Pharmacodynamic Profile of Ertapenem against Klebsiella pneumoniae and Escherichia coli in a Murine Thigh Model

Abstract

The pharmacodynamic profile of ertapenem was evaluated in a neutropenic mouse thigh infection model. Extended-spectrum beta-lactamase (ESBL)-positive and ESBL-negative clinical strains of Escherichia coli and Klebsiella pneumoniae were studied. MICs ranged from 0.0078 to 0.06 μg/ml with standard inoculum tests. Ertapenem doses were administered once to five times daily to achieve various exposures, reported as the percentage of the dosing interval that the concentration of free ertapenem was in excess of the MIC (%T>MIC_free). Mean values for the static exposure and 80% maximally effective exposure (ED₈₀) were 19% (range, 2 to 38%) and 33% (range, 13 to 65%) T>MIC_free, respectively. Differences in exposure requirements based on the presence of an ESBL resistance mechanism or bacterial species were not evident. In addition, experiments using a 100-fold higher inoculum did not decrease the magnitude of the reduction in bacterial density from baseline achieved compared to lower-inoculum studies. The pharmacodynamic parameter of %T>MIC_free correlated well with bactericidal activity for all isolates, and the static and ED₈₀ exposures are consistent with those reported previously for carbapenems.

Ertapenem is presently approved in the United States to treat complicated intra-abdominal infections, complicated skin and skin structure infections, community-acquired pneumonia, complicated urinary tract infections, and acute pelvic infections (Invanz package insert; Merck & Co., Inc, Whitehouse Station, N.J.). These infections are often characterized as polymicrobial in nature. In addition, convincing data have associated inadequate empirical therapy with increased failure rates and increased mortality for many of these infections (8). Therefore, initiation of treatment with a potent, broad-spectrum agent is essential.

Ertapenem has demonstrated considerable in vitro potency against a wide range of pathogenic gram-positive and gram-negative organisms, excluding nonfermentating organisms. In addition, the in vivo activity of ertapenem against Streptococcus pneumoniae has previously been described (10). However, many fewer data exist to describe the in vivo pharmacodynamics of this agent in gram-negative infections. The objective of this study was to evaluate the killing activity of ertapenem against Escherichia coli and Klebsiella pneumoniae clinical isolates, two of the most commonly recovered pathogens in the above-mentioned infections, in a neutropenic thigh infection model. Four isolates were extended-spectrum beta-lactamase (ESBL) producers. We focused the analysis on the percentage of the dosing interval that the concentration of free ertapenem was in excess of the MIC (%T>MIC), since this parameter has been repeatedly identified as the parameter correlating best with carbapenem efficacy (9). In vivo efficacy at a high inoculum (10⁷ CFU/ml) was additionally evaluated in a subset of isolates to assess the impact of a larger inoculum size on the pharmacodynamic characteristics of ertapenem.

MATERIALS AND METHODS

Antimicrobial agents.

Ertapenem analytical-grade powder (Merck Research Laboratory, Rahway, N.J.) was used for all in vitro and in vivo testing. Ertapenem was reconstituted with 10 mM 3-(N-morpholino)propanesulfonic acid (MOPS) as per the manufacturer's specifications for all in vitro testing. For animal dosing, an ertapenem solution was prepared with sterile water for injection.

Bacterial isolates and susceptibilities.

Eight clinical isolates, four E. coli and four K. pneumoniae, were used throughout the study. Two each of the E. coli and K. pneumonia isolates were well-characterized ESBL producers and were kindly supplied by J. Quinn (Rush University, Chicago, Ill.). The ertapenem MICs for all isolates were determined by broth microdilution as per the NCCLS with a minimum of three independent tests (7).

Thigh infection model.

Specific-pathogen-free female ICR mice weighing approximately 25 g were obtained from Harlan Sprague Dawley, Inc. (Indianapolis, Ind.). The animals were maintained and utilized in accordance with National Research Council recommendations and were provided food and water ad libitum.

Mice were rendered neutropenic by intraperitoneal injection of cyclophosphamide at 150 mg/kg of body weight at 4 days and 100 mg/kg at 1 day prior to inoculation (3). Renal impairment was induced with a single injection of uranyl nitrate at 5 mg/kg 3 days prior to inoculation (1). A suspension of each test isolate was prepared from a fresh subculture that had been incubated for less than 20 h and diluted to achieve an inoculum of 10⁶ or 10⁸ CFU/ml. Final inoculum concentrations were confirmed with serial dilution and plating techniques. Thigh infection was produced by injection of 0.1 ml of the prepared inoculum into each mouse thigh 2 h prior to antimicrobial therapy.

Pharmacokinetic studies and dosage regimen determination.

Mice were prepared as described above for the thigh infection model. E. coli ATCC 25922 was used to produce thigh infection for the pharmacokinetic studies. Single doses of ertapenem at 10, 40, and 100 mg/kg in 0.2-ml volumes were administered subcutaneously 2 h after thigh inoculation. Blood samples were collected by intracardiac puncture from six mice per time point for a total of six to eight time points per regimen over 6 h. Serum samples were separated after centrifugation and transferred into polypropylene tubes containing an equal volume of 0.1 M morpholineethanesulfonic acid (MES)-ethylene glycol (1:1, vol/vol) as the stabilizing agent. Samples were stored at −80°C until being assayed.

Ertapenem concentrations in murine serum samples were determined using a previously validated high-performance liquid chromatography assay (10). The assay was linear over a range of 0.2 to 50 μg/ml (r² = 0.99). Intraday coefficients of variation for the low (1 μg/ml) and high (40 μg/ml) quality control samples were 6.06 and 6.47%, respectively. Interday coefficients of variation for the quality control samples were 5.37 and 3.43%, respectively.

Pharmacokinetic parameters for individual doses were calculated with first-order elimination by a nonlinear least-squares technique (PCNONLIN version 4.2; Statistical Consultants, Lexington, Ky.). Compartmental selection was based on visual inspection of the fit and use of the correlation between the observed and calculated concentrations. Mean pharmacokinetic parameters were calculated from the individual parameter estimates. PCNONLIN was used to simulate a variety of ertapenem exposures expressed as %T>MIC based on the mean parameter estimates.

Protein binding studies previously conducted using infected mice of the same species resulted in a value for %T>MIC_free of 95.5% over the range of doses employed (10). Therefore, efficacy in this study was correlated to ertapenem exposure expressed as %T>MIC_free.

Treatment regimens were selected to span a range of ertapenem exposures as %T>MIC_free. Animals were administered subcutaneous 0.2-ml doses of ertapenem or sterile water as a control beginning 2 h after inoculation of thighs. Seventy-eight groups of at least three mice per group received treatment with ertapenem over 24 h. Seventeen groups of at least three mice per group received sterile water in the same volume and on the same schedule as the treatment groups. Ertapenem treatment ranged from 0.3 to 375 mg/kg per day as one to five divided doses.

Efficacy as assessed by bacterial density.

Two hours after infection was established and simultaneous to initiation of dosing (0 h), control mice (three per group) were sacrificed. After 24 h of treatment, untreated controls and treatment groups were sacrificed. Animals were euthanized by CO₂ inhalation followed by cervical dislocation. Both thighs were removed and individually homogenized in 5 ml of normal saline. Serial dilutions were plated on Trypticase soy agar with 5% sheep blood for CFU determinations. Efficacy (change in bacterial density) was calculated by subtracting the mean log₁₀ CFU per thigh of the 0-h control mice sacrificed just prior to dosing from the mean log₁₀ CFU per thigh of the untreated controls and treatment groups at the end of 24 h of therapy.

Data analysis.

The change in the bacterial density of the thigh for ertapenem-treated and placebo-treated control animals is reported using descriptive statistics. An inhibitory sigmoid E_max dose-effect model derived from the Hill equation was used to characterize the relationship between %T>MIC_free and ertapenem efficacy. Exposures were calculated from MICs obtained in standard (10⁵-CFU/ml) inoculum tests. The exposures that resulted in bacteriostasis and in 80% maximal efficacy were used as reference points to compare the results obtained among isolates. The Student's t test was used to compare differences in ertapenem efficacy between E. coli and K. pneumoniae species as well as between ESBL-negative and ESBL-positive isolates.

RESULTS

In vitro susceptibility.

Ertapenem MICs for the eight study isolates are shown in Table 1. Several isolates were tested at both a standard inoculum and a 100-fold higher (10⁷-CFU/ml) inoculum. An in vitro inoculum effect was observed for all isolates, regardless of the presence of an ESBL genetic determinant. However, ertapenem MICs for all ESBL-positive and -negative isolates remained in the susceptible range when tested at an elevated inoculum, consistent with previous studies (5). These MICs represent a relatively wide range of susceptibilities for ertapenem, thus enabling pharmacodynamic evaluation over a wide range of exposures.

TABLE 1.

Characterization of E. coli and K. pneumoniae test isolates and results of in vivo ertapenem treatment at standard inoculum

Isolatea	ESBL(s) produced	MIC at standard inoculum/MIC at high inoculum (μg/ml)c	Static exposure (%T>MICfree)d	ED80 (%T>MICfree)e
EC 54	Noneb	0.0078/0.015	38	65
EC 120	Noneb	0.015/0.5	12	15
EC 243	TEM-26	0.06/0.125	5	50
EC 285	TEM-10	0.03/0.125	2	13
KP 134	Noneb	0.015/NT	15	19
KP 135	Noneb	0.015/NT	19	22
KP 227	TEM-10	0.06/0.25	25	38
KP 230	TEM-26, unspecified SHV	0.06/NT	37	46

^aInternal strain designations.

^bScreened negative for ESBL production as per NCCLS guidelines.

^cStandard inoculum, 10⁵ CFU/ml; high inoculum, 10⁷ CFU/ml. NT, not tested.

^dP was 0.344 for E. coli versus K. pneumoniae isolates and 0.725 for non-ESBL-producing versus ESBL-producing isolates.

^eP was 0.766 for E. coli versus K. pneumoniae isolates and 0.666 for non-ESBL-producing versus ESBL-producing isolates.

Pharmacokinetics.

Ertapenem pharmacokinetics were linear over the range of doses tested. The concentration-versus-time profile of total ertapenem concentrations in mouse serum samples was characterized with a one-compartment model. Mean peak ertapenem concentrations of 33.89 (coefficient of variation [CV], 3.56%), 77.71 (CV, 14.52%), and 161.02 (CV, 6.00%) μg/ml were obtained after single doses of 10, 40, and 100 mg/kg, respectively. Similar rates of elimination were observed for these doses, with values of 1.29, 1.12, and 1.05 h⁻¹ for each dose. Mean parameter estimates were calculated from the individual pharmacokinetic profiles. These mean estimates were utilized to simulate the ertapenem regimens used throughout the study. The observed serum concentration data correlate well with the profiles predicted from the mean parameter estimates for the three doses tested and are presented in Fig. 1.

FIG. 1.

Pharmacokinetic profile of total ertapenem concentrations in mice. Symbols represent actual mean data points for each regimen; solid and dashed lines represent predicted profiles for each regimen.

Efficacy as assessed by bacterial density.

Excellent recovery of bacteria was obtained within infected thighs for both E. coli and K. pneumoniae strains. At initiation of treatment (0 h), bacterial densities ranged from 5.19 to 6.09 (mean, 5.56 ± 0.30) and 5.40 to 5.77 (mean, 5.59 ± 0.13) log₁₀ CFU/thigh for E. coli and K. pneumoniae isolates, respectively. The mean increase in bacterial density over 24 h in untreated controls was 1.93 ± 1.67 and 2.76 ± 0.59 log₁₀ CFU/thigh for E. coli and K. pneumoniae isolates, respectively.

Dose-response relationships for E. coli and K. pneumoniae isolates are presented in Fig. 2 and 3, respectively. As expected, excellent correlations between %T>MIC_free and efficacy were observed for all species (r² range, 0.84 to 0.98). Treatment efficacy in terms of static exposure, defined as the exposure that results in no net change in bacterial density after 24 h of treatment from the starting inoculum, and exposure that produced 80% maximum effect (ED₈₀) were compared among the eight isolates tested. Static exposures ranged from 2 to 38% T>MIC_free, with a mean value of 19%. The mean ertapenem exposure required for 80% maximum efficacy was 33% T>MIC_free (range, 13 to 65%) as shown in Table 1. No significant difference in exposure was required to achieve either a bacteriostatic effect or the ED₈₀ between E. coli and K. pneumoniae isolates (P values of 0.344 and 0.766, respectively). In addition, there were no differences in exposure requirements between ESBL-negative and ESBL-positive isolates with regard to static exposure or ED₈₀ (P values of 0.725 and 0.666, respectively). These data are presented in Table 1. Free ertapenem concentrations in excess of the MIC for 36.3% ± 16.6 of the dosing interval produced significant reductions in bacterial density (≥1 log CFU/thigh from the start of therapy).

FIG. 2.

Relationship between %T>MIC_free of ertapenem and change in bacterial density following infection with non-ESBL-producing (A) and ESBL-producing (B) E. coli at standard inoculum.

FIG. 3.

Relationship between %T>MIC_free of ertapenem and change in bacterial density following infection with non-ESBL-producing (A) and ESBL-producing (B) K. pneumoniae at standard inoculum.

Efficacy as assessed by bacterial density at an increased inoculum.

Ertapenem treatment for two E. coli isolates, EC 120 (ESBL negative) and EC 285 (ESBL positive), was evaluated using a 100-fold higher inoculum for comparison with standard inoculum studies. Consistent recovery of bacterial density within 7 (mean, 7.47 ± 0.21) log₁₀ CFU/thigh at initiation of treatment (0 h) was achieved for the E. coli isolates evaluated in this portion of the study. None of the animals in the untreated control group infected with EC 120 survived after 24 h, consistent with previous work in an identical model where zero or only one of three animals receiving the control treatment survived 24 h postinoculation at 10⁷ CFU/ml with EC 120. Since available data for surviving animals infected with 10⁷ CFU of this isolate/ml have revealed a mean increase of 2.21 ± 0.03 log₁₀ CFU/thigh in untreated controls at 24 h, this value will be applied to estimate the growth potential and characterize the curve under identical conditions in this study. In contrast, a decrease of 0.27 ± 0.11 log₁₀ CFU/thigh occurred after 24 h in untreated controls infected with EC 285, again consistent with previous studies using this isolate (D. Maglio, C. Ong, M. A. Banevicius, Q. Geng, C. H. Nightingale, and D. P. Nicolau, Abstr. 43rd Intersci. Conf. Antimicrob. Agents Chemother., abstr. A1317, 2003). Treatment with ertapenem regimens of which the dosing and schedule were identical to those used in the standard inoculum tests for both isolates was evaluated. Figure 4 depicts the relationship between ertapenem exposure and change in bacterial density at a 100-fold higher inoculum compared to the data obtained with the 10⁵ inoculum for the ESBL-negative (Fig. 4A) and ESBL-positive (Fig. 4B) strains of E. coli tested. Exposure is represented as total daily dose. In both cases, the same doses (in mg/kg) produced similar or greater reductions in bacterial density for the higher-inoculum studies in spite of higher MICs.

FIG. 4.

Relationship between total daily dose of ertapenem and change in bacterial density following infection with ESBL-negative E. coli 120 (A) and ESBL-positive E. coli 285 (B) at 10⁵ and 10⁷ CFU/ml.

DISCUSSION

Knowledge of the relationship between drug exposure and bactericidal effect is increasingly utilized in the rational design of treatment regimens and prediction of antimicrobial success or failure. The objective of this study was to quantify this relationship for ertapenem and two of the most commonly isolated pathogens which this drug is intended to treat. To characterize the dose effects of ertapenem, we used the well-established E_max model and found excellent correlation between bactericidal effect and exposure represented as %T>MIC_free. Two points of reference calculated from this model allowed comparison between groups of isolates.

Previous studies have identified a target exposure for carbapenems of 20 to 30% T>MIC that is required to achieve bacterial stasis and optimize outcome (R. Walker, D. Andes, J. Conklin, S. Ebert, and W. Craig, Abstr. 34th Intersci. Conf. Antimicrob. Agents Chemother., abstr. A91, 1994; W. Craig, S. Ebert, and Y. Watanabe, Abstr. 33rd Intersci. Conf. Antimicrob. Agents Chemother., abstr. 86, 1993). A wide range of static and 80% maximally effective ertapenem exposures were observed among the eight isolates evaluated in this study. These studies were conducted using two bacterial species with and without ESBL resistance to allow pharmacodynamic profiling of a somewhat heterogeneous population. Thus, the range of exposures noted highlights the variable characteristics for each unique drug and pathogen combination. However, the mean static exposure of 19% T>MIC_free is broadly in agreement with those data reported previously for carbapenems. Growth in untreated controls for two of the isolates, EC 243 and EC 285, was less than that observed for the other strains (i.e., less than 1 log CFU). Thus, the poor growth characteristics may explain the comparatively low static exposures calculated for both of these strains. These differences in growth characteristics did not appear to have any effect on ED₈₀. Moreover, it is important to note that differences in exposure requirements to achieve stasis or the ED₈₀ did not appear to result from the presence of an ESBL resistance mechanism. This finding is consistent with previous in vivo studies conducted with a number of E. coli treated with cefepime (Maglio et al., 43rd ICAAC, abstr. A1317). Nor were any differences on the basis of bacterial species noted.

Recent work by our group, referenced above, also revealed that increased MICs obtained with 100-fold higher inocula did not necessitate increased drug exposure to produce similar bactericidal effects for ESBL-positive isolates treated with the beta-lactam cefepime. Accordingly, it did not seem likely that increased exposures of ertapenem would be necessary to produce similar bacterial killing for ESBL-positive isolates at higher inocula, particularly since in vitro decreases in susceptibility were modest and ertapenem MICs remained within the susceptible range. These results are consistent with previous observations showing that ertapenem MICs remained at or below 1 μg/ml for ESBL-producing E. coli and K. pneumoniae with high-inoculum tests (5). Our results for a subset of two E. coli isolates (with and without ESBL) tested at the 10⁷-CFU/ml inoculum do support this prediction, as the same doses (in mg/kg) of ertapenem produced similar bactericidal activity or activity enhanced by as much as 1 log₁₀ CFU/ml in higher-inoculum tests. The inoculum size of 10⁷ CFU/ml was chosen because apparent manifestation of ESBL-mediated (or perhaps other) resistance has been observed at this magnitude of increase in in vitro tests with many beta-lactam agents. As inoculum sizes of even higher orders have required increased drug exposures for other drug and pathogen combinations, studies designed to evaluate ertapenem efficacy beyond an inoculum of 10⁷ CFU/ml would be of interest (4).

A previous study has demonstrated substantial in vivo bactericidal activity with ertapenem for various gram-negative pathogens, although not in the context of clinically used exposures (2). Ertapenem is approved for use in the United States at a dose of 1 g daily, administered intravenously or intramuscularly. The calculated exposure achieved with this presently approved regimen results in a %T>MIC_free value of 49.7 to 95.9% for the sample of isolates used in this study (9). This range of exposures is well in excess of the static exposure found in this study for the eight isolates tested, including the ESBL-producing strains. These exposures also exceed those required to achieve 80% of the maximum effect. Moreover, the eight isolates used in this study reflect a relatively wide range of MICs for ertapenem against gram-negative pathogens. These results are in agreement with the substantial bactericidal activity of ertapenem against ESBL-positive strains of K. pneumoniae in an in vitro time-kill model using steady-state free concentrations simulated in humans (D. Burgess and R. Hall II, Abstr. Annu. Mtg. Am. Coll. Clin. Pharm., abstr. 128, 2003). Current ertapenem MIC₉₀ of ≤0.03 and ≤0.06 have been reported for E. coli and K. pneumoniae, respectively (6). In addition, MIC₉₀ of ≤0.25 have recently been reported for ESBL-positive and AmpC-hyperproducing strains of Enterobacteriaceae (E. Cercenado, O. Cuevas, V. Garcia-Arias, M. Sanchez-Somolinos, J. Guinea, and E. Bouza, Abstr. 43rd Intersci. Conf. Antimicrob. Agents Chemother., abstr. E-2007, 2003). These MICs fall within the range of MICs represented in this study, and therefore, a similar %T>MIC value for ertapenem in serum could be predicted with these values. At present, the pharmacodynamic data derived from the present study support the successful use of ertapenem for infection due to E. coli and K. pneumoniae regardless of the potential for ESBL-containing genotypes.