome of strains was related to the plentiful chromosomal and plasmid-encoded antibiotic resistance genes. The data revealed that the emergence of carbapenemase genes and ESBLs was the primary resistance mechanism. Among them, KPC-2 and CTX-M-9 group genes were the most prevalent in this study and were reported to be the most easily transmitted horizontally in KP (Fuga et al., 2019;Perez, 2019;Vargas et al., 2019). In addition, DHA was reported to be the most common gene in KP producing AmpC β-lactamases in many studies, which was consistent with our study (Akata et al., 2019;Kazemian et al., 2019). Notably, under antibiotic selective pressure, the pathogen continuously accumulates multiple drug resistance genes by acquiring plasmids and transferable genetic elements, ultimately leading to XDR strains harboring a 'super resistome' (Navon-Venezia et al., 2017).
MLST could be used to understand the epidemiology at a large scale in local outbreak situations. In the current study, ST11 was the most predominant epidemic type among the six different ST types identified, concordant with the report that ST11 was the major sequence type in Mainland China (Fu et al., 2019;Yu et al., 2019), and different from the international hyper-epidemic lineage ST258. Previous studies have shown that the hyperepidemic clonal complex of multidrug-resistant K. pneumoniae in ST258 clone may be spreading in the northeastern region of Hungary (Toth et al., 2010), which were different from our results. Interestingly, only one of the seven housekeeping genes between ST11 (gapA,infB,mdh,pgi,phoE,rpoB,tonB: 3,3,1,1,1,1,4) and ST258 (gapA,infB,mdh,pgi,phoE,rpoB,tonB: 3,3,1,1,1,1,79) was different, suggesting that they were closely related to each other . Notably, all strains of ST11 XDR-KP isolates harbored the KPC-2 gene in our study, suggesting that the KPC-2 gene was involved in the epidemic dissemination of XDR-KP strains, which has frequently been reported worldwide in the interspecific transmission and dissemination of Enterobacteriaceae (Gao et al., 2019). If not effectively controlled, the ST11 clone could possibly spread widely in the future and even lead to the outbreak of XDR-KP strains.
Univariate analyses revealed that ICU admission, transferring from another hospital, respiratory diseases, pulmonary infection, tracheal cannula and exposure to carbapenem and glycopeptide antibiotics were strongly correlated with XDR-KP infection, which has also been identified by other studies. In addition, ICU admission, tracheal cannula, and carbapenem exposure were identified as independent risk factors for infection with XDR-KP in our study. Moreover, these ICU patients with concomitant infections had more severe underlying diseases and lower immunity, which might have made them more vulnerable to acquiring XDR-KP infection, and invasive procedures such as tracheal cannula and mechanical ventilation would have increased the possibility of drug-resistant bacteria adhesion, which ultimately led to infection with XDR-KP strains (Kiddee et al., 2019;. In addition, the selection pressure of antibiotics played an important role in the development of multiple drug resistance of bacteria. It was generally believed that under the pressure of carbapenem, susceptible strains were inhibited or killed, leading to the proliferation of many drug-resistant strains as dominant bacteria. Moreover, drugresistant genes carried by drug-resistant bacteria could have been transmitted horizontally to susceptible strains by the acquisition of plasmids and transferable genetic elements, thus causing more serious epidemics of drug-resistant bacteria (Lee et al., 2013;Lopez-Camacho et al., 2014;Zhang et al., 2018). Therefore, we should select antibiotics according to drug sensitivity test results, and the abuse of carbapenem antibiotics should be avoided to reduce the production of XDR-KP strains.
Our study highlights the high mortality associated with XDR-KP infection. We found that solid tumors and septic shock were independent risk factors for death caused by XDR-KP infection. Most cancer patients usually have low immunity after radiotherapy or chemotherapy. After infection with "superbacteria" such as XDR-KP, these patients often received inappropriate initial antimicrobial treatment, which might damage prognosis and increase mortality. In addition, drug-resistant bacteria often cause serious infections, resulting in bacteremia or even sepsis in patients, causing septic shock that leads to multiple organ failure or even death (Falcone et al., 2016;Christodoulou et al., 2017;Machuca et al., 2017). Therefore, the surveillance of cancer and septic shock patients infected with XDR-KP should be strengthened to reduce their mortality.

limitations
Given that this was a retrospective, case-control study, it is not possible to determine the absolute risk associated with the identified risk factors. In addition, this retrospective study was conducted at a single medical centre, and our sample size was relatively small, which might have led to errors in statistical analysis and the omission of some other risk factors.

COnClUSIOnS
This retrospective study was the first to systematically analyze the clinical characteristics, molecular epidemiology, and risk factors of XDR-KP infection and the predictors of mortality. Our findings showed that ST11 was the predominant epidemic type of XDR-KP producing KPC-2 in Southwestern China. In addition, ICU admission, tracheal cannula, and carbapenem exposure were identified as independent risk factors for infection with XDR-KP in this study. Moreover, another important finding of this study was that solid tumors and septic shock were independent risk factors for death caused by XDR-KP infection. Therefore, physicians should be aware of these high-risk patients with notable predictive factors for XDR-KP infection to reduce their mortality rate.

DATA AVAIlABIlITY STATEMEnT
The raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher.

EThICS STATEMEnT
The study was approved by the Ethics Committee of Renji Hospital, Chinese Academy of Sciences University, Chongqing, China. The ethics committee waived the need for written informed consent provided by participants, and no additional informed consent was required.