Pharmacodynamics of Ceftolozane plus Tazobactam Studied in an In Vitro Pharmacokinetic Model of Infection

Abstract

Ceftolozane plus tazobactam is an antipseudomonal cephalosporin combined with tazobactam, an established beta-lactamase inhibitor, and has in vitro potency against a range of clinically important β-lactamase-producing bacteria, including most extended-spectrum-β-lactamase (ESBL)-positive Enterobacteriaceae. The pharmacodynamics of β-lactam–β-lactamase inhibitor combinations presents a number of theoretical and practical challenges, including modeling different half-lives of the compounds. In this study, we studied the pharmacodynamics of ceftolozane plus tazobactam against Escherichia coli and Pseudomonas aeruginosa using an in vitro pharmacokinetic model of infection. Five strains of E. coli, including three clinical strains plus two CTX-M-15 (one high and one moderate) producers, and five strains of P. aeruginosa, including two with OprD overexpression and AmpC β-lactamases, were employed. Ceftolozane MICs (E. coli, 0.12 to 0.25 mg/liter, and P. aeruginosa, 0.38 to 8 mg/liter) were determined in the presence of 4 mg/liter tazobactam. Dose ranging of ceftolozane (percentage of time in which the free-drug concentration exceeds the MIC [fT>MIC], 0 to 100%) plus tazobactam (human pharmacokinetics) was simulated every 8 hours, with half-lives (t_1/2) of 2.5 and 1 h, respectively. Ceftolozane and tazobactam concentrations were confirmed by high-performance liquid chromatography (HPLC). The ceftolozane-plus-tazobactam fT>MIC values at 24 h for a static effect and a 1-log and 2-log drop in initial inoculum for E. coli were 27.8% ± 5.6%, 33.0% ± 5.6%, and 39.6% ± 8.5%, respectively. CTX-M-15 production did not affect the 24-h fT>MIC for E. coli strains. The ceftolozane-plus-tazobactam fT>MIC values for a 24-h static effect and a 1-log and 2-log drop for P. aeruginosa were 24.9% ± 3.0%, 26.6% ± 3.9%, and 31.2% ± 3.6%. Despite a wide range of absolute MICs, the killing remained predictable as long as the MICs were normalized to the corresponding fT>MIC. Emergence of resistance on 4× MIC plates and 8× MIC plates occurred maximally at an fT>MIC of 10 to 30% and increased as time of exposure increased. The fT>MIC for a static effect for ceftolozane plus tazobactam is less than that observed with other cephalosporins against E. coli and P. aeruginosa and is more similar to the fT>MIC reported for carbapenems.

INTRODUCTION

Ceftolozane plus tazobactam is a novel antibacterial consisting of a combination of the new cephalosporin ceftolozane with a well-established β-lactamase inhibitor, i.e., tazobactam. The combination has completed phase 3 clinical trials for complicated intra-abdominal and complicated urinary tract infections (1, 2). It was recently approved in the United States for these indications and is undergoing regulatory review in the European Union.

The in vitro potency of ceftolozane plus 4 mg/liter tazobactam has been studied against Pseudomonas aeruginosa and members of the Enterobacteriaceae from the United States and Europe (3–5). Ceftolozane plus tazobactam was the most active β-lactam agent when tested against multidrug-resistant P. aeruginosa and Enterobacteriaceae (6); however, due to ceftolozane's stability to pseudomonal AmpC beta-lactamase, the addition of tazobactam produced only a marginal improvement in potency against P. aeruginosa. Similar observations were made with multiresistant and β-lactam-resistant clinical isolates (7, 8). In contrast, against extended-spectrum-β-lactamase (ESBL)-producing and/or ceftazidime-resistant Escherichia coli, Klebsiella pneumoniae, and Proteus sp., MIC₅₀ values of ≥32 mg/liter for ceftolozane alone were reduced to 1 to 4 mg/liter by the addition of tazobactam. The pharmacodynamics of ceftolozane plus tazobactam has been studied in a neutropenic murine thigh infection model and in vitro models. The dominant pharmacodynamic index was the percentage of time in which the free-drug concentration exceeded the MIC (fT>MIC), and the fT>MIC for 24-h static effect was 26.3 ± 2.1% for Enterobacteriaceae (4 strains) and 24.0 ± 3.3% for P. aeruginosa (4 strains). The equivalent magnitudes for a 1-log kill were 31.6 ± 1.6% and 31.5 ± 3.9%, respectively (9). These values are lower than the 30 to 40% fT>MIC bacteriostatic targets often associated with cephalosporins and may be related to ceftolozane's penicillin-binding protein affinity, with the increased bactericidal activity being more like that of carbapenems.

The objective of this study was to determine the fT>MIC targets for antibacterial effect, as measured by the extent of bacterial killing and changes in population profiles at 24 h, 48 h, and 72 h of ceftolozane plus tazobactam against a range of Enterobacteriaceae and P. aeruginosa isolates with and without production of ESBL and other β-lactamases.

MATERIALS AND METHODS

Pharmacokinetics.

A FerMac 310 fermentation system (Electrolab, Tewkesbury, England) in vitro pharmacokinetic model (as previously described [24]) was used to simulate fT>MIC (0 to 100%) of free (unbound) ceftolozane (TOL) for each strain over 72 h when administered every 8 h, i.e., total of 9 doses (10). The pharmacokinetic profile from ceftolozane plus tazobactam (TAZ) as a 2:1 mg:mg ratio produces a TOL/TAZ serum concentration ratio of 2.9:1 (11). Initially, in the dose-ranging experiments, tazobactam was dosed at a fixed ceftolozane-tazobactam ratio of 2.9:1. In subsequent experiments, tazobactam given intravenously every 8 h was dosed at human pharmacokinetic concentrations associated with a 500 mg dose (maximum concentration of free/unbound drug of 18.0 mg/liter) (12). Due to the difference in half-lives between ceftolozane and tazobactam (2.5 h and 1 h, respectively), the model was supplemented with ceftolozane (12.5 ml/h) throughout each dosing period via a separate dosing chamber to achieve the required concentration-time profile. The ceftolozane free-drug concentrations used were based on total drug concentration-time profiles from healthy volunteers (protein binding, 20%) (11). Tazobactam free-drug concentrations were based on protein binding of 20% (12). The simulated half-lives of ceftolozane and tazobactam were 2.5 h and 1 h, respectively.

Media and bacterial counting.

Mueller-Hinton broth (100%; Thermo Fisher, Basingstoke, England) was used in all experiments. Nutrient agar plates (Thermo Fisher, Basingstoke, England) were used to recover bacterial strains from the model. Carbapenemase (kindly supplied by J. Spencer, Bristol University, Bristol, United Kingdom) was used to neutralize ceftolozane. The carbapenemase neutralized ceftolozane up to a concentration of 50 μg/ml. Aliquots of 1 ml each were taken from the central compartment at 0, 1, 2, 3, 4, 5, 6, 7, 12, 24, 25, 26, 27, 28, 29, 30, 31, 36, 48, 60, and 72 h, treated with 10 μl of carbapenemase, left for 20 min, and then plated on nutrient agar plates using a spiral plater (Don Whitley Scientific, Yorkshire, United Kingdom) for determination of total bacterial counts. A second aliquot not treated with carbapenemase was plated on nutrient agar plates containing 1×, 2×, 4×, and 8× the MIC of ceftolozane and tazobactam (4 mg/liter) for the strain tested to detect changes in population profile. Any isolates grown on these plates were stored at −70°C and subsequently had their ceftolozane-tazobactam (4 mg/liter) MIC determined by agar dilution using the CLSI method, which was modified to produce a linear series of dilutions (13). Ceftolozane-tazobactam MICs were assessed by concurrent agar dilution MIC testing of single survivors at 72 h and of parental (naive) strains. The risk of emergence of resistance was stratified by the percent fT>MIC.

Bacteria.

Five strains of E. coli and five strains of P. aeruginosa were tested (Table 1). The initial inoculum used was 1 × 10⁶ CFU/ml. MICs were determined in duplicate on all strains as described by CLSI using microdilution methodology with the exception that linear dilutions were used instead of traditional 2-fold dilutions. Tazobactam was used at a fixed concentration of 4 mg/liter. E. coli strains 47202 (CTX-M-15 hyperproducer) and 49439 (CTX-M-15 moderate producer) were supplied by JMI Laboratories, North Liberty, IA, USA; E. coli 47861 (NIHJ 10909) and E. coli strains 44913, 47861, and 46961 (clinical strains) are held in the collection at the Microbiology Department, Southmead Hospital. P. aeruginosa 17286 (meropenem isogenic mutant; meropenem MIC, 6 mg/liter), 38475, and 39150 are clinical strains held at the Microbiology Department, Southmead Hospital. None of the clinical E. coli or P. aeruginosa strains have been genetically characterized. P. aeruginosa 47235 and 47237 were supplied by Cubist Pharmaceuticals, Inc. (Lexington, MA, USA).

TABLE 1.

MICs of ceftolozane with and without tazobactam (4 mg/liter) and resistance mechanisms (where known)

Bacterium	Strain	Resistance mechanism	MIC (mg/liter)
			Ceftolozane	Ceftolozane-tazobactam
E. coli	44913	None	0.12	0.12
	47861	None	0.12	0.12
	46961	None	0.19	0.19
	49439	CTX-M-15	8	0.19
	47202	CTX-M-15	64	0.25
P. aeruginosa	17286	Meropenem isogenic mutant	0.5	0.38
	38475	None	0.5	0.5
	39150	None	1	1
	47235	OprD porin mutation, AmpC overexpression	6	6
	47237	OprD porin mutation, AmpC overexpression	8	8

Antibiotic assay.

Ceftolozane and tazobactam concentrations were measured by HPLC. The stationary phase for both compounds was Gemini-NX 5uC18 (100 by 4.6 mm; Phenomenex, Macclesfield, United Kingdom). The mobile phase composition was 98% phosphate buffer, 1% acetonitrile, 1% orthophosphoric acid. The wavelength was 254 nm for ceftolozane and 220 nm for tazobactam; detection was by UV absorbance using a Dionex UltiMate 3000 variable-wavelength detector (Thermo Fisher, Hemel Hempstead, United Kingdom), and the flow rate was 1.8 ml/min. Aqueous samples (10 μl) were injected, and the retention was approximately 5 to 6 min for ceftolozane and 3 to 4 min for tazobactam. The standard curve concentration range was 0 to 50 mg/liter for both ceftolozane and tazobactam.

Data analysis.

The relationship between fT>MIC and antibacterial effect for each strain as described by log reduction in viable count at 24, 48, and 72 h were delineated using a Boltzmann sigmoid Emax equation: y = bottom + (top − bottom)/{1 + exp[(V50 − x)/slope]} using the software package GraphPad Prism version 4 (San Diego, CA, USA). The combined data for all E. coli strains and the combined P. aeruginosa strains were plotted using GraphPad Prism giving a pooled fT>MIC where possible, using a top-to-bottom variable-slope equation: y = bottom + (top − bottom)/{1 + 10̂[(log EC₅₀ − x) × Hill slope]}.

Emergence of resistance.

Changes in population analysis profiles for each strain were assessed by growth on plates containing 2×, 4×, and 8× the ceftolozane MIC plus 4 mg/liter tazobactam. Colonies growing on the 4× and 8× MIC plates were subsequently assessed for MIC change at 72 h.

RESULTS

MICs.

The MICs for the individual E. coli strains and P. aeruginosa strains with the β-lactamases present are shown in Table 1.

Relationship between methods of tazobactam dosing and antibacterial effect.

Dosing tazobactam at a fixed ratio (2.9:1) with ceftolozane to achieve a T>MIC range of 0 to 100% against an ESBL-producing E. coli strain (49439) resulted in unexpectedly high T>MIC values (81.9%). It was concluded that there was insufficient tazobactam present to protect the strain from the β-lactamase present. Dosing tazobactam as a continuous infusion at 4 mg/liter reduced the T>MIC to 26.2%; using mean tazobactam human pharmacokinetics provided sufficient protection (T>MIC of 24.4%). Subsequent experiments all used mean tazobactam human pharmacokinetic profiling.

Pharmacokinetic curves.

There was good correlation between the target and measured ceftolozane and tazobactam concentrations; the interday coefficient of variation (CV) was 4.2% for ceftolozane and 7.7% for tazobactam, showing that we were able to accurately concurrently model the different half-lives of the two drugs. The lower limit of detection was 1.0 mg/liter for both compounds. Figures 1 and 2 show scatter plots of the predicted and observed ceftolozane and tazobactam concentrations. Linearity of the ceftolozane assay is shown by the equation y = 5.5318x + 2.5709, and r² was 0.9746; for tazobactam, y = 1.8432x + 0.0602, and r² was 0.9996. The actual half-life simulated was 2.7 h for ceftolozane and 1.064 h for tazobactam.

FIG 1.

Scatter plot of predicted and observed ceftolozane concentrations, with 95% confidence intervals.

FIG 2.

Scatter plot of predicted and observed tazobactam concentrations, with 95% confidence intervals.

Relationship between fT>MIC and antibacterial effect.

There was a clear relationship between antibacterial effect and fT>MIC for all strains. The fT>MIC values for a static effect and for 1-log, 2-log, and 3-log reductions in counts of each strain plus mean data for E. coli and P. aeruginosa strains are shown in Tables 2 and 3. The fT>MIC relationship to antibacterial effect (log change in viable count compared) for the combined E. coli strains (n = 5) at 24, 48, and 72 h is shown in Table 2, and that for all P. aeruginosa strains (n = 5) is shown in Table 3.

TABLE 2.

fT>MIC relationship to antibacterial effect for ceftolozane plus tazobactam against E. coli after 24, 48, and 72 h

Strain	MIC (mg/liter)	Antibacterial effect (fT>MIC)
		24 h				48 h				72 h
		Static effect	1-log drop	2-log drop	3-log drop	Static effect	1-log drop	2-log drop	3-log drop	Static effect	1-log drop	2-log drop	3-log drop
44913	0.12	22.0	24.2	26.3	29.2	57.2	60.8			98.7
47861	0.12	26.8	33.7	40.1	46.3	38.1	39.5	39.7	40.3	44.5	45.8	45.8	47.0
46961	0.19	36.6	37.9	39.6	45.3	57.5	58.2	60.0		79.3	80.3	81.3	83.3
49439	0.19	24.4	31.6	39.8	56.1	54.0	61.7			81.3	95.0
47202	0.25	29.0	37.5	51.4		44.3	49.2			48.9
Mean		27.5	31.9	36.5	44.2	51.7	55.1	49.9	40.3	76.0	73.7	63.6	65.2
SD		6.41	5.73	6.77	11.14	9.20	10.47			22.70	25.26

TABLE 3.

fT> MIC relationship to antibacterial effect for ceftolozane plus tazobactam against P. aeruginosa after 24, 48, and 72 h

Strain	MIC (mg/liter)	Antibacterial effect (fT>MIC)
		24 h				48 h				72 ha
		Static effect	1-log drop	2-log drop	3-log drop	Static effect	1-log drop	2-log drop	3-log drop	Static effect	1-log drop	2-log drop	3-log drop
17286	0.38	20.7	20.8			47.8	56.7
38475	0.50	23.0	25.3	30.5		26.3	29.0	33.2		23.0	24.5
39150	1.00	27.8	28.0	28.2	29.2	25.3	25.5
47235	6.00	25.6	27.8	29.6
47237	8.00	27.5	31.3	36.3	53.8	20.8	22.0			NT	NT	NT	NT
Mean		24.9	26.6	31.2	41.5	24.1	25.5	33.2		23.0	24.5
SD		3.04	3.90	3.56		2.93	3.50

^aNT, not tested.

There was an excellent curve fit between fT>MIC and antibacterial effect, with r² values for the pooled E. coli strains being greater than 0.9700 at 24, 48, and 72 h. For P. aeruginosa at 24 and 48 h, all r² values were greater than 0.93 except for P. aeruginosa 47235 (0.6698 at 48 h). At 72 h, the relationship between antibacterial effect and fT>MIC was variable, and this was reflected in the curve values, which ranged from 0.3014 to 0.9919.

At 24 h, the fT>MIC for a static effect for E. coli strains ranged from 22.0 to 36.6%. The E. coli strain with higher ceftolozane-tazobactam MICs (0.25 mg/liter) had an antibacterial effect response similar to that of the strains with MICs of 0.12 mg/liter. The mean fT>MIC for a static effect and for 1-, 2-, and 3-log reductions in viable count increased from 27.8% ± 5.6% to 33.0% ± 5.5% and from 39.4% ± 8.9% to 44.2% ± 11.2%, respectively. The mean fT>MIC for static effect increased from 27.8% ± 5.6% at 24 h to 50.2% ± 8.6% at 48 h, and as time increased, the fT>MIC increased for each antibacterial effect measure (Table 2; Fig. 2). Neither the resistance mechanism nor the MIC affected the fT>MIC values; strains 47202 and 49439, containing CTX-M ESBLs, showed an antibacterial effect comparable to that seen in the strains without these enzymes (Fig. 3 and 4).

FIG 3.

T>MIC versus antibacterial effect for E. coli 49439, a CTX-M-15 moderate producer.

FIG 4.

T>MIC versus antibacterial effect for E. coli 47202, a CTX-M-15 hyperproducer.

A similar pattern of exposure response was observed with P. aeruginosa strains at 24 h; the fT>MIC for a static effect ranged from 20.7 to 27.8% (Table 3). The two P. aeruginosa strains with higher ceftolozane-tazobactam MICs (6.0 and 8.0 mg/liter) had antibacterial effect responses similar to those of strains with MICs of 0.38, 0.5, and 1 mg/liter, demonstrating that the T>MIC is consistent across a range of MICs. The mean fT>MIC for a static effect and for 1- and 2-log reductions in viable count at 24 h increased from 24.9% ± 3.0% to 26.6% ± 3.9% to 31.2% ± 3.7%, respectively (Table 2). Only two of the five strains tested demonstrated a 3-log reduction in viable count.

Increased exposure had a variable effect on the fT>MIC; the static mean fT>MIC decreased slightly from 24.9% at 24 h to 24.1% at 48 h. For strain 47235 at 48 h, the combination did not produce a static effect, and the fT>MIC increased from 20.7% to 47.8% for strain 17286. At 48 h, the ceftolozane- tazobactam combination achieved a 2-log reduction in viable count against strain 38475 alone, and at 72 h this was reduced to a static effect and a 1-log reduction in bacterial count (Table 2).

Emergence of resistance.

Tables 4 and 5 show the risk of emergence as stratified by fT>MIC over time (24 to 72 h) for E. coli and P. aeruginosa, respectively. As time increased, the number of CFU/ml increased; at 24 h, E. coli strains produced growth on 4× MIC plates at fT>MIC values of 10 to 30%. No growth was observed at a fT>MIC of ≥40%; no growth occurred on 8× MIC plates. At 48 h, growth was noted on 8× MIC plates up to a fT>MIC of 40%. At 72 h, a maximum risk of resistance was seen at a fT>MIC of 30%. Growth on 8× MIC plates was seen at 60% in 1/4 experiments. P. aeruginosa experiments showed increased risk of resistance at fT>MIC values of 10 to 60% at 24 h on 4× MIC plates (2/5 experiments; 4 log₁₀) (Table 5); 1/5 experiments showed growth on 8× MIC plates at a fT>MIC of 40%. At 48 h, growth was observed at a fT>MIC of 100% on 4× and 8× MIC plates in 1/5 experiments; maximum growth was seen at a fT>MIC of 20 to 40%. A similar pattern was seen at 72 h.

TABLE 4.

Risk of emergence of resistance in E. coli after 24, 48, and 72 h

T>MIC (%)	24 h						48 h						72 h
	No. of expts/total	Growth on 4× MIC plates (log CFU/ml)		No. of expts/total	Growth on 8× MIC plates (log CFU/ml)		No. of expts/total	Growth on 4× MIC plates (log CFU/ml)		No. of expts/total	Growth on 8× MIC plates (log CFU/ml)		No. of expts/total	Growth on 4× MIC plates (log CFU/ml)		No. of expts/total	Growth on 8× MIC plates (log CFU/ml)
		Mean	SD		Mean	SD		Mean	SD		Mean	SD		Mean	SD		Mean	SD
0	0/5	<2		0/5	<2		2/5	3.40		1/5	3.92		1/5	3.89		1/5	2.92
10	4/5	3.59	1.42	0/5	<2		4/5	5.44	1.45	2/4	3.57		4/4	4.73	1.38	2/4	2.35
20	2/5	4.30		0/5	<2		5/5	4.08	1.80	3/4	3.26	0.23	5/5	5.16	1.83	3/4	5.22	0.90
30	1/5	5.70		0/5	<2		3/4	5.12	2.34	3/4	3.00	1.13	4/5	5.70	1.36	3/4	3.50	1.15
40	0/5	<2		0/5	<2		4/5	5.17	1.42	2/4	3.93		2/5	5.31		2/4	3.79
60	0/5	<2		0/5	<2		1/5	4.18		0/5	<2		2/5	4.69		1/4	4.85
80	0/5	<2		0/5	<2		0/4	<2		0/5	<2		1/3	6.76		0/4	<2
100	0/5	<2		0/5	<2		0/5	<2		0/5	<2		0/4	<2		0/4	<2

TABLE 5.

Risk of emergence of resistance in P. aeruginosa after 24, 48, and 72 h

T>MIC (%)	24 h						48 h						72 h
	No. of expts/total	Growth on 4× MIC plates (log CFU/ml)		No. of expts/total	Growth on 8× MIC plates (log CFU/ml)		No. of expts/total	Growth on 4× MIC plates (log CFU/ml)		No. of expts/total	Growth on 8× MIC plates (log CFU/ml)		No. of expts/total	Growth on 4× MIC plates (log CFU/ml)		No. of expts/total	Growth on 8× MIC plates (log CFU/ml)
		Mean	SD		Mean	SD		Mean	SD		Mean	SD		Mean	SD		Mean	SD
0	0/5	<2		0/5	<2		0/5	<2		0/5	<2		0/4	<2		0/4	<2
10	2/5	4.68		1/5	4.38		2/4	7.37		2/5	6.53		3/4	6.97	1/3	7.23
20	2/5	4.45		0/5	<2		3/5	3.83	2.43	1/5	7.00		2/4	3.55		1/3	7.40
30	2/5	4.24		1/5	4.30		3/5	3.35	1.29	2/5	6.71		1/4	7.65		1/3	7.28
40	1/5	4.49		1/5	3.20		3/5	4.64	1.16	1/5	7.00		2/4	7.55		2/4	6.89
60	2/5	4.44		0/5	<2		2/4	6.45		2/4	6.54		1/3	6.32		1/3	5.81
80	0/5	<2		0/5	<2		1/5	6.65		2/5	5.16		1/3	7.18		1/2	6.04
100	0/5	<2		0/5	<2		1/5	7.26		1/5	5.60		1/3	7.48		1/3	7.36

MICs were determined for the parent strains and progeny growing on 4× or 8× MIC plates at 72 h. Testing by agar dilution MIC showed increases in the ceftolozane MIC alone and in the ceftolozane-tazobactam MIC of the progeny ranging from 3× to more than 16× the original MIC. Strain 47202 (CTX-M-15) showed the greatest increase in MIC (16×). It is unclear what the change in resistance mechanism(s) is, as we have not characterized the strains.

DISCUSSION

The pharmacodynamics of cephalosporins against E. coli and Pseudomonas aeruginosa is well described, with fT>MIC (as a percentage) for a static effect being the established pharmacodynamic index relating to clinical efficacy (14–16). It is accepted that a T>MIC of 35 to 40% is associated with a 24-h bacteriostatic effect for aerobic Gram-negative rods and that higher T>MIC values are linked with bactericidal drug activity for cephalosporins (www.eucast.org).

The data in this study reaffirm this relationship, with an excellent correlation between fT>MIC and antibacterial effect for ceftolozane plus tazobactam with E. coli and P. aeruginosa. The fT>MIC values for ceftolozane plus tazobactam at 24 h for a static effect and for 1-log₁₀ and 2-log₁₀ drops in initial inoculums for E. coli were 27.8% ± 5.6%, 33.0% ± 5.6%, and 39.6% ± 8.6%, respectively. The ceftolozane-tazobactam fT>MIC values for a 24-h static effect and for 1-log₁₀ and 2-log₁₀ drops in CFU count for P. aeruginosa were 24.9 ± 3.0%, 26.6 ± 3.9%, and 31.2 ± 3.6%. These data support data obtained by Craig and Andes (9) in a murine neutropenic thigh model using susceptible Gram-negative bacilli and ESBL-containing E. coli. They reported a mean fT>MIC of 26.3% ± 2.1% for a static effect with cephalosporin-susceptible strains, and for ESBL-containing strains, a fT>MIC of 31.1% ± 4.4% for a static effect was observed (9). In our study, we observed similar 24-h static effect and 1- and 2-log₁₀ kill values for E. coli strains containing ESBLs and non-ESBL strains. In similarly designed experiments with razupenem and Enterobacteriaceae, the fT>MIC values for the Enterobacteriaceae strains we tested were 27.5% to 47% (34.2% ± 7.6%) for a bacteriostatic effect at 24 h and 42.3% to 64.1% (42.5% ± 7.8%) for a 2-log-unit reduction in count, which is in keeping with the published literature on carbapenems (17). The increase in the fT>MIC threshold as time increases has been observed in previous studies with other cephalosporins; e.g., previous data obtained by using ceftaroline against four strains of E. coli showed an increase in the static fT>MIC from 35.0% ± 6.3% at 24 h to 40.7% ± 4.7% at 48 h, 47.1% ± 1.6% at 72 h, and 48.1% ± 16.1% at 96 h (K. E. Bowker, unpublished data). The increase in T>MIC may be due to initial killing of more susceptible colonies within the population.

The fT>MIC values for a static effect for P. aeruginosa were similar to those for E. coli with a mean value of 24.9%. The 1-log and 2-log₁₀ drops in CFU were lower than those for E. coli at 26.6% and 31.2%, respectively; again, these figures are similar to those reported by Craig and Andes (9), who cited values of 24% ± 3.3% and 31.5% ± 3.9% (n = 4), but somewhat lower than those reported by Lepak et al. (18), who tested 12 strains of P. aeruginosa and reported a T>MIC for static effect of 31.2% ± 6.99% and a value for a 1-log drop in count of 39.4% ± 7.53%. The fT>MIC threshold did not change with change in MIC. For P. aeruginosa 34845, the fT>MIC for static effect was 23.0%, markedly lower than the value (37%) we previously reported for doripenem for the same strain (22). In that study, the mean fT>MIC for 3 strains was 25.0% ± 11% for a static effect. Ceftolozane plus tazobactam appears to be more carbapenem-like than cephalosporin-like in its activity. Bulik et al. (23), using a range of Gram-negative bacteria, including ESBL producers and P. aeruginosa (ceftolozane-tazobactam MICs of 1 to 64 mg/liter and piperacillin-tazobactam MICs of 8 to 64 mg/liter), reported that a T>MIC of ≥37.5% resulted in a 1- to 3-log₁₀ reduction in bacterial count after 24 h for strains with MICs of ≤16 mg/liter. However, they concluded that ceftolozane plus tazobactam had improved efficacy over that of piperacillin plus tazobactam. The higher fT>MIC target in their study could be due to study design; however, our own data with piperacillin-tazobactam against P. aeruginosa, obtained with the model used in this study, indicated fT>MIC values of 39.8% ± 3.3% for a static effect and 62.8% ± 17.7% for a 2-log reduction in inoculum, suggesting that ceftolozane plus tazobactam is pharmacodynamically superior to piperacillin plus tazobactam (19).

VanScoy et al. took a different approach, focusing on the protective capacity of tazobactam (20). They reported times above the tazobactam threshold of 35, 50, and 75%, respectively, for a static effect and for 1-log and 2-log reductions in counts of isogenic CTX-M-15-producing E. coli at 24 h. Our initial data concur that if insufficient tazobactam was present, it would not protect against CTX-M-producing strains.

In this study, we tested only E. coli; however, we recently observed that the fT>MIC exposure varies within the Enterobacteriaceae (21). fT>MIC values for a 24-h bacteriostatic effect were similar for ceftaroline against E. coli (35.0% ± 6.3%), K. pneumoniae (36.1% ± 8.3%), and Proteus mirabilis (39.1% ± 26.4%). Values were higher with Citrobacter and Serratia spp. In addition there was much more strain-to-strain variation within the Proteus strains than E. coli and K. pneumoniae, while K. pneumoniae required much higher fT>MIC values for a 2-log or 3-log kill than E. coli. Hence, it is not clear to what extent observations on E. coli can be extrapolated to the entire family of Enterobacteriaceae.

The target required for prevention of emergence of resistance is much greater than that required for a static effect or a 1-log₁₀ reduction and also increases as time increases. The maximum risk of resistance generally occurred at a fT>MIC around 30% for E. coli, which is similar to the target static fT>MIC. The standard dosing of ceftolozane plus tazobactam would achieve a fT>MIC much higher than this figure. Population analysis profiles with razupenem and Enterobacteriaceae indicated growth on recovery media with 2×, 4×, and 8× MIC at fT>MICs in the range of 1% to 69% but rarely at values of ≥70%. In our previous study using doripenem and P. aeruginosa, we were unable to suppress resistance even using doses of 4× to 6× those used in human studies. Similarly, in this study we were unable to suppress resistance even with a fT>MIC of 100% for P. aeruginosa. The relationship between fT>MIC and changes in population profiles for both E. coli and P. aeruginosa species and ceftolozane-tazobactam is similar to that seen in previous studies with E. coli and P. aeruginosa using our in vitro model (17, 19).

In conclusion, this study confirms that fT>MIC predicts the ceftolozane-tazobactam antibacterial effect and that the pharmacokinetic/pharmacodynamic targets are similar for E. coli and P. aeruginosa. The target required for static effect is lower than the 30 to 40% fT>MIC targets often associated with cephalosporins and may be related to ceftolozane's increased bactericidal activity, which is more like that of carbapenems. Furthermore, ESBL production and strain differences in MICs did not affect fT>MIC. Finally, the method used for assessing the effect of a β-lactam inhibitor against strains containing an enzyme(s) is important and may affect data and conclusions.