A possible link between improper use of carbapenems and prevalence of carbapenem-resistant Klebsiella Pneumoniae in a hospital

Background: The emergence of carbapenem-resistant Enterobacteriaceae made the treatment dicult, which has become a major issue of public health. A sharp increase of carbapenem-resistance rate in Klebsiella pneumoniae was observed in a maternity and child health care hospital in Zunyi, China, in 2014.To investigate the cause and epidemiology of the carbapenem-resistant Klebsiella pneumoniae (CRKp) in the affected hospital. Methods: In 2015-2016, CRKp isolated from all the clinical samples were analyzed to identify the carbapenem-resistance genes. They were then ngerprinted in order to determine their genetic relationship. Clinical data such as usage of imipenem in 2012-2016 and the nosocomial infection surveillance data were analyzed Results: Thirty ve isolates of CRKp out of 4328 various pathogens were obtained and bla NDM-1 was identied to be the most common resistant gene present in the CRKp isolates. The ngerprint analysis identied 15 major clusters of CRKp isolates. The bacteria with close proximity relationship tended to be from the same wards. However, a few CRKp isolates from different wards were found to be genetically highly related. The clinical data showed a signicantly higher usage of carbapenems in 2012-2013, before the CRKp rate sharply increased in 2014. The nosocomial infection surveillance showed unexpected high rate of failures to meet the requirement of the hospital environment hygiene and hand hygiene in the neonatal ward. Conclusion: The increasing isolation rate of CRKp was associated with poorly regulated usage of carbapenems, impropriate medical practices and the poor hospital environmental hygiene and hand hygiene.

sample of 2015-2016 were analyzed to determine the genes responsible for carbapenem resistance as well as ngerprinting analysis to establish the genetic relationship among these isolates.

Materials
Columbia Blood Agar and Mueller-Hinton agar were purchased from Zhengzhou Beiruite Biotechnology.
Ezup Column Bacteria Genomic DNA Puri cation Kit (Sangon Biotech Shanghai. Inc), ABI 7500 Real Time PCR System (Thermo Fisher Scienti c. Inc.), PowerPac Basic, Mini-Sub cell and Gel Doc™ XR+ system (Bio-Rad Laboratories. Inc) were purchased from speci ed companies.

Clinical specimens
All the non-duplicate CRKp isolates were isolated from the clinical samples collected from various wards by growing on Columbia Blood Agar during 2015~2016. The resistance to antibiotic imipenem was detected by Vitek 2 Compact, then con rmed by imipenem E-test. These CRKp, with their imipenem minimum inhibitory concentrations (MICs) being ≥4μg /ml, were stored in 10% skim milk at -70°C before analysis.

Clinical data collection
The clinical data were retrieved from the database recorded in the department of pathology and other relevant departments of the hospital for analysis. These included the inpatients in 2012 to 2016, and their detection of 4 imipenem-resistant Enterobacteriaceae pathogens, Escherichia coli(E. coli), Klebsiella pneumoniae(K. pneumoniae), Enterobacter cloacae(E.cloacae) and Klebsiella oxytoca(K. oxytoca). Data were used for analysis to determine whether these patients having received imipenem treatment, and other β-lactam in addition to imipenem, had their infection con rmed by isolation of pathogens as well as whether those pathogens were Extended-Spectrum Beta-Lactamases(ESBLs)-Producing, which is one of the criteria to justify the use of imipenem.

Nosocomial infection surveillance data
Nosocomial infection surveillance data in 2012 to 2016 were retrieved from the hospital database, which include the bacteria count of air and doctors' and nurses' hands from the wards in which CRKp cases were identi ed.

Data Analysis
The ngerprint bands of CRKp produced by ERIC-PCR were analyzed, the cluster data table and the relative distance and the Dendrogram were generated by using SPSS 19.0. The differences of genotypes are displayed according to relative distance among strains. representing an overall ~0.81% detection rate. Most of the CRKp isolates (25/35, ~71%) were from the neonatal ward with 6 from the PICU (~17%) and 4 (~11%) from the two pediatric wards. Almost all the CRKp strains were isolated from the sputum samples, except for two with one from tracheal intubation (ID# 9 from neonatal ward) and the other from blood (ID#35 from PICU), as enlisted in Table 2.
Twenty six out of 35 CRKp isolates were detected to contain bla NDM-1 gene (74%); 3 had bla NDM-1 and bla KPC genes (9%); 2 had both bla NDM-1 and bla VIM genes (6%) ; 1 had both bla NDM-1 and bla IMP genes (2.5%) ; 1 had bla SME alone (2.5%); 2 samples were negative for any of the genes in question (6%) ( Table   2). Moreover, 10 randomly chosen bla NDM-1 PCR products were sequenced and the results con rmed that they were all correct speci c products of the bla NDM-1 gene.  PICU. It was noted that most of the genetically related isolated were from the same wards, but there were a few clusters of isolates found in different wards, such as the clusters C, G and J being found in PICU and neonatal wards.

Imipenem overuse prior to the increase of CRKp prevalence
The imipenem treatment of the patients and the imipenem resistance of the pathogens isolated from these patients were retrospectively analyzed. As shown in Figure 2, the imipenem-resistant K. pneumoniae were barely detected in 2012 and 2013, but >16% isolated K. pneumoniae were found to be imipenem resistant in 2014, and the imipenem resistance rate in K. pneumoniae isolates remained high, at ~10%, in the following 2 years. The imipenem resistant E. coli were also found to be increased in 2014 and 2015 (<2%) and further increased to >6% in 2016, while no obvious imipenem resistance was found in the E. cloacae and K. oxytoca isolates in 2012-2016.
Analysis of the clinical data recorded in 2012-2016 (Table 3) shows that in 2012 and 2013, there were 116 and 137 inpatients, respectively, who had been treated with imipenem. Among them, 89 and 77 were either bacterial pathogen negative or not tested for bacterial pathogens at all, representing 77% and 56%, respectively, of the patients having received the imipenem treatment. Among the pathogen positive and imipenem-treated inpatients, approximately 37% and 31% of them in 2012 and 2013, respectively, had their pathogens determined to be ESBLs producing, which should be considered to be quali ed for the use of imipenem. In the following years, the overall number of imipenem-treated inpatients decreased to 92, 55 and 50 in years 2014, 2015 and 2016, respectively. Among them, although the percentage of the pathogen-positive inpatients increased to 64%, 46% and 54%, respectively, the percentage of those with their pathogens determined to be ESBLs producing remained similar, being 42%, 32% and 33%, respectively (Table 3).

Discussion
The growth of the imipenem resistant rate in K. pneumoniae is the fastest among the all species of the Enterobacteriaceae in the hospital in recent years, followed by E. coli (Figure 2). The carbapenem resistance can be caused by the acquisition of plasmids with various carbapenamase genes. Six carbapenemase candidate genes were analyzed in the CRKp isolates in this study. The most frequently identi ed gene is bla NDM-1 , which is different from the previous ndings in China that the common epidemic genotype is found to be bla KPC [7,8].
In 2 CRKp isolates (ID # 35 and ID # 45) ( Table 2), none of these 6 carbapenemase genes were detected, probably because their carbapenem resistance was conferred by other than these 6 resistance genes, or by other mechanisms of resistance, such as porin loss and/or e ux pump activation [9]. Carbapenemases belong to β-lactamases with the capacity to hydrolyze or inactivate carbapenems. As per the Ambler's molecular classi cation, the carbapenemases are divided into class A, B, and D [10]. Serratia marcescens enzyme (SME) and K.pneumoniae carbapenemase (KPC) belong to Class A carbapenemases, their hydrolytic mechanism confers resistance to many β-lactam antibiotics including carbapenems. Verona imipenemase (VIM), Imipenemase (IMP) and New Delhi metallo-β-lactamase (NDM) belong to Class B carbapenemases, which are metallo-β-lactamases that use a Zn 2+ cation for hydrolysis of the β-lactam ring, and resistant to clavulanic acid, tazobactam, and sulbtactam. Class D carbapenemases have been described among four subfamilies of Oxacilinase(OXA)-type β-lactamases, the intrinsically weaker carbapenemase activity is augmented by coupling β-lactamase production with an additional resistance mechanism, such as decreased membrane permeability or increased active e ux [11,12].
Doctors can choose the appropriate treatment according to the prevalence of the speci c carbapenemase gene [13]. The behavior of the bacterial pathogens can change as a result of genetic mutations by insertion/deletion of DNA fragment, homologous recombination, spontaneous induction of the SOS response and replication-transcription con ict to adapt to the environment and to resist attack of antibiotics [14,15], consequently leading to generation of antibiotic resistance including resistance to carbapenem. This is a slow process and it is estimated that the spontaneous mutations occur at a rate of 10 -10 to 10 -9 per nucleotide per generation for most of bacteria under certain growth conditions [14]. In this hospital, the resistance rates of CRKp jumped swiftly from 0.0% in 2013 to 16.9% in 2014 (Figure 2), which was an usual event and warranted an urgent study to identify the mechanism of this growth in order to develop an effective strategy to prevent further spreading, ultimately reducing the presence of CRKp in the affected hospital.
ESBLs are enzymes produced by some bacterial pathogens that are able to hydrolyze 3 rd generation antibiotics such as cephalosporins and aztreonam, leading to nasty resistance to commonly used antibiotics. Carbapenems are beta-lactam antibiotics that have a broad spectrum of activities against many gram-positive and gram-negative, aerobic and anaerobic organisms. Imipenem is the rst antibiotic of carbapenems approved as a potent broad-spectrum antibiotic for treatment in 1985. Subsequently, the other carbapenems such as Meropenem, Ertapenem were developed [16]. They are highly stable to resist the hydrolysis by the ESBLs and AmpC beta-lactamases, which are considered to be the "gold standard" treatment for serious ESBL producing pathogen infections [17] and infection of complex bacteria with multi-drug resistance (MDR). Unfortunately, they were tended to be prescribed for conditions suspected to be infection but without pathogens identi ed or for cases which were not con rmed whether an infection was involved. According to the analysis of the clinical data in this study (Table 3), approximately 1% of all the discharged patients were treated with imipenem in 2012 and 2013, among whom, most of them had no pathogen detected or no test performed to detect pathogens. Even for the pathogens isolated from a small percentage of patients who received imipenem treatment, only approximately one third of the pathogens identi ed were con rmed to be ESBLs-producing and most of them were non-ESBLsproducing bacteria or non-enterobacteriaceae bacteria, such as Haemophilus in uenzae, Streptococcus pneumoniae, etc(Additional le 1), which should have been treated by antibiotics other than carbapenem.
During this study, irregular combination of medications were also found, including cases who received imipenem combined with other β-lactams. It appeared that prescription of carbapenem antibiotics was not well restricted and there was lack of robust justi cation for using carbapenem via collective discussion with experts during those years. Fortunately, the relevant guidelines are now in place. However, the overuse of antibiotics not only occurred at this hospital, but also in other hospitals in China and other countries as having been previously reported [18,19,20]. The Average Annual Growth Rate (AAGR) of De ned Daily Doses (DDDs) of carbapenem antibiotics of 151 grade A tertiary general hospitals in China from 2011 to 2014 was 17.67 % [18]. At the same time, the rates of imipenem resistance in K. pneumoniae trended to increase from 8% to 10.5% in China [19]. In Thailand, the quarterly CRE incidence increased signi cantly from 3.37 per 100,000 patient-days in the last quarter of 2011 to 32.49 per 100,000 patient-days in the last quarter of 2016. The quarterly hospital-wide carbapenem consumption increased 1.58 DDDs per 1,000 patient-days in the corresponding time [20]. Obviously, heavy use of this antibiotic is considered to favor selecting resistance mutation and lead to the emergence of CRKp. Schroeder JW has detailed the mechanism of antibiotic resistance previously [14]. A meta-Analysis has revealed that antibiotic usage can lead to resistance mutation, even in animal husbandry where there is a positive association between bacterial resistance and antibiotic consumption in farm animals [21,22,23,24].
Furthermore, our data analysis has shown that, most of the CRKp were isolated from neonatal ward.
However, the largest number of prescriptions for imipenem was found to be from the pediatrics words in the early days, and the neonatal ward had the second largest number of prescriptions (data not shown).
The clinical data also showed that, rstly, most of the inpatients in the neonatal wards had no corresponding clinical symptoms, albeit they had the highest rate of CRKp pathogen isolation. This suggests that the CRKp in the neonatal patients are likely colonizing in the body, but not responsible for the infection. Secondly, most of the bacteria were isolated from the second time sputum detection from the same inpatient, with the rst sputum examination being negative for pathogens. This is highly likely that they acquired the bacteria in the hospital, and these bacteria probably were spreading in the same ward, even between the different wards. The speculation has been supported by the cluster analysis of the pathogen genetic ngerprinting results. Cluster analysis showed that most of the highly similar genotypes bacteria were isolated from the same wards, especially sub-clusters a, b, c, and clusters J and M which were most closely related, suggesting that they were likely spreading between the different inpatients at the same ward. Moreover, the highly related pathogens were also isolated from different wards, such as clusters C, G and J, they were isolated from Neonatal ward and PICU wards. The highly close genetic relationship among these pathogens suggest that they were spreading between the different wards as a result of reproduction of the same bacterial strain. This speculation is supported by the observation that most of these pathogens carry the same resistant gene bla NDM-1 .
It is known that bacteria can acquire external genetic material through three main approaches, transformation (incorporation of naked DNA), transduction (phage mediated) and conjugation (bacterial "sex"). Conjugation is, in particular, a very e cient approach of inter-bacterial gene transfer that involves cell-to-cell contact [25], found to be a key mechanism for the transfer of antibiotic resistant genes between different bacteria. Observations and personal conversations indicated that the medical staff tended to not do hand disinfection nor changed gloves between examinations of patients. It is highly likely that the hands contaminated with pathogens would transmit them to everywhere they touched, including the patients, other medical staff, medical devices, even objects around the hospital. The transmission can likely further increase the chance for the antibiotic resistant genes to be transferred among the bacteria existing in the hospital environment. The limitation of this study is that other environmental substances in and around the hospital were not thoroughly examined for the existence of bacteria with antibiotic resistance genes, such as soil, water and waste, which could be important contributing factors [27].

Conclusions
In conclusion, this study illustrates the molecular characterization and prevalence of CRKp in the maternity and child health care hospital in Zunyi, China. Base on the analysis, the swift increase in the CRKp was associated with poorly regulated usage of carbapenems, impropriate medical practices and the poor hospital environmental hygiene and hand hygiene, as well as the bla NDM-1 gene is the major contributor for the prevalence of the antibiotic resistance in the pathogen bacteria found in the hospital.
The conclusive information should be useful in assisting optimizing the guidelines for controlling the bacterial infection in hospitals as well as helpful for the doctors to make decision in treating patients with infections.