Emergence of Resistance in Klebsiella aerogenes to Piperacillin-Tazobactam and Ceftriaxone

We examined the effects of piperacillin-tazobactam (TZP) concentration and bacterial inoculum on in vitro killing and the emergence of resistance in Klebsiella aerogenes. The MICs for 15 clinical respiratory isolates were determined by broth microdilution for TZP and by Etest for ceftriaxone (CRO) and cefepime (FEP). The presence of resistance in TZP-susceptible isolates (n = 10) was determined by serial passes over increasing concentrations of TZP-containing and CRO-containing agar plates.

ABSTRACT

We examined the effects of piperacillin-tazobactam (TZP) concentration and bacterial inoculum on in vitro killing and the emergence of resistance in Klebsiella aerogenes. The MICs for 15 clinical respiratory isolates were determined by broth microdilution for TZP and by Etest for ceftriaxone (CRO) and cefepime (FEP). The presence of resistance in TZP-susceptible isolates (n = 10) was determined by serial passes over increasing concentrations of TZP-containing and CRO-containing agar plates. Isolates with growth on TZP 16/4-μg/ml and CRO 8-μg/ml plates (n = 5) were tested in high-inoculum (HI; 7.0 log₁₀ CFU/ml) and low-inoculum (LI; 5.0 log₁₀ CFU/ml) time-kill studies. Antibiotic concentrations were selected to approximate TZP 3.375 g every 8 h (q8h) via a 4-h prolonged-infusion free peak concentration (40 μg/ml [TZP40]), peak epithelial lining fluid (ELF) concentrations, and average AUC_0–24 values for TZP (20 μg/ml [TZP20] and 10 μg/ml [TZP10], respectively), the ELF FEP concentration (14 μg/ml), and the average AUC_0–24 CRO concentration (6 μg/ml). For HI, FEP exposure significantly reduced 24-h inocula against all comparators (P ≤ 0.05) with a reduction of 4.93 ± 0.64 log₁₀ CFU/ml. Exposure to TZP40, TZP20, and TZP10 reduced inocula by 0.81 ± 0.43, 0.21 ± 0.18, and 0.05 ± 0.16 log₁₀ CFU/ml, respectively. CRO-exposed isolates demonstrated an increase of 0.42 ± 0.39 log₁₀ CFU/ml compared to the starting inocula, with four of five CRO-exposed isolates demonstrating TZP-nonsusceptibility. At LI after 24 h of exposure to TZP20 and TZP10, the starting inoculum decreased by averages of 2.24 ± 1.98 and 2.91 ± 0.50 log₁₀ CFU/ml, respectively. TZP demonstrated significant inoculum-dependent killing, warranting dose optimization studies.

INTRODUCTION

AmpC β-lactamases are clinically important hydrolyzing enzymes produced by certain Enterobacteriaceae that confer resistance to many β-lactams, including third-generation cephalosporins (e.g., ceftriaxone [CRO]) (1). Some clinically relevant members of the order Enterobacterales (e.g., Klebsiella [formerly Enterobacter] aerogenes) are particularly difficult to manage because they harbor AmpC β-lactamase on chromosomes, which may not be detected via routine susceptibility testing. High-level expression of AmpC β-lactamases can occur by either induction or, more often, by selection of hyperproducing subpopulations (i.e., derepressed mutants) in the presence of certain β-lactams. Concern for resistance via selection with CRO has led to the use of carbapenem antibiotics for AmpC producers due to the structural stability of carbapenems against AmpC β-lactamases and clinical efficacy (2–5). There is great interest in carbapenem-sparing options with the continuing threat of antibiotic resistance. Cefepime (FEP), a zwitterionic cephalosporin with enhanced structural stability against AmpC β-lactamases, has been suggested as a carbapenem-sparing agent due to its low potential for induction and clinical efficacy (6–9). Piperacillin-tazobactam (TZP) is a ureidopenicillin/β-lactamase inhibitor combination that has also been suggested as a carbapenem-sparing option for AmpC-producing Enterobacteriaceae (10, 11).

While TZP is considered a weak inducer of AmpC (12–14), piperacillin does not maintain stability in the presence of high AmpC production, and tazobactam does not reliably inhibit these β-lactamases. In addition, TZP MICs or the serum concentration may play a role in the selection of resistant subpopulations of K. aerogenes. This is particularly concerning for infections such as pneumonia, where TZP concentrations in lung epithelial lining fluid (ELF) are only 40 to 50% compared to serum concentrations (15, 16), and yet the high bacterial burden increases the opportunity for emergence of bacterial strains via mutational derepression (17). The pharmacodynamics of β-lactams have been well elucidated with clinical success described as time-dependent (%T > MIC) and optimal bactericidal activity at concentrations 3 to 4 times the MIC (18). Current strategies for dosing β-lactams such as TZP focus on prolonging the %T > MIC. Whether such dosing strategies subject isolates to a β-lactam “mutant selection window” is of major concern, especially in high-inoculum infections (19, 20). The objective of this study is to determine the effect of various TZP concentrations and inoculum size on the emergence of resistant subpopulations among clinical respiratory isolates of K. aerogenes.

RESULTS

Susceptibility testing.

Of the 15 K. aerogenes isolates, 10 were susceptible to TZP (MIC ≤ 16/4 μg/ml based on the piperacillin component) by broth microdilution, with MICs ranging from 2/4 to 64/4 μg/ml, 9 were susceptible to CRO (MIC ≤ 1 μg/ml), and 13 were susceptible to FEP (MIC ≤ 2 μg/ml) both by Etest methodology (Table 1) (21). The MICs obtained by manual broth microdilution (BMD) demonstrated essential agreement for 9 of the 12 initial isolates for which essential agreement could be determined. Three isolates had a BD Phoenix TZP MIC > 64/4 μg/ml, and one had a TZP MIC of 64/4 μg/ml, so the essential agreement with manual BMD is unknown.

TABLE 1.

MICs determined by BD Phoenix and laboratory methods for TZP, CRO, and FEP

Isolatea	MIC (μg/ml)b
	TZP		CRO		FEP
	BD Phoenix	BMD	BD Phoenix	Etest	BD Phoenix	Etest
N576	>64	64	>32	>16	≤1	0.5
N577*	4	2	≤0.5	0.094	≤1	0.047
N578	>64	32	>32	16	>16	4
N579	>64	64	>32	>16	≤1	0.25
N582	4	4	≤1	0.125	≤1	0.0625
N583	64	32	32	16	≤1	0.19
N604*	4	4	≤1	0.19	≤1	0.047
N607*	4	4	≤1	0.125	≤1	0.047
N609*	8	8	≤1	0.125	≤1	0.047
N611	4	8	≤1	0.25	≤1	0.047
N612	>64	64	>32	>16	≤1	0.25
N614	32	4	8	4	≤1	4
N615	16	2	≤1	0.094	≤1	0.023
N616*	4	2	≤1	0.125	≤1	0.047
N617	4	8	≤1	0.125	≤1	0.047
S/NS (%)	67/33	60/40	67/33	67/33	87/13	80/20

^a*, Strains selected for time-kill studies. S/NS, susceptible/nonsusceptible defined according to 2018 CLSI breakpoints (TZP S ≤ 16/4, NS ≥ 32/4; CRO S ≤ 1, NS ≥ 2; FEP S ≤ 2, SDD 4–8, R ≥ 16).

^bMICs were determined by broth microdilution for piperacillin-tazobactam (TZP) and by Etest for ceftriaxone (CRO) and cefepime (FEP).

Resistance selection using sequential antibiotic-supplemented agar plates.

Of the 10 TZP-susceptible isolates, 5 demonstrated the ability to grow on the highest concentration of both CRO (8 μg/ml) and TZP (16/4 μg/ml) plates. Only one isolate, N615, failed to demonstrate the ability to grow on the CRO 1-μg/ml plate. The five isolates that were able to grow on the highest concentration of TZP and CRO plates were selected for use in time-kill studies.

Time-kill studies.

The mean starting inoculum was 7.86 ± 0.13 log₁₀ CFU/ml in the high-inoculum time-kill study. Bactericidal activity was observed with FEP with colony counts reductions averaging 4.93 ± 0.64 log₁₀ CFU/ml from the starting inoculum compared to the TZP and CRO groups. Minimal reductions in starting inoculum were also noted in the TZP groups, with average reductions of 0.81 ± 0.43 (P < 0.05), 0.21 ± 0.18 (P = 0.12), and 0.05 ± 0.16 (P = 0.09) log₁₀ CFU/ml for the 40 (TZP40), 20 (TZP20), and 10 μg/ml (TZP10) TZP concentrations, respectively. Four of the five isolates exposed to CRO demonstrated an average increase from the starting inoculum of 0.42 ± 0.39 log₁₀ CFU/ml. The fifth strain demonstrated a modest reduction of 0.18 ± 0.19 log₁₀ CFU/ml. In addition, each isolate demonstrated an initial reduction in starting inoculum for the first 8 h of CRO exposure, followed by regrowth, as depicted by the time-kill study data presented in Fig. 1. Changes in the bacterial inoculum when exposed to the antibiotics are presented in Table 2.

FIG 1.

High-inoculum time-kill for N577.

TABLE 2.

Mean 24-h changes in high-inoculum isolates

Isolate	Mean 24-h change in inoculum (log10 CFU/ml ± SD)a
	GC	TZP40	TZP20	TZP10	CRO	FEP
N577	1.11 ± 0.00	−1.08 ± 0.20	−0.31 ± 0.03	0.10 ± 0.14	0.64 ± 0.06	−4.62 ± 0.99
N604	1.20 ± 0.01	−1.30 ± 0.44	−0.37 ± 0.25	−0.18 ± 0.27	0.70 ± 0.20	−4.96 ± 0.16
N607	1.09 ± 0.04	−0.37 ± 0.22	−0.17 ± 0.08	−0.05 ± 0.03	0.20 ± 0.04	−5.59 ± 0.36
N609	1.06 ± 0.16	−0.61 ± 0.27	0.00 ± 0.22	−0.13 ± 0.06	−0.18 ± 0.19	−4.21 ± 0.08
N616	1.15 ± 0.22	−0.67 ± 0.38	−0.19 ± 0.00	0.01 ± 0.18	0.75 ± 0.07	−5.29 ± 0.48

^aGC, growth control; TZP40, 40 μg/ml piperacillin-tazobactam; TZP20, 20 μg/ml piperacillin-tazobactam; TZP10, 10 μg/ml piperacillin-tazobactam; CRO, ceftriaxone; FEP, cefepime.

MICs were determined at the end of the 24-h high-inoculum time-kill studies. MIC determinations were performed twice for each isolate-antibiotic combination and time-kill studies were conducted in duplicate. CRO exposure resulted in TZP nonsusceptibility in four of the five isolates. The single TZP-susceptible isolate demonstrated a two-dilution increase in TZP MIC from baseline. Of the approximated peak TZP ELF concentrations (i.e., TZP10 and TZP20), all five of the TZP20-exposed isolates remained TZP susceptible, with only N577 demonstrating an increase in TZP MIC by >1 dilution from baseline. Three of the five TZP10-exposed isolates failed to demonstrate within-run categorical agreements in postexposure TZP MIC despite multiple repeats being conducted. For isolates where MIC agreement did not exist between replicates, both MICs are reported with the corresponding alpha-numeric identifier (e.g., N604A and N604B) (Table 3).

TABLE 3.

Changes in TZP MICs before and after high-inoculum time-kill

Isolate	Antibiotic	MIC (μg/ml)a
		Initial	TZP20 exposure	TZP10 exposure	CRO exposure
N577	TZP	2	16	4	32
	CRO	0.0625	0.125	0.125	12
N604	TZP	4	4	A-32; B-4	32
	CRO	0.19	0.19	A-12; B-0.19	8
N607	TZP	4	8	8	16
	CRO	0.125	0.19	8	24
N609	TZP	8	4	A-4; B-64	64
	CRO	0.125	0.125	A-0.125; B-12	16
N616	TZP	2	2	A-2; B-32	32
	CRO	0.125	0.125	A-0.19; B-8	48

^aA and B postrun MICs are shown for isolates not having within-run categorical agreement; higher postrun MICs are shown for isolates with essential agreement.

The mean starting inoculum for the low-inoculum time-kill study was 5.17 ± 0.52 log₁₀ CFU/ml. At a low inoculum, TZP10 and TZP20 significantly reduced the starting inoculum for all five isolates and four of the five isolates, respectively (P < 0.05). The enhanced bacterial killing is shown in Fig. 2 for the representative strain N577. At lower inoculum, CRO significantly inhibited bacterial growth compared to the starting inoculum, as well as the growth controls for all five isolates (Table 4).

FIG 2.

Low-inoculum time-kill for N577.

TABLE 4.

Mean 24-h changes in high-inoculum isolates

Isolate	Mean 24-h change in inoculum (log10 CFU/ml ± SD)a
	GC	TZP20	TZP10	CRO
N577	3.93 ± 0.14	−2.86 ± 0.04	−2.77 ± 0.01	−2.85 ± 0.10
N604	3.94 ± 0.14	−2.75 ± 0.07	−2.68 ± 0.09	−2.75 ± 0.02
N607	3.57 ± 0.81	1.13 ± 2.39	−2.66 ± 0.20	−2.05 ± 3.21
N609	3.21 ± 0.38	−3.37 ± 0.61	−3.06 ± 1.23	−2.26 ± 1.93
N616	3.23 ± 0.23	−3.35 ± 0.40	−3.36 ± 0.09	−1.12 ± 2.13

^aGC, growth control; TZP40, 40 μg/ml piperacillin-tazobactam; TZP20, 20 μg/ml piperacillin-tazobactam; TZP10, 10 μg/ml piperacillin-tazobactam; CRO, ceftriaxone.

DISCUSSION

Clinical epidemiologic observations have placed the occurrence of AmpC selection in Enterobacter spp. at 8 to 19% (22–24), and the emergence of resistance appears to be contingent upon antimicrobial exposure and the bacterial burden of infection. For example, if patients were empirically started on FEP, then the emergence of resistance may be less likely regardless of bacterial inoculum due to FEP’s stability against AmpC β-lactamases. This is unlike CRO or TZP, where the emergence of resistance in high-inoculum disease states and mutational derepression are more likely to occur compared to diseases with low inocula. In the present study, one-third of our clinical K. aerogenes isolates demonstrated the ability to grow in the presence of both TZP and CRO at increasing concentrations despite having a susceptible phenotype. This demonstrates that the presence of resistant mutants was evident in the starting population but may not be apparent via manual and automated susceptibility testing methods. In addition, there appeared to be no correlation with initial MIC and the emergence of resistance during the static time-kill studies. The aim of the high-inoculum time-kill was to determine whether clinically achievable concentrations of TZP would be enough to suppress the growth of resistant mutants within the starting inoculum. To our surprise, the lack of TZP bactericidal activity in the high-inoculum time-kill study did not correlate with a resistant subpopulation in the ELF-simulated TZP20 since the MICs were still TZP susceptible according to the Clinical and Laboratory Standards Institute (CLSI) breakpoints. The occurrence of elevated MICs in TZP10-exposed organisms may indicate the ability of low TZP concentrations to select for resistance, which has been described in Pseudomonas aeruginosa (19). The results of our low- and high-inoculum time-kill studies suggest that the efficacy of TZP is inversely proportional to bacterial burden. Cheng et al. found no difference between patients treated with TZP or those treated with FEP or carbapenems for bacteremias caused by organisms with potential AmpC derepression in their retrospective cohort (11). However, there were relatively few occurrences of lower respiratory tract infections, and the association of the site of infection on mortality was not reported. In addition, the TZP dose used in the study was not explicitly stated and could have influenced the performance of TZP against such organisms. In contrast, the Merino Trial demonstrated that treatment with TZP compared to meropenem did not result in a noninferior 30-day mortality in patients with bacteremias caused by E. coli or K. pneumoniae (25). It is unclear from this trial whether pharmacodynamic dose optimization of TZP using extended infusion may have resulted in improved outcomes for patients treated in the TZP arm of the study. In the present study, although TZP40 had a significant decrease in log₁₀ CFU/ml at 24 h compared to the baseline, it failed to demonstrate bactericidal activity. Since TZP was significantly less effective than FEP at reducing the bacterial burden of high-inoculum K. aerogenes, FEP may be the preferred carbapenem-sparing agent compared to TZP for pneumonia.

We acknowledge that the present study has some limitations. While selection of derepressed mutants may be the cause of reduced susceptibility based on the prevalence of inducible AmpC in K. aerogenes, we did not perform genotypic testing to confirm that AmpC β-lactamase production was the definitive mechanism of resistance to TZP. However, in clinical practice, genotypic testing for AmpC is not routinely performed. Second, the run time of the time-kill models represent a 24-h time period, which may not be long enough to allow for overgrowth of resistant subpopulations. In addition, this study utilized a static model which does not capture the physiologic exposure of bacteria to fluctuating concentrations of antibiotics in vivo. However, this would suggest that the increase of derepressed mutants may be greater with dynamic concentrations during times where antibiotic concentrations are less than the MIC. Since %T > MIC is the best indicator of β-lactam activity, this model should have resulted in better TZP activity because the model maintains a 100% T > MIC. Lastly, the concentrations of TZP were based on the concentrations achievable with a 3.375-g intravenous (i.v.) prolonged infusion regimen administered every 8 h (q8h) and extrapolating a percentage of ELF-to-serum penetration from previous studies that used intermittent and continuous infusions (15, 16). However, these ELF concentrations are similar to published data obtained during 13.5-g (12.7 μg/ml) and 18-g (19.1 μg/ml) continuous infusions of TZP (15). In studies evaluating parameters associated with rise of TZP-resistance in Pseudomonas aeruginosa, an inverted “U” phenomenon was described wherein intermittent boluses (higher C_max) allowed for a lower C_min:MIC ratio before the emergence of resistance occurred compared to prolonged infusion regimens (19). P. aeruginosa has many resistance mechanisms and may not be the most appropriate comparator for K. aerogenes. Enterobacter cloacae, a more similar organism, was evaluated in a murine model evaluating ceftizoxime exposure on the emergence of ceftizoxime resistance wherein the AUC/MIC parameter was determined to be best correlated with resistance (26). It may be of interest to examine whether a higher C_max or AUC_0–24 has a similar protective effect with regard to AmpC-derepression in K. aerogenes since TZP10 exposure demonstrated some resistance which was not observed with TZP20 exposure. Studies evaluating dynamic concentrations and differing dosing regimens (e.g., intermittent compared to prolonged infusion) should be undertaken to determine whether TZP is appropriate for high-inoculum infections and what pharmacodynamics index predicts success against K. aerogenes.

In conclusion, with the increasing incidence of carbapenem resistance worldwide, clinicians need to be cognizant of carbapenem-sparing options for organisms with inducible AmpC production, especially in disease states where antibiotic concentrations are variable and routine drug monitoring of β-lactam antibiotics is seldom performed. In this study, FEP demonstrated superior antibacterial activity against K. aerogenes compared to CRO or various concentrations of TZP at a high bacterial inoculum. TZP did, however, demonstrate greater reductions with a lower initial bacterial inoculum compared to a high inoculum. CRO demonstrated the ability to reliably select for resistance with a high inoculum of ostensibly susceptible isolates. Given the variability in the initial inoculum of pneumonia and the clinical and economic threat posed by multidrug-resistant Gram-negative pathogens, it may be prudent to avoid CRO and consider FEP over TZP for definitive therapy of K. aerogenes, even when these agents are phenotypically sensitive.

MATERIALS AND METHODS

Bacterial strains.

Fifteen clinical respiratory K. aerogenes isolates from patients hospitalized at the University of New Mexico Hospitals (UNMH) in Albuquerque, NM, during the period from 12 August 2017 to 31 December 2017 were selected. Strains were obtained in critically ill patients diagnosed with lung processes from either sputum, bronchoalveolar lavage, or endotracheal tube sampling.

Antibiotics and media.

CRO and FEP analytical powder were obtained from the USP (Rockville, MD), and Etests were obtained from bioMérieux, Inc. (Durham, NC). TZP powder for injection (Wyeth Pharmaceuticals, Philadelphia, PA) was obtained from the UNMH pharmacy department. Supplemented Mueller-Hinton broth (SMHB) obtained from Becton Dickinson (Sparks, MD) was used in the BMD, MIC, and time-kill experiments. Mueller-Hinton agar (MHA; Becton Dickinson) was used for Etest studies, antibiotic-supplemented plate studies, and bacterial enumeration for time-kill studies.

Susceptibility testing.

Initial MICs were determined using BMD for TZP according to standard procedures and read by study personnel. Of note, the commercial product with a 1:8 TZP ratio was utilized in BMD instead of a fixed concentration of tazobactam. Etests were performed for CRO and FEP and read by study personnel (21). Since there were no U.S. Food and Drug Administration-approved Etests for TZP at the time of this study, BMD was used to determine MICs. The MICs for all antibiotics were determined prior to study initiation, and TZP and CRO MICs were determined again after antibiotic exposure in the high-inoculum time-kill study. The same method for MIC determination was used for each antibiotic for all subsequent MIC evaluations according to standard procedures. These initial MICs were compared to the MICs determined by BD Phoenix susceptibility panels (Becton Dickinson, Franklin Lakes, NJ) performed by TriCore Reference Labs. Strains were considered to demonstrate essential agreement if the BD Phoenix MIC was within one dilution of the laboratory-determined MIC.

Resistance selection using sequential antibiotic-supplemented agar plates.

The presence of resistant mutants within K. aerogenes (n = 10) that were initially reported as susceptible was determined via positive selection using solid media supplemented with increasing concentrations of TZP (4/4 to 16/4 μg/ml, based on the piperacillin component) and CRO (1 to 8 μg/ml) (27). These antibiotics were chosen because resistance has been demonstrated to these agents during therapy (22). The initial plate contained a test isolate cultured onto antibiotic-free media and incubated overnight. A 0.5 MacFarland suspension was prepared. Direct inoculation occurred by pipetting 500 μl of this suspension onto a plate containing the lowest concentration of antibiotic and spreading the suspension across the plate using a sterile glass spreader. This amount of isolate (∼7.5 × 10⁷ CFU) was chosen as the inverse of the estimated frequency of mutational deletion of regulatory ampD (27, 28), which is necessary to ensure that naturally occurring mutants would be present at baseline (29). These plates were incubated for 48 h at 37°C. Any growth of test isolate was then used to prepare a 0.5 MacFarland suspension that was spread onto a plate containing a higher antibiotic concentration, as described above, until the isolate failed to grow or the maximal predetermined concentration of antibiotic had been reached.

Time-kill studies.

Isolates demonstrating growth on the maximum TZP- and CRO-containing plates (n = 5) were evaluated in duplicate via time-kill experiments at a starting inoculum goal of 7.0 log₁₀ CFU/ml. Colonies of different morphology were selected from plates incubated for 24 h and adjusted to appropriate inoculum. All time-kill experiments were performed at 37°C for 24 h. This bacterial density was chosen to represent the high bacterial burden present in pneumonia (30–32). Since no clinical ELF data existed for CRO, 100% ELF penetration was assumed and 6 μg/ml was used to represent the AUC_0–24 for a 2-g i.v. q24h dosing regimen (33). FEP was studied at 14 μg/ml, which was derived from a previously published ELF study of a 2-g i.v. bolus, followed by 4 g i.v. over 24 h (34). Three TZP concentrations were studied to approximate the median serum C_max (40 μg/ml) and a 50% reduction in the C_max to represent the ELF concentration (20 μg/ml), and the ELF AUC_0–24 (10 μg/ml) was obtained after administering a 3.375-g i.v. q8h with a 4-h infusion (35). Test tubes containing no antibiotic were used for growth control. All antibiotic concentrations and growth controls were tested in duplicate for each isolate. Samples were collected at 0, 2, 4, 8, and 24 h and then serially diluted in normal saline, plated in triplicate onto MHA plates, and incubated for 24 h at 37°C. Colony counts were determined by visual inspection to determine the log₁₀ CFU/ml; the lower limit of reliable detection using these methods is 2.0 log₁₀ CFU/ml. If excessive variance in any one time-kill precluded statistical analysis, then that specific isolate/antibiotic run and a growth control were repeated in duplicate. Isolates exposed to CRO and ELF-simulated TZP concentrations had MICs to CRO and TZP determined after the high-inoculum time-kill study.

After the high-inoculum time-kill study, a low-inoculum time-kill study with a goal starting inoculum of 5.0 log₁₀ CFU/ml was tested for all isolates. ELF-mimicked concentrations were used for CRO (6 μg/ml) and TZP (20 and 10 μg/ml), as described above. FEP was omitted from the low-inoculum study since it demonstrated significant killing in the high-inoculum study. Samples were collected at 0, 3, 6, and 24 h and then serially diluted in normal saline, plated in triplicate onto MHA plates, and incubated for 24 h at 37°C. Colony counts were again determined by visual inspection to determine the log₁₀ CFU/ml with the lower limit of reliable detection of 2.0 log₁₀ CFU/ml.

Time-kill curves were generated by plotting mean colony counts (log₁₀ CFU/ml) versus time to compare 24-h killing effects (Fig. 1 and 2). Bactericidal activity was defined as a ≥3.0 log₁₀ CFU/ml reduction from baseline. Within-run categorical agreement was defined as having both MICs classified as susceptible, intermediate, or resistant.

Statistical analysis.

Differences in log₁₀ CFU/ml between multiple groups were compared by one-way analysis of variance with a Holm-Sidak post hoc test. Reductions in inoculum between two groups were compared using Welch’s t test. A P value of ≤0.05 was considered significant for all tests. All statistical analyses were performed with Sigma Plot (version 14.0; Systat Software, Inc., San Jose, CA).