Sigmoid E_max Modeling To Define the Fixed Concentration of Enmetazobactam for MIC Testing in Combination with Cefepime

ABSTRACT

The use of carbapenem antibiotics to treat infections caused by Enterobacterales expressing increasingly aggressive extended-spectrum β-lactamases (ESBLs) has contributed to the emergence of carbapenem resistance. Enmetazobactam is a novel ESBL inhibitor being developed in combination with cefepime as a carbapenem-sparing option for infections caused by ESBL-producing Enterobacterales. Cefepime-enmetazobactam checkerboard MIC profiles were obtained for a challenge panel of cefepime-resistant ESBL-producing clinical isolates of Klebsiella pneumoniae. Sigmoid maximum effect (E_max) modeling described cefepime MICs as a function of enmetazobactam concentration with no bias. A concentration of 8 μg/ml enmetazobactam proved sufficient to restore >95% of cefepime antibacterial activity in vitro against >95% of isolates tested. These results support a fixed concentration of 8 μg/ml of enmetazobactam for MIC testing.

TEXT

Resistance to carbapenem antibiotics by Gram-negative bacteria, notably Enterobacterales, Pseudomonas aeruginosa, and Acinetobacter baumannii, constitutes a significant global threat, and the development of novel therapeutic modalities addressing carbapenem resistance is recognized as a priority by the World Health Organization (1). The overuse of carbapenems to treat infections caused by extended-spectrum β-lactamase (ESBL)-producing Enterobacterales has undoubtedly contributed to the increase in carbapenem resistance (2). The magnitude of the health care burden posed by Enterobacterales resistant to third- and fourth-generation cephalosporins, mediated mainly by ESBLs, continues to rise. At least 10-fold more infections from Enterobacterales resistant to third-generation cephalosporins, and five times more deaths, occur in the United States and Europe alone compared to those from carbapenem-resistant Enterobacterales (3, 4). In the context of antimicrobial stewardship precepts, β-lactam–β-lactamase inhibitor (BL-BLI) combinations such as piperacillin-tazobactam or amoxicillin-clavulanate have been proposed as carbapenem-sparing options, though their efficacy has diminished against emergent ESBLs increasingly refractory to inhibition by BLIs (5, 6). There remains a critical need for novel carbapenem-sparing therapies targeting ESBL-producing Enterobacterales.

Therapy of infections attributable to ESBL-producing Klebsiella pneumoniae has proven particularly challenging. Although the non-carbapenem-containing BL-BLI ceftazidime-avibactam affords excellent coverage of ESBL-producing K. pneumoniae, this agent is recommended for carbapenem-resistant Enterobacterales expressing K. pneumoniae carbapenemases (KPCs). The activity of ceftolozane-tazobactam against ESBL-producing K. pneumoniae is reportedly poor, and this combination should be reserved for difficult-to-treat P. aeruginosa infections (7–9).

Enmetazobactam is a novel extended-spectrum β-lactamase inhibitor developed in combination with cefepime as an alternative to carbapenems for treating Gram-negative infections where ESBLs are likely to be the predominant β-lactam resistance mechanism (10, 11). Cefepime-enmetazobactam exhibited potent in vitro and in vivo activities against ESBL-producing isolates of Enterobacterales and demonstrated superiority over piperacillin-tazobactam in a phase 3 randomized controlled trial in patients with complicated urinary tract infections and/or acute pyelonephritis (7, 12–15).

In the present study, cefepime-enmetazobactam checkerboard MIC profiles were obtained for a challenge panel of cefepime-resistant ESBL-producing K. pneumoniae clinical isolates. Sigmoid maximum effect (E_max) modeling of the data elucidated the relationship between enmetazobactam concentration and restoration of cefepime activity and identified the optimal enmetazobactam concentration for in vitro testing.

RESULTS

Enmetazobactam exerts a concentration-dependent effect on the antibacterial activity of cefepime against ESBL-producing K. pneumoniae isolates.

Cefepime-enmetazobactam checkerboard MIC profiles were determined for the collection of cefepime-resistant (MIC ≥ 32 μg/ml) K. pneumoniae clinical isolates (n = 84) obtained from a surveillance study with confirmed genotypes for one or more ESBLs, with some coharboring the OXA-48 carbapenemase (15). Enmetazobactam exerted a concentration-dependent effect on growth inhibition by cefepime, evidenced by decreasing MIC₉₀ values for cefepime with increasing concentrations of fixed enmetazobactam (Table 1). Cefepime MIC₉₀ values were lower for ESBL-only producers than for isolates coproducing OXA-48, since enmetazobactam does not inhibit the weak hydrolysis of cefepime by OXA-48 (10). Epidemiological cutoff (ECOFF) values for the complete set of isolates for different enmetazobactam concentrations likewise showed that higher enmetazobactam concentrations decreased the ECOFF (Table 2). Enmetazobactam alone did not exhibit intrinsic antibacterial activity against the panel of isolates over the concentration range tested (i.e., the MIC₁₀₀ was >16 μg/ml) (data not shown).

TABLE 1.

Phenotypes and MIC₉₀s determined for the ESBL-producing isolates of K. pneumoniae used in this study

β-Lactamase genotype	No. of isolates	Cefepime MIC90 (μg/ml) at fixed enmetazobactam concn (μg/ml) of:
		0	0.25	0.5	1	2	4	8	16
CTX-M±SHVa	74	>32	>32	4	1	1	0.5	0.5	0.25
CTX-M±SHV+OXA-48b	10	>32	>32	>32	>32	32	16	8	4
Combined	84	>32	>32	32	32	8	2	1	1

^aCTX-M-15 was detected in 65/74 of isolates (87.8%), followed by CTX-M-14 and CTX-M-27 (2 isolates each) and by CTX-M-1, CTX-M-9, CTX-M-22, and CTX-M-55 (1 isolate each). Three isolates harbored an SHV-type ESBL, one together with CTX-M-15 and one together with CTX-M-27.

^bSix isolates coharbored CTX-M-15, 2 isolates coharbored CTX-M-14, and 2 isolates coharbored CTX-M-9 together with an SHV-type ESBL.

TABLE 2.

ECOFF values determined for the ESBL-producing isolates of K. pneumoniae used in this study

β-Lactamase genotype	No. of isolates	ECOFF (μg/ml) at fixed enmetazobactam concn (μg/ml) of:
		0	0.25	0.5	1	2	4	8	16
CTX-M±SHV±OXA-48	84	NDa	ND	4	1	0.5	0.5	0.25	0.063

^aND, not determinable.

Sigmoid E_max modeling adequately describes the relationship between enmetazobactam concentration and cefepime activity against ESBL-producing K. pneumoniae.

Sigmoid E_max models with a baseline parameter effect were computed for all K. pneumoniae isolates from their MIC checkerboard profiles; curves for three different isolates, representative of the patterns seen across all 84 isolates examined, are shown in Fig. 1. Predicted versus observed plots developed from all isolates indicate that the model describes the data with no bias at enmetazobactam concentrations of 0.5 μg/ml or higher (Fig. 2). For the complete set of isolates tested (n = 84), the median enmetazobactam concentration giving half of the asymptotic maximum effect (IC₅₀) was 0.02 μg/ml, the median of the Hill coefficient, γ (describing the steepness of the model at the IC₅₀), was 1.4, and the median of the baseline effect (the cefepime MIC at an infinite concentration of enmetazobactam) was 0.06 μg/ml (Table 3). The broad distribution of all parameter magnitudes may be associated with differences in type and combination of β-lactamases, though restricting the analysis to isolates expressing CTX-M-15 as the sole ESBL (n = 64) did not narrow the distribution of model parameters (Table 3 and Fig. 1a and b). The observed parameter differences may reflect differences between isolates in β-lactamase expression, porin functionality, efflux, or other factors and requires further study. Isolates coharboring OXA-48 and CTX-M-15 (n = 6) had an IC₅₀ distribution similar to that for the CTX-M-15 population. However, OXA-48 affected the baseline effect parameter, resulting in higher terminal cefepime MIC values at maximal enmetazobactam concentrations and lowered γ coefficients, leading to more gradual declines of the curves (Table 3 and Fig. 1C). These observations corroborate the potent activity of enmetazobactam against ESBLs and emphasize the importance of investigating MIC checkerboard profiles for challenge panels of target pathogens to quantify strain-to-strain variability.

FIG 1.

Sigmoid E_max with baseline-effect models for three representative checkerboard MIC data sets. The x axis shows the fixed concentration of enmetazobactam (linear scale), and the y axes show cefepime MICs (log₂ scale). Crosses depict observed values, and lines depict the model-predicted curves. (a) CTX-M-15-producing isolate: IC₅₀ = 0.11 μg/ml, γ = 8.8, MIC_Cemt(∞) = 0.03 μg/ml. (b) CTX-M-15-producing isolate: IC₅₀ = 1.0 μg/ml, γ = 6.3, MIC_Cemt(∞) = 0.28 μg/ml. (c) CTX-M-15+OXA-48-coproducing isolate: IC₅₀ = 0.9 μg/ml, γ = 2.1, MIC_Cemt(∞) = 2.0 μg/ml.

FIG 2.

Checkerboard MIC plots of observed and predicted cefepime MICs against ESBL-producing isolates of K. pneumoniae over a range of fixed enmetazobactam concentrations. x axes represent predicted MICs, and y axes represent observed MICs, both at log₂ scale. Data points are shown as crosses, and linear regressions are shown as solid lines. One isolate tested in the checkerboard assay failed to confirm the cefepime-only MIC at ≥32 μg/ml (top left).

TABLE 3.

Enmetazobactam sigmoid Emax model parameters determined for the ESBL-producing isolates of K. pneumoniae used in this study

Subset of isolates	IC50 (μg/ml)		γ		MICCemt(∞) (μg/ml)
	Range	Median	Range	Median	Range	Median
Complete set of isolates (n = 84)	1.0 × 10−8 to 3.9 × 10−8	0.02	0.35 to 13.6	1.4	0.03 to 8.0	0.06
CTX-M-15-only producers (n = 64)	1.0 × 10−8 to 3.4 × 10−8	0.01	0.44 to 13.5	1.4	0.03 to 1.0	0.06
CTX-M-15+OXA-48 producers (n = 6)	2.5 × 10−6 to 3.9 × 10−6	0.01	0.35 to 3.6	1.1	0.03 to 8.0	0.1

Enmetazobactam at a fixed concentration of 8 μg/ml restores the activity of cefepime against ESBL-producing isolates of K. pneumoniae.

The optimal concentration of enmetazobactam for in vitro testing should restore cefepime antibacterial activity against ESBL-producing isolates tested and account for the variability expected within a population of ESBL-producing pathogens. For these purposes, an enmetazobactam concentration providing >95% of the asymptotic maximum effect (IC₉₅) was defined as the lower limit of cefepime activity restoration, and the 95th percentile was chosen as the minimal threshold to account for variability among the population of pathogens.

Model parameters determined for each isolate were used to calculate enmetazobactam concentrations as a function of asymptotic maximum effects, and corresponding percentiles are plotted in Fig. 3. The 95th percentile of the IC₉₅ of enmetazobactam calculated across all K. pneumoniae isolates examined here was 4.2 μg/ml. This value increased to 5.2 μg/ml at the 95th percentile of the IC_97.5, and to 8.7 μg/ml at the 97.5th percentile of the IC_97.5. These results indicate that an enmetazobactam concentration of 8 μg/ml (corresponding to the next highest log₂ concentration exceeding 4.2 μg/ml) reliably restores cefepime activity in vitro against diverse ESBL-producing K. pneumoniae isolates.

FIG 3.

Restoration of cefepime activity by enmetazobactam against ESBL-producing isolates of K. pneumoniae. The x axis represents the fraction of the asymptotic maximum effect (IC_xx), and the y axis depicts the modeled concentration of enmetazobactam. Percentiles are indicated using different line styles, and the enmetazobactam concentration range required to restore 95% to 97.5% of cefepime activity for 95% to 97.5% of isolates is shaded in gray.

DISCUSSION

Direct-acting antibiotics exert antibacterial growth effects as a function of the antibiotic concentration, where the MIC is defined as the lowest antibiotic concentration that prevents visible bacterial growth. Since BLIs largely lack intrinsic antibacterial activity, their activity profile is assessed indirectly through protection from hydrolysis afforded by partner antibiotics, quantifiable in vitro as a reduction in the MIC. Selection of β-lactamase inhibitor concentrations for MIC testing has used different approaches (16, 17). While most recently approved BL-BLI combinations use the BLI part as a fixed concentration for MIC testing, few combinations use fixed BL-BLI ratios. A previous study employed E_max modeling to investigate the relationship between β-lactamase inhibitor concentration and restoration of β-lactam antibacterial activity against resistant pathogens (18). The present study applied this method to establish the relationship between cefepime MIC and enmetazobactam concentration for a challenge panel of ESBL-producing K. pneumoniae clinical isolates. A fixed enmetazobactam concentration of 8 μg/ml restored >95% of cefepime activity (as defined above) against >95% of isolates tested. The ECOFF distinguishes microorganisms without (wild type) and with phenotypically detectable acquired resistance mechanisms (non-wild type) to the agent in question (19). An ECOFF of 0.125 μg/ml was obtained for cefepime against a diverse collection of K. pneumoniae clinical isolates (n = 799) in a surveillance program (15). The fixed enmetazobactam concentration of 8 μg/ml also effectively restored the ECOFF of cefepime against the subset of ESBL-producing K. pneumoniae isolates in this panel. This demonstrates that the phenotypically detectable acquired resistance mechanism (ESBL production) was largely inactivated and that enmetazobactam at this concentration can shift the MIC distribution of these isolates back to wild type. Cefepime in combination with an enmetazobactam concentration fixed at 8 μg/ml has subsequently been used to establish broth microdilution MIC quality control ranges and guide disk mass selection for disk diffusion testing (7, 15, 20). In preclinical pharmacokinetic-pharmacodynamic (PK-PD) studies with ESBL-producing Enterobacterales isolates, time above a free enmetazobactam threshold concentration (fT > C_T) of 2 μg/ml was identified as the PK-PD index predictive of efficacy (12, 13, 21). By comparison, a fixed concentration of 4 μg/ml of either avibactam or tazobactam is used for ceftazidime-avibactam and ceftolozane-tazobactam MIC testing, whereas an avibactam fT > C_T of 1 μg/ml and tazobactam fT > C_T of 0.5 μg/ml are predictive of efficacy (22–25). The PK-PD relationships determined from preclinical and clinical data sets, together with the current findings, will help establish susceptibility breakpoints for cefepime-enmetazobactam at a fixed enmetazobactam concentration of 8 μg/ml (26, 27).

MATERIALS AND METHODS

Susceptibility testing.

Checkerboard analyses were performed by broth microdilution MIC testing using CLSI susceptibility testing standards (28). Enmetazobactam was dissolved at 5,120 μg/ml in water, whereas cefepime was dissolved at 5,120 μg/ml in 0.1 M phosphate buffer (pH 6.0). Twofold serial dilutions at 4-fold final concentrations were prepared, and 25-μl aliquots of cefepime and enmetazobactam were combined into wells of 96-well plates. Bacterial inocula were prepared at approximately 1 × 10⁶ CFU/ml by suspending cells in cation-adjusted Mueller-Hinton broth at 0.5 McFarland unit and diluting the cells 100-fold in N-[tris(hydroxymethyl)methyl]-2-aminoethanesulfonic acid buffer. Aliquots of cefepime-enmetazobactam in 96-well plates were inoculated with 50 μl of cell suspension, resulting in a final inoculum size of 5 × 10⁵ CFU/ml. Final cefepime concentrations ranged from 0.03 μg/ml to 32 μg/ml, whereas final enmetazobactam concentrations ranged from 0.25 to 16 μg/ml. Plates were incubated for 16 to 20 h at 37°C in ambient air, and MIC values were read at the first well with no visible growth.

Data analysis.

ECOFF values were determined as described previously (19) using the ECOFFinder_XL_2010_v2.1 file (http://www.eucast.org/mic_distributions_and_ecoffs/), reporting the ECOFF 99% rounded up to the next MIC. The MIC of cefepime as a function of enmetazobactam concentration (MIC_Cemt) was approximated for each tested isolate using a sigmoid E_max curve with a baseline effect parameter (equation 1), where C_emt is the enmetazobactam concentration variable, MIC_Cemt(0) is the MIC of cefepime in the absence of enmetazobactam, MIC_Cemt(∞) is the projected MIC of cefepime at an infinite amount of enmetazobactam (baseline effect parameter), IC₅₀ is the concentration giving half of the asymptotic maximum effect, and the Hill parameter γ indicates the steepness of the model at the IC₅₀ (29).

$${\text{MIC}}_{\text{Cemt}}={\text{MIC}}_{\text{Cemt(0)}}-\frac{({\text{MIC}}_{\text{Cemt(0)}}-{\text{MIC}}_{\text{Cemt(}\infty \text{)}})\times {\text{C}}_{\text{emt}}^{\text{γ}}}{{\text{C}}_{\text{emt}}^{\text{γ}}\hspace{0.17em}+\hspace{0.17em}{\text{IC}}_{\text{50}}^{\text{γ}}}$$ (1)

Each checkerboard data set was plotted in x/y format, and the sum of squared residuals (SSR) was computed and minimized using Microsoft Excel's Solver add-in to obtain the set of parameter values best describing the experimental data (30). For modeling purposes, cefepime checkerboard MIC values of >32 μg/ml were set to 64 μg/ml. Weighted SSRs (SSR_W) were used for minimization (equation 2), where ${\text{MIC}}_{\text{Cemt}}^{\text{Obs}}$ is the observed cefepime MIC, and ${\text{MIC}}_{\text{Cemt}}^{\text{Cpred}}$ is the predicted cefepime MIC as a function of enmetazobactam.

$${\text{SSR}}_{\text{W}}={\left(\frac{{\text{MIC}}_{\text{Cemt}}^{\text{Obs}}\hspace{0.17em}-\hspace{0.17em}{\text{MIC}}_{\text{Cemt}}^{\text{Pred}}}{{\text{MIC}}_{\text{Cemt}}^{\text{Obs}}}\right)}^2$$ (2)