Klebsiella pneumoniae (K. pneumoniae) is a common opportunistic pathogen of community and healthcare associated infections^[1]. Multidrug-resistant K. pneumoniaeis spreading globally and producing serious infections associated with septicemia, pnumonia, urinarytract infection, intra-abdominal infection, and bacterial meningitis^[2]. However, there are very limited therapeutic options available for multidrug-resistant K. pneumoniae, causing an unprecedented public health crisis^[2-3].

As a useful therapeutic agent, carbapenems (CPMs) play an important role in treating infections of multidrug-resistant K. pneumoniae, particularly those carrying genes for extended-spectrumβ-lactamases^[4]. However, with therapeutic use of CPMs in hospitals, CPM-resistant K. pneumoniae has also been emergent and increased rapidly^[5-6]. Resistance to CPMs of K. pneumoniae typically involves several mechanisms, and the production of carbapenemases (CPMases) is an important one^[7-9]. Besides CPM, CPMases can also hydrolyze the majority of β-lactam antibiotics, including penicillins, cephalosporins and cephamycins. Traditionally, CPMases have been assigned to Ambler classes A (KPC-type), B (MBL), and D (OXA-type)^{[2, 10]}. The CPMases were encoded by CPMase-producing genes, such as bla_IMP, bla_VIM, bla_KPC^[2].

Recently, spread of CPMase-producing K. pneumoniae has been reported worldwide, and the phenotypic and genotypic characteristics of the strains from many countries have been identified^{[2, 8-9]}. However, the phenotypic and genotypic studies on clinical isolates of CPMase-producing K. pneumoniae in China were very few and all of these investigations were performed in a very short period^[11-13]. Herein, we conducted a five-year surveillance of CPMase-producing K. pneumoniae in one of the most famous children's hospitals of China, to reveal both phenotypic and genotypic characteristics of clinical isolates of CPMase-producing K. pneumoniae from pediatric patients between 2010 and 2014. Our results demonstrated that a more resistant genotypic subtype, namely New Delhi-Metallo-1 (NDM-1), were gradually increasing and becoming predominant in the pediatric patients, which deserved the attention of both pediatricians and public health officials.

Beijing Children's Hospital, Capital Medical University, is a general pediatric teaching hospital with 970 beds. The hospital has more than 70, 000 admissions per year. All the patients were recruited as inpatients in Beijing Children's Hospital from January 2010 to December 2014.

1, 730 strains of K. pneumoniae were isolated from nasopharyngeal secretions, tracheal aspirations, blood, pus secretions, mid-stream catheterized urine, pleural effusions, catheter and ascites. Routine methods and the Phoenix100 automated microbiology analyzer (BD Research Inc. USA) were employed to identify the bacterial strains.

The CPM-non-susceptible isolates of K. pneumoniae were identified by disk diffusion method using imipenem or meropenem disks. Zone Diameter ≤ 22 mm was defined as non-susceptible to the drugs.

Phenotypic test for the CPMase-producing strains was conducted by using the modified Hodge test and the double-disk synergy test (imipenem/ imipenem+EDTA and meropenem/meropenem+ EDTA).

The Minimal Inhibition Concentration (MIC) values for 15 antibiotics were determined by a broth microdilution method (BD Research Inc. USA). The antibiotics were as follows: ampicillin, cefazolin, cefotaxime, ceftazidime, cefepime, aztreonam, amoxicillin-clavulanic, piperacillin-tazobactam, imipenem, meropenem, trimethoprim-sulfamethoxazole, gentamicin, amikacin, ciprofloxacin, and colistin. The interpretive break points of Clinical and Laboratory Standards Institute (CLSI) M100-S24 were used to interpret the MIC results for antimicrobial agents^[14]. The Escherichia coli ATCC25922 and K. pneumoniae ATCC BAA-1705 strains were adopted as the standards for quality control (QC). WHONET 5.6 software (http://www.whonet.org/) was used to perform resistance analysis.

To detect CPMase-producing genes in the CPM-non-susceptible strains, PCR amplification was performed on a PTC-100-type PCR instrument (MJ Research Inc. USA). The CPMase-producing genes included bla_IMP, bla_VIM, bla_SPM, bla_GIM, bla_NDM-1, bla_SIM, bla_KPC, bla_GES, bla_IMI/NMC-A, bla_SME, and bla_OXA, and the primers were shown in Table 1 ^{[12, 15-17]}. The reaction system (total volume, 50 µL) consisted of: PCR Master Mix (2 × PCR buffer, 3 mmol/L Mg²⁺, 200 µmol/L dNTPs, and 0.1 U Taq DNA polymerase), 50 pmol each of the two primers, and DNA template obtained by heating a bacterial suspension. The reaction conditions were as follows: pre-denaturation at 95 ℃ for 3 min, followed by 35 amplification cycles of 95 ℃ for 1 min, 55 ℃ for 1 min and 72 ℃ for 1 min, with a final extension step of 72 ℃ for 5 min.

The PCR products from the above reactions were purified and sequenced using the ABI PRISM TM377 with the dideoxy-mediated chain-termination method. Sequences obtained were compared with the preexisting sequences via NCBI BLAST to determine the type of CPMase. The nucleotide sequence containing the bla_IMP-4, bla_IMP-8, bla_NDM-1, and bla_KPC-2 open reading frame has been assigned the EMBO/GenBank accession number FJ384365.1, EU368856.1, and KP036457.1 and KR108243.1.

SPSS 12.0 was used to perform regression analysis. Trends of resistance rates to each antimicrobial agent for microorganisms were performed using Chi-square statistics. A P < 0.05 was considered statistically significant.

During 2010-2014, 1, 730 non-duplicated strains of K. pneumoniae were isolated from different inpatients in our hospital. Among those strains, 179 (10.3%) were identified as non-susceptible (Zone Diameter ≤ 22 mm) to imipenem or meropenem via disk-diffusion method. The percentages of non-susceptible isolates were lower than 10% between 2010 and 2012, namely 7.8% (28/358) in 2010, 8.9% (37/415) in 2011, 7.4% (28/379) in 2012. However, it increased to 11.2% (40/356) in 2013 and further increased to 20.7% (46/222) in 2014.

The clinical features of CPM-non-susceptible K. pneumoniae isolates were shown in Table 2. The isolates were main obtained from Internal Medicine (39.7%, 71/179) and Haematology (29.6%, 53/179). Most of the patients were under 7 years old (76.0%, 136/179). The majority of the strains were isolated from tracheal aspirations (50.8%, 91/179) and blood (31.3%, 56/179). Notably, the isolates originated from blood increased from 14.2% to 41.3% during the study period.

Phenotypic analysis based on modified Hodge test and double-disk synergy test showed that none of CPM-non-susceptible isolates of K. pneumoniae had negative results in both tests, indicating all of the 179 isolates were CPMase producers. Among them, 152 (84.9%) strains were positive in the modified Hodge test and 174 (97.2%) positive in the double-disk synergy test.

The results of the disk-diffusion experiments showed that 156 (87.1%) of K. pneumoniae isolates were meropenem resistant, 122 (68.1%) exhibited imipenem resistance, and 120 (67.0%) were resistant to both imipenem and meropenem.

The results of broth microdilution experiments conducted for all of the 179 isolates were shown in Table 3. The isolates demonstrated very high resistance rates against ampicillin (100%), cefazolin (100%), cefotaxime (100%), amoxicillin-clavulanic (100%), and ceftazidime (99.4%). There were only two drugs with susceptibility rates higher than 50%, namely amikacin (88.2%) and ciprofloxacin (65.9%). An increase of resistance was seen for imipenem and meropenem from 14.3% to 89.1% and from 67.9% to 100%, respectively. No isolate of K. pneumoniae was found resistant to colistin.

The PCR results showed that 166 (92.7%) strains harbored the CPMase-producing genes detected in this study. The bla_IMP and bla_NDM-1genes were present in 95 (57.2%) and 65 (39.2%) strains, respectively, and the other 6 (3.6%) strains had bla_KPCgene.

Subsequent sequencing of the 166 PCR-positive strains further identified the genotypes of the respective CPMase genes. Among the 95 strains harboring the bla_IMP gene, 92 (96.8%) and 3 (3.2%) strains were IMP-4 and IMP-8 producers, respectively. The sequencing results also confirmed that the 65 strains with PCR positive results of bla_NDM-1 gene were NDM-1 producers. In addition, all sequenced KPC producers (6 strains) had bla_KPC-2 gene.

During our study period from 2010 to 2014, the most predominant CPMase-producing isolates were IMP-4 producers (51.4%, 92/179). NDM-1 producer also occupied a very large percentage (36.3%, 65/179). Only 3.4% (6/179) and 1.7% (3/179) were IMP-8 and KPC-2 producers, respectively. The annual distributions revealed an increase of NDM-1 producers and decrease of IMP-4 producers from 2010 to 2014 (Table 4).

The results of broth microdilution experiments indicated that different CPMase producers had different profiles of antimicrobial resistance (Table 5). Compare to IMP-4 producers, the isolates from KPC-2 producers had higher resistant rates frequencies in terms of resistance to imipenem (P = 0.017), amikacin (P = 0.010) and ciprofloxacin (P= 0.013). There were also significant differences in the frequencies of drug resistance between the isolates from NDM-1 producers and IMP-4 producers for piperacillin-tazobactam (P = 0.000), imipenem (P = 0.000) and meropenem (P = 0.000). The isolates from KPC-2 producers and NDM-1 producers had significantly higher rates of drug resistance to antibiotics than those from the IMP-4 producers (P < 0.05).

As useful antimicrobial agents, the β-lactam antibiotics play an important role in treating infection in the pediatric population, in which, CPMs have the broadest spectrum and greatest stability against hydrolysis by β-lactamases^[4]. CPMs have been regarded as a dependable drug for treating K. pneumoniae infections. However, with its increasing clinical usage, CPM-resistant K. pneumoniae are becoming more and more pervasive throughout the world^[18], but few phenotypic and genotypic investigations on CPM-resistance K. pneumoniae has been reported in China^[11-13]. This is the first study describing the 5-year prevalence of CPM-non-susceptible K. pneumoniae and their genetic characteristics responsible for CPMase production in a pediatric hospital in China.

Our study showed that the percentages of CPM-non-susceptible K. pneumoniae increased from 7.8% in 2010 to 20.7% in 2014 in our hospital. In addition, an increase in resistance was observed for imipenem and meropenem in our study, which is similar to the CHINET (antimicrobial resistance surveillance network in China) data^[19]. Furthermore, we compared our data and the data from the children's hospitals in CHINET^[20]. The increasing trend of drug resistant rates of imipenem and meropenem are similar, but the resistant rates against imipenem and meropenem varied in different hospitals and the rates in our hospital are higher than the others^[20-22]. The possible interpretation for this difference is that patients in different hospitals have different clinical needs, which leads to different prescribing habits of doctors in different hospitals. Because most patients in our hospital have been treated in other hospital before and have limited antibiotic choices, imipenem and meropenem might be utilized more frequently in our hospital, resulting in higher resistant rates of these drugs.

The most common mechanism for CPM resistance is the production of CPMases. In recent years, there have been many reports of CPMase-producing K. pneumoniae worldwide and the main types of CPMases reported in Asia and Europeare are IMP, VIM, and NDM^{[9, 23-24]}. Two genotypes, namely IMP-4 and IMP-8, are prevalent in China^[25-26]. In our hospital, IMP-4 type K. pneumoniae was the predominant genotype from 2010 to 2012, but NDM-1 genotype became the predominant one in 2013 and 2014. NDM-1 type strains have been recently described and are calling more and more attentions because of its resistance to most antibiotics^[27-29].

In the present study, the CPM-non-susceptible K. pneumoniae isolates showed a high rate of drug resistance to β-lactam antibiotics except aztreonam and piperacillin-tazobactam. It is worthwhile to note that 179 nosocomial K. pneumoniae isolates from selected pediatric departments have high rates of intermediate and resistance to CPM. The frequencies of resistance to imipenem and meropenem were 68.2% and 87.2%, and the rates of the intermediate to them were 26.8% and 8.4%, respectively. Therefore, it is suggested that imipenem and meropenem should be assessed at one time in order to find more CPM-non-susceptible K. pneumoniae. The CPM-non-susceptible K. pneumoniae isolates in our study showed the lowest resistance to amikacin (10.1%), followed by ciprofloxacin (20.7%), which may be due to these two drugs were rarely used in children because of their major side-effects.

In summary, the present study is generally to report on the characteristics of CPM-non-susceptible K. pneumoniae isolates from a pediatric hospital in China. The production of MBL, including two genotypes IMP and NDM, is one of the important mechanisms of CPM-non-susceptible of K. pneumoniae isolates in our pediatric setting. It is necessary to strengthen the surveillance of drug resistance and to gain an understanding of the characteristics and mechanisms of CPM-non-susceptible K. pneumoniae infections among the local pediatric population, which may help us better understand the prevalence of CPMase-producing K. pneumoniae and contribute to effective controls of the drug-resistant strains in pediatric clinics.

We are grateful to all members of Beijing Children's Hospital for their cooperation and technical help.

Protocol was approved by Ethics Committee of Beijing Children's Hospital, Capital Medical University. Informed consent was waived because we used existed strains and did not pose risk to patients.