Pharmacokinetic/Pharmacodynamic Evaluation of Aztreonam/Amoxicillin/Clavulanate Combination against New Delhi Metallo-β-Lactamase and Serine-β-Lactamase Co-Producing Escherichia coli and Klebsiella pneumoniae

This study aimed to examine specific niches and usage for the aztreonam/amoxicillin/clavulanate combination and to use population pharmacokinetic simulations of clinical dosing regimens to predict the impact of this combination on restricting mutant selection. The in vitro susceptibility of 19 New-Delhi metallo-β-lactamase (NDM)-producing clinical isolates to amoxicillin/clavulanate and aztreonam alone and in co-administration was determined based on the minimum inhibitory concentration (MIC) and mutant prevention concentration (MPC). The fractions of a 24-h duration that the free drug concentration was within the mutant selection window (fTMSW) and above the MPC (fT>MPC) in both plasma and epithelial lining fluid were determined from simulations of 10,000 subject profiles based on regimens by renal function categories. This combination reduced the MIC of aztreonam and amoxicillin/clavulanate to values below their clinical breakpoint in 7/9 K. pneumoniae and 8/9 E. coli, depending on the β-lactamase genes detected in the isolate. In the majority of the tested isolates, the combination resulted in fT>MPC > 90% and fTMSW < 10% for both aztreonam and amoxicillin/clavulanate. Clinical dosing regimens of aztreonam and amoxicillin/clavulanate were sufficient to provide mutant restriction coverage for MPC and MIC ≤ 4 mg/L. This combination has limited coverage against NDM- and extended-spectrum β-lactamase co-producing E. coli and K. pneumoniae and is not effective against isolates carrying plasmid-mediated AmpC and KPC-2. This study offers a potential scope and limitations as to where the aztreonam/amoxicillin/clavulanate combination may succeed or fail.


Introduction
Klebsiella pneumoniae and Escherichia coli are common multidrug-resistant (MDR) Gramnegative bacilli associated with hospital-acquired and ventilator-associated pneumonia (HAP/VAP) in critically ill patients [1,2]. Infections due to MDR bacteria are responsible for at least 700,000 deaths annually [3]. Even though polymyxins are emerging as the antibiotics of last resort to treat MDR infections including HAP/VAP, this class of drug is associated with toxicity and poor therapeutic outcomes, including a rapid rise of heteroresistance in monotherapy [4]. The combination of polymyxins with other antibiotics in the treatment of

Bacterial Isolates and Susceptibility Testing
All clinical isolates of K. pneumoniae and E. coli were provided by the affiliated hospital of Qingdao University. Next-generation sequencing was used to determine the types of β-lactamase gene for each strain, as previously described [11]. Eighteen clinical isolates that carry bla NDM were selected for further testing. E. coli ATCC 25922 and ATCC 35218, K. pneumoniae ATCC 700603 and C. braakii HDC438 were used as quality control strains.
A checkerboard method using broth microdilution was used to determine the susceptibility profile of each clinical isolate to aztreonam alone, amoxicillin/clavulanate, and aztreonam/amoxicillin/clavulanate. MIC determination was conducted in triplicate and modal MIC was used to characterize the susceptibility of each isolate to the antibiotics.
Four K. pneumoniae and three E. coli strains were selected to determine the antimicrobial's mutant prevention concentration (MPC) against each strain. A series of Mueller-Hinton agar plates containing an antimicrobial agent or combination at 1×, 2×, 4×, 8×, 16× and 32 × MIC were prepared. A high-density inoculum of ≥ 10 10 cfu/mL was cultured and confirmed by absorbance at 660 nm OD (Milton Roy Spectronic 21D spectrophotometer) to ensure the emergence of more resistant bacteria. The prepared Mueller-Hinton agar was inoculated with a high inoculum density, evenly coated, and cultured at 37 • C for 48 h. The Mueller-Hinton agar was observed and recorded every 24 h. The lowest antimicrobial concentration that completely prevented bacterial growth was determined as the MPC.

Time-Kill Kinetic Assay
In order to study the effects of amoxicillin/clavulanate and aztreonam alone or in combination on the in vitro time-course of bacterial growth and kill dynamics, one strain of each bacteria type was selected to run the time-kill kinetic experiment at MIC and two-fold MIC. There were four treatment groups for each strain: control, amoxicillin/clavulanate, aztreonam, and aztreonam plus amoxicillin/clavulanate. K. pneumoniae LW5 and E. coli 37 were cultured in Mueller-Hinton broth at 37 • C for an hour to ensure that logarithmic growth was achieved. The initial inoculum size was approximately 5 × 10 5 cfu/mL. There was no drug added to the control group. For K. pneumoniae LW5, the drug concentrations at MIC were 64/32 mg/L amoxicillin/clavulanate, 32 mg/L aztreonam, 1/0.5 mg/L amoxicillin/clavulanate plus 0.25 mg/L aztreonam. For the E. coli isolate 37, the MICs were 128/64 mg/L amoxicillin/clavulanate, 32 mg/L aztreonam, 1/0.5 mg/L amoxicillin/clavulanate plus 0.25 mg/L aztreonam. Twice the previous drug concentrations were added in the two-fold MIC treatment arm. The culture flask was incubated at 37 • C and constant shaking at 120 rpm. At 0, 2, 4, 6 and 8 h post-drug administration, 20 µL of the liquid was sampled and diluted in 10-fold increments in fresh sterile saline at room temperature and spread 100 µL volumes from each dilution in 10 spots on agar plates. Colonies were counted following incubation at 37 • C for 18 to 24 h. All time-kill experiments were performed in triplicate on separate occasions. The average concentration of bacteria at each time point was calculated and recorded.

Population Pharmacokinetic Simulations
The dosing regimens of aztreonam and amoxicillin/clavulanate used in the simulation are listed in Table 1. The aztreonam high dose did not exceed 6 g daily doses, whereas the amoxicillin/clavulanate dosing regimens were based on that recommended in the package insert. A prolonged 3 h infusion was used to achieve higher f T >MIC . The same dosing interval was used for both aztreonam and amoxicillin/clavulanate, assuming that these antimicrobials are administered simultaneously.
A detailed description of the population pharmacokinetic models and setup for simulations are provided in the Supplementary Materials. The demographics for the virtual patient population consisted of males and females in equal proportion with various degrees of renal impairment. Creatinine clearance (CrCL) was simulated using a uniform distribution ranging from 10 to 150 mL/min and normalized to 1.73 m 2 body surface area. The lower limit of CrCL was set at 10 mL/min to exclude patients who undergo hemodialysis, which can change the clearance of drugs. The steady-state concentrationtime profiles of aztreonam, amoxicillin and clavulanate were simulated from the population pharmacokinetic models described below.
The model for aztreonam in patients with renal impairment and complicated intraabdominal infection is a two-compartment model [25,26] wherein its clearance (CL in L/h) is dependent on CrCL such that CL = 4.73 * (CrCL/100) 0.43 with 24.1% coefficient of variation (CV%). Central volume (V C in L) is dependent on body weight (WT) such that V C = 7.43 * (WT/70) 1.99 with 50.9% CV. Peripheral volume (V P ) is 6.44 L (27.7% CV) and intercompartmental clearance (Q) is 9.26 L/h. The plasma protein binding of aztreonam is 56%.
The drug concentration in the lung was approximated based on epithelial lining fluid (ELF) penetration rates. These rates were 55% and 30% for aztreonam and amoxicillin/clavulanate, respectively [28,29]. No mucin binding was assumed for drug exposures in the ELF.

Pharmacodynamic Metrics to Quantify Suppression of Emergent Mutant
The restriction of the resistant mutant selection was measured by two PD parameters, f T MSW and f T >MPC , for the fraction of time over the 24 hours wherein the free drug concentration was within the MSW and the fraction of time over the same period wherein the free drug concentration exceeded the MPC, respectively. Limiting the selection of resistant mutants can be achieved by maintaining drug concentrations above MPC during therapy and decreasing the time that the bacteria spent within MSW [17,30,31].
Then, 10,000 concentration-time profiles at steady-state over a 24 h period were simulated for each drug to compute these two PD parameters. Both f T >MIC and f T >MPC were determined for each profile. f T MSW was subsequently computed as the difference between f T >MIC and f T >MPC . If f T >MPC is 0, f T MSW is not computed. For each of the selected isolates, descriptive statistics of the PD parameters were listed.

Target Pharmacodynamic Indices and Probability of Target Attainment
The previous two PD indices associated with the inhibition of drug resistance are dependent on establishing sufficient probability of target attainment (PTA) at the MIC of the combination therapy. For aztreonam, the target PD index associated with 2 log kill in E. coli using a single compartment dilutional model was 60% f T >MIC [32]. The predefined PD target for amoxicillin/clavulanate was 50% f T >MIC [27]. The ability of each dosing regimen to achieve its predefined PD target was determined by PTA at incremental MIC. Dosing simulations for clavulanic acid were not undertaken given that its PD target is not established [27].

Software
All PK simulations and PD analyses were carried out in R (4.1.2) using the RxODE package and user-defined functions, respectively.  Table 2). The control E. coli ATCC 25922 was susceptible to both aztreonam and amoxicillin/clavulanate, whereas the control E. coli ATCC 35218 strain that contained TEM-1 was susceptible to aztreonam, but not amoxicillin/clavulanate. The control K. pneumoniae strain was resistant to both aztreonam and amoxicillin/clavulanate. The other positive control C. braakii HDC438 tested positive for KPC-2 and plasmid-mediated AmpC (CMY-79) genes. The MIC values of aztreonam and amoxicillin/clavulanate alone were all greater than or at least 64 mg/L, indicating that these isolates were very resistant to both aztreonam and amoxicillin/clavulanate. The aztreonam/amoxicillin/clavulanate combination has a significant synergistic effect on the control K. pneumoniae strain (FICI = 0.0937). This combination reduced the MIC of aztreonam and amoxicillin/clavulanate to values below their clinical breakpoint in 7/9 K. pneumoniae and 8/9 E. coli [24]. In three isolates, the combination did not significantly reduce the MICs compared to that of monotherapy.

In Vitro Antimicrobial Susceptibility
The MPC values of aztreonam and amoxicillin/clavulanate alone or in combination against three K. pneumoniae and four E. coli isolates are shown in Table 3. The MPC values of aztreonam alone were all > 256 mg/L, whereas the MPCs of amoxicillin/clavulanate alone were ≥128 mg/L. The combination consisting of aztreonam/amoxicillin/clavulanate significantly reduced the MPC values to ≤8 mg/L for these seven isolates. The reduction in MPC was at least 32-fold.

Time-Kill Kinetics
The results of the time-kill kinetics are shown in Figure 1. Amoxicillin/clavulanate as monotherapy at 128/64 and 256/128 mg/L did not reduce the bacterial density of the K. pneumoniae LW5 isolate, whereas aztreonam treatment at 32 and 64 mg/L resulted in an initial kill, but rebounded at 4 and 8 h, respectively. For E. coli 37, the bacterial burden in the amoxicillin/clavulanate treatment group continued to rise and was not effectively suppressed at 128/64 mg/L, whereas bacteriostasis was observed at 256/128 mg/L. Bacteriostasis was observed at 8h when treated with aztreonam alone at MIC and at two-fold MIC. We did not evaluate the extremely high concentration of amoxicillin/clavulanate or aztreonam required to exert a bactericidal effect against these two isolates. Previous experience indicated that aztreonam MIC against extremely resistant E. coli harboring NDM-1 could be as high as 2048 mg/L [33].
Treatment with aztreonam/amoxicillin/clavulanate showed that a 2 log 10 -kill was achieved at around 4 h. These isolates did not exhibit a growth rebound at 8 h when treated with aztreonam/amoxicillin/clavulanate, except for K. pneumoniae LW5 exposed to amoxicillin/clavulanate plus aztreonam at MIC. Aztreonam/amoxicillin/clavulanate resulted in a bactericidal effect at 8 h even though the concentrations of the combination are much lower than aztreonam or amoxicillin/clavulanate alone.

Pharmacodynamic Analysis of Resistant Mutant Selection in Blood
The clinical dosing regimens of amoxicillin/clavulanate and aztreonam by renal function categories listed in Table 1 were used to evaluate the effects of combination therapy on the PD parameters associated with restricting the emergence of resistant mutants. The doses were expected to achieve ≥ 90% PTA for 60% f T >MIC of ≤8 mg/L for aztreonam and 50% f T >MIC of ≤ 8 mg/L for amoxicillin/clavulanate in the plasma (Figure 2). A PTA of approximately 80% at 8 mg/L MIC was achieved in patients with creatinine clearance > 50 mL/min using a dosing regimen 1/0.2 g amoxicillin/clavulanate q6h (Figure 2).
A hypothesis was tested such that the aztreonam/amoxicillin/clavulanate combination will reduce the f T MSW and increase f T >MPC , compared to the corresponding monotherapy. For this hypothesis to work, sufficient PTA (i.e., ≥90%) should be achieved at the isolate combination MICs [34]. These two PD parameters were only determinable in the combination setting because the MIC and MPC in the monotherapy were too high. As shown in Table 4, aztreonam f T >MPC was, on average, ≥90%, whereas f T MSW was <10% in the plasma. Two E. coli isolates, 13 and 48, had higher combination MPC and MIC than other isolates, resulting in lower f T >MPC and higher f T MSW .
In the combination therapy, the overall f T >MPC of amoxicillin/clavulanate was also > 90% except for the two E. coli isolates 13 and 48, which had the highest MPC of 8 mg/L. The f T MSW were very limited, indicating that the opportunity for the selection of resistant mutants is restricted in the combination therapy.

Pharmacodynamic Analysis of Resistant Mutant Selection in ELF
A 40% penetration [28] into the ELF was assumed for aztreonam; the two PD indices in the ELF were determined based on a lower drug exposure in the ELF. Apart from E. coli isolates 13 and 48, the f T MSW was close to 0 for the other isolates (Table 5). For these two isolates, the f T MSW ranged from 4 to 10% of the dosing interval. Aztreonam f T >MPC was, on average, ≥90% in the ELF. We assumed an ELF penetration rate of 30% for amoxicillin/clavulanate [29,35]. The f T MSW was <15% and f T >MPC was >60% for all isolates, except E. coli 13 and 48 (Table 5). For these two isolates, the f T MSW ranged from 10 to >40% of the dosing interval. The combination significantly reduced the likelihood for resistant mutants to emerge in the ELF by restricting the time that the bacteria spent in MSW.    Table 1. Probability of target attainment values were computed based on steady-state drug concentrations in the blood.  Table 1. Probability of target attainment values were computed based on steady-state drug concentrations in the blood.  In the computation of PTA for the dosing regimens of aztreonam and amoxicillin/ clavulanate in the ELF, we assumed that aztreonam and amoxicillin/clavulanate have very little binding to mucin. Figure 3 shows that aztreonam dosing regimens were expected to achieve ≥ 90% PTA for 60% f T >MIC of ≤8 mg/L in the ELF; this breakpoint is the same as that in the plasma. Because amoxicillin and clavulanate have a lower ELF penetration rate, the dosing regimens of amoxicillin/clavulanate were expected to achieve ≥ 90% PTA for 50% f T >MIC of ≤4 mg/L in patients with creatinine clearance < 50 mL/min. For the dosing regimen in patients with creatinine clearance > 50 mL/min, a ≥90% PTA for 50% f T >MIC was expected to be achieved at MIC of ≤2 mg/L. In the computation of PTA for the dosing regimens of aztreonam and amoxicillin/clavulanate in the ELF, we assumed that aztreonam and amoxicillin/clavulanate have very little binding to mucin. Figure 3 shows that aztreonam dosing regimens were expected to achieve ≥ 90% PTA for 60% fT>MIC of ≤8 mg/L in the ELF; this breakpoint is the same as that in the plasma. Because amoxicillin and clavulanate have a lower ELF penetration rate, the dosing regimens of amoxicillin/clavulanate were expected to achieve ≥ 90% PTA for 50% fT>MIC of ≤4 mg/L in patients with creatinine clearance < 50 mL/min. For the dosing regimen in patients with creatinine clearance > 50 mL/min, a ≥90% PTA for 50% fT>MIC was expected to be achieved at MIC of ≤2 mg/L.

Discussion
The acyl enzyme and β-lactamase inhibitor complex that is formed from nucleophilic attack by clavulanic acid is more stable than β-lactams since clavulanic acid is a suicidal inhibitor, whereas β-lactam antibiotics are hydrolyzed by the enzyme. In contrast, avibactam does not interact via an acyl group, but rather by a carbamyl moiety, resulting in a high potency against various β-lactamases including AmpC, P99, KPC-2, TEM and SHV wherein clavulanic acid potencies against these enzymes are limited [20,36]. The majority of the ESBL genes in our collection of clinical isolates are plasmid-encoded and can be acquired by horizontal gene transfer; these isolates are likely clonally-related, given that they were collected from the same hospital and within a period of 5 years. Most of the plasmidencoded β-lactamase genes were for SHV, TEM and CTM-M. CMY-42, encoded by the plasmid-mediated AmpC gene [37], may be responsible for the resistance of E. coli isolate 9 to aztreonam/amoxicillin/clavulanate combination. The C. braakii isolate expressing KPC-2 and CMY-79 enzymes was resistant to the triple antibiotic combination. These two isolates have the ESBL gene wherein clavulanic acid has limited potency; the IC 50 of clavulanic acid against AmpC and KPC-2 enzymes was >100 nM (Table 6) [20]. As for the K. pneumoniae LW1 and LW15 isolates, the identified ESBLs were supposed to be inhibited by clavulanic acid. Resistance to the triple combination in these two isolates could be due to the hyperproduction of these plasmid-encoded β-lactamases [38,39]; this mechanism was not evaluated in our study. In our previous study, the aztreonam/avibactam combination was potent against these four isolates [11], indicating that the deficiency in the OmpF and/or OmpC porins is unlikely the mechanism for their resistance to aztreonam/amoxicillin/clavulanate combination.
For other isolates, the effect of the aztreonam/amoxicillin/clavulanate combination was comparable to the combination of aztreonam/ceftazidime/avibactam in terms of antimicrobial susceptibility [11]. The results suggest that this combination can be used as a limited alternative to aztreonam/ceftazidime/avibactam. Emeraud also measured the in vitro susceptibility of three combinations of aztreonam with ceftazidime/avibactam, ceftazidime/tazobactam, and amoxicillin/clavulanate to multi-drug resistant bacteria, and their results showed that aztreonam/amoxicillin/clavulanate was similar to aztreonam/ceftazidime/avibactam against some bacterial strains and also more cost-effective [40]. The combination of aztreonam and clavulanate may also be promising against MLB-producing Enterobacterales, but we did not evaluate this combination given that this combination or clavulanate alone is not available commercially. We evaluated aztreonam/sulbactam combination against these isolates, but did not observe synergistic activities for this combination (data not shown).
Amoxicillin is the preferred partnering drug for clavulanate over ticarcillin, since clavulanate induces the expression of AmpC and antagonizes the antibacterial activity of ticarcillin in Pseudomonas aeruginosa [41]. The aztreonam/amoxicillin/clavulanate combination, however, is not useful against P. aeruginosa (data not shown). Chromosomal AmpC hyperproduction in P. aeruginosa, being the main mechanism driving β-lactam resistance, is not inhibited by clavulanic acid. In contrast, aztreonam/avibactam could inhibit AmpC and was active against AmpC-producing P. aeruginosa [33,42]. The coverage of the aztreonam/amoxicillin/clavulanate combination is limited to E. coli and K. pneumoniae, depending on the types of β-lactamases expressed in the pathogen, whereas aztreonam/ceftazidime/avibactam may provide broader spectrum coverage of β-lactamases and pathogens.
The current study differentiates from other studies evaluating synergistic activities of aztreonam/amoxicillin/clavulanate combination against NDM-and ESBL-co-producing Enterobacterales in that we examined the notion that effective combination should not only produce synergism, but also restrict the emergence of resistant mutants by reducing the MSW and maximizing f T MSW . Xu et al. conducted checkerboard microdilution and time-kill assays on 56 strains of carbapenem-resistant Gram-negative pathogens. The combination of amoxicillin/clavulanate and aztreonam was found to have a synergistic effect of up to 97% of E. coli [23]. Ract et al. compared the synergistic activity of combinations consisting of aztreonam/amoxicillin/clavulanate or piperacillin/tazobactam in 22 strains of K. pneumoniae and 14 strains of E. coli. They concluded that the combination of aztreonam/amoxicillin/clavulanate was more effective than that with piperacillin/tazobactam, and this combination could be an attractive unconventional treatment [43]. These existing studies have only evaluated synergistic activities using MIC and compared them with clinical breakpoints.
To infer the clinical effects of this combination, we conducted simulations using tested and recommended clinical dosing regimens. For aztreonam, we chose the dosing regimens from the REJUVENATE clinical trial rather than the recommended regimens in the package insert; the safety of aztreonam for these dosing regimens has been demonstrated in the trial [26]. For amoxicillin/clavulanate, we chose the clinically recommended dosing regimens. In this in vitro and simulation study, the PD parameters, f T MSW and f T >MPC , were calculated to extrapolate the clinical efficacy of aztreonam/amoxicillin/clavulanate combination and its capacity to restrict the emergence of drug-resistant mutation against bacteria already harboring NDM. In select isolates wherein combination therapy could bring their MICs to their breakpoints, the f T >MPC values are greater than 90% and all of the f T MSW values are less than 6% in the blood, except for amoxicillin/clavulanate in patients with renal clearance >50 to 150 mL/min. A larger dose of amoxicillin/clavulanate can result in better suppression of further resistance development so that f T >MPC is sufficiently large and the f T MSW is reduced to <10% of the 24 h duration, especially in the case wherein MPC is 8 mg/L and MIC is 4 mg/L for amoxicillin/clavulanate. Much of the literature suggested that the clinically recommended dose of amoxicillin/clavulanate is too conservative for patients with high renal clearance [27,44].

Conclusions
This combination could play an important, but limited, niche in the clinical treatment of multidrug-resistant E. coliand K. pneumonia carrying MBLs and other β-lactamases. Our study offers a potential scope and limitations as to where the aztreonam/amoxicillin/ clavulanate combination may succeed or fail. Funding: This work was supported by grants from Shandong Provincial Natural Science Foundation (ZR2019BC025).