Determination of the In Vivo Pharmacodynamic Profile of Cefepime against Extended-Spectrum-Beta-Lactamase-Producing Escherichia coli at Various Inocula

Abstract

Cefepime was evaluated in vivo against two inoculum sizes of four strains of Escherichia coli that produced extended-spectrum beta-lactamases (ESBLs) in a murine neutropenic thigh infection model to characterize the pharmacodynamic activity of cefepime in the presence of ESBL-producing bacteria and to evaluate if differences in lengths of cefepime exposure are required with various inocula. Three strains possessed a single enzyme each: TEM-10, TEM-12, and TEM-26. The fourth strain possessed two TEM-derived ESBLs and a third uncharacterized enzyme. Two non-ESBL-producing E. coli strains were included for comparison. Mice received various doses of cefepime to achieve a spectrum of percentages of time the drug was above the MIC (%T>MICs) for each isolate at both inocula. No significant difference in cefepime exposure was required to achieve similar bactericidal effects for ESBL- and non-ESBL-producing isolates when the starting inoculum was 10⁵ CFU of E. coli per thigh. The increased MICs observed in vitro for the ESBL-producing strains at 10⁷ CFU/ml did not predict the amount of exposure required to achieve a comparable level of bactericidal activity in vivo at the corresponding starting inoculum of 10⁷ CFU/thigh. Compared to the cefepime exposure in tests with the lower inoculum (10⁵ CFU/thigh), less exposure was required when the starting inoculum was 10⁷ CFU/thigh (%T>MIC, 6% versus 26%), such that similar doses (in milligrams per kilogram of body weight) produced similar bactericidal effects with both inocula of ESBL-producing isolates. Equivalent exposures of cefepime produced similar effects against the microorganisms regardless of the presence of ESBL production. Pharmacodynamic profiling undertaken with conventional cefepime MIC determinations predicted in vivo microbial outcomes at both inoculum sizes for the ESBL-producing isolates evaluated in this study. These data support the use of conventional MIC determinations in the pharmacodynamic assessment of cefepime.

Extended-spectrum-beta-lactamase (ESBL) production among gram-negative pathogens presents a significant concern for several reasons that include wide substrate specificity and the apparent presence of an inoculum effect (6). All beta-lactam drugs except carbapenems and cephamycins are subject to inactivation by these enzymes (2). Several outcome studies have reported clinical failures when beta-lactams were used to treat serious infections with pathogens that produce ESBLs (9, 11). However, the dosage regimens and drug exposures used are not provided in these reports, and the specific ESBLs produced by the isolates are not well described. In addition, outcome data from studies using many cephalosporins other than ceftazidime are limited due to a relatively small number of well-documented cases.

The inoculum effect that has been associated with these pathogens, i.e., the severalfold increase in MIC observed with a 100-fold increase in inoculum, has been cited as a potential reason for clinical failures in spite of seemingly appropriate treatment for pathogens that appear to be susceptible in vitro. The presence of an ESBL-related drug resistance mechanism affects extended-spectrum cephalosporins to various degrees (2). Certain enzymes are more efficient than others in their ability to hydrolyze particular extended-spectrum cephalosporins. To further complicate the issue, ESBLs may coexist with other resistance determinants and enzymes within an organism capable of inducing various levels of ESBL expression (2). Therefore, it is possible that certain cephalosporins may be clinically effective in the presence of certain ESBLs; however, this possibility remains highly theoretical. Unfortunately, the lack of sufficient outcome data and the lack of obvious markers to predict a pathogen's capacity for hydrolyzing a given cephalosporin have severely limited treatment options for these organisms.

However, a recent outcome study demonstrated favorable clinical responses to non-ceftazidime extended-spectrum cephalosporins for the treatment of pathogens that produced TEM-6 or TEM-12 ESBLs (13). Thus, there is a need to better understand the impact of the presence of ESBLs on the efficacies of various treatment options. The time-dependent pharmacodynamic profile of cephalosporins has been quite reliable in predicting treatment outcomes in various infection models (4) and, to a more limited extent, in clinical practice (5). However, its utility in the presence of ESBL production is less clear due, in part, to the assumption that MICs obtained by standard in vitro testing are unreliable. Thus, one objective of this study was to characterize the pharmacodynamic relationship between cefepime and Escherichia coli isolates that produce ESBLs compared to isolates without ESBLs in a murine thigh infection model to determine if the presence of ESBL enzymes affects the percentage of time the drug is above the MIC (%T>MIC) required for efficacy. In addition, the effectiveness of cefepime was evaluated against an elevated inoculum of 10⁷ CFU/thigh and compared to its effectiveness against an inoculum size of 10⁵ CFU/thigh to determine if the inoculum effect observed in vitro with ESBL-producing isolates correlates with a need for greater cefepime exposure in vivo when a higher inoculum is present.

MATERIALS AND METHODS

Antimicrobial agents.

Standard analytical-grade cefepime (Bristol-Myers Squibb, Princeton, N.J.) was used for all in vitro testing. Cefepime commercial powder for injection (Bristol-Myers Squibb) was purchased for in vivo experiments. Cefepime powder (1-g vials) was reconstituted with 10 ml of normal saline per the manufacturer's instructions, and final concentrations were diluted to achieve the desired dosages prior to each experiment.

Bacterial isolates.

Six E. coli isolates were used throughout the in vivo study. Four of the isolates produced ESBLs. The ESBLs were characterized previously for three of these isolates, which were kindly supplied by J. Quinn (Rush University, Chicago, Ill.). The fourth ESBL-producing isolate was a clinical isolate obtained at Hartford Hospital. Isoelectric-focusing studies of this isolate identified the presence of two TEM enzymes and a third enzyme presumptively identified as an SHV. Of the two comparator E. coli isolates that did not produce ESBLs, one was a clinical isolate and the other was E. coli ATCC 25922. The impact of an inoculum effect on 16 additional isolates (8 ESBL- and 8 non-ESBL-producing strains) was also evaluated by in vitro testing.

In vitro susceptibility testing.

The MICs of cefepime were determined for all isolates with broth microdilution techniques per the NCCLS by using a minimum of three independent tests (10). For each isolate, MICs were determined for inoculum sizes of 10⁵ and 10⁷ CFU/ml.

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 injections of cyclophosphamide (150 mg/kg at 4 days and 100 mg/kg 1 day prior to inoculation) (8). Renal impairment was induced with a single injection of uranyl nitrate (5 mg/kg) 3 days prior to inoculation (1). A suspension of E. coli was prepared from a fresh subculture that had been incubated for less than 20 h and diluted to achieve inocula of 10⁶ and 10⁸ CFU/ml for the standard- and higher-inoculum bacterial-density studies, respectively. Final inoculum concentrations were confirmed by 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 infections for the pharmacokinetic studies. Single doses of cefepime at 25, 75, and 300 mg/kg were administered subcutaneously in 0.2-ml volumes 2 h after thigh inoculation. Blood samples were collected by intracardiac puncture from six mice per time point for a total of eight time points over 12 h. Serum was separated after centrifugation and stored in polypropylene tubes at −80°C until analysis.

The validation of an assay to determine cefepime concentrations in mouse serum was performed according to a previously established method for use with human serum (7). The assay was linear over a range of 0.5 to 50 μg/ml (r² = 0.99). The intraday coefficients of variation for the low (2-μg/ml) and high (40-μg/ml) check samples were 2.99 and 1.70%, respectively. The interday coefficients of variation for the low and high check samples were 4.01 and 2.41%, respectively.

The pharmacokinetic parameters for individual doses, including terminal-phase elimination rate constant, elimination half-life, apparent volume of the central compartment, apparent steady-state volume of distribution, area under the serum drug concentration-time curve, and total-body clearance, 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. The mean parameter estimates were used to simulate a variety of cefepime exposures, expressed as %T>MICs.

Treatment regimens.

Two hours after thigh infection was established as described above, treatment was initiated with subcutaneous doses of cefepime or, for untreated controls, normal saline. Treatment regimens were selected to provide a spectrum of cefepime exposures, expressed as %T>MICs, for each isolate. The %T>MIC was calculated using MICs obtained with both the standard inoculum (10⁵ CFU/thigh) and the 100-fold-higher inoculum (10⁷ CFU/thigh) for the low- and high-inoculum in vivo studies, respectively. Seventy-five groups of at least six mice per group received treatment with cefepime over 24 h. Seventeen groups (six mice per group) of untreated controls received normal saline in the same volume and on the same schedule as for the treatment regimens. Cefepime doses ranged from 0.05 to 1,500 mg/kg/day, divided into one to five doses.

In addition, the evaluation of cefepime treatment was extended to 48 h for an additional six treatment groups. The 48-h regimens were evaluated for three isolates (two ESBL-producing strains and one non-ESBL-producing strain) at both inoculum sizes to assess whether cefepime efficacy was maintained in the presence of ESBL production.

Efficacy as assessed by bacterial density.

Two hours after infection was established and just prior to dosing (0 h), control mice (six per group) were sacrificed. After 24 h of treatment, both the untreated controls and the treated mice were sacrificed. The 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 counting. Efficacy (change in bacterial density) was calculated by subtracting the mean log₁₀ number of CFU per thigh of the control mice sacrificed just prior to treatment initiation (0 h) from the mean log₁₀ number of CFU per thigh of untreated controls and treatment groups at the end of 24 h of therapy.

Assessment of efficacy after 48 h of treatment was undertaken as described above for the 24-h assessment. Efficacy (change in bacterial density) was calculated by subtracting the mean log₁₀ number of CFU per thigh of the control mice sacrificed just prior to treatment initiation (0 h) from the mean log₁₀ number of CFU per thigh of untreated controls and treatment groups at the end of 48 h of therapy.

Data analysis.

The changes in bacterial density in the thigh, expressed as changes in log₁₀ numbers of CFU for both cefepime-treated and untreated control animals, were reported using descriptive statistics. The sigmoid maximum dose-effect (E_max) model using the maximum effective concentration was used to characterize the relationship between the %T>MIC and cefepime efficacy. To compare magnitudes of efficacy among the isolates tested, predicted efficacy at a %T>MIC of 70% was calculated for each strain, as the efficacies of cephalosporins have been shown to be maximized at this point. The %T>MIC required to produce 80% maximum bactericidal efficacy (80% effective dose [ED₈₀]) was used as a reference point to compare the degrees of cefepime exposure required at 10⁷ and 10⁵ CFU/thigh. The efficacies of cefepime for the treatment of isolates with and without ESBL enzymes were compared using Student's t test.

RESULTS

In vitro susceptibility.

Cefepime MICs for the study isolates were determined at two bacterial densities (Table 1). The cefepime MICs for all isolates except one of the ESBL producers, EC 243, at the standard inoculum (10⁵ CFU/ml) were in the range indicating susceptibility. With the exception of E. coli ATCC 25922, MIC determinations using a 100-fold higher inoculum resulted in cefepime MICs in the resistant range for all isolates, including EC 120, the clinical isolate that was not identified as an ESBL producer. The MICs for two of the ESBL-producing isolates, EC 243 and EC 285, at 10⁷ CFU/ml were quite high. Therefore, these isolates were evaluated only in vivo at the standard inoculum size, since the MICs obtained with the 10⁷-CFU/ml inoculum exceeded the serum cefepime concentrations achievable with the dosage regimens used in our model.

TABLE 1.

Characterization of E. coli test isolates and results of in vivo cefepime treatment at the standard and a high inoculum

Isolatea	ESBL producedb	Median MIC (μg/ml)			%T>MIC needed to achieve ED80d		CFU change at a %T>MIC of 70%e
		105 CFU/ml	107 CFU/ml	Ratio (high/standard)	105 CFU/thigh	107 CFU/thigh
ATCC 25922	None	0.06	1	17	23	29	−2.56
EC 120	None	0.5	64	128	23	6	−1.75
EC 242	TEM-12	0.75	32	43	26	6	−1.51
EC 243	TEM-26	256	>1,024	>4	11	NA	−2.26
EC 273	ESBLc	2	256	128	41	4	−1.75
EC 285	TEM-10	4	1,024	256	24	NA	−1.44

^aInternal strain designation.

^bNone, screened negative for ESBL production per NCCLS guidelines.

^cESBL production was confirmed according to NCCLS guidelines, and subsequent isoelectric focusing suggested that this isolate carries two TEM enzymes and possibly an SHV enzyme.

^dP = 0.8 for values for non-ESBL-compared to ESBL-producing isolates. NA, not applicable.

^eP = 0.328 for values for non-ESBL-compared to ESBL-producing isolates.

The large jump in MIC from 0.5 to 64 μg/ml with the 100-fold-higher inoculum for the ESBL-negative isolate EC 120 prompted the in vitro evaluation of several additional isolates to further examine the magnitude of an in vitro inoculum effect among isolates with and without ESBL enzymes. The cefepime MICs for these 16 additional isolates (8 with and 8 without ESBLs) are presented in Table 2. At the higher inoculum, a greater magnitude of increase in MIC was observed for the non-ESBL- than for the ESBL-producing isolates.

TABLE 2.

MICs for additional E. coli isolates at the standard and a high inoculum

ESBL-positive isolatesa				ESBL-negative isolatesb
Isolate	Median MIC (μg/ml)			Isolate	Median MIC (μg/ml)
	105 CFU/ml	107 CFU/ml	Ratio (high/standard)		105 CFU/ml	107 CFU/ml	Ratio (high/standard)
EC 29	16	256	16	EC 54	0.03	2	64
EC 32	16	512	32	EC 56	0.03	2	64
EC 33	4	128	32	EC 109	0.03	1	32
EC 34	128	2,048	16	EC 139	0.015	2	128
EC 35	16	512	32	EC 63	0.125	0.125	1
EC 38	1	16	16	EC 65	0.5	64	128
EC 39	8	128	16	EC 70	0.06	2	32
EC 40	8	256	32	EC 186	0.5	256	512

^aPreviously characterized isolates containing TEM- or SHV-derived ESBLs.

^bScreened negative for ESBL production per NCCLS guidelines.

Pharmacokinetics.

Cefepime displayed linear pharmacokinetics over the range of doses tested. A one-compartment model was used to characterize the cefepime concentration-versus-time profile of mice. Mean (percent coefficient of variation) peak cefepime concentrations of 64.2 (9.9%), 175.7 (6.4%), and 510.5 (9.4%) μg/ml were obtained after single doses of 25, 75, and 300 mg/kg, respectively, were administered. Similar rates of elimination were observed for these doses, with values of 0.56, 0.64, and 0.46 h⁻¹, respectively. Mean parameter estimates were calculated from the data measuring serum drug concentration versus time. These mean estimates were utilized to simulate the cefepime regimens used throughout the study. The serum drug concentrations observed as well as the profiles predicted from the mean parameter estimates for the three doses tested are presented in Fig. 1.

FIG. 1.

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

Efficacy as assessed by bacterial density (standard inoculum).

The number of E. coli organisms recovered from the thighs of the infected animals just prior to the initiation of treatment (0 h) ranged from 5.28 to 5.57 log₁₀ CFU/thigh (mean, 5.41 ± 0.11 log₁₀ CFU/thigh). Growth in the untreated controls over 24 h ranged from 0.06 to 3.77 log₁₀ CFU/thigh (mean, 1.88 ± 1.54 log₁₀ CFU/thigh). None of the animals infected with E. coli ATCC 25922 in the untreated control groups survived the 24-h study period. Therefore, a group of untreated animals infected with this isolate was sacrificed at 12 h to confirm the persistence of infection in the absence of treatment and to characterize the growth potential of this isolate. The mean increase in bacterial density for this isolate at 12 h was 3.12 ± 0.09 log₁₀ CFU/thigh.

The dose-response relationships for ESBL- and non-ESBL-producing isolates are presented in Fig. 2. The degrees of cefepime exposure (%T>MIC) required to achieve the ED₈₀ were compared between ESBL- and non-ESBL-producing isolates to see if the presence of these enzymes was associated with greater drug exposure requirements. We noted no significant differences in exposure requirements to achieve the ED₈₀ among the isolates studied (P = 0.80). In addition, using a %T>MIC of 70% as a reference point for comparison, no significant difference in predicted magnitude of kill was observed between ESBL- and non-ESBL-producing E. coli isolates (P = 0.33). These data are presented in Table 1.

FIG. 2.

Relationship between the %T>MIC and the change in bacterial density following infection with non-ESBL-producing (A) and ESBL-producing (B) E. coli isolates at the standard inoculum size (10⁵ CFU/mouse thigh).

Efficacy as assessed by bacterial density (higher inoculum).

A consistent recovery of bacterial density within 7 log₁₀ CFU/thigh at the initiation of treatment (0 h) was achieved with the subset of four E. coli isolates evaluated in this portion of the study (mean, 7.47 ± 0.3 log₁₀ CFU/thigh). Bacterial density increased by a mean of 1.61 ± 1.1 log₁₀ CFU/thigh in untreated controls over 24 h. Again, for E. coli ATCC 25922, none of the animals in the untreated control groups survived the 24-h study period. Therefore, a group of untreated animals sacrificed at 6 h demonstrated an increase of 0.82 ± 0.69 log₁₀ CFU of this isolate. Figures 3 and 4 depict the relationship between the degree of cefepime exposure and the change in bacterial density at the 100-fold-higher inoculum compared to the bacterial densities obtained at the lower inoculum (10⁵ CFU/thigh) for non-ESBL- and ESBL-producing isolates, respectively. For ATCC 25922, similar values for %T>MIC, and therefore higher milligram-per-kilogram doses, were required to achieve the ED₈₀ at the higher inoculum, as predicted by in vitro susceptibility tests. However, for the ESBL-producing isolates and the non-ESBL-producing clinical isolate a smaller value for %T>MIC was required to achieve the ED₈₀ at the higher inoculum (Table 1). Interestingly, the total milligram-per-kilogram doses that corresponded to the exposures needed to achieve the ED₈₀ at both inoculum sizes were similar for the ESBL-producing isolates. In light of this finding, we administered two of the same cefepime doses that were previously used in the standard-inoculum in vivo tests to mice infected with 10⁷ CFU for EC 285/thigh. The cefepime MIC for this isolate at 10⁷ CFU/ml was 1,024 μg/ml, largely in excess of achievable peak drug concentrations for this study. Therefore, the %T>MIC achieved with these doses at 10⁷ CFU/thigh was 0. However, 5 mg of cefepime/kg every 24 h reduced bacterial density by 0.71 and 0.37 log₁₀ CFU, and 75 mg/kg every 12 h reduced bacterial density by 1.49 and 1.48 log₁₀ CFU with the standard and 100-fold-higher inocula, respectively (Fig. 5).

FIG. 3.

Relationship between the %T>MIC and the change in bacterial density following infection with non-ESBL-producing E. coli strains ATCC 25922 and EC 120 at the standard inoculum (A) and a 100-fold-higher inoculum (B).

FIG. 4.

Relationship between the %T>MIC and the change in bacterial density following infection with ESBL-producing E. coli strains EC 242 and EC 273 at the standard inoculum (A) and a 100-fold-higher inoculum (B).

FIG. 5.

The same milligram-per-kilogram doses produce similar bactericidal effects at two different inoculum sizes for ESBL-producing E. coli EC 285.

In addition, changes in bacterial density were evaluated at 48 h for three isolates to see if the results obtained at 24 h would be maintained. Reductions in bacterial density of equal or slightly greater magnitudes were maintained for all isolates (ATCC 25922, EC 242, and EC 273) at both inoculum sizes.

DISCUSSION

Controversy still exists as to the appropriateness of advanced-generation cephalosporins and certain beta-lactam and beta-lactamase inhibitor combinations (i.e., piperacillin-tazobactam) in the treatment of infections due to ESBL-producing bacteria. Pharmacodynamics can serve as a useful tool for understanding the significance of in vivo ESBL production on treatment options. To compare the pharmacodynamic profiles of isolates with and without ESBLs, the degree of exposure required to achieve 80% maximum effect and the magnitudes of bacterial kill achieved at similar drug exposures were used as reference points. Pharmacodynamic studies with cephalosporins have demonstrated that maximum bactericidal activity occurs when concentrations exceed the MIC for 60 to 70% of the dosing interval for gram-negative pathogens (3). Therefore, we chose to use a %T>MIC of 70% as a reference point to compare magnitudes of effect. Based on these surrogate markers, the time-dependent pharmacodynamic profile of cefepime against ESBL-producing isolates was similar to the profile obtained with non-ESBL-producing isolates in standard-inoculum in vivo tests when exposures were normalized to the MIC. For isolates such as EC 243, for which the cefepime MIC is 256 μg/ml, normalization of cefepime exposures to this MIC results in concentrations far in excess of those achieved clinically. In spite of its seeming clinical irrelevance, this methodology was employed for all isolates to evaluate the predictive value of pharmacodynamics for isolates capable of producing ESBLs, which to our knowledge has not previously been established.

It is noteworthy that the ESBL-producing isolates appeared quite different from ATCC 25922 with regard to predicted efficacy at a %T>MIC of 70%. The fact that no significant difference was observed between ESBL- and non-ESBL-producing isolates collectively may be due to differences in efficacy between the non-ESBL-producing comparator isolates themselves. A larger sample size of isolates would clearly provide a better representation of cefepime activity among clinical isolates without ESBLs for which the cefepime MICs varied. However, the similarity between EC 120 and the ESBL-producing isolates indicates that the results observed with the ESBL-producing isolates are not based on ESBL production in and of itself.

A significant concern in regard to ESBL-producing isolates is that they appear to affect extended-spectrum cephalosporins to various degrees in vitro at high inocula. In our study we found a significant increase in cefepime MICs for increased inocula of nine E. coli isolates that did not produce ESBLs. Furthermore, the magnitude of an in vitro inoculum effect appeared greater among ESBL-negative than among ESBL-positive E. coli isolates. However, MICs for the majority of the ESBL-negative isolates at 10⁵ CFU/ml were quite low and remained in the range indicating susceptibility when the inoculum was 10⁷ CFU/ml, regardless of the magnitude of the increase. More important, for both ESBL-positive and -negative pathogens, the varied increases in MICs in vitro at higher inocula may not be predictive of in vivo potency for all cephalosporins.

Previously, cefepime was effective in a rat abscess model in which the animals were infected with a high inoculum (>10⁷ CFU) of Klebsiella pneumoniae that harbored the TEM-26 enzyme (12). Cefepime MICs were in the range indicating susceptibility when the inoculum was 10⁵ CFU/ml but resistant when the inoculum was 10⁷ CFU/ml. Using simulated human exposures, the authors noted that the greater efficacy observed with cefepime than with the other beta-lactams tested was surprising since cefepime appeared least potent in vitro against an inoculum of 10⁷ CFU/ml in comparison to the other agents. However, efficacy seems to be explained in that study by MICs obtained for an inoculum of 10⁵ CFU/ml, against which cefepime was most potent in comparison. Another study that evaluated K. pneumoniae with various TEM-derived ESBLs noted that static doses of several cephalosporins were similar at low and high inocula (D. Andes, G. Jacoby, and W. A. Craig, Abstr. 35th Intersci. Conf. Antimicrob. Agents Chemother., abstr. A62, 1995).

Our study supports these findings with several clinical E. coli isolates that harbor various TEM-derived ESBL enzymes. We evaluated equivalent cefepime exposures (adjusted for MICs obtained with both inoculum sizes) in vivo at both the standard inoculum and a 100-fold-higher inoculum with the presumption that similar degrees of exposure, and thus higher doses, would be required to provide similar bactericidal effects at higher inocula. Interestingly, this pattern was evident only for E. coli ATCC 25922. In contrast, the dose-response curves for the ESBL-producing isolates indicate that similar milligram-per-kilogram doses would achieve similar reductions in bacterial density with both inocula. Administration of the same milligram-per-kilogram doses used in standard-inoculum in vivo tests to animals infected with 10⁷ CFU of EC 285/thigh further supports this prediction, as the same doses produced similar reductions in bacterial density at both the high and low inocula. That these results were obtained from the high-inoculum infection in spite of a %T>MIC of 0 calls into question the validity of a presumed in vivo inoculum effect for cefepime against E. coli isolates with the ESBLs evaluated in this study.

It is possible that the presence of other ESBL enzymes, such as SHV-derived enzymes, the combination of various other resistance mechanisms, and different species (i.e., Klebsiella spp.) may produce different results. The number of isolates evaluated in this study was small and not representative of all ESBL- or non-ESBL-producing E. coli isolates. However, the similarities noted between these resistant pathogens and the non-ESBL-producing isolate EC 120 seem to indicate that the presence of ESBLs is not an independent predictor of efficacy. Rather, the cefepime MIC obtained with standard susceptibility tests was more predictive of in vivo outcome, regardless of the inoculum size. These data support the use of conventional MIC determinations in the pharmacodynamic assessment of cefepime.