Pharmacokinetics-Pharmacodynamics of Tazobactam in Combination with Piperacillin in an In Vitro Infection Model

Abstract

We have previously demonstrated the pharmacokinetic-pharmacodynamic (PK-PD) index best associated with the efficacy of tazobactam when used in combination with ceftolozane to be the percentage of the dosing interval during which tazobactam concentrations remained above a threshold value (%time>threshold). Using an in vitro infection model and the same isogenic CTX-M-15-producing Escherichia coli triplet set genetically engineered to transcribe different levels of bla_CTX-M-15, herein we describe dose fractionation studies designed to evaluate the PK-PD index associated with tazobactam efficacy, when given in combination with piperacillin, and the impact of the presence of a different β-lactam agent, or different bla_CTX-M-15 transcription levels, on the magnitude of the tazobactam PK-PD index necessary for efficacy. The recombinant strains demonstrated piperacillin MIC values of 128, >256, and >256 μg/ml for the low-, moderate-, and high-level CTX-M-15-producing E. coli strains, respectively. The MIC value for piperacillin in the presence of 4 μg/ml of tazobactam was 2 μg/ml for all three strains. The PK-PD index associated with tazobactam efficacy was confirmed to be %time>threshold, regardless of β-lactamase transcription (r² = 0.839). The tazobactam concentration thresholds, however, changed with the CTX-M-15 transcription level and were 0.25, 0.5, and 2 μg/ml for the low-, moderate-, and high-level CTX-M-15-producing strains, respectively (r² = 0.921, 0.773, and 0.875, respectively). The %time>threshold values for tazobactam necessary for net bacterial stasis and a 1- and 2-log₁₀-unit CFU/ml decrease from baseline at 24 h were 44.9, 62.9, and 84.9%, respectively. In addition to verifying our previous study results, these results also demonstrated that the magnitude of bacterial-cell killing associated with a β-lactam–β-lactamase inhibitor combination is dependent on the amount of β-lactamase produced. These data provide important information for the development of β-lactam–β-lactamase inhibitor combination agents.

INTRODUCTION

Piperacillin, a ureidopenicillin, has a spectrum of activity that includes both Gram-positive and -negative bacteria. However, due to the emergence of β-lactamase-producing Pseudomonas aeruginosa and Enterobacteriaceae, piperacillin was combined with tazobactam, a penicillanic acid sulfone β-lactamase inhibitor. Tazobactam extends piperacillin's spectrum of activity to include bacteria producing many Ambler class A β-lactamases, including narrow- and extended-spectrum (TEM-, SHV-, and CTX-M-type) β-lactamases, and some class C (AmpC-type) β-lactamases (1, 2). Since piperacillin-tazobactam was introduced into clinical practice in 1993 (3), it has been relied upon for the treatment of serious Gram-negative infections, including pneumonia and intra-abdominal infections.

While potent in vitro activity is an important feature of any antimicrobial agent or combination agent, appropriate dosing based on an understanding of the pharmacokinetic-pharmacodynamic (PK-PD) characteristics of the prescribed agent can both optimize and maintain the activity of the agent. Optimization of the PK-PD profile of an antibiotic agent is an important treatment strategy, especially in light of the globally recognized rise in multidrug-resistant Gram-negative bacteria in the face of decreasing numbers of novel antimicrobials in clinical development (4, 5). Maximizing PK-PD profiles may be the most readily available option for ensuring clinical success and potentially reducing the development of antimicrobial resistance (6). In the case of β-lactams, many clinicians utilize prolonged or continuous infusions to maximize the percentage of time in which the β-lactam concentration remains above the MIC for the pathogen (%time>MIC) (7, 8), which is the PK-PD index associated with efficacy (9).

Despite the established use of β-lactam–β-lactamase inhibitors in clinical practice, knowledge of the PK-PD index associated with efficacy for the β-lactamase inhibitor component is lacking. By utilizing PK-PD in vitro infection models, the time course of activity of an antimicrobial, including the exposure-response relationship, can be elucidated. In a previous study, we demonstrated that the PK-PD index associated with the efficacy of tazobactam, when used in combination with ceftolozane, was the percentage of the time in which the tazobactam concentration remained above a threshold concentration (%time>threshold) (10). The challenge organisms utilized in the ceftolozane-tazobactam studies were an isogenic set of three Escherichia coli strains, each expressing different levels of the CTX-M-15 β-lactamase enzyme (10). While that study was the first to fully characterize the PK-PD index of a β-lactamase inhibitor, the evaluation of tazobactam's activity when combined with another β-lactam agent against the same isogenic set of E. coli strains provides an opportunity to confirm the PK-PD index associated with tazobactam efficacy. The objectives of the studies described herein were to identify the PK-PD index (e.g., area under the concentration-time curve [AUC], maximum concentration of drug in serum [C_max], or %time>threshold) that best predicts the efficacy of tazobactam when used in combination with piperacillin against the same isogenic CTX-M-15-producing E. coli set; to determine the magnitude of the tazobactam PK-PD index associated with various levels of efficacy; and to determine the impact of the presence of a different β-lactam agent, or different bla_CTX-M-15 transcription levels, on the magnitude of the tazobactam PK-PD index necessary for efficacy.

MATERIALS AND METHODS

Bacteria, antimicrobial, and β-lactamase inhibitor.

Piperacillin was purchased from Sigma-Aldrich (St. Louis, MO), and tazobactam was provided by Cubist Pharmaceuticals (Lexington, MA). E. coli ATCC 25922 was used as the internal control in all susceptibility studies. Three isogenic E. coli strains producing differing levels of CTX-M-15 were utilized in these studies. The construction of the strains has been described previously (10). Briefly, the bla_CTX-M-15 gene and various upstream promoter regions were inserted into a promoterless cloning vector. Three distinct upstream promoter regions were designed to provide different levels of bla_CTX-M-15 transcription. The different levels of bla_CTX-M-15 transcription resulted in different amounts of CTX-M-15 production. These recombinant vectors were then transformed into a clinical wild-type piperacillin-susceptible E. coli strain belonging to multilocus sequence type 131 (10). The levels of bla_CTX-M-15 transcription in all three isolates were determined by quantitative real-time (qRT)-PCR assays using an endogenous reference gene (rpsL). CTX-M-15 transcription levels were compared to those of the E. coli strain demonstrating the lowest level of enzyme production, MIC result, and β-lactam hydrolytic activity. During a 2-min interval, the hydrolytic activity of the CTX-M-15 enzyme produced by each recombinant strain was measured by observing the changes in absorbance due to the opening of the β-lactam ring in an Ultrospec 3300 pro UV/visible-light spectrophotometer (GE Healthcare Biosciences, Pittsburgh, PA). The degree of hydrolytic activity was calculated using the following formula: hydrolytic activity = [(Δabsorbance/min)/protein concentration] × −1,000 (factor), with the protein concentration given in micrograms per milliliter.

Media and in vitro susceptibility studies.

Susceptibility studies were conducted using cation-adjusted Mueller-Hinton broth (BD Laboratories, Franklin Lakes, NJ) following a microdilution methodology in accordance with Clinical and Laboratory Standards Institute guidelines (11). The susceptibility of each strain to piperacillin was determined alone and in combination with a fixed tazobactam concentration (4 μg/ml). Susceptibility studies were performed in triplicate over a 2-day period and represented as the modal value.

PK-PD in vitro model and sample processing.

The one-compartment PK-PD in vitro infection model utilized in these studies has been described previously (10). The model consists of a central infection compartment that contains growth medium, the challenge strain, and a magnetic stir bar, which ensures the homogeneity of the drug(s) within the central compartment. The central infection compartment is placed atop a stir plate, and the entire unit is placed within an incubator, set at 35°C, that is temperature and humidity controlled. Drug-free growth medium is pumped into the central infection compartment via a computer-controlled peristaltic pump. The growth medium is simultaneously removed through an exit port and collected in a waste container. The challenge isolate is aseptically inoculated directly into the central infection compartment, and the peristaltic pump is set at a flow rate that allows for the simulation of the human concentration-time profile of the drug(s) under study. The test drug(s) is infused using computer-controlled syringe pumps that allow simulation of the properties of the test drug(s) (e.g., desired half-lives, dosing frequencies, and concentrations). Specimens for CFU determination and drug concentration assays are collected at predetermined time points from the central infection compartment using a sterile syringe and needle through a rubber septum.

In order to maintain the integrity of the plasmid-mediated β-lactamase in each of the challenge strains, cultures were grown overnight on Mueller-Hinton agar (BD Laboratories) containing subinhibitory concentrations of 1 μg/ml ceftriaxone (Sigma-Aldrich, St. Louis, MO). For each study described herein, initial inocula of 1.0 × 10⁶ CFU/ml were prepared by taking colonies from the overnight cultures and growing the inocula to mid-logarithmic phase in a flask of Mueller-Hinton broth set in a water shaker bath (35°C and 125 rotations per minute). The bacterial concentration within the flask of Mueller-Hinton broth was determined using a spectrophotometer and a previously confirmed growth curve for each challenge strain.

Bacteria within the central infection compartment were then exposed to changing free-drug concentrations of piperacillin and tazobactam, based on protein binding estimates of 30% and human half-lives of 1 h for both drugs. One-milliliter samples were collected for CFU determination at 0, 2, 4, 6, 8, 12, and 24 h. Each sample was centrifuged and resuspended twice with sterile normal saline to prevent drug carryover. The samples were then cultured onto Trypticase soy agar enriched with 5% sheep blood and incubated at 35°C for 24 h. One-milliliter samples for drug assay were collected at 0.5, 1, 3, 5, 6.5, 7, 8.5, 11, 12.5, and 24 h and were immediately frozen at −80°C until they were assayed for drug concentration.

Dose-ranging studies.

Dose-ranging studies were conducted over 24 h using each isogenic E. coli strain to evaluate the relationships between the tazobactam concentration and the change from the baseline log₁₀ CFU/ml at 24 h in the presence of piperacillin for each of the strains. In these studies, piperacillin and tazobactam, alone or in combination, were administered as 30-min infusions every 6 h (4 doses over 24 h). Piperacillin was administered as either 2,000 or 4,000 mg, while the tazobactam dose ranged between 31.25 and 1,000 mg.

Dose fractionation studies.

Dose fractionation studies were conducted over 24 h. The tazobactam dosing regimens that were selected for evaluation in the dose fractionation studies were those from the dose-ranging studies that were associated with one-half the maximal effect of the drug combination. The total daily tazobactam exposure, as measured by the free-drug AUC from 0 to 24 h (AUC_0–24), was fractionated into every 6-, 8-, 12-, and 24-h dosing interval. In these studies, piperacillin at 2,000 mg or 4,000 mg was administered using a dosing interval of 6 h (4 doses over 24 h). All dose fractionation studies were performed in duplicate for each of the three isogenic strains.

Analytical method.

All samples were assayed by liquid chromatography followed by tandem mass spectrometry (LC–MS-MS) (MicroConstants, San Diego, CA). Standard curves ranged from 0.5 to 500 μg/ml for piperacillin and from 0.1 to 100 μg/ml for tazobactam. Both standard curves were linear over their respective ranges (r² = 1.00 and 1.00, for piperacillin and tazobactam, respectively). The lower limit of quantification was 0.5 μg/ml for piperacillin and 0.1 μg/ml for tazobactam. The intraday coefficient of variation (CV) for piperacillin ranged from 3.38 to 4.05%, and that for tazobactam ranged from 2.91 to 5.00%. The interday CV for tazobactam ranged from 2.82 to 10.8%, and that for piperacillin ranged from 1.01 to 5.97%.

Pharmacokinetic-pharmacodynamic analysis.

Data from the dose fractionation studies were evaluated using Hill-type models and nonlinear least-squares regression. The data were weighted using the inverse of the estimated measurement variance. Relationships between the change from baseline in log₁₀ CFU/ml at 24 h and the tazobactam AUC, C_max, and %time>threshold were evaluated. The %time>threshold for tazobactam was identified through an iterative process in which candidate concentration thresholds of 0.01, 0.05, 0.1, 0.25, 0.5, 0.75, and 1 μg/ml were evaluated. Concentration threshold discrimination was based on the evaluation of the dispersion of data along the exposure axis and optimization of r² values for the relationship between change from baseline in log₁₀ CFU/ml at 24 h and %time>threshold for tazobactam.

Nucleotide sequence accession numbers.

The nucleotide sequences of bla_CTX-M-15 and upstream promoter regions designed for constructing the plasmid vectors utilized here were previously submitted to the EMBL/GenBank/DDBJ sequence databases and assigned accession numbers KC355190, KC355191, and KC355192.

RESULTS

In vitro susceptibility testing.

The MIC values for piperacillin alone were determined to be 128, >256, and >256 μg/ml for the low-, moderate-, and high-level CTX-M-15-producing E. coli strains, respectively (Table 1). The MIC value for piperacillin in the presence of 4 μg/ml of tazobactam was 2 μg/ml for all three challenge strains. The hydrolytic activity rates, represented as the amount of ceftazidime, a surrogate for piperacillin, hydrolyzed per minute per milligram of protein, were 35, 85, and 560 for the low-, moderate-, and high-level CTX-M-15-producing strains, respectively. The levels of bla_CTX-M-15 transcription detected in the low-, moderate-, and high-level CTX-M-15-producing strains, relative to the E. coli strain that demonstrated the smallest amount of enzyme production (control), were 1-, 8.3-, and 43.9-fold higher, respectively.

TABLE 1.

Susceptibility test results for piperacillin and piperacillin combined with tazobactam, ceftazidime hydrolytic activity rates against E. coli strains producing different levels of CTX-M-15, and transcription levels of bla_CTX-M-15

E. coli straina	MIC value (μg/ml)		Hydrolytic activityb	qRT-PCRc
	Piperacillin alone	Piperacillin + tazobactam (4 μg/ml)
Control	1	1	−3d	ND
Low	128	2	35	1
Moderate	>256	2	85	8.3
High	>256	2	560	43.9

^aLow, moderate and high represent low-, moderate-, and high-level CTX-M-15-producing E. coli strains, respectively.

^bHydrolytic activity rates are expressed as the amount of ceftazidime hydrolyzed per minute per milligram of protein.

^cThe qRT-PCR results represent the transcription of bla_CTX-M-15 relative to that of the E. coli strain demonstrating the lowest CTX-M-15 production based upon MIC results and hydrolysis assays for β-lactams. ND, not detectable.

^dA negative value for the control strain indicates the absence of hydrolytic activity.

Pharmacokinetics.

The targeted piperacillin and tazobactam pharmacokinetic profiles were well simulated within the PK-PD in vitro model (Fig. 1) as evidenced by the agreement between the observed and targeted concentration-time profiles for each drug (r² = 0.984, slope = 1.18, intercept = −1.9 for piperacillin; r² = 0.992, slope = 0.927, and intercept = −0.301 for tazobactam).

FIG 1.

Relationships between the observed and targeted piperacillin (A) and tazobactam (B) concentrations for regimens studied.

Dose-ranging studies.

In each dose-ranging study, the no-treatment growth control grew well, as evidenced by an average bacterial burden of 1.2 × 10⁸ CFU/ml by 24 h. The dosing regimen that employed a low tazobactam exposure (e.g., tazobactam at 31.25 mg every 6 h [q6h]) resulted in a bacterial burden similar to that of the growth control at 24 h, while the dosing regimens that employed high tazobactam exposures (e.g., tazobactam at 125 mg q6h, 250 mg q6h, 500 mg q6h, and 1,000 mg q6h) resulted in reductions from baseline in the bacterial burden of greater than 2 log₁₀ CFU/ml at 24 h.

Upon review of the dose-ranging data, tazobactam dosages were selected for fractionation for each of the three isogenic strains. For the low- and moderate-level CTX-M-15-producing strains, a total daily dose of 1,000 mg of tazobactam was chosen for evaluation. For the high-level CTX-M-15-producing strain, a total daily dose of 2,000 mg of tazobactam was chosen for evaluation. The piperacillin total daily dose was 8 g/day divided evenly every 6 h for the low-level CTX-M-15-producing strain and 16 g/day divided evenly every 6 h for the moderate- and high-level CTX-M-15-producing strains.

Dose fractionation studies.

The results from the dose fractionation studies are presented in Fig. 2 for each of the three isogenic strains. As shown in Fig. 2, the growth control grew well in each experiment, reaching an average bacterial density of 1.79 × 10⁸ CFU/ml at 24 h. As evidenced by a less than 1-log₁₀ CFU/ml decrease from baseline at 24 h, maximal efficacy was not achieved for the high-level CTX-M-15-producing strain. As a result, the experiments for the high-level CTX-M-15-producing strain were repeated with larger tazobactam doses. As shown in Fig. 3, the larger tazobactam doses resulted in maximal effect. Based on these results and the original results for the low- and moderate-level CTX-M-15-producing strains, a >3-log₁₀ CFU/ml decrease in bacterial burden was observed for each of the three isogenic strains over the first 8 h in each of the dose fractionation studies. After the first 8 h, a clear demarcation in bacterial growth was seen among the dosing intervals. For the majority (6/8) of the dose fractionation studies, regardless of the amount of CTX-M-15 enzyme produced, the every-6-h tazobactam dosing regimen resulted in the largest decrease in bacterial burden (>1-log₁₀ CFU/ml reduction from baseline) at 24 h. As shown in Fig. 2A and B and 3, the every-12-h and every-24-h tazobactam dosing regimens performed worse than the every-6-h dosing regimen, regardless of the amount of CTX-M-15 produced. The performance of the every-8-h regimens was less predictable than that of the every-6-, -12-, or -24-h dosing regimen.

FIG 2.

Dose fractionation study results for the low-level (A), moderate-level (B), and high-level (C) CTX-M-15-producing E. coli strains. The effect of each active regimen is shown relative to the no-treatment controls. PIP, piperacillin; TAZ, tazobactam.

FIG 3.

Averaged repeated dose fractionation study results for high-level CTX-M-15-producing E. coli strain. The effect of each active regimen is shown relative to the no-treatment controls.

Pharmacokinetic-pharmacodynamic analysis.

The relationships between change from baseline in log₁₀ CFU/ml at 24 h and the three tazobactam PK-PD indices, AUC, C_max, and %time>threshold, are presented in Fig. 4. The %time>threshold for tazobactam was the PK-PD index that was most closely associated with efficacy (r² = 0.839). The tazobactam concentration threshold changed with the amount of CTX-M-15 transcription and was 0.25, 0.5, and 2 μg/ml for the low-, moderate-, and high-level CTX-M-15-producing strains, respectively (r² = 0.921, 0.773, and 0.875, respectively). The parameter estimates (percent standard error of the estimate [%SEE]) for the relationship between change from baseline in log₁₀ CFU/ml and %time>threshold for tazobactam were as follows: Hill's constant, 1.53 (0.69); maximum effect (E_max), 8.83 (7.41); 50% effective concentration (EC₅₀), 85.6 (88.8); and E₀, 2.39 (0.23). The %time>threshold values for tazobactam necessary for net bacterial stasis and a 1- and 2-log₁₀ CFU/ml decrease from baseline at 24 h did not vary by enzyme production and were 44.9, 62.9, and 84.9%, respectively.

FIG 4.

Relationships between the tazobactam PK-PD indices, AUC, C_max, and %time>threshold, and the change in log₁₀ CFU/ml from baseline at 24 h of isogenic CTX-M-15-producing E. coli in a PK-PD in vitro infection model.

DISCUSSION

The objectives of this study were (i) to identify the PK-PD index that best predicted the efficacy of tazobactam in combination with piperacillin against an isogenic CTX-M-15-producing E. coli triplet set that differed solely in the bla_CTX-M-15 transcription levels; (ii) to determine the magnitude of the tazobactam PK-PD index necessary for efficacy; and (iii) to determine the impact of the presence of a different β-lactam agent, or different bla_CTX-M-15 transcription levels, on the tazobactam PK-PD index necessary for efficacy.

The results of the studies described herein demonstrated that %time>threshold was the PK-PD index that was most closely associated with tazobactam efficacy. The tazobactam concentration threshold increased with increasing bla_CTX-M-15 transcription such that the concentration thresholds were 0.25, 0.5, and 2 μg/ml for the low-, moderate-, and high-level CTX-M-15-producing strains, respectively. Given that the MIC is well correlated with bla_CTX-M-15 transcription, the increase in tazobactam concentration threshold with increasing enzyme transcription is not a surprising finding. The critical implication of this finding is that there is no single target tazobactam concentration threshold across a population of β-lactamase-producing isolates.

The results of this study corroborate those of a previous study we conducted, the objective of which was to identify the PK-PD index associated with the efficacy of tazobactam when used in combination with ceftolozane (10). In that study, we determined that %time>threshold was the PK-PD index most closely associated with the efficacy of tazobactam, in the presence of ceftolozane, against the same isogenic challenge set of CTX-M-15-producing E. coli strains. Also consistent with the earlier investigations, the results described here showed that the tazobactam concentration threshold was dependent upon CTX-M-15 transcription levels, where the tazobactam concentration threshold increased with increasing CTX-M-15 transcription levels. In our previous study, we demonstrated that the magnitudes of the %time>threshold for tazobactam, when given in combination with ceftolozane, associated with net bacterial stasis and 1- and 2-log₁₀ CFU/ml reductions from baseline at 24 h were approximately 35, 50, and 70%, respectively (10). However, the results of the current study demonstrated that the magnitudes of the %time>threshold values for tazobactam, when given in combination with piperacillin, were higher than those for the combination of tazobactam with ceftolozane (%time>threshold values for tazobactam of 45, 63, and 85% were associated with net bacterial stasis and 1- and 2-log₁₀ CFU/ml reductions from baseline at 24 h, respectively) (10). This increase in %time>threshold for tazobactam, when given in combination with piperacillin, is likely due to changes in the binding of the β-lactamase to the combination of piperacillin-tazobactam and the stability of the β-lactamase.

Despite the correlation of results between studies, the current study has some limitations. First, the test strains were laboratory constructs that produced a known single β-lactamase enzyme, the amount of which was a unique controlled variable. We believe that if we had used clinical isolates in our studies, %time>threshold for tazobactam would still have been identified as the PK-PD index most closely associated with the efficacy of tazobactam in combination with piperacillin. However, an increase in the magnitude of the tazobactam concentration threshold would have likely been observed due to the presence of additional resistance determinants. This was observed in the study by VanScoy et al., which showed that %time>threshold for tazobactam was the PK-PD index most closely associated with the efficacy of tazobactam in combination with ceftolozane against clinical isolates with multiple resistance determinants (12). A second limitation is that we simulated the piperacillin-tazobactam concentration-time profiles associated with 30-min infusions. As extended or prolonged infusions of β-lactam antimicrobials have gained wide acceptance in the clinical arena, future studies should include evaluation of the pharmacokinetic profile associated with an extended infusion of piperacillin-tazobactam (4 h) and an evaluation of the impact of infusion time on %time>threshold for tazobactam associated with efficacy. A final limitation of our study was the absence of an evaluation of how the immune system would impact the magnitude of %time>threshold for tazobactam necessary for β-lactamase inhibitor efficacy. It is expected that the results observed in this in vitro model would be enhanced in an immunocompetent host with an intact complement system (13). An in vivo study utilizing an immunocompetent murine infection model may be the most appropriate way to evaluate the impact of the immune system.

In conclusion, using a one-compartment PK-PD in vitro infection model, we were able to confirm that %time>threshold for tazobactam is the PK-PD index that is most closely associated with the efficacy of tazobactam, when given in combination with piperacillin, against selected E. coli strains that differed in the amounts of β-lactamase produced. We also demonstrated that the tazobactam concentration threshold increased with an increasing bla_CTX-M-15 transcription level. These data, which are consistent with the findings of our previous work with ceftolozane-tazobactam, provide essential information that will inform dosing strategies for tazobactam when used in combination with other β-lactam agents.