Antibiotic Susceptibility of NDM-Producing Enterobacterales Collected in the United States in 2017 and 2018

Carbapenem-resistant Enterobacterales (CRE) are an urgent public health threat (1). In 2017, CRE caused an estimated 13,100 cases in hospitalized patients, 1,100 deaths, and $130 million in estimated attributable health care costs in the United States (1). Carbapenem resistance in Enterobacterales can develop via several mechanisms, the most concerning of which is the presence of carbapenemases. Carbapenemases are often carried on mobile genetic elements that can spread easily between bacteria (2). Treatment options for these infections are limited, and mortality associated with CRE infections is higher than for carbapenem-susceptible Enterobacterales infections (3, 4). Since 2015, newly available β-lactam combination agents such as ceftazidime-avibactam, meropenem-vaborbactam, and imipenem-relebactam have shown promise in treating CRE infections, especially with CRE that produce the Klebsiella pneumoniae carbapenemase (KPC) (5–7). However, none of these newer β-lactam combination agents exhibit significant activity against the metallo-β-lactamases (MBLs), including the New Delhi MBL (NDM) (8).

Since the initial reports of NDM in 2009, NDM-producing CRE have disseminated worldwide, and they are endemic in India, Pakistan, and Bangladesh (9–11). Isolates with bla_NDM were found in the United States in specimens collected as early as 2009 (11). Since then, the detection of carbapenemase-producing CRE has been facilitated by the establishment of the Antibiotic Resistance Laboratory Network (AR Lab Network) (https://www.cdc.gov/drugresistance/solutions-initiative/ar-lab-networks.html). Established in 2016, the AR Lab Network is composed of 55 public health laboratories across the United States with the aim to rapidly detect and contain antibiotic resistance. A report of the first nine months of testing in the AR Lab Network revealed that 3.2% of CRE tested were positive for bla_NDM (12). Thus, while NDM appeared to be the most common MBL in the United States, Verona integron-encoded MBL (VIM) and active-on-imipenem MBL (IMP) were also found to be present in 0.3% and 0.4% of Enterobacterales, respectively (12). Data on the percentage of CRE isolates that were positive for MBLs in the AR Lab Network during 2017 and 2018 are now available (https://arpsp.cdc.gov/profile/antibiotic?tab=ar-lab-network). The emergence of MBL-producing isolates is a concerning development, as these isolates are especially resistant to antibiotics, including the newer β-lactam combination agents, and have displayed pan-resistance (13).

In this study, we sought to determine the antimicrobial susceptibility testing (AST) results for NDM-producing Enterobacterales in the United States against a set of 32 currently available antibiotics and 3 combination agents that are not yet available. To accomplish this, we requested NDM-positive isolates tested by the AR Lab Network between January 2017 and November 2018. These data will help inform clinicians, public health authorities, and policy makers about possible treatment options for infections caused by NDM-producing Enterobacterales.

A total of 275 unique Enterobacterales isolates carrying bla_NDM collected through November 2018 were submitted from 30 AR Lab Network laboratories to the Centers for Disease Control and Prevention (CDC). Isolates were sent from all seven regions of the AR Lab Network (https://www.cdc.gov/drugresistance/laboratories/AR-lab-network-testing-details.html): West (n = 67), Northeast (n = 59), Midwest (n = 44), Southeast (n = 33), Mountain (n = 31), Mid-Atlantic (n = 28), and Central (n = 13). The isolates comprised eight different species, including 125 Klebsiella pneumoniae, 115 Escherichia coli, 28 Enterobacter cloacae complex, 3 Proteus mirabilis, 1 Citrobacter freundii, 1 Klebsiella aerogenes, 1 Klebsiella oxytoca, and 1 Providencia stuartii isolate. The isolates were recovered from a diverse range of primary specimens, including blood (n = 29), respiratory (n = 11), stool or rectal swab (n = 45), urine (n = 144), and other (n = 46) specimens.

All isolates were confirmed to harbor bla_NDM. The modified carbapenem inactivation method (mCIM) was positive for carbapenemase production in all isolates. The broth microdilution (BMD) metallo-β-lactamase screen was positive for 270 isolates (98.2%), negative for 4 isolates (1.5%), and indeterminate for 1 isolate (0.4%).

All isolates were resistant to ertapenem, cefoxitin, cefotaxime, ceftriaxone, ceftazidime, ampicillin, ampicillin-sulbactam, and ceftolozane-tazobactam using Clinical and Laboratory Standards Institute (CLSI) interpretive criteria (14). The MIC ranges, MIC₅₀, MIC₉₀, and categorical interpretations for the remaining drugs are shown in Table 1. No more than 18.2% of isolates were susceptible to any β-lactam or fluoroquinolone (Table 1). Isolate susceptibilities to aminoglycosides, tetracyclines, trimethoprim-sulfamethoxazole, chloramphenicol, and nitrofurantoin varied (Table 1). Nitrofurantoin data were similar for isolates from urinary sources and the overall isolate set (Table 1). For colistin, 99.1% of E. coli isolates, 92.8% of K. pneumoniae isolates, and 96.4% of E. cloacae complex isolates were wild type (WT) based on a CLSI-established epidemiological cutoff value of ≤2 μg/ml (14). All quality control results were within approved ranges (14).

All 275 isolates were multidrug resistant (MDR), and 116 (42.2%) were extensively drug resistant (XDR) (34.8% of E. coli and 46.4% of K. pneumoniae isolates); 207 (75.3%) isolates displayed difficult-to-treat resistance (80.9% of E. coli and 76.0% of K. pneumoniae isolates). Thirteen isolates (4.7%) were categorized as pan-not susceptible (intermediate or resistant) to all 31 drugs with CLSI or FDA breakpoints (14, 15). Four of these 13 isolates (30.8%) also displayed a non-wild-type colistin MIC of ≥4 μg/ml. Three isolates (1.1%) displayed resistance to all 31 drugs and were categorized as pan-resistant; one of these also showed a colistin non-wild-type MIC.

MIC ranges and MIC₅₀ and MIC₉₀ values for aztreonam-avibactam, cefepime-taniborbactam, and cefepime-zidebactam, which are not FDA approved, are provided in Table 1 (16). There were 23 isolates (8.4%) with an aztreonam-avibactam MIC of >4/4 μg/ml and 8 isolates (2.9%) with an aztreonam-avibactam MIC of >8/4 μg/ml. There were 80 isolates (29.1%) with a cefepime-taniborbactam MIC of >8/4 μg/ml, and there were 8 isolates (2.9%) with a cefepime-zidebactam MIC of >8/8 μg/ml.

This report describes the AST results of 275 Enterobacterales isolates harboring bla_NDM collected in the United States during 2017 and 2018. The high proportion of resistance to all currently FDA-approved antibiotics highlights the difficulty clinicians face when attempting to treat infections caused by NDM-producing Enterobacterales. The proportion of susceptible isolates did not exceed 90% for any of the FDA-approved antibiotics tested.

β-Lactams and fluoroquinolones are considered the first-line agents for the treatment of serious Gram-negative infections (17). In this collection, the proportion of isolates with resistance to these agents was very high. Aztreonam showed the highest in vitro activity of all FDA-approved β-lactams, with 18.2% of isolates displaying susceptibility. This was expected, as aztreonam is not hydrolyzed by NDM but is by extended-spectrum β-lactamases (ESBLs) and AmpC β-lactamases; thus, the high number of isolates that were not susceptible to aztreonam likely represents the coproduction of other β-lactamases in addition to NDM. Similar levels of aztreonam activity against NDM-producing bacteria have been described by others (18, 19).

Aminoglycosides are not considered first-line agents due to their potential to cause nephrotoxicity, but their use has been advocated for the treatment of CRE infections (20). A newer agent, plazomicin, has been demonstrated to be superior to colistin for the treatment of infections caused by CRE (21). In the present evaluation, plazomicin and amikacin demonstrated similar activities, as both drugs generated susceptible results for 52.7% of the NDM-producing isolates tested. The observed susceptibility to plazomicin was lower than reported from a recent study in which 66% of NDM-positive isolates were susceptible to plazomicin (22). In previous studies, resistance to plazomicin has been attributed to 16S rRNA methylases that confer high-level resistance to all aminoglycosides, including reported plasmids that harbor both bla_NDM and 16S rRNA methylase genes (11, 23, 24). In our study, none of the aminoglycosides tested showed reliable activity against NDM-producing bacteria; nevertheless, these agents may have a role to play in the treatment of some of these infections when AST is performed and susceptibility is confirmed.

Eravacycline was approved by the FDA in August 2018 for the treatment of complicated intra-abdominal infections. Eravacycline and tigecycline exhibited similar MIC₅₀ and MIC₉₀ values for this collection of isolates. The higher susceptibility observed with tigecycline than with eravacycline is due to the different breakpoints (≤0.5 μg/ml is susceptible for eravacycline and ≤2 μg/ml is susceptible for tigecycline) established by FDA. Neither agent has CLSI breakpoints, and the European Committee on Antimicrobial Susceptibility Testing (EUCAST) breakpoint for tigecycline is lower than the FDA breakpoint (≤0.5 μg/ml is susceptible per EUCAST and ≤2 μg/ml is susceptible per the FDA) (15, 25). Using the more conservative EUCAST breakpoints, 61.8% and 66.2% of the NDM-producing isolates would be categorized as susceptible to tigecycline and eravacycline, respectively. Regardless of the breakpoints applied, these two drugs displayed the highest in vitro activities of all FDA-approved antibiotics tested in this study. However, limitations exist for both drugs. A pooling of tigecycline noninferiority studies demonstrated that tigecycline was associated with excess deaths compared with other drugs (26). Also, tigecycline only has FDA-approved indications for complicated skin and skin structure infections, complicated intra-abdominal infections, and community-acquired bacterial pneumonia (27). It has proven inferior to comparator antibiotics for ventilator-associated pneumonia and urinary tract infections (27, 28). Eravacycline is currently indicated only for the treatment of complicated intra-abdominal infections and is not indicated for the treatment of complicated urinary tract infections (8).

Omadacycline was approved by the FDA in October 2018 for the treatment of community-acquired bacterial pneumonia and acute bacterial skin and skin structure infections; 59.6% of the isolates in this study were susceptible to omadacycline. This agent may have a role in the treatment of some infections with NDM-producing bacteria when AST is performed and susceptibility is confirmed.

Colistin, which has been suggested as a treatment option for CRE (20), displayed in vitro activity against the NDM producers in this collection; 94.5% of isolates tested had a colistin wild-type MIC. Nonetheless, colistin has consistently been found to be inferior to other active agents for the treatment of CRE, so its role in the treatment of NDM producers remains to be determined (5, 21, 28). In addition to questionable efficacy, colistin treatment is associated with high rates of nephrotoxicity and performing colistin AST is problematic (20, 29). Colistin disk AST and gradient diffusion AST are unreliable, leaving broth microdilution, colistin broth disk elution, and the colistin agar test as the only reliable methods of testing (29, 30).

Nitrofurantoin lacks a role in the treatment of serious invasive infections due to its limited systemic absorption, but it may be useful in the treatment of uncomplicated urinary tract infections (31). Among all E. coli isolates with bla_NDM in this collection, 77.4% displayed susceptibility to nitrofurantoin; thus, it remains an agent with reasonable activity for uncomplicated urinary tract infections. For Enterobacterales isolates other than E. coli, only 5.0% displayed susceptibility.

AST was also performed with some agents that are not FDA approved, namely, cefepime-taniborbactam, cefepime-zidebactam, and aztreonam-avibactam. Testing with cefepime-taniborbactam and cefepime-zidebactam resulted in MICs severalfold lower than for cefepime alone. The activity of cefepime-zidebactam against metallo-β-lactamase producers has been described previously by Sader et al., who found the MIC₅₀ and MIC₉₀ to be 0.5/0.5 and 8/8 μg/ml, respectively, for 20 MBL-producing strains (32). Importantly, no clinical breakpoints currently exist for these investigational agents; thus, MIC results cannot be construed to infer susceptibility or potential clinical utility.

Aztreonam-avibactam is intriguing, as it can currently be administered to patients by combining two FDA-approved drugs, aztreonam and ceftazidime-avibactam. There are several case reports documenting the use of ceftazidime-avibactam with aztreonam (33–35). Clinical trials will be needed to confirm the efficacy of this combination. In our study, the MIC₉₀ of aztreonam-avibactam for all isolates tested was 4/4 μg/ml, although higher MICs were associated with the E. coli than the K. pneumoniae isolates (MIC₉₀ of 8/4 μg/ml and MIC₉₀ of 0.5/4 μg/ml, respectively). For clinicians considering the combination of ceftazidime-avibactam and aztreonam for patients with highly resistant infections caused by Enterobacterales, AST for aztreonam-avibactam is available through the AR Lab Network’s Expanded Antimicrobial Susceptibility Testing for Hard-to-Treat Infections (ExAST) program (https://www.cdc.gov/drugresistance/laboratories/ar-lab-network-testing-details/expanded-ast.html). Caution should be exercised when interpreting MIC results since no clinical breakpoints have been established.

One limitation of this study is that the AST profiles presented may not be representative of all NDM-producing isolates in the United States because the isolates tested were a convenience sample. Another limitation is that cefiderocol, which is an FDA-approved agent with activity against NDM-producing Enterobacterales, was not evaluated (36, 37). Lastly, whole-genome sequence analysis would be required to determine the genetic relatedness of the isolates and the complete set of acquired resistance determinants present. Strengths of this study include both the large number of contemporaneous isolates with a rare resistance mechanism collected from geographically diverse regions in the United States and the large number of antibiotics tested, including almost all currently available therapeutic options. Caution should be exercised when interpreting the results for the non-FDA-approved drug combinations given that no clinical efficacy data are available and clinical breakpoints have not been established. Pharmacokinetic and pharmacodynamic data also remain unknown for these non-FDA-approved combinations. Further research would be necessary to determine the genetic explanation for the observed MIC values. Future research could also explore fosfomycin AST, which must be performed by agar dilution, and cefiderocol AST by broth microdilution, which requires a special formulation of iron-depleted cation-adjusted Mueller-Hinton broth (14).

In summary, NDM-producing Enterobacterales cause infections that present substantial treatment challenges. The proportion of susceptible isolates did not exceed 90% for any of the FDA-approved antibiotics tested. This study may help inform clinicians of which agents retain the most efficacy against these highly resistant organisms. The difficulty in treating these infections highlights the importance of rapid detection, containment, and infection control efforts to prevent these infections from becoming more common.

In September 2018, the CDC requested that public health laboratories in the AR Lab Network submit all Enterobacterales isolates with bla_NDM collected since 1 January 2017. The AR Lab Network includes the public health laboratories of all 50 states, four large cities, and Puerto Rico. Isolates were deduplicated so that only one isolate per species per patient was analyzed. If multiple isolates of the same species were submitted from the same patient, the isolate with the earliest collection date was chosen for inclusion in the study. If multiple isolates of the same species from the same patient were collected on the same day, the most invasive isolate was chosen for inclusion in the study.

Organism identification of all isolates was confirmed by matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) mass spectrometry using a Biotyper 3.1 MALDI-TOF system (Bruker Daltonics, Billerica, MA). A multiplex PCR was performed on all isolates to confirm the presence of bla_NDM and to detect bla_KPC (11). The mCIM test was performed on all isolates in accordance with CLSI guidance to confirm that bla_NDM was conferring carbapenemase production (14). The BMD metallo-β-lactamase screen, which is an imipenem MIC compared to an imipenem MIC with a fixed concentration of EDTA and 1,10-phenanthroline, was performed on all isolates; an MIC decrease of ≥2 doubling dilutions was considered positive (11, 38).

AST of isolates was performed for 32 FDA-approved antibiotics and 3 combination agents that are not FDA approved using in-house-developed frozen reference BMD panels according to CLSI guidance (39). Panels were prepared using 96-well sterile, U-bottom polystyrene plates (Caplugs, Buffalo, NY) with cation-adjusted Mueller-Hinton broth (BD Difco, Sparks, MD). The inoculum was prepared by adjusting a suspension of isolated colonies from overnight growth in 0.85% saline to a turbidity equivalent to a 0.5 McFarland standard. The panels were inoculated with a 95-pin sterile inoculator (10 μl pickup; Caplugs). E. coli (ATCC 25922), Pseudomonas aeruginosa (ATCC 27853), K. pneumoniae (BAA-2146, ATCC 700603, and BAA-2814), and Acinetobacter baumannii (NCTC 13304) were used for quality control on each day of testing (American Type Culture Collection, Manassas, VA, and National Collection of Type Cultures, London, United Kingdom). In addition, E. coli NCTC 13353 was used as an initial quality control of each lot of cefepime-taniborbactam-containing panels. Aztreonam-avibactam MICs were determined using a fixed concentration of avibactam (4 μg/ml), cefepime-taniborbactam MICs were determined using a fixed concentration of taniborbactam (4 μg/ml), and cefepime-zidebactam MICs were determined using a 1:1 concentration of cefepime to zidebactam. All categorical interpretations were determined according to CLSI guidance (14). When no CLSI guidance was available, FDA interpretive criteria were used (15). AST was performed once for each isolate.

Each isolate was categorized as MDR or XDR according to the criteria described by Magiorakos et al. (40). Briefly, an isolate was defined as MDR if intermediate or resistant to at least one agent in three or more antibiotic categories; an isolate was defined as XDR if intermediate or resistant to at least one agent in all but two or fewer categories. For the purposes of MDR or XDR classification, CLSI interpretive criteria were applied when available; FDA interpretive criteria were applied for tigecycline, and European Committee on Antimicrobial Susceptibility Testing (EUCAST) criteria were applied for colistin (MIC of ≤2 μg/ml is susceptible and MIC of >2 μg/ml is resistant) (25). Isolates were also categorized as difficult to treat based on the criteria described by Kadri et al. (17). Briefly, difficult-to-treat isolates were defined as isolates that were intermediate or resistant to all β-lactams and fluoroquinolones (excluding recently approved β-lactam combination agents).

The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the CDC.

This study was performed under General Protocol number 7218 for ethical approval by the Institutional Review Board of the Centers for Disease Control and Prevention. We thank Achaogen, Melinta Therapeutics, Inc., Merck Sharp & Dohme Corp., Paratek Pharmaceuticals, Inc., Tetraphase Pharmaceuticals, Inc., Venatorx Pharmaceuticals, Inc., and Wockhardt Bio, AG, for supplying the drug powders for plazomicin, vaborbactam, ceftolozane, relebactam, omadacycline, eravacycline, cefepime–VNRX-5133 (now named cefepime-taniborbactam), and zidebactam. We also thank Cipla, USA, Inc. (the current provider of plazomicin).