Carbapenemases in Enterobacteriaceae: Detection and Antimicrobial Therapy

Carbapenem-resistant Enterobacteriaceae (CRE) have spread rapidly around the world in the past few years, posing great challenges to human health. The plasmid-mediated horizontal transmission of carbapenem-resistance genes is the main cause of the surge in the prevalence of CRE. Therefore, the timely and accurate detection of CRE, especially carbapenemase-producing Enterobacteriaceae, is very important for the clinical prevention and treatment of these infections. A variety of methods for the rapid detection of CRE phenotypes and genotypes have been developed for use in clinical microbiology laboratories. To overcome the lack of efficient antibiotics, CRE infections are often treated with combination therapies. Moreover, novel drugs and emerging strategies appeared successively and in various stages of development. In this article, we summarized the global distribution of various carbapenemases. And we focused on summarizing and comparing the advantages and limitations of the detection methods and the therapeutic strategies of CRE primarily.


INTRODUCTION
Carbapenem antibiotics are generally considered the most effective antibacterial agents for the treatment of multidrug-resistant bacterial infections. However, with the widespread use of carbapenem antibiotics, the prevalence of carbapenem-resistant Enterobacteriaceae (CRE) has increased rapidly, and has become a serious threat to public health. The production of carbapenemases is the major mechanism underlying carbapenem resistance in CRE throughout the world. Carbapenemases are a kind of β-lactamase that can hydrolyze carbapenem antibiotics. According to the Ambler classification method, carbapenemases can be divided into classes A, B, and D. Class A and class D carbapenemases are serine β-lactamases, and class B carbapenemases are metallo-β-lactamases (MBLs) (Ambler, 1980). There is a large overlap between CRE and carbapenemase-producing Enterobacteriaceae (CPE), but the difference is that they were named according to the carbapenem-resistant phenotype and the resistance mechanism (carbapenemase production), respectively. The correct distinction of CRE and CPE and the rapid detection of CPE are important in the treatment and management of clinical infections. This article summarizes the epidemiology of CRE, the detection of CPE, and the status of clinical treatments.
In the past few years, cases of multiple carbapenemases in the same Enterobacteriaceae isolate have been reported. For example, bla NDM−1 and bla IMP−4 coexisted in KPN , Enterobacter cloacae or Citrobacter freundii carried both bla KPC and bla NDM (Feng et al., 2015;Du et al., 2016a;Yang et al., 2018). Besides, there was a Klebsiella oxytoca isolate coexpressing three carbapenemases, KPC-2, NDM-1, and IMP-4, which was identified in 2017, and the plasmids containing these three resistance genes have emerged in most other members of the family Enterobacteriaceae, including E. coli, E. cloacae, and Klebsiella species (Wang et al., 2017).

RAPID DETECTION OF CARBAPENEMASES
Initial susceptibility testing like broth microdilution techniques, the Kirby-Bauer disk diffusion method and automatic analysis systems were standardized and simple. But using the screening breakpoints recommended by the CLSI or EUCAST guidelines will miss the inefficient carbapenemases like KPC variants and OXA-48 (Fattouh et al., 2016;Gagetti et al., 2016). Automated systems may cause discrepancies in the detection of all types of carbapenemase producers (Woodford et al., 2010). Therefore, phenotypic assays and molecular-based techniques are the two main methods currently used to detect carbapenemases.

Phenotypic Detection Assays
The modified Hodge test (MHT) is a common phenotypic method for the detection of CPE. It is based on whether the growth of the indicator strain is enhanced at the junction of the inhibition zone and the growth line produced by the indicator strain and the test strain, respectively, and estimates whether the test strain has an inactivation effect on antibacterial drugs (Girlich et al., 2012). The method has high sensitivity and specificity in detecting KPC-producing CRE but poor sensitivity in detecting class B β-lactamases (<50%). However, this limitation can be overcome by the addition of Triton X-100, which was proposed and called the Triton Hodge FIGURE 1 | The global distribution of various carbapenemases in CPE. Carbapenemases have emerged in majority regions all over the world. KPCs are the most common carbapenemases and mainly prevalent in China, the Unite States, Italy, and the majority regions of South America; NDMs are mainly prevalent in China, Pakistan, India, and Bangladesh, and widely spread around the world; IMPs are mainly prevalent in Japan and Taiwan, China; VIMs are mainly prevalent in Greece; OXA mainly refers to OXA-48, and is mainly prevalent in Turkey, Morocco, and European countries (France, Germany, Netherlands, Italy, the United Kingdom, and so on); and various carbapenemases locally spread in Europe.
test. This method increased the sensitivity of the detection of NDM-producing clinical isolates to >90% and improved its performance in detecting other carbapenemases at the same time . But the false-positive and false-negative results will affect clinical judgment (Carvalhaes et al., 2010). Nordmann et al. (2012) subsequently developed a colorimetric assay, the Carba NP test, which is faster and has lower falsepositive rate than MHT. In this test, the change in the pH of the reaction system caused by the carbapenemase hydrolysis of imipenem is monitored as the concomitant change in the color of phenol red, which is judged subjectively by the operator in the laboratory. Moreover, this method could preliminarily identify carbapenemases types based on tazobactam and EDTA . And then Pires et al. (2013) replaced phenol red with bromothymol blue as the pH indicator when they developed the Blue-Carba test, which improved the assay sensitivity from 93.3 to 100% (Novais et al., 2015). Bogaerts et al. (2016) proposed an electrochemical method derived from the traditional assay, and designated it the Bogaerts-Yunus-Glupczynski (BYG) Carba test. This test reduces the time required from 2.5 h to about 30 min, and resulting from the real-time curve results, this test offers a real-time objective measurement of carbapenemase-producing isolates (Bogaerts et al., 2016). Various commercialized products are also available, such as Rapidec Carba NP (bioMérieux), Rosco Rapid Carb Screen, and the Rapid Carb Blue Kit. A study suggested that most manual and commercial rapid colorimetric assays are insufficiently sensitive for the detection of OXA-48-type producers (Tamma et al., 2017).
In 2018, another study demonstrated that the MBT STAR-Carba kit (Bruker Daltonics), which is based on bicarbonate, displays higher sensitivity in the detection of OXA, but still cannot avoid undetected errors (Rapp et al., 2018).
The carbapenem inactivation method (CIM) is another effective phenotypic test. This method determines the carbapenemase activity of the tested bacteria by measuring the diameter of the inhibition zone of E. coli ATCC 25922 after the carbapenem disk is inactivated by the test bacterium. The results are highly consistent with the presence of carbapenemase genes, including those encoding KPC, NDM, VIM, IMP, OXA-48, and OXA-23, detected with polymerase chain reaction (PCR) (100% agreement for Enterobacteriaceae) (van der Zwaluw et al., 2015). The modified CIM (mCIM) became the CLSIrecommended method in 2017. A study indicated that both the sensitivity and specificity of mCIM were 100% (Kuchibiro et al., 2018). Because of its simplicity, clear criteria, cost-effectiveness, and availability in any laboratory, the mCIM has become a useful tool in microbiology laboratories.
Many tests that rely on directly monitoring the hydrolysis of β-lactamases to detect CPE have been reported, including a spectrophotometric method (Bernabeu et al., 2012), which is regarded as a reliable detection assay. But extracting the carbapenemases from the bacterial cells is time-consuming, and there were various factors reducing the veracity of the results. To overcome these limitations, Takeuchi et al. (2018) developed a dual-wavelength measurement which could measure the hydrolytic activity of carbapenemases using bacterial cells directly. On the one hand, this method is time saving (about 40 min for preparation and incubation, but the time of detecting OXA should be prolonged appropriately). On the other hand, this method showed higher sensitivity and specificity than carbaNP at the same incubation time, and obtained consistent results upon mCIM. However, the requirement for a specific instrument (spectrophotometer) and the small sample size limit its clinical application (Takeuchi et al., 2018).
In 2011, Hrabák et al. (2011) proposed that matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) could be used to screen CPE by detecting the by-products of the hydrolyzed carbapenem. Since then, other groups have developed various MALDI-TOF-based methods to improve the sensitivity of the procedure, reduce the detection time, and facilitate the interpretation of the results (Johansson et al., 2014;Knox et al., 2014;Sauget et al., 2014;Lasserre et al., 2015;Papagiannitsis et al., 2015). For example, aiming at the low sensitivity mainly resulting from the false-negative results obtained with OXA-48-type producers, Papagiannitsis et al. (2015) added NH 4 HCO 3 to the reaction buffer, which improved its sensitivity from 76 to 98%. To save time, Lasserre et al. (2015) developed a MALDI-TOF-based method that directly detects resistant Enterobacteriaceae from primary culture plates in <30 min and ensures high sensitivity and specificity. In 2018, a survey demonstrated that a MALDI-TOF-MS-based ertapenem hydrolysis assay rapidly and accurately detected the carbapenemase activity of Enterobacteriaceae strains in positive blood cultures (Yu et al., 2018a). Although the costs of measurement using MALDI-TOF MS are low, the equipment remains expensive, which limits the wide application of this method in clinic (Lasserre et al., 2015).
As well as all these methods, carbapenemase-inhibitorbased disc tests have been shown to detect carbapenemases (Tsakris et al., 2010). For example, combining boronic acid with an ertapenem or meropenem disk has been applied in detecting production of KPC (Doi et al., 2008). Adding ethylenediaminetetraacetic acid to a carbapenem disk makes it a useful compound in detecting MBLs (Franklin et al., 2006). Ote et al. developed an immunochromatographic assay to directly detect OXA-48-like carbapenemase using a monoclonal antibody, and the results were obtained in a very short time . A bioluminescence-based carbapenem susceptibility detection assay was reported in 2018 that allows carbapenemase-producing CRE and non-carbapenemaseproducing CRE to be distinguished with a sensitivity of 99% and a specificity of 98% (Van Almsick et al., 2018).

Molecular-Based Detection Methods
Tests based on molecular techniques are considered the gold standards for the identification of carbapenemase genes (Nordmann et al., 2011), the advantages and limitations have been summarized in Table 1. PCR is the most commonly used traditional molecular genotyping method. However, the traditional PCR method for identifying a single gene is timeconsuming. Therefore, multiple PCR that was time-saving with high levels of sensitivity and specificity (Ellington et al., 2016) was proposed and developed. From 2006 to 2012, the multiplex real-time PCR systems have been initially established for the rapid detection of most carbapenemases like KPC, OXA-48 (Swayne et al., 2011), VIM, IMP (Mendes et al., 2006), and NDM (Monteiro et al., 2012). Furthermore, various modified methods were proposed to overcome the inaccuracy caused by the diversity of OXA-48-like carbapenemases , such as a real-time PCR assay based on a high-resolution melt analysis , and a multiplex PCR assay using peptide-nucleic acid probes, which could identify resistance genes in a mixture of Enterobacteriaceae isolates with highly efficient (Jeong et al., 2015). As well as the methods described above, several other molecular methods are used to detect CPE. For example, Walker et al. (2016) combined nested PCR, real-time PCR, and microfluidics to identify the common carbapenemases genes. A PCR-based method in a cartridge format developed to detect CPE in rectal swabs, which is run on the GeneXpert platform, displayed high sensitivity (96.6%) and specificity (98.6%) within a short time (32-48 min) (Tato et al., 2016). Srisrattakarn et al. (2017) developed a loop-mediated isothermal amplification method with hydroxynaphthol blue dye (LAMP-HNB), which was highly efficient (100% sensitivity and specificity). In 2018, the microfluidic chip technology which allows the rapid detection of pathogens and their resistance genes (Kim et al., 2017) was used to detect carbapenem-resistance genes, with high sensitivity and specificity (both >90.0%), and fully met the requirements for clinical diagnoses (Zhang G. et al., 2018). Verigene Gram-negative blood culture assay, the microarray-based commercialized products, was available to identify the carbapenemases (Ledeboer et al., 2015). But the materials cost is a little bit expensive approximately $60-80 per test (Hill et al., 2014). In addition, whole genome sequencing is the most reliable method for the detection of carbapenemase genes, but the high cost, long turnaround time, and difficult data management limit the routine clinical application of this method (Patel, 2016). Yu et al. (2018b) also developed a novel multiplex PCR amplification reaction to directly and rapidly identify the epidemic CRKP ST258/ST11 strain. The advantages and limitations of common detection methods have been shown in Table 1.

TREATMENT OF CRE INFECTIONS
To the bests of our knowledge, almost all β-lactam antibiotics have limited effects on the treatment of CRE infections, and carbapenemases cannot be inhibited by traditional β-lactamase inhibitors . Some restricted drugs, such as polymyxins, tigecycline, and fosfomycin, may be active. A proportion of CRE strains producing KPC and OXA-48 are also sensitive to aminoglycosides (gentamicin and amikacin). However, there are significant deficiencies in the use of monotherapy to treat CRE infections with these antibiotics. Polymyxin has significant nephrotoxicity and neurotoxicity (van Duin et al., 2013), and the optimal dose for treatment is unknown. This antibiotic has also been challenged by the emergence and global spread of mobilized colistin resistance (mcr) determinants. The presence of both mcr-1 and various bla NDM has been reported in Enterobacteriaceae isolates (Du et al., 2016b;Yao et al., 2016;Zheng et al., 2017;Li et al., 2018). The increased mortality risk conferred by tigecycline (Cai et al., 2010;Shen et al., 2015;Ni et al., 2016) has led to warnings by the Food and Drug Administration (FDA, 2013). Furthermore, reports of clinical tigecycline resistance were published soon after its first use in medical practice. The resistance mechanisms that have been reported including mutations in tet (Linkevicius et al., 2016;He et al., 2019) and the increased expression of RND efflux pumps (Nicoloff and Andersson, 2013;Fang et al., 2016). Besides, tigecycline tends to inducing resistance during therapy (Spanu et al., 2012;van Duin et al., 2014;Du et al., 2018). The therapeutic effects of aminoglycosides in CRE infections can be affected by rmtB which confers high-level and widespread resistance (Cheng et al., 2016). The efficacy of fosfomycin is limited and resistance to this drug develops rapidly during treatment (Karageorgopoulos et al., 2012). Moreover, fosfomycin-modified genes play the key role in fosfomycin resistance. It is noteworthy that a carbapenem-, colistin-, and tigecycline-resistant E. coli strain carrying the fosA3 was reported in China in 2018 , which poses a great threat to public health.
For the reasons described above, several methods have been proposed to enhance the efficacies of these antibiotics, including aerosolized antibiotics for treatment with colistin (Valachis et al., 2015) and higher maintenance doses of colistin and tigecycline (Falagas et al., 2014;Trifi et al., 2016). These regimens did improve the therapeutic effects, but convincing evidence is sparse. In this context, combination therapies have been recommended to treat multidrug-resistant CRE infections. Not only the retrospective studies but also the in vitro tests and clinical applications have proved that the combination therapies were effective for the treatment of CRE (Cprek and Gallagher, 2015;Oliva et al., 2015;Ku et al., 2017). And the mortality rates associated with combination therapies especially the carbapenem-containing combinations were lower than those associated with monotherapy (Mataseje et al., 2016). By combining previous researches on combination therapies (Entenza and Moreillon, 2009;Tzouvelekis et al., 2012;Falagas et al., 2014;Pontikis et al., 2014;Chinese XDR Consensus Working Group et al., 2016), several regimens were proposed in Table 2. However, the mechanistic basis of the synergy has not yet been established for most commonly used combination therapies (Baym et al., 2016). As well as the antibiotic combination treatments, novel β-lactamase inhibitors and antimicrobial therapeutics were developed to treat CRE infections and eliminate colonization. Avibactam (AVI) is a novel β-lactamase inhibitor that inhibits KPC, ESBL, AmpC, and OXA-48 (van Duin and Bonomo, 2016). Ceftazidime-AVI (CAZ-AVI) has been used in clinical treatments in the United States since 2015 and was recommended by CLSI in 2018. These combination is effective not only for strains producing KPC and OXA-48 , but also for hypervirulent KPN carrying bla KPC−2 (Yu et al., 2018c). CAZ-AVI combined with ertapenem also successfully treated a patient infected with NDM-producing KPN (Camargo et al., 2015). And clinical reports indicated that CAZ-AVI showed commendable therapeutic effect in treating complicated urinary tract or intra-abdominal infections (Tuon et al., 2018). Comparing with colistin, CAZ-AVI showed  Shields et al., 2017). To broaden the antibacterial spectrum, aztreonam-AVI was proposed, and effectively inhibited a variety of class A, B, and D carbapenemases (Vasoo et al., 2015). Another two novel carbapenem-β-lactamase  (Lob et al., 2017) 3. Well tolerated (Sims et al., 2017) 4. Few adverse evens (Zhanel et al., 2018) 1. Poor inhibition of MBLs and OXA (Lapuebla et al., 2015a) 2. Insufficient phase III clinical trials data (Sims et al., 2017) 1. Low expression of OmpK36 (Hecker et al., 2015) Meropenemvaborbactam 1. Inhibition of KPC (Lapuebla et al., 2015b) 2. Well tolerated 3. Few adverse evens (Zhanel et al., 2018)  inhibitor combinations, imipenem-relebactam and meropenemvaborbactam, were developed to treat CPE infections. And the latter has been recommend by FDA 1 . In vitro data have indicated that the two combinations are highly active against KPC-producing Enterobacteriaceae but poorly susceptive against MBLs and OXA-type carbapenemases (Lapuebla et al., 2015a,b). And exact efficacy and safety must be defined with further clinical data (Zhanel et al., 2018). Besides, meropenem-nacubactam during clinical development have shown promising in vitro activity against KPC and MBL-producing CRE Mushtaq et al., 2019). Moreover, cefepime-zidebactam could inhibit CRE producing carbapenemases of classes A, B, and D (Thomson et al., 2019), other cefepime-β-lactam enhancer such as cefepime-enmetazobactam (AAI101)/WCK-5153, etc. which were in earlier stages of development may represent a novel carbapenem-sparing option (Giacobbe et al., 2018;Moya et al., 2019;Papp-Wallace et al., 2019). Several other new drugs such as plazomicin, eravacycline, and cefiderocol developed to treat CRE infections are in various stages of development (Kohira et al., 2016;Thaden et al., 2017;Rodríguez-Baño et al., 2018), among which plazomicin performed more potent effect and lower side effects than other aminoglycosides (Livermore et al., 2011;Riddle et al., 2012;Walkty et al., 2014;Castanheira et al., 2018) and eravacycline showed favorable clinical response and had well pharmacokinetics, tolerability, and in vitro activity (Zhanel et al., 2016;Thaden et al., 2017). The application of cefiderocol needs further clinical data. In 2018, the injection products of plazomicin and eravacycline have been recommend by FDA 2 . However, due to the emergence of resistant isolates (Livermore et al., 2011;Castanheira et al., 2018;He et al., 2019), enough attention should be paid to the development of drug resistance. The advantages, limitations, and mechanisms of resistance of novel antimicrobial therapeutics have been shown in Table 3.
As well as novel drugs, various strategies for the management of carbapenem resistance have recently emerged. For example, based on studies of fecal microbiota transplantation (FMT) and enteric pathogens (Wang et al., 2014;Caballero et al., 2015), FMT was hypothetically suggested to be used as a clearance method for CRE colonized patients, but the feasibility requires further study (Wang et al., 2016;Qazi et al., 2017). Based on research into the mechanisms of antibiotic cytotoxicity (Cheng et al., 2014;Citorik et al., 2014;Dwyer et al., 2014), novel synthetic tools developed for the precise removal of genomic islands have been proposed to replace antibiotic treatments (Vercoe et al., 2013). Immunological-based therapies, such as monoclonal antibodies targeting poly-(-1,6)-N-acetyl glucosamine (Skurnik et al., 2016) and cationic antimicrobial peptides (Pan et al., 2015), are also under investigation as substitutes for traditional antibiotics (DiGiandomenico and Sellman, 2015). The ability of predatory bacteria to reduce the primary pathogen in mammalian system has been demonstrated, which suggested the application prospect in clinic (Shatzkes et al., 2016). The advantages and limitations of these main novel strategies have been summarized in Table 4.

SUMMARY
In recent decades, CRE have spread widely in various medical institutions around the world, and due to the time-consuming detection methods and limited treatment regimens, the mortality rates among patients are high. Therefore, the timely and accurate detection of CRE, especially CPE, is essential for the clinical treatment and prevention of infections. A variety of phenotypic methods and gene-based methods are available for the rapid detection of carbapenemases, and these are expected to be used routinely in clinical microbiology laboratories. At present, novel antibacterial drugs and emerging strategies which have been recommend or during development, with good activity and safety profiles, are expected to be applied to the clinical treatment of these infections in the near future.

AUTHOR CONTRIBUTIONS
HD designed the study. XC and HZ performed data analysis and prepared the manuscript. All authors approved the final manuscript.