In vitro activity of the siderophore cephalosporin, cefiderocol, against molecularly characterized, carbapenem-non-susceptible Gram-negative bacteria from Europe

Abstract Objectives Many carbapenem-resistant (CR) Gram-negative (GN) pathogens exhibit MDR, meaning few therapeutic options are available for CR-GN infections. Cefiderocol, a siderophore cephalosporin, has demonstrated in vitro efficacy against CR-GN bacteria. In the SIDERO-CR-2014–2016 surveillance study, European clinical isolates comprising carbapenem-non-susceptible (CarbNS) Enterobacterales and MDR non-fermenters were tested against cefiderocol and comparators. Methods Cefiderocol MICs were determined using iron-depleted CAMHB, and comparators using CAMHB, per recommended CLSI methodology. Carbapenemase gene profiles were determined using PCR. Results Isolates (N = 870) from 23 European countries comprised CarbNS Enterobacterales (n = 457), MDR Pseudomonas aeruginosa (n = 177) and MDR Acinetobacter baumannii (n = 236). The most common carbapenemases were KPC (52%), OXA-48-like (19%), VIM (14%) and NDM (8%) in Enterobacterales, VIM (41%) in P. aeruginosa and OXA-23-like (57%) and OXA-24/40-like (37%) in A. baumannii. Most carbapenemase-producing isolates (65%) co-carried ESBLs. Approximately half of P. aeruginosa isolates were negative for carbapenemases, compared with 10% of Enterobacterales and 3% of A. baumannii. A similar proportion of Enterobacterales were susceptible to cefiderocol (81.6%; 79.0% of VIM producers; 51.4% of NDM producers; based on EUCAST breakpoint values) compared with comparator antimicrobial agents, including colistin (76.4%; 93.5% of VIM producers; 78.4% of NDM producers) and ceftazidime/avibactam (76.6%; 1.6% of VIM producers; 2.7% of NDM producers). Of P. aeruginosa isolates, 98.3% were susceptible to cefiderocol (100% of VIM producers), similar to colistin (100%). Against A. baumannii, 94.9% had cefiderocol MIC ≤2 mg/L and 93.6% of isolates were susceptible to colistin. Conclusions Cefiderocol demonstrated potent activity against CarbNS and MDR GN bacteria, including non-fermenters and a wide variety of MBL- and serine-β-lactamase-producing strains.


Introduction
Carbapenems are a treatment of choice for many infections caused by Gram-negative (GN) bacteria. Globally, the incidence of carbapenem-resistant (CR) GN bacteria is increasing, 1,2 with resistance largely due to the production of b-lactamase enzymes. 3 There are very few therapeutic options for CR-GN bacterial infections, as many CR pathogens exhibit MDR, including resistance to alternative b-lactams (e.g. cephalosporins and penicillins), as well as other common drug classes such as aminoglycosides and fluoroquinolones. 1,3 At the time of treatment initiation, and even after pathogen identification, molecular characterization of the mechanism of carbapenem resistance is not available and the identity of a specific b-lactamase conferring resistance is likely to be unknown. These unknowns can lead to a delay in the time taken to begin appropriate antibiotic therapy, which can negatively impact the length of hospital stay, in-hospital costs and mortality. 4 Cefiderocol is a novel siderophore cephalosporin, which was developed for the treatment of MDR GN bacteria, including those resistant to carbapenems. Cefiderocol has recently been approved in Europe for the treatment of infections due to aerobic GN organisms in adults with limited treatment options 5 and in the USA for the treatment of complicated urinary tract infections (cUTIs), including pyelonephritis, caused by susceptible GN organisms in adult patients with limited treatment options. 6 The structure of cefiderocol is based around a cephalosporin backbone with the addition of a catechol moiety at the 3-position side chain. The cephalosporin core enables cefiderocol to act like other cephalosporins, binding primarily to PBPs and killing bacterial cells by inhibition of peptidoglycan cell wall biosynthesis. Cefiderocol differs from other cephalosporins in that the catechol moiety chelates ferric (Fe-III) iron, mimicking natural siderophores, allowing cefiderocol to exploit the bacteria's own active receptormediated iron transport system to cross the outer membrane. 7,8 The resulting increase in periplasmic concentration circumvents non-specific resistance due to porin loss or efflux and enhances cefiderocol's activity relative to carbapenems, other cephalosporins and b-lactam/b-lactamase inhibitor combinations. 6,9,10 Cefiderocol is active against CR-GN bacteria harbouring b-lactamases from all Ambler classes, including KPC, VIM, IMP, NDM and OXA carbapenemases, and is active against ESBL-producing Escherichia coli and Klebsiella pneumoniae, as well as meropenem-resistant Pseudomonas aeruginosa and Acinetobacter baumannii. [11][12][13] Susceptibility testing of cefiderocol by broth microdilution requires an iron-depleted medium to promote the natural production of siderophores by bacterial cells. Iron-depleted in vitro conditions are essential in order to mimic the hypoferremic conditions encountered by bacteria in the human body during infection. 7,11,14 Determination of cefiderocol MICs using iron-depleted (iron concentration ,0.03 mg/L) CAMHB (ID-CAMHB) 15 has been shown to give reproducible results that correlate with in vivo efficacy 16 and ID-CAMHB has been recommended as the standard medium for cefiderocol MIC determination by CLSI. 17 In the SIDERO-CR surveillance study, CR and MDR clinical isolates of GN bacteria collected from patients between 2014 and 2016 were tested against cefiderocol and comparators using recommended CLSI broth microdilution methodology. 18 In this report, we focus on the clinical isolates provided for the SIDERO-CR study by European hospitals.

Methods
Full methodology for the SIDERO-CR study and molecular characterization using PCR has been published previously (see Supplementary materials, available as Supplementary data at JAC-AMR Online). 19,20 The SIDERO-CR isolates were taken from a 2014-16 International Health Management Associates, Inc. (IHMA; Schaumburg, IL, USA) surveillance collection based on their known antimicrobial susceptibility phenotypes and/or their species identification. 19 Isolates were screened for the presence of clinically relevant b-lactamase genes, including Ambler class A ESBLs (e.g. SHV, CTX-M) and carbapenemases (e.g. KPC, GES), class B MBLs (e.g. NDM, VIM), class C plasmid-mediated AmpC-type b-lactamases and class D b-lactamases (e.g. OXA-type). 20  MICs were determined for cefiderocol, cefepime, ceftazidime/ avibactam, ceftolozane/tazobactam, ciprofloxacin, colistin and meropenem by broth microdilution according to CLSI guidelines. 18 Susceptibilities of all antibiotics, with the exception of cefepime when tested against A. baumannii (CLSI breakpoint 8 mg/L), 18 were interpreted using EUCAST breakpoints. 5,21 Enterobacterales and P. aeruginosa are considered susceptible to cefiderocol at MIC 2 mg/L (resistant .2 mg/L). 5 Cefiderocol was tested using ID-CAMHB, while all other antimicrobial agents were tested using standard CAMHB. Quality control testing was performed on each day of testing.
The activity of cefiderocol and comparators by carbapenemase is summarized in Against both VIM-and GES-producing P. aeruginosa, cefiderocol and colistin demonstrated potent activity, with 100% of isolates being susceptible to both agents. Cefiderocol and colistin were also active against non-carbapenemase-producing MDR P. aeruginosa, with 96.6% of isolates being cefiderocol susceptible and 100% being colistin susceptible.
The wide variety of carbapenemases produced by European Enterobacterales, P. aeruginosa and A. baumannii isolates is apparent in Figure 3. Cefiderocol MIC distributions across key carbapenemase classes are provided in Tables S2-S4. A range of b-lactamases were identified across 99 isolates with cefiderocol MIC values of .2 mg/L, including isolates with co-carriage of multiple b-lactamases. The most common carbapenemase was KPC (n " 39), followed by NDM (n " 19), VIM (n " 13), OXA-48-like (n " 10), OXA-23-like (n " 6) and OXA-24-like (n " 5). Twelve isolates carried either OSBLs/ESBLs or no known b-lactamase, with eight isolates harbouring the PER-type b-lactamase.   25 In particular, the removal of the intermediate category by EUCAST has an impact on interpreting levels of resistance. This is illustrated by the rate of cefiderocol resistance in KPC-producing CarbNS Enterobacterales, which differs by .10-fold across the breakpoint-determining bodies: 16.4% resistant using EUCAST, 1.7% resistant using FDA and 0% resistant using CLSI breakpoints. This demonstrates that an order of magnitude difference in resistance rates can be perceived, despite the same susceptibility breakpoint being Longshaw et al.

Discussion
used by EUCAST and FDA, as an artefact of the lack of an intermediate breakpoint and the MIC at which the resistance breakpoint is set. This is magnified with MBLs, e.g. NDM: 48.6% (18/37) isolates are classed as cefiderocol resistant by EUCAST but only 5.4% (2/37) by FDA and 2.7% (1/37) by CLSI breakpoints. Therefore, it seems that the perceived activity of cefiderocol against KPC-and NDM-producing isolates, for example, may be affected by the differences in interpretive breakpoints for resistance.
Although there is currently some disparity between published breakpoints, pharmacokinetic/pharmacodynamic modelling predicts .90% probability of achieving 75% fT .MIC with the recommended cefiderocol dosing regimen up to an MIC of 4 mg/L for patients with normal renal function. 26 In Phase 3 clinical studies, the mean concentration of unbound cefiderocol exceeded 4 mg/L for the whole dosing period.
Similar isolate studies involving cefiderocol have used provisional CLSI breakpoints. In a study including 1086 CR isolates from the USA (737 KPC producers), MICs were higher for isolates of Enterobacterales with b-lactamases compared with those without, but no clear association was found between the type of b-lactamase and the MIC. 27 Based on the CLSI provisional susceptibility breakpoint of 4 mg/L, 90.5% of Enterobacterales isolates were susceptible to cefiderocol. Notably, as with the European SIDERO-CR-2014-2016 isolates, the MIC 90 for cefiderocol in Enterobacterales was 4 mg/L and the cefiderocol MIC distribution was similar to Figure 3(a). For non-fermenters, cefiderocol activity was independent of the presence of b-lactamases. A recent local study carried out in southern Spain using 231 MDR-GN isolates, including KPC, OXA, ESBL and MBL producers, found that 98% of isolates were susceptible to cefiderocol based on CLSI breakpoints. 28 In this study, 100% of KPC-producing isolates (n " 50) and carbapenemase-producing K. pneumoniae (n " 107) were cefiderocol susceptible.
A key limitation of the SIDERO-CR study was the geographical spread and number of collection sites within countries. Isolates were selected by a limited number of sites per country based on the MDR/CR phenotype, therefore the relative frequency observed may not reflect the national prevalence. Additionally, the number of sites and isolates per country were not proportionate to population. Consequently, the isolate collection is not necessarily representative, as there may be considerable heterogeneity in the numbers and types of pathogens and mechanisms of resistance provided by different locales. However, from SIDERO-CR-2014-2016 isolate characterization, it is apparent that carbapenem resistance across Europe is associated with a diverse range of b-lactamases. Of the CR isolates included, 83.8% (729/870) produced at least one carbapenemase, with 65.2% (475/729) of carbapenemase-producing isolates also carrying ESBLs and/or OSBLs.
Current treatment options for infections caused by MDR-GN bacteria are limited. Throughout Europe, considerable heterogeneity exists across countries with respect to the types of carbapenemases contributing to CarbNS and MDR phenotypes. Even for countries where the dominant carbapenemase is KPC (e.g. Italy, Greece), recent reports have described an increasing proportion of non-KPC MBL-mediated resistance. A Rapid Risk Assessment report from the ECDC described an increase in NDM-containing carbapenemase-producing Enterobacterales in Italy, 29 while an outbreak of XDR K. pneumoniae with genes encoding for OXA-48 and NDM-1 has recently been reported in the Mecklenburg-Western Pomerania state of Germany. 30 A large outbreak of NDM-1producing CarbNS K. pneumoniae was investigated across eight Greek hospitals between 2013 and 2016. 31 The vast majority of these NDM-1-producing K. pneumoniae isolates were of a clonal type (ST11) similar to those identified in Bulgaria in 2015-16, indicating that the Balkan region is also at risk of increasing prevalence of MBL-producing Enterobacterales. 32 Additionally, it has been reported that the use of ceftazidime/avibactam to treat KPCproducing K. pneumoniae infection has led to a shift in the carbapenemase landscape in Greece, with the incidence of MBLs increasing between 2015-17 (12.0%) and 2018 (51.1%), mainly due to VIM-producing K. pneumoniae becoming more prevalent. 33 This shift from KPC-to MBL-based resistance is of concern as these Includes two isolates co-carrying KPC and VIM carbapenemases, eight isolates co-carrying OXA-48 and NDM carbapenemases, one isolate co-carrying GES and VIM carbapenemases, one isolate co-carrying GES and NDM carbapenemases, two isolates co-carrying OXA-24/40-like, OXA-58 and NDM carbapenemases and one isolate co-carrying OXA-23-like and NDM carbapenemases. MIC n , minimum concentration inhibiting n% of isolates tested; NA, not applicable (,10 isolates).
Cefiderocol and carbapenem-non-susceptible Gram-negatives JAR   (Table 1). Colistin, often considered a treatment of last resort, has a broad spectrum of GN activity and is frequently used for CR-and MDR-GN infections, particularly for MDR A. baumannii. 37 However, colistin is undesirable for first-line empirical use due to associated renal toxicity. 38,39 In addition, colistin-resistant strains are emerging, somewhat in P. aeruginosa and A. baumannii, but particularly in K. pneumoniae (including KPC, VIM, NDM and OXA producers), mediated, to some extent, by transmissible mcr-1 resistance. 40 An ECDC Rapid Risk Assessment report was published in response to the increasing trend in MCR-1-mediated colistin resistance in Enterobacterales. 41 Among SIDERO-CR-2014-2016 isolates, 123 (14.1%) were colistin non-susceptible (Figure 1), including 23.6% of Enterobacterales (28.3% of K. pneumoniae) and 6.4% of A. baumannii, with 100% of P. aeruginosa isolates being colistin susceptible. These results do not necessarily reflect the incidence of colistin resistance in specific geographical locations or from specific infection sources; for example, the incidence of colistinresistant A. baumannii was reported to be 47.7% in a set of A. baumannii RTI isolates from Greece, Italy and Spain (n " 65). 42 In vitro assessment of CR-GN isolates from 18 hospitals in Greece demonstrated that 40% of K. pneumoniae and A. baumannii isolates were resistant to colistin; 100% of these isolates had cefiderocol MICs of 4 mg/L. 43 Increasing colistin resistance is a concern for clinicians as it severely restricts an already limited set of treatment options. Patients with serious CR-GN bacterial infections have a poor prognosis and require more effective, less toxic treatments than colistin, covering all MBLs and serine-b-lactamases as well as other mechanisms of resistance. Therefore, there is an urgent need for empirical treatments that cover a range of GN pathogens, coupled with rapid diagnostic techniques to identify pathogen phenotypes quickly, which will allow the prompt delivery of effective therapy.
In this study, the proportions of isolates susceptible to cefiderocol are similar to comparators in Enterobacterales and generally similar to colistin but greater than other comparators in P. aeruginosa. Cefiderocol retained activity against the majority of isolates harbouring MBLs, with only 3.0% (3/99) of MBL-positive CarbNS Enterobacterales isolates having MIC values .4 mg/L. The in vitro activity (MIC 2 mg/L) of cefiderocol against a range of serine-b-lactamases varied from 100% in OXA-58-producing isolates to 83.6% in KPC producers. In addition, 91.5% of isolates with no carbapenemase and 97.5% with no b-lactamase had a cefiderocol MIC 2 mg/L, demonstrating potent cefiderocol activity where the mechanism of resistance was not clear.
Cefiderocol demonstrated potency against the MDR P. aeruginosa isolates (98.3% cefiderocol susceptible), yet half carried no carbapenemase. The low prevalence of carbapenemases in SIDERO-CR-2014-2016 P. aeruginosa isolates aligns with recently reported data describing the prevalence of XDR P. aeruginosa; of 1445 P. aeruginosa isolates collected from 51 Spanish hospitals, 252 (17.3%) were classified as XDR and only 3.1% carried either carbapenemases or ESBLs. 44 Resistance in non-b-lactamase-producing P. aeruginosa isolates in SIDERO-CR-2014-2016 is likely to be due to porin-and efflux pump-mediated mechanisms. 45 Cefiderocol's mechanism of entry into the bacterial cell allows it to evade these mechanisms of resistance, retaining in vitro activity against strains with alterations in outer membrane porins or overexpressed efflux pumps. 7,10 In conclusion, the SIDERO-CR European dataset, coupled with the rise of resistance to existing agents and the shift towards MBL-from serine-b-lactamase-producers, demonstrates the diverse and dynamic nature of the European carbapenemase Longshaw et al.
landscape. In SIDERO-CR-2014-2016, cefiderocol exhibited potent in vitro activity against a broad range of CarbNS and MDR pathogens. The isolates tested included b-lactamase-producing strains from all Ambler classes, with cefiderocol demonstrating activity against the key MBLs VIM and NDM, as well as against clinically important serine-b-lactamases KPC, GES and OXA, and isolates cocarrying ESBLs and OSBLs. Consequently, cefiderocol represents a key addition to the limited armamentarium available for the treatment of infections caused by CR-and MDR-GN organisms and could be a particularly valuable and timely treatment option for when resistance is apparent but the mechanism is unknown.