The Growing Threat of NDM-Producing Escherichia coli With Penicillin-Binding Protein 3 Mutations in the United States—Is There a Potential Role for Durlobactam?

Abstract

We report identification of 5 patients with infections caused by NDM-5-producing Escherichia coli harboring PBP3 mutations that showed reduced susceptibility to aztreonam-avibactam and cefiderocol. Durlobactam, a novel diazabicyclooctane β-lactamase inhibitor, demonstrated minimum inhibitory concentrations ranging from 0.5 to 2 µg/mL supporting future investigations into a potential role in clinical management.

BACKGROUND

Escherichia coli harboring New Delhi metallo-β-lactamases (NDM) are a global threat. Current guidance suggests that either cefiderocol or ceftazidime-avibactam in combination with aztreonam are preferred treatments for serious infections [1]. Both exert activity through inhibition of penicillin binding protein (PBP) 3; however, NDM-harboring E. coli with PBP3 insertions are increasingly reported [2]. These insert variants have decreased binding affinity for both cefiderocol and aztreonam leading to reduced susceptibility [3]. Moreover, E. coli harboring mutant PBP3 often co-harbors transferrable β-lactamases that further compromise the activity of preferred β-lactams [3, 4]. Alternatives to β-lactams are both less effective and more toxic [5, 6]. Thus, identifying effective antimicrobials for NDM E. coli with PBP3 insertions is an urgent need.

We recently cared for 5 epidemiologically unrelated patients at 2 centers who presented with intra-abdominal infections caused by NDM-producing E. coli with reduced susceptibility to aztreonam-avibactam [3, 7]. In each case, patients had surgical or interventional radiology-based source control and were clinically cured (Table 1). Three patients were successfully treated with tetracycline-based regimens, whereas the other 2 patients, each with secondary bloodstream infections, responded to prolonged courses of ceftazidime-avibactam plus aztreonam. Given the limited therapeutic options, and reports of therapeutic failure with cefiderocol and ceftazidime-avibactam plus aztreonam for similar cases [4], we explored the in vitro activity of alternative agents.

Table 1.

Clinical Characteristics, Antibiotic Susceptibility Testing, and Whole-Genome Sequencing for NDM-Producing E. coli Isolates Collected From 5 Patients

Patient	Age/Sex	Underlying Diseases	Infection Type	Sequence Type	NDM Variant	Other β-lactamases	PBP3 (ftsI)	Aztreonam-avibactam MICa (µg/mL)	Cefiderocol MICb (µg/mL)	Durlobactam MIC (µg/mL)c	Antibiotic Treatment (d)	Source Control	Treatment Outcomed
1	42/M	Crohn's	Enterocutaneous fistula with obstructed drain	2659	NDM-5	CMY-42	Q227H, 334insYRIN, E353K, I536L	8	2	1	Tigecycline (5)	Drain repositioning	Clinical cure
2	72/M	Myelodysplastic syndrome, chronic neutropenia	Neutropenic enterocolitis with perirectal abscess	167	NDM-5	CTX-M-15	334insYRIN, E353K, I536L	2	2	0.5	Ceftazidime-avibactam plus aztreonam, tigecycline (28)	Incision and drainage, seton drain placement, neutrophil recovery	Clinical cure
3	70/M	Diabetes, chronic sacral wound with colostomy, cholecystitis, complex intra-abdominal infections	Biliary abscess with paracolic fistula	167	NDM-5	CTX-M-15	Q227H, 334insYRIN, E353K, I536L	4	>32	2	Tigecycline (indefinite)	Repositioning of percutaneous cholecystostomy tube, aspiration of intra-abdominal abscess	Clinical improvement, ongoing infection
4	48/M	Alcohol-related cirrhosis s/p liver transplant	Bloodstream infection	361	NDM-5	CMY-145	Q227H, 334insYRIN, E353K, I536L	16	8	2	Ceftazidime-avibactam plus aztreonam (35)	Surgical repair of biliary leak	Clinical cure
5	72/M	Metastatic pancreatic cancer s/p liver resection	Bloodstream infection	410	NDM-5	CMY-141	334insYRIK, A413V	16	1	2	Ceftazidime-avibactam plus aztreonam (28)	Pancreatic duct stent replacement	Clinical cure

Abbreviations: MIC, minimum inhibitory concentration (µg/mL); WT, wild type; YRIK, Tyr-Arg-Ile-Lys; YRIN, Tyr-Arg-Ile-Asn.

^aTested in the presence of 4 µg/mL of avibactam.

^bTested using iron-depleted, cation-adjusted Mueller-Hinton broth.

^cTested using doubling dilutions ranging from 0.06 to 64 µg/mL.

^dClinical cure was defined as resolution of signs and symptoms of infection without recurrence of NDM-producing E. coli infection.

Durlobactam is a novel diazabicyclooctane β-lactamase inhibitor with a spectrum that includes OXA-type β-lactamases present in carbapenem-resistant Acinetobacter baumannii. Durlobactam was recently approved by the Food and Drug Administration (FDA), in combination with sulbactam, as a treatment for hospital-acquired or ventilator-associated pneumonia caused by susceptible isolates of A. baumannii. Interestingly, durlobactam has independent in vitro activity against Enterobacterales. This activity is mediated by PBP2 inhibition resulting in potent killing against Enterobacterales in hollow-fiber infection models [8, 9]. As durlobactam is not meaningfully degraded by β-lactamase enzymes [10], we hypothesized it would retain in vitro activity against NDM-harboring E. coli with PBP3 inserts. No IRB approval was needed for this case series per institutional standards.

RESULTS

Initial screening of NDM-producing E. coli isolates was conducted by sulbactam-durlobactam disk diffusion testing (Supplementary Figure 1). Zones of inhibition against clinical isolates were similar to those against E. coli ATCC 25922, a control strain without exogenous β-lactamases or PBP mutations and with uniformly low antibiotic minimum inhibitory concentrations (MICs). Given that sulbactam does not have clinically relevant activity against E. coli, is hydrolyzed by NDM enzymes [11], and ampicillin-sulbactam MICs were ≥16 µg/mL against all isolates, we anticipated that inhibition of bacterial growth was mediated solely by durlobactam.

To confirm these findings, we performed broth microdilution susceptibility testing and whole genome sequencing (WGS) of bacterial isolates. Susceptibility testing was performed in triplicate by standard methods and results were interpreted according to Clinical and Laboratory Standards Institute (CLSI) breakpoints (https://clsi.org/m100free). Durlobactam was tested across a range of 0.06 to 64 µg/mL without the addition of sulbactam. No susceptibility breakpoints are available for aztreonam-avibactam and durlobactam, and therefore MIC values are reported without interpretation. WGS was performed on an Illumina NextSeq platform (San Diego, California, USA). Sequences were assembled using SPAdes v3.15.5 and antibiotic resistance genes were identified by ResFinder. Protein sequences were compared to E. coli K-12 MG1655.

Susceptibility testing and WGS results are shown in the Table 1. Aztreonam-avibactam and cefiderocol MICs ranged from 2 to 16 µg/mL and 1 to >32 µg/mL, respectively. Isolates from each of the 5 patients were genetically unrelated and separated by >10 000 core-genome single nucleotide polymorphisms; each harbored bla_NDM-5. Three of 5 isolates carried bla_CMY variants, whereas the remaining 2 carried bla_CTX-M-15. Sequence analysis of PBP3 showed a 4 amino acid insertion after position 333 for all isolates (YRIN [Tyr-Arg-Ile-Asn] in 4, YRIK [Tyr-Arg-Ile-Lys] in 1). Other non-synonymous substitutions in PBP3 were identified in all isolates. Durlobactam MICs ranged from 0.5 to 2 µg/mL confirming initial observations from disk diffusion testing.

DISCUSSION

Herein we highlight the limited treatment options for NDM-producing E. coli with PBP3 inserts and reduced susceptibility to aztreonam-avibactam and cefiderocol. All isolates harbored other β-lactamases, including CMY variants known to attenuate the activity of aztreonam-avibactam [3]. To explore novel therapeuticss, we assessed in vitro activity of durlobactam, which does not inhibit NDM enzymes but is stable to hydrolysis and demonstrates potent in vitro activity against E. coli due to intrinsic PBP2 inhibition [8]. We showed in vitro activity of durlobactam against all evaluated isolates. These data support future investigations into the potential role of durlobactam against NDM-producing E. coli with PBP3 inserts. Fortunately, all patients in our study responded to conventional regimens following surgical management; however, such options may not be viable for all patients. Indeed, it is unlikely that pharmacodynamic targets can be achieved with ceftazidime-avibactam plus aztreonam once aztreonam-avibactam MICs exceed 4 µg/mL [12].

NDM-producing E. coli harboring PBP3 insertions are increasingly reported [2–4] and are now endemic in many regions globally [3, 13]. This is particularly problematic considering PBP3 insertions at residue 333 result in structural changes to the transpeptidase binding pocket which limit the accessibility of several β-lactams—including aztreonam, cefepime, cefiderocol, and other cephalosporins [2]. Although the activity of carbapenems is generally preserved due to their high affinity for PBP2, they are readily hydrolyzed by NDM β-lactamases mitigating clinical utility. Further complicating matters, many NDM-producing E. coli with PBP3 insertions co-harbor CTX-M or CMY variants that hydrolyze aztreonam, and in the case of CMY-42 and CMY-145, overcome inhibition by avibactam [2, 3]. These complementary mechanisms of resistance cannot be overcome by inhibition of β-lactamases alone. For example, taniborbactam, a cyclic boronate NDM inhibitor does not restore cefepime activity against NDM-producing E. coli with PBP3 insertions due to the inability of cefepime to bind PBP3 [14]. Ergo, effective β-lactam combinations depend upon the use of agents that do not exclusively target PBP3 and are not hydrolyzed by β-lactamases.

Durlobactam, a novel diazabicyclooctane β-lactamase inhibitor, adds complimentary activity through potent inhibition of serine β-lactamases and PBP2 [8]. In our experience, durlobactam MICs were ≤2 µg/mL for all isolates, a concentration below the minimum concentration achieved in patients receiving standard dosing of 1 gram every 6 hours [15]. Potent in vitro killing has also been observed against carbapenemase-producing E. coli isolates tested by static time-kill analyses simulating steady-state concentrations of durlobactam ∼17 µg/mL [9]. The most significant concern related to using durlobactam as a PBP2 inhibitor clinically is the reported high frequency of resistance identified for PBP2 inhibitors like mecillinam and other newer-generation diazabicyclooctane β-lactamase inhibitors [16–18]. Resistance to these agents appears to be unrelated to drug inactivation or PBP mutations but instead manifests through upregulation of stress response and other regulatory pathways impacting in vivo efficacy. How these alterations impact pathogen virulence or in vivo efficacy of PBP inhibitors is unclear. Thus, further investigation is needed to determine the pharmacokinetic and pharmacodynamic targets for durlobactam efficacy as an active agent alone, and in combination with β-lactams.

The combination of cefepime-zidebactam also holds promise for future treatment of gram-negative pathogens with PBP3 inserts [14, 19]. Against 28 NDM-producing E. coli with PBP3 inserts, cefepime-zidebactam MICs were ≤0.125 µg/mL [14]. This activity is likely driven by zidebactam considering that cefepime's activity is compromised in PBP3 mutants, and zidebactam alone demonstrated MICs ≤1 µg/mL. Both zidebactam and the combination of cefepime-zidebactam demonstrated in vivo efficacy in a neutropenic murine pneumonia model against carbapenemase-producing Pseudomonas aeruginosa with PBP3 insertions [19]. These data attest to the potential utility of PBP2 inhibitors like durlobactam in combination with β-lactams that target complimentary PBPs. Similarly, mecillinam and pivmecillinam are PBP2-specific β-lactam agents widely used in Scandinavia for the treatment for urinary tract infections and are now under FDA evaluation.

Our experience also highlights the limitations of available susceptibility testing and molecular diagnostics for multidrug-resistant pathogens in clinical microbiology laboratories. Although aztreonam-avibactam susceptibility testing is available through the Centers for Disease Control and Prevention’s (CDC's) Antimicrobial Resistance Laboratory Network, this testing is not always practical in real time. Alternative methods for evaluating synergy of ceftazidime-avibactam plus aztreonam are used in research laboratories but are not guided by quality-control and are subject to inter-user variability. The CLSI has developed a broth disk elution method that performs well against Enterobacterales but has not been widely implemented [20]. In each of our cases, rapid diagnostic tests identified the presence of NDM enzymes, but did not distinguish between NDM alleles. Furthermore, the presence of CMY and/or PBP3 variants are not included on commercial diagnostics or otherwise routinely assessed. Thus, if cefiderocol or ceftazidime-avibactam plus aztreonam is considered for treatment, clinicians should pursue susceptibility testing, particularly for NDM-producing E. coli. Notably, PBP3 insertions have not been widely reported for other NDM-producing Enterobacterales [3].

In conclusion, we report 5 cases of NDM-5-producing E. coli with PBP3 alterations demonstrating reduced susceptibility to aztreonam-avibactam and cefiderocol. Durlobactam demonstrated in vitro activity against each of the isolates, however, its role in clinical management is unclear until further studies, including those to identify optimal dosing and partner compounds, have been conducted. Clinicians should be aware of the potential for reduced activity of the front-line treatment options among NDM-producing E. coli, and ideally guide therapy according to accurate susceptibility results.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Financial support. No specific funding was received for this work. Durlobactam powder was kindly provided by Innoviva Specialty Therapeutics.