Epidemiological characteristics of carbapenem-resistant Enterobacteriaceae collected from 17 hospitals in Nanjing district of China

Objective In total, 97 carbapenem-resistant Enterobacteriaceae (CRE) were collected from 17 hospitals located in Nanjing, Southeast China, and analyzed for epidemiological characteristics. Methods Antimicrobial susceptibility was determined; followed by determination of the prevalence of resistance determinants, including extended-spectrum β-lactamase (ESBLs), plasmid-mediated AmpC enzyme (pAmpCs), plasmid-mediated quinolone resistance genes (PMQRs), fosfomycin resistance gene and exogenously acquired 16S rRNA methyltransferase (16S-RMTase) using PCR and DNA sequencing. The sequence types (STs) of CRE were determined by multi-locus sequence typing (MLST). The plasmid profiles were detected by PCR-based replicon typing (PBRT). Results All the CRE strains displayed high MIC50 and MIC90 for nearly all clinical available antibiotics, except for aztreonam/avibactam, minocycline, ceftazidime/avibactam, tigecycline, and colistin. KPC-2 (79.4%) and NDM (19.6%) were the main carbapenemases, CTX-M (76.3%) and SHV (60.8%) were the predominant ESBLs. In addition, oqxAB (70.1%) and qnr (63.9%) were the major PMQRs; rmtB (47.4%) was the main 16S-RMTase; fosA (76.3%) and fosA3 (37.1%) were the fosfomycin resistance gene. PBRT analysis showed presence of IncR (66.0%) and IncFII (64.9%) replicon types in the majority of the isolates, followed by IncFIB (46.4%) and IncX3 (16.5%). The IncFII and IncR replicon-types were found mainly in K. pneumoniae (68.8%), whereas the IncX3 replicons dominated in E. coli isolates (100.0%). The three dominating MLST-types ST11, ST15 and ST268 comprised 68.0% of the 77 K. pneumoniae. Seven distinct STs were identified among 8 E. coli. Conclusions The treatment for infections caused by CRE isolates is challenged by the presence of multiple resistance determinants and plasmid replicons. Our results highlighted the expansion of blaKPC-2 carrying K. pneumoniae ST11, the new emergency of single blaNDM-5 carrying K. oxytoca ST36, as well as blaIMP-4 and blaNDM-1 co-carrying E. cloacae ST418, which alert us on the urgency for antimicrobial resistant surveillance, to prevent dissemination of these highly transmissible and dangerous lineages.


Introduction
With the wide use of carbapenem antibiotics in clinical therapy, carbapenem-resistant Enterobacteriaceae (CRE) has dramatically increased and become a serious public health issue [1]. CRE constitutes a large group of bacteria with different mechanisms for drug resistance [2]. Among them, carbapenem resistant Klebsiella pneumoniae (CRKP) accounts for approximately 60%, followed by Escherichia coli and Enterobacter cloacae [3,4]. Epidemiological studies have shown that production of carbapenemase is the main mechanism for carbapenem resistance [5], with blaKPC and blaNDM being the most prevalent ones in the CRKP and carbapenem resistant E. coli (CREC), respectively [6,7]. Recently, the cooccurrence of multiple resistance determinants, even more than 2 carbapenemases in one strain, has been frequently reported [8,9], alerting us on the importance of resistance surveillance, since the resistant determinants are mainly plasmid-borne with highly transmissible nature [9,10]. Additionally, the production of extendedspectrum β-lactamases (ESBLs) and/or AmpC enzymes in combination with decreased permeability by mutations in outer membrane proteins OmpK35 and OmpK36, as well as over-expression of efflux pumps play roles in the carbapenem resistance [11,12]. It is noteworthy that CRE infections are closely associated with high mortality because of limited antimicrobial use [13]. To date, fosfomycin, minocycline, tigecycline, colistin, ceftazidime/avibactam and aztreonam/avibactam have been recommended for treatment of infections caused by CRE [14]. However, data on susceptibilities of these antibiotics on CRE are still limited.
Although CRE has been frequently reported in China [8,15], most of the analyzed CRE were collected from the Third-Class A General Hospitals, and the information on CRE isolates collected from specialized hospitals, Children's hospital and level II hospitals was less available. Furthermore, data on non-K. pneumoniae CRE are also less available, owing to the focus of globally disseminated K. pneumoniae.
In this study, 97 CRE strains were collected from 17 hospitals, including specialized hospitals, Children's hospital and level II hospitals. The antimicrobial susceptibility, resistant determinants, sequence types (STs), as well as plasmid replicons, were analyzed to investigate the epidemiological characteristics of these CRE.

MIC determination
In total, 27 antimicrobial agents were used to test the susceptibilities of the CRE. Among them, 25 were tested by micro-broth dilution method, including ertapenem, imipenem, meropenem, cefepime, ceftazidime, cefotaxime, cefuroxime, cefazolin, cefmetazole, piperacillin/tazobactam, amikacin, gentamicin, funantuoyin, trimethoprim and sulphame-thoxazole, aztreonam, piperacillin, ciprofloxacin, levofloxacin, aztreonam/avibactam, ceftazidime/avibactam, tigecycline, and colistin. The susceptibilities toward fosfomycin and minocycline were determined by Kirby-Bauer method. Results were interpreted according to the guideline of CLSI 2019 [16]. Considering the absence of CLSI breakpoints for interpretation of tigecycline and colistin results, the current European Committee on Antimicrobial Susceptibility Testing (EUCAST) (www.eucast.org) guidelines were used. For tigecycline, a cutoff MIC of ≤1 μg/ml and > 2 μg/ml was used as the susceptibility and resistance breakpoints, respectively. For colistin, a cutoff MIC of ≤2 μg/ml was taken as susceptibility breakpoint. The E. coli ATCC25922 was used as quality control. Carbapenem resistance was defined as a MIC of ≥2 μg/ml for ertapenem or a MIC of ≥4 μg/ml for imipenem or meropenem.

Statistical analysis
SPSS software (20.0) was used to implement statistical analysis. The differences on distribution of resistant determinants/plasmid replicons between bacteria were analyzed by Chi-square test, and the differences were considered to be significant when p value was less than 0.05; The differences on distribution of antimicrobial resistance determinants and plasmid replicons among CRE were analyzed by McNemar test. The distribution rates were considered to be the same when p value was more than 0.05.

Discussion
In this study, we provided data on the antimicrobial resistance profiles, STs, and plasmid replicons profiles of CRE isolates collected from specialized hospitals, Children's hospital and level II hospitals in Nanjing district, Southeast China.
Most of our CRE strains displayed high resistance against the commonly used antimicrobial agents, which was consistent with previous report, providing evidence that the CRE strains are usually resistant to many other classes of antibiotics in clinical practice [24]. Fortunately, most of the CRE strains were still susceptible towards tigecycline, colistin, ceftazidime/avibactam, and aztreonam/avibactam. Resistance to aztreonam/avibactam and ceftazidime/avibactam have appeared in our study, which may be attributed to mutations in the carbepenamase KPC and NDM [25], suggesting a rapid resistance development of K. pneumoniae. None carbepenemase encoding genes were detected among the 5 CRE strains in our study, where the production of ESBLs variants and/or AmpC enzymes in combination of overexpressed efflux pumps or the decreased permeability might contribute to the carbapenem resistance.
The high prevalence of blaKPC-2 and many other resistance related genes, such as blaESBLs, PMQRs, as well as 16S-RMTase was consistent with our previous report [8], indicating the frequent co-existence or coevolution of antibiotic resistance genes, which might lead to the emergence of untreatable K. pneumoniae infections. Among them, the quite high prevalence of blaESBL among the blaKPC-2 producing CRE strains in our study represented crippling and urgent threats to public health, because multiple copies of blaCTX-M and Table 4 The distribution of major plasmid replicons among carbapenem-resistant Enterobacteriaceae  blaKPC within plasmids could be integrated and disseminated into chromosome [26]. In that case, the spread of such strains would be horizontally and vertically accelerated within hospitals. To date, the co-occurrence of blaCTX-M-65 and blaKPC-2 in our study has been previously reported in K. pneumoniae [27]. Notably, we found quite high prevalence of fosA in blaKPC-2 producing K. pneumoniae. fosA has been reported to be chromosomally encoded by clinically relevant Gram-negative species and contributes to intrinsic fosfomycin resistance [28]. However, some strains carrying fosA in our study displayed susceptibility to fosfomycin, which needs to be further investigated. Furthermore, the wide distribution of fosA and fosA3, as well as the emergence of fosA5, indicated that fosfomycin should be cautiously used for treating the CRKP infections, since the combination of plasmid-borne fosA3 and blaKPC-2 could accelerate the spread of antibiotic resistance [29]. Another interesting point is that, blaCTX-M-45 used to be identified by an algorithm [30]. However, this is the first time that we found this gene in a clinical C. freundii isolate. Among these SHV encoding genes, the blaSHV-13 was discovered in K. pneumoniae isolates in Amsterdam [31], whereas, the blaSHV-67, blaSHV-172, blaSHV-182, and blaSHV-190 are brand new, and have never been reported previously. OXA-10-type class D β-lactamases (previously shown to be weak carbapenemases) was firstly identified in our clinical C. freundii [32]. Moreover, CMY-34 was identified in C. freundii isolate in Danish army recruits, and was also found in our study [33].
To the best of our knowledge, the two blaCMY-65 producing C. freundii strains, one blaCMY-77 producing K. pneumoniae isolate, one blaCMY-77 producing K. oxytoca and a blaACT-16 producing E. cloacae found in our study have not been reported previously. For the first time, we found the co-occurrence of blaDHA-1 and blaCMY-65 in C. freundii, and the co-occurrence of blaCMY-77 and blaDHA-1 in K. oxytoca. The expansion of ST11 for the KPC-2 producing K. pneumoniae was in accordance with previous report indicating that the ST11 is the dominating epidemic clone among CRKP [15]. Albeit the KPC-2 producing E. coli ST410 has been reported, as far as we know, the blaNDM-5 and blaKPC-2 co-carrying E. coli ST167, blaNDM-5 carrying K. oxytoca ST36, as well as blaIMP-4 and blaNDM-1 co-carrying E. cloacae ST418 identified in our study have not been reported previously. Additionally, the blaKPC-2 has been identified in C. freundii and E. cloacae, and blaNDM-1 in C. freundii. This is the first time that we identified blaKPC-2 in C. freundii ST116 and E. cloacae ST723, and blaNDM-1 in C. freundii ST36.
Plasmids are extra-chromosomal DNA elements representing major reservoirs for horizontal transmission of antibiotic resistance among bacteria [34]. To date, multiple plasmids have been found to be the vesicles for spread of carbapenemase, ESBLs, and PMQRs [34]. The high prevalence of IncFII, IncR and IncFIB plasmid replicons in our study alert us on the urgency of implementing antimicrobial resistance surveillance, since IncFII-type plasmids are highly distributed vesicles for resistant determinants in Enterobacteriaceae [35]; Moreover, IncR plasmid is an important reservoir of multidrug resistance in Enterobacteriaceae strains, because the conserved IncR backbones include the multidrug resistant (MDR) regions [36]; and IncFIB plasmids were reported to be associated with majority of the antimicrobial resistance genes [37]. Similarly, some of the distribution correlations between various plasmid replicons and multiple resistant genes found in our study are consistent with the previous reports [38][39][40]. Among them, IncX3 plasmid is a narrow-spectrum plasmid widely distributed in E. coli from wildlife in Europe [38], which has also been found to be predominantly associated with fluoroquinolone resistance genes and β-lactam resistance genes. This is the similar situation in our CRE strains, where IncX3 was also found to be associated with blaNDM and qnrB4. However, the situation in K. pneumoniae is still uncertain, because few IncX3 plasmid replicons were found. In addition, plasmids encoding carbapenemases have been demonstrated to play a core role in the rapid spread of CRE [39]. In our study, the distribution correlations between blaNDM and plasmid replicons IncX3, IncFIA and IncFII K also confirmed the proposal, since the spread of blaNDM is involved in diverse and heterogeneous plasmids [40]. As we know, IncX3 plasmid has been an important vehicle with high mobility in worldwide dissemination of blaNDM [41], and IncFIA-type conjugative plasmid encoding blaNDM-1 has been involved in the outbreak caused by K. pneumoniae strains in Tunisia [11]. The IncN and IncH1 plasmids in our study are less common, this is in accordance with the previous study, which revealed that IncN and IncH1 plasmids are host-specific, and are more predominant in livestock and horses in Denmark, respectively [42]. However, the appearance of such plasmid replicons in our study may suggest the spread of plasmids among human and farm livestock.
Five isolates lacked any plasmid type in our study, which may be due to the limitations in the PBRT protocol [43]. It's however noteworthy that the co-existence of multiple plasmid replicons in our CRE strains is consistent with previous study [39], which may result from a conserved backbone responsible for regulation and mating pair stabilization [44].

Conclusion
In addition to wide distribution of CRE isolates in Nanjing district, Southeast China, the high-level carriage of carbapenemases, ESBLs, pAmpCs, 16S-RMTase, PMQRs and major plasmid replicons in our study might be a potential challenge regarding the transmissible capability. Moreover, the expansion of ST11 blaKPC-2 carrying K. pneumoniae, the new emergence of blaNDM-5 carrying K. oxytoca ST36, as well as blaIMP-4 and blaNDM-1 co-carrying E. cloacae ST418 are worrying, efficient and sustained control measures are urgently required.