Characteristics of NDM-1-producing Klebsiella pneumoniae ST234 and ST1412 isolates spread in a neonatal unit

Background The emergence of carbapenem-resistant Klebsiella pneumoniae (CR-KP) has become a significant problem worldwide and also being a major threat to children and newborns. Here we report an outbreak of NDM-1-producing K. pneumoniae in a neonatal unit. Results Six CR-KP strains, isolated from neonates with symptoms of infection, were identified using a VITEK-2 compact system, and the clinical data were retrieved from the electronic case records. In vitro susceptibility testing with broth dilution method showed that all six K. pneumoniae isolates were resistant to carbapenems and susceptible to colistin, aminoglycosides, fluoroquinolones and tigecycline. Based on the polymerase chain reaction results, each isolate was found to be blaNDM-1 gene positive. Clonal relationships were analysed using pulsed-field gel electrophoresis (PFGE) and multilocus sequence typing (MLST) and showed that two different PFGE patterns were formed, which belonged to sequence types ST234 and ST1412. Plasmids carrying blaNDM-1 were successfully transferred from four of the six isolates to an Escherichia coli recipient through conjugative assays. S1-PFGE and Southern blot hybridization showed that four NDM-1-producing K. pneumoniae were clonal and carried blaNDM-1 on the same plasmid. The outbreak was effectively controlled by reducing the potential infection sources. All the patients were successfully treated and recovered after receiving an increased dose of carbapenems. Although the source of this outbreak was not clear, comprehensive measures were carried out and the outbreak was effectively controlled. Conclusions ST234 and ST1412 of NDM-1-producing Klebsiella pneumoniae are the resistant clone spread in the neonatal unit, comprehensive infection control measures and optimized carbapenem therapy played an important role in controlling this NDM-1-producing K. pneumoniae outbreak.

For different care model and limited drug selection for neonates, effective infection control and treatment should be further studied [10]. Here, we present an outbreak of NDM-1-producing K. pneumoniae in the neonatal ward of a university hospital in China. The molecular characteristics and transmission mechanism of NDM-1-producing K. pneumoniae was also studied.

Bacterial isolates and patients
K. pneumoniae strains were collected from neonates with symptoms of infection who were admitted to the neonatal ward of a university hospital in Nanjing between June and August of 2016. Identification and in vitro susceptibility tests were carried out and carbapenem-resistance was analyzed with VITEK-2 compact system (bioMérieux, Marcy-l'Étoile, France). The electronic case records including patient demographics, antimicrobial treatment and clinical outcomes were retrieved and reviewed. In order to investigate the source of nosocomial infection, related specimen were collected from incubator surface, head circumference tape, blood machine surface and healthcare worker, isolates named KP7 to KP10 were analyzed.

Molecular typing
NDM-1-producing strains were genotyped using PFGE and MLST [16]. The allelic profiles and sequence types (ST) were available in the MLST database (http:// bigsdb.pasteur.fr/klebsiella/klebsiella.html). Clonal relationships were analysed using PFGE and bacterial DNA was digested with XbaI endonuclease (TaKaRa, Dalian, China) as previously described [7]. Salmonella enterica serotype H9812 was used as a marker. The PFGE patterns were compared using BioNumerics software (Applied Maths, Kortrijk, Belgium), and a phylogenetic tree was built for cluster analysis. Clusters were defined as DNA patterns sharing > 90% similarity.

Conjugation experiments
Conjugative assays were performed using sodium azide-resistant E. coli J53 as the recipient strain (donated by professor Wenen Liu, Central South University) [17]. Transconjugants were selected on MacConkey agar plates containing 150 mg/L sodium azide (Sigma Aldrich, St Louis, MO) and 30 mg/L ceftazidime (Sigma-Aldrich) for 24 h at 35°C. Presence of the bla NDM-1 gene in the transconjugants was confirmed by PCR analysis, and their antimicrobial susceptibilities were determined as well.
Plasmid analysis S1-PFGE and Southern blotting were performed to analyse the size of the NDM-1-carrying plasmids in the K. pneumoniae strains as previously described and the results were analyzed according to the criteria proposed by Tenover et al. [18,19] Plasmid DNA of the isolates embedded in agarose gel plugs were digested with S1 nuclease and separated by PFGE. Plasmids obtained by PFGE were then transferred to a positively charged nylon membrane. The membrane was hybridized with digoxigenin-labelled bla NDM-1 -specific probes and the signals were detected using an NBT/BCIP colour detection kit (Roche Applied Sciences, Penzberg, Germany).

Clinical characteristics of the K. pneumoniae isolates
The clinical profiles of the six NDM-1-positive K. pneumoniae isolates (KP1 to KP6) are shown in Table 1, including the patient demographics, the date of specimen isolation, specimen source and clinical diagnosis, antimicrobial treatment and clinical outcomes. The six NDM-1-positive K. pneumoniae isolates were obtained from blood (n = 5) and sputum (n = 1). Five patients had sepsis, and one had neonatal respiratory distress syndrome. No infants died from their infections after effective treatment. The newborns experienced long hospitalization periods (mean, 62 days), and treatment with increased doses of carbapenem showed good prognoses.

Molecular epidemiology
Two distinct PFGE patterns were obtained from the XbaI DNA digests among the six CR-KP isolates: type A (n = 2) and type B (n = 4). Carbapenem-susceptible K. pneumoniae isolated from the incubator surface, head circumference tape, blood machine surface, and medical workers showed no homology with the six CR-KP isolates (Fig. 1a). Two distinct sequence types were obtained from the six isolates:KP1 together with KP2 are ST234 and KP3, KP4, KP5 and KP6 are ST1412. Comparison of these results showed that all PFGE type A isolates corresponded to ST234, and the type B isolates corresponded to ST1412 (Fig. 1b).

Transfer of carbapenemase resistance
Transconjugants were obtained in the conjugation experiments, with a success rate of 67%. In the conjugation experiments, the bla NDM-1 genes were transferred successfully from the donor K. pneumoniae isolates (except KP1 and KP2 isolates) to the recipient E. coli J53 via plasmids. Antimicrobial susceptibility detection showed  Plasmid analysis S1-PFGE and Southern blot hybridization analysis indicated that the bla NDM-1 genes were transferred via the same plasmids (with approximate size 50 kb), except for in the KP1 and KP2 strains (Fig. 2b). Although the KP1 and KP 2 strains carried plasmid DNA (with approximate size 110 kb) (Fig. 2a), they did not seem to transmit the bla NDM-1 gene via these plasmids (Fig. 2b). Additionally, conjugation experiments and Southern blotting hybridization showed that the bla NDM-1 gene probably falls within the chromosome of KP1 and KP2 isolates.

Discussion
Infectious diseases caused by NDM-1-producing Enterobacteriaceae are known to be associated with high morbidity and mortality worldwide. This is a great challenge for paediatricians due to the transferability of the gene and the poor prognosis in children. Of the carbapenem resistance genes found in carbapenem-resistant Enterobacteriaceae in children, NDM genes (including coding genes of NDM-1 and NDM-6) are more common than KPC and VIM [10]. Compared to adults, children are more limited in antibiotic use. Although these isolates were all susceptible to tigecycline and colistin in vitro, the paediatrician had not used either of these treatments because of their potential side effects in children. Aminoglycosides and fluoroquinolones are also not used for children due to their nephrotoxicity and ototoxicity [6]. After careful consideration, the paediatrician in this report chose to use an increased dose of carbapenems (meropenem 0.5 g IVD qd; imipenem 1.0 g IVD qd), and no side effects were reported during the antimicrobial therapy. Fortunately, all of the neonates in this report had positive prognoses. However, the proposed doses of meropenem and imipenem for neonates who are greater than 28-days-old are 120 mg/kg/day and 60-100 mg/kg/ day, respectively, according to the Sanford Guide to Antimicrobial Therapy. In the present study, the doses used are more than 4 times the amount suggested by the guide. Thus, a recent study showed that monotherapy for treatment of CRE infection in adults was associated with higher mortality than that of combination therapy, and limited data were available for children [10,20,21]. Although we had a good prognosis this time, further dosage optimization of carbapenem and medication for antimicrobial therapy for NDM-1-producing Enterobacteriaceae should be investigated. In our study, PFGE analysis showed two clusters of the NDM-1-producing strains that corresponded to ST234 and ST1412. Among these, none were found to be clonally related to the carbapenem-susceptible K. pneumoniae isolated from the surface of incubators, blood machines, neonatal head circumference tapes and medical workers. Regarding the epidemiology of this spread, 2 isolates (5, 6) were identified in June, and 4 isolates (1, 2, 3, and 4) were identified in August. This means that the ST1412 isolates remained in the neonatal ward for 2 months and reappeared in August, along with a new strain. It is unfortunate that although we have not found the source of this outbreak, we cannot deny that the hospital environment was likely the source of these infections. After all, the incubator water and the sharing of breast milk has been reported to be 'hotspots' for bacterial transmission in neonatal wards [7,22]. When we first confirmed the outbreak, our hospital control team took active measures to sterilize the source of infection in the environment and restrict the admission of newborns to the unit. Finally, the outbreak event was successfully controlled and all infected patients recovered and were discharged. NDM-1-producing K. pneumonia have been reported as belonging to various types of MLST: ST11, ST14, ST17, ST25, ST37, ST76, ST105, ST147, ST149, ST231, ST340, and ST1043 [6,7,23]. Our data indicate that NDM-1-producing K. pneumoniae strains belong to two independent types, ST234 and ST1412, which differ from most types reported previously, particularly for the never reported NDM-1 producing ST234 strains, which have only been reported as KPC-producing K. pneumonia [24].
The plasmid analysis in this study showed that the bla NDM-1 genes transferred using the same plasmids (with approximate size 50 kb), except for the KP1 and KP2 strains. We analysed the genomic DNA of four transconjugants using PCR, and the results showed that only bla NDM-1 was present. This result indicates that bla SHV-148 was not on the same plasmid as bla NDM-1 . For the KP1 and KP2 strains, we speculated that the bla NDM-1 gene may be in the chromosome or may be a high-molecular-weight non-conjugative plasmid that was not detected by the methodology used here. Further studies should be investigate this gene. The source of the NDM-1 has not been determined. It has been studied in previous bla NDM variants from sewage water in hospitals, which serves as an important source of the spread of antibiotic resistance genes [25]. Guidance from the European Centre for Disease Prevention and Control suggested that intervention measures should be taken on antimicrobial stewardship, clean hospital settings, equipment reprocessing, staff education and microbiological capacity to minimize risks of spread of CRE [26]. All we can do is strengthen the unity of the clinical departments to prevent further infection.

Conclusions
ST234 and ST1412 of NDM-1-producing Klebsiella pneumoniae are the resistant clone spread in the neonatal unit, comprehensive infection control measures and optimized carbapenem therapy played an important role in controlling this NDM-1-producing K. pneumoniae outbreak.