Molecular epidemiological analysis and risk factors for acquisition of carbapenemase-producing Enterobacter cloacae complex in a Japanese university hospital

Background To clarify the molecular epidemiology of carbapenem-resistant Enterobacter cloacae complex (CREC) and the risk factors for acquisition of carbapenemase-producing E. cloacae complex (CPEC). Methods Using clinical CREC isolates detected in a Japanese university hospital over 4 years, carbapenemase production was screened with phenotypic methods. Carbapenemase genes were analysed by PCR and sequencing. Molecular epidemiological analyses were conducted with repetitive extragenic palindromic (REP)-PCR and multilocus sequence typing (MLST). CRECs were identified to the subspecies level by hsp60 sequencing. Whole-genome sequencing of plasmids was conducted. A case-control study was performed to identify risk factors for acquisition of CPEC among patients with CREC. Results Thirty-nine CRECs including 20 CPECs carrying blaIMP-1 were identified. Patients with CPEC had longer hospital stay before detection (26.5 days vs. 12 days, p = 0.008), a urinary catheter (odds ratio [OR], 5.36; 95% confidence interval [CI], 1.14–30.9; p = 0.023), or intubation (OR, 7.53; 95% CI, 1.47–53.8; p = 0.008) compared to patients without CPEC. Four genetically closely related CPEC clusters were observed, which showed that three of four CPEC clusters corresponded to E. asburiae (ST 53), E. hormaechei subsp. steigerwaltii (ST 113 and ST 1047) and E. cloacae subsp. cloacae (ST 513) by MLST and hsp60 sequencing. Seven representative plasmids shared structures with class I integron containing blaIMP-1 and IncHI2A replicon type. Conclusions A longer hospital stay, presence of a urinary catheter, and intubation are risk factors for CPEC acquisition. In addition to horizontal transmission of genetically indistinguishable CPECs, IncHI2A plasmid carrying blaIMP-1 appeared to be transferred among genetically different ECs.


Background
The emergence of carbapenem-resistant Enterobacteriaceae (CRE), especially carbapenemase-producing Enterobacteriaceae (CPE), has been increasing worldwide and is a clinical and public health threat [1]. Since September 2014, CRE infectious diseases have been involved in Category V Infectious Diseases identified by the Act on Prevention of Infectious Diseases and Medical Care for Patients Suffering Infectious Diseases in Japan. The National Epidemiological Surveillance of Infectious Diseases and Japan Nosocomial Infections Surveillance (JANIS) by the Ministry of Health, Labor and Welfare of Japan have been implemented for CRE using the criteria of either meropenem-minimum inhibitory concentration (MIC) ≥2 μg/mL or cefmetazole-MIC ≥64 μg/mL, in addition to imipenem-MIC ≥2 μg/mL. The annual report from JANIS 2017 showed that the prevalence of CRE is relatively low (0.27%). About half of CRE involves Enterobacter species (mainly Enterobacter cloacae complex (EC)). IMP-type β-lactamase is the most common carbapenemase in Japan [2,3]. The latest results of surveillance of Japanese CRE in 2017 revealed that 239 CPE strains (28%) were detected, and 227 CPE strains carried bla IMP, of which EC was the most common (74 strains). The IMP genotype was assessed in some strains, and IMP-1 (44%) and IMP-6 (56%) were detected [4].
CRE has been analysed predominantly in Klebsiella pneumoniae (KP) because of the initial global spread of CRE of KPC-type β-lactamase-producing KP strains. Most CRE research has focused on frequently detected species such as KP and Escherichia coli [5]. Among CRE, CPE bacteremia has a fourfold higher mortality rate within 14 days comparing to non-CPE CRE bacteremia [6]. It has been suggested that CPE is more likely to spread than non-CPE, and it is listed as one of the important research themes for infection control [7]. Recent meta-analyses of the clinical epidemiology of CRE showed that CREC is the second most common species among the studies that focused on a single species of the Enterobacteriaceae family [8]. In Japan as well, CREC is the second major species of CRE, and molecular characterization of CREC clinical isolates has been reported [9][10][11]. Further understanding of the molecular epidemiology of CREC and investigation of carbapenemase gene-carrying plasmids, which is the most important resistance mechanism of transmission, are needed to prevent the spread of CREC.
This study aimed to clarify the molecular epidemiology of CPEC isolates and their plasmids carrying carbapenemase genes detected in Nagoya University Hospital (NUH) and to analyse the risk factors for CPEC acquisition compared with CREC without carbapenemase production.

Study design and population
This was a single-centre, retrospective, observational study of hospitalized patients with positive cultures of CREC from April 1, 2012 to March 31, 2016 at NUH, a 1,035-bed tertiary acute care hospital in Japan. The first CREC isolate from a patient at NUH during the study period that met the CRE surveillance definition in Japan was included [3]. A case-control study was conducted to identify risk factors for acquisition of CPEC between patients who acquired CPEC and those who acquired non-CPEC.

Data collection
Patient information was retrieved from patient electronic medical records. The parameters included demographics, background conditions and comorbidities, recent health care-associated exposure (such as stays in health care facilities), invasive procedures, the presence of indwelling devices, exposure to antimicrobials within 3 months prior to isolation of CREC, and the clinical outcome. Infectious clinical diagnosis was determined according to the information present in medical charts recorded by the attending doctor. The patients were determined to be colonisers if they did not have any signs or symptoms of infection based on information in their medical charts.

Microbiological methods
Primary identification of bacterial species and antimicrobial susceptibilities was performed using an automated identification and susceptibility testing system (MicroScan WalkAway; Beckman Coulter, Brea, CA, USA) according to Clinical and Laboratory Standards Institute guidelines (document M100-S22), with dry plates (Beckman Coulter) in each case. EC isolates that showed either meropenem-MIC ≥2 μg/ mL or cefmetazole-MIC ≥64 μg/mL, in addition to imipenem-MIC ≥2 μg/mL, according to the definition by the Act on Prevention of Infectious Diseases and Medical Care for Patients Suffering Infectious Diseases for CRE in Japan, were included [3].

Screening of carbapenemase and sequencing of the carbapenemase gene
CRECs were screened to detect carbapenemase production using disc synergy tests, specific inhibitors, and the modified carbapenemase inactivation method (mCIM) [12]. All mCIM-positive CRECs were screened for the presence of bla IMP-1 , bla IMP-2 , bla NDM-1 , bla VIM-2 , and bla KPC genes using PCR with primers as previously described [13][14][15]. When an isolate tested positive for the carbapenemase gene, its PCR product was sequenced at a commercial laboratory (Eurofin Genomics, Tokyo, Japan) and assembled with Sequencher DNA sequence analysis software (Gene Codes, Ann Arbor, MI, USA). Using its consensus sequence, the type of IMP was determined with BLAST (https://blast.ncbi.nlm.nih.gov/ Blast.cgi).
Repetitive extragenic palindromic polymerase chain reaction (REP-PCR) DNA was isolated with an UltraClean Microbial DNA isolation kit (MoBio, San Diego, CA, USA) and used for experiments with the Enterobacter spp. fingerprinting kit (bioMerieux Japan, Tokyo, Japan) per the manufacturer's procedure. PCR products were separated by electrophoresis using microfluidic lab-on-a-chip (Agilent Bioanalyzer 2100; Agilent, Santa Clara, CA, USA). Results were analysed using DiversiLab (bioMerieux Japan) on-line software employing the Pearson correlation method, which places more emphasis on the presence or absence of bands than on their intensity. ECs with a similarity of fingerprinting less than 95% were considered genetically different, and isolates with a similarity of > 98% were considered indistinguishable [16]. Isolates with a similarity between these values were judged manually using the pattern overlay option in the software. Isolates that were indistinguishable by DiversiLab testing belonged to clusters, and horizontal transmission was defined when an indistinguishable CREC was detected in different patients.

hsp60 PCR and sequencing
Amplification of hsp60 was accomplished using previously described primers and conditions [17] with DNA extracted with the Cica Geneus DNA Extraction Reagent (Kanto Chemical, Tokyo, Japan). DNA sequencing was performed at a commercial laboratory (Eurofin Genomics) and assembled with Sequencher DNA sequence analysis software (Gene Codes). Based on the neighbour-joining tree of the hsp60 sequences, detailed bacterial names were determined from 12 genetic clusters and an unstable sequence crowd using ClustalW (http://clustalw.ddbj.nig.ac.jp/). Reference strains and type strains were used for neighbourjoining tree of the hsp60 sequences, and the genotypes were determined according to Hoffmann et al [17]. To identify members of E. cloacae complex to subspecies level, species and subspecies were referred to Chavda et al based on hsp60 cluster result [18].

Multilocus sequence typing (MLST)
DNA was isolated with Cica Geneus DNA Extraction Reagent (Kanto Chemical). Seven housekeeping genes were amplified using primer sets according to the method previously reported [18,19]. DNA sequencing was performed at a commercial laboratory (Eurofin Genomics) and assembled with Sequencher DNA sequence analysis software (Gene Codes). Using its consensus sequence, the sequence type (ST) was determined with the Enterobacter cloacae locus/sequence definitions database (https://pubmlst.org/bigsdb?db=pubmlst_ecloa cae_seqdef ). If the sequence did not match the existing ST, new alleles and MLST profiles were registered in the Enterobacter cloacae locus/sequence definitions database (https://pubmlst.org/bigsdb?db=pubmlst_ecloa cae_seqdef ).

Statistical analysis
All analyses were performed using EZR [23]. The association of categorical variables with CREC patients was performed using Fisher's exact test. For continuous data, Mann-Whitney tests were applied appropriately. Statistical significance was considered when the p-value was less than 0.05.

Results
A total of 39 patients with non-duplicate CREC isolates were identified during the study period. Twenty CPECs among 39 CRECs were revealed with mCIM, and bla IMP-1 was identified in all CPEC isolates with negative PCR results for bla IMP-2 , bla NDM-1 , bla VIM-2 , and bla KPC. The annual incidence of the CPEC cases among CREC cases is shown in Fig. 1. The number of CREC cases was six in the first year and ten to eleven cases per year in the subsequent 3 years. About half of CREC was CPEC every year during the study period.
Baseline characteristics and prior health care exposure of the patients with CPEC and non-CPEC are shown in Table 1. CREC isolates were detected most commonly in patients aged 65-79 years old. Almost all individuals had at least one underlying comorbid condition, and approximately 75% had a history of surgery. Patients with CPEC were more likely to have had a longer hospital stay (26.5 days vs. 12 days, p = 0.008), to have had a urinary catheter (odds ratio [OR], 5.36; 95% confidence interval [CI], 1.14-30.9; p = 0.023), and to have been intubated (OR, 7.53; 95% CI, 1.47-53.8; p = 0.008) before a positive culture result.
About half of CREC patients (17/39; 43.6%) required a stay in the intensive care unit within 7 days after CREC was cultured, but no significant difference was found between CPEC and non-CPEC cases. Regardless of the culture source, the mortality rate within 30 days or during hospitalization did not differ significantly between CPEC cases and non-CPEC cases.
Six plasmids except for pNUH14_ECL028_1 had almost identical conjugation elements, mobile elements, and a class 1 integron containing bla IMP-1 , aac(6′)-IIc, and sul1 (Fig. 3). pNUH14_ECL028_1 also had a class 1 integron containing bla IMP-1 , sul1, and aac (6′)-1a instead of aac (6′)-IIc, and did not have identical conjugation elements and mobile elements compared with the other six plasmids. The insertion sequence ISKpn7 was located upstream of a class 1 integron, and the insertion sequence IS1 was located downstream of the integron in five plasmids (pNUH12_ECL006_1, pNUH13_ECL030_1, pNUH15_ ECL035_1, pNUH15_ECL032_1 and pNUH15_ECL020_ 1). All seven plasmids carry heavy metal resistance genes for mercury, arsenic and tellurite. All plasmids except for pNUH15_ECL011_1 carry HipBA toxin/antitoxin system.
The data have been deposited with links to BioProject accession number PRJDB7521 in the DDBJ Bio-Project database. BioSample metadata are available in the DDBJ BioSample database under accession numbers SAMD00143514 to SAMD00143520. The sequence data of plasmids are available in the DDBJ/ EMBL/GenBank databases under the accession numbers AP019382 to AP019388.

Discussion
Molecular epidemiological analysis and genetic characterization of CREC and risk factors for CPEC acquisition over 4 years in a Japanese university-affiliated hospital were performed. Among the patients in whom CREC was detected, long-term hospitalization, the presence of an indwelling urinary catheter, and intubation were associated with isolation of CPEC. Comparison of the fingerprinting patterns by REP-PCR with MLST and hsp60-based genotyping showed general concordance except for isolate No. 39. REP-PCR experiments were conducted repeatedly, but the results were same. The reason why only this isolate did not show concordance is unclear. Molecular epidemiological analysis revealed that genetically indistinguishable CPECs were detected from patients without any epidemiological relationship who were hospitalized in different years, suggesting horizontal transmission, whereas non-CPECs were mostly genetically distinguishable. All CPECs carried bla IMP-1 , and interestingly, even genetically distinct CPECs carried highly similar IncHI2A plasmids harboring bla IMP-1 . These results suggest that the spread of CPECs in the hospital was due to complex mechanisms of a combination of clonal and plasmid transmission. Although we   containing bla IMP-1 and heavy metal resistance genes and toxin/antitoxin system. Intriguingly, this structure was highly similar to that of pMTY11043 IncHI2 detected from Enterobacter hormaechei (GenBank accession number AP081352.1) reported in Tokyo, Japan [9]. As reported in the study [9], the IncHI2A plasmids may be more likely to have common genes for toxinantitoxin systems and heavy metal resistance, resulting in maintenance of the plasmids, especially in hospital environments [24,25]. As few studies about CPECs have analysed their plasmids [9,26,27], further investigation of molecular epidemiology and plasmid distribution of CPECs at a regional or countrywide level is warranted. The tendency for clonal transmission of CPECs presented in this study supported the validity of infection control policies focusing on CPE according to the CPE Toolkits published by the Centres for Disease Control and Prevention of England [28,29]. In our hospital, according to the instructions from the Ministry of Health, Labor and Welfare in Japan, once a CPE not non-CPE CRE, was detected, active surveillance culture of the inpatients in the same ward as the index case was conducted for early detection of asymptomatic carriers. In addition, stringent contact precautions were implemented in both colonised and infected patients to prevent further transmission [30]. However the results of transmission of CPECs between patients hospitalized in different periods and without any apparent epidemiological linkage pose some challenges for infection control in our hospital. Hidden environmental sources such as sink drainage could exist [31,32], although previous environmental cultures at the responsible intensive care unit and wards found no contamination (data not shown).
This study has some limitations. First, the study was performed at a single centre, and the results cannot be generalized to other institutions. Second, only the E. cloacae complex was targeted in this study among CRE. Whether this result can be applied to other Enterobacteriaceae is unknown.

Conclusions
Molecular epidemiology and the genetic background of plasmids conferring carbapenem resistance showed Fig. 3 Linear comparison of bla IMP-1 -carrying plasmid sequences from Enterobacter cloacae complex strains was performed with BLAST and visualized with Easyfig (http://mjsull.github.io/Easyfig/). Red, yellow, green, and blue arrows indicate the bla IMP-1 gene, other antimicrobial resistant genes (ARG), type IV secretion system (T4SS)-associated genes for conjugation, and mobile gene elements (MGE) detected from CDS annotations, respectively. Structural features of a class 1 integron containing bla IMP-1 are shown, and the structures are almost identical in six plasmids, except for pNUH14_ECL028_1 horizontal transmission of some clones and plasmids with a common backbone of the IncHI2A replicon type and class I integron containing bla IMP-1 . Risk factors for CPEC acquisition are a longer hospital stay and use of indwelling devices, especially intubation and a urinary catheter. Early detection of CPEC and strict infection control measures upon detection including active surveillance culture for asymptomatic carriers are necessary to minimize the transmission of CPECs.