Comparative Pharmacodynamics and Antimutant Potentials of Doripenem and Imipenem with Ciprofloxacin-Resistant Pseudomonas aeruginosa in an In Vitro Model

Abstract

To compare the antipseudomonal efficacy of doripenem and imipenem as well as their abilities to restrict the enrichment of resistant Pseudomonas aeruginosa, multiple-dosing regimens of each drug were simulated at comparable values of the cumulative percentages of a 24-h period that the drug concentration exceeds the MIC under steady-state pharmacokinetic conditions (T_>MIC) and ratios of the 24-hour area under the curve (AUC₂₄) to the MIC. Three clinical isolates of ciprofloxacin-resistant P. aeruginosa (MIC of doripenem, 1 μg/ml; MICs of imipenem, 1, 2, and 2 μg/ml) were exposed to thrice-daily doripenem or imipenem for 3 days at AUC₂₄/MIC ratios of from 50 to 170 h (doripenem) and from 30 to 140 h (imipenem). The antimicrobial effects for susceptible and resistant subpopulations of bacteria were expressed by the areas between control growth and time-kill curves (I_Es) and areas under the bacterial mutant concentration curves (AUBC_Ms), respectively. With each antibiotic, the I_E and AUBC_M versus log AUC₂₄/MIC relationships were bacterial strain independent. At similar AUC₂₄/MIC ratios, doripenem was slightly less efficient than imipenem against susceptible and resistant subpopulations of bacteria. However, doripenem appeared to be somewhat more efficient than imipenem at clinically achievable AUC₂₄s related to the means of the MICs for the three studied strains and had higher antimutant potentials for two of the three strains.

INTRODUCTION

Pseudomonas aeruginosa remains an important cause of nosocomial infections of the lung, urinary tract, and bloodstream, especially within intensive care units (13, 19). This organism is particularly problematic for critically ill, neutropenic, immunocompromised, and mechanically ventilated patients (21). Over the past decade, fluoroquinolone resistance among these organisms has increased, often requiring other antibiotics for successful treatment (20). Recent studies have suggested that beta-lactam monotherapy for serious pseudomonal infections is probably as effective as combination therapy, even though in vitro synergism can sometimes be demonstrated with the combinations (4, 15).

Doripenem is a new carbapenem antibiotic active against Pseudomonas aeruginosa. The pharmacodynamics of doripenem have been studied in vivo using a murine thigh model of infection with 10 Gram-negative organisms (1) and a rabbit model of pseudomonal pneumonia (3). A species-independent relationship between changes in the logarithm of the number of CFU of Gram-negative organisms per thigh and the cumulative percentages of a 24-h period that the drug concentration exceeds the MIC under steady-state pharmacokinetic conditions (T_>MIC) was established in the former study, and similar efficacies of doripenem, imipenem, and meropenem were reported in the latter. T_>MIC relationships of doripenem and imipenem antipseudomonal effects were also observed with each of three isogenic strains in an in vitro study using a dynamic model (14). Because the simulated ranges of the T_>MICs and the ratios of carbapenem dose (D) to the MIC overlapped only minimally, the antipseudomonal effects of doripenem and imipenem could not be compared under identical conditions. To make possible such a comparison, we studied the pharmacodynamics of doripenem and imipenem with ciprofloxacin-resistant strains of P. aeruginosa while simulating the pharmacokinetics at comparable T_>MICs and ratios of the 24-hour area under the curve (AUC₂₄) to the MIC.

(This study was partly presented at the 50th Interscience Conference on Antimicrobial Agents and Chemotherapy, Boston, MA, 2010, abstr. A1-1142.)

MATERIALS AND METHODS

Antimicrobial agents, bacterial strains, and susceptibility testing.

Doripenem was provided by Johnson & Johnson (New Brunswick, NJ), and imipenem was provided by LKT Laboratories, Inc. (St. Paul, MN). Three clinical isolates of ciprofloxacin-resistant P. aeruginosa (MIC 4 μg/ml), strains 8996, 8997, and 14051, were supplied by Johnson & Johnson.

Susceptibility testing was performed by broth microdilution techniques with organisms grown in Ca²⁺- and Mg²⁺-supplemented Mueller-Hinton broth (MHB/S) at an inoculum size of 5 × 10⁵ CFU/ml. The MIC of doripenem for all three strains was 1 μg/ml. The MICs of imipenem were 2 μg/ml with P. aeruginosa 8996 and 8997 and 1 μg/ml with 14051.

The mutant prevention concentrations (MPCs) (6) were determined as described elsewhere (9). Briefly, the tested microorganisms were cultured in two 50-ml aliquots of fresh brain heart infusion broth and incubated overnight at 36°C. The two 50-ml aliquots were then centrifuged at 2,000 rpm for 30 min. The supernatant was discarded, and the remaining pellets were resuspended in 3.0 ml of fresh MHB/S. The 3.0-ml aliquots were thoroughly vortexed and combined in one tube to yield a concentration of 10¹⁰ CFU/ml. From the combined 6.0-ml suspension, 8 1:10 serial dilution steps were made in sterile saline and 4 50-μl drops from each dilution step were plated on antibiotic-containing Mueller-Hinton agar (MHA) plates that were prepared in our laboratory. Concentrations for the doripenem plates ranged from 1 to 32 μg/ml, and for imipenem they ranged from 128 to 2,048 μg/ml. The inoculated plates were incubated for up to 144 h at 36°C and visually screened for growth. The MPC was taken as the lowest antibiotic concentration that completely inhibited growth. The MPCs for all strains were 8 μg/ml for doripenem and 1,024 μg/ml for imipenem.

In vitro dynamic model and simulated pharmacokinetic profiles.

A previously described dynamic model (2) was used in the study. Briefly, this two-compartment model consists of the central compartment and three bioreactors, artificial chambers (Fibercell Systems Inc., Frederick, MD) connected in series, which represent the peripheral compartments. For all experiments, the bacterial inoculum was prepared from previously frozen inocula by thawing, diluting with MHB/S, and then incubating at 36°C to bring the organisms into growth phase.

This mixture was then inoculated via an entry port into each peripheral compartment, which also contained MHB/S, and incubated until a density of approximately 10⁸ CFU/ml was achieved, at which time the antibiotic was introduced into the central compartment (time zero). Given a 20-ml volume of the peripheral compartment, the total number of organisms in the starting inoculum reached approximately 2 × 10⁹ CFU. Control experiments without antimicrobial agent were performed to characterize growth kinetics.

A series of monoexponential profiles that mimic thrice-daily administration (60-min infusion) of doripenem and imipenem (half-life of each, 1 h) (18, 22) was simulated for 3 consecutive days.

Simulated AUC₂₄/MIC ratios varied from 60 to 180 h for doripenem and from 30 to 120 h for imipenem. These ranges incorporate the clinically attainable AUC₂₄/MIC ratios that can be achieved in humans infected by P. aeruginosa isolates with MICs equal to the geometric mean of MICs for the three studied strains after thrice-daily administration of 0.5 g doripenem and imipenem.

As the antimicrobial effect depends on antibiotic concentrations in the peripheral compartments (where the organisms come into contact with antibiotic), peripheral compartments were sampled to determine antibiotic concentrations using an agar well diffusion bioassay. Difco antibiotic medium 5 seeded with Staphylococcus aureus ATCC 29213 was used to measure the doripenem and imipenem concentrations.

Quantification of antimicrobial effect and susceptibility changes.

In each experiment, the peripheral compartments were sampled to determine bacterial concentrations. The numbers of surviving organisms were determined by serial dilution of sample in cold sterile saline and inoculating 20 μl in triplicate onto commercially available MHA. The same samples were also inoculated at 0, 24, 48, 72, and (if needed) 96 h onto plates containing 2×, 4×, 8×, and 16× MIC of the antibiotic being tested during the experiment. After overnight incubation at 36°C, the resulting bacterial colonies were counted, and the numbers of CFU/ml were calculated. The detection limit was 167 CFU/ml. The duration of the experiments was at least 72 h.

Based on time-kill data, the intensity of the antimicrobial effect (the area between control growth and time-kill curves [I_E]) (8, 11) from time zero to the time after the last antibiotic dose at which the number of antibiotic-exposed bacteria reached 10⁹ CFU/ml was determined.

To delineate AUC₂₄/MIC relationships with resistance, areas under the bacterial mutant concentration curves (AUBC_Ms) (9) were determined for subpopulations resistant to 2×, 4×, 8×, and 16× MIC of antibiotic from the beginning of treatment to 72 h (Fig. 1). Changes in susceptibility of P. aeruginosa were examined by MIC determinations using bacteria from drug-free plates inoculated at 0, 24, 48, 72, and (if applicable) 96 h during each experiment. Stability of the observed resistance was determined daily by consecutive passaging of P. aeruginosa on antibiotic-free agar plates for 5 consecutive days.

Fig 1.

AUBC_M analysis of population data (P. aeruginosa 8996 exposed to doripenem at AUC₂₄/MIC of 50 h). AUBC_Ms are indicated by shaded areas.

To reconstruct AUC₂₄/MIC-response and maximum concentration of drug in plasma (C_max)/MIC-response relationships with doripenem from the above-mentioned in vivo study (1), T_>MIC curves of the 24-hour change in the log number of CFU per thigh (Δlog N₂₄) in mice infected with Escherichia coli, Enterobacter cloacae, and P. aeruginosa were scanned and digitized (Grafula 3 software, version 2.10). Data reported with Klebsiella pneumoniae could not be properly distinguished and, therefore, were excluded from our analysis. Because of pronounced scattering of the points, log AUC₂₄/MIC, log C_max/MIC, and T_>MIC versus Δlog N₂₄ data sets were fitted by a linear regression rather than a sigmoid equation. Scanned and digitized time-kill data that were reported from the in vitro study (14) were used to combine data obtained with three organisms and to reconstruct the respective T_>MIC and log D/MIC relationships of log N₂₄.

RESULTS

Carbapenem pharmacodynamics with susceptible P. aeruginosa.

Simulated pharmacokinetic profiles and killing kinetics of one of the studied strains (P. aeruginosa 8997) exposed to doripenem and imipenem are shown in Fig. 2. As seen in the figure, an approximately 1.5-fold range of the T_>MIC for doripenem was provided by the 3.4-fold ranges of the AUC₂₄/MIC and C_max/MIC ratios. With imipenem, a 1.6-fold range of the T_>MIC was provided by the 4.5- to 4.7-fold ranges of the AUC₂₄/MIC and C_max/MIC ratios. The determined antibiotic concentrations were close to the designed values. With doripenem, the observed AUC₂₄/MIC ratios (50, 90, and 170 h) closely approximated the targeted AUC₂₄/MIC ratios (60, 90, and 180 h, respectively). Similar concordance between the observed and designed concentrations was observed in simulations with imipenem (30, 70, and 140 h versus 30, 60, and 120 h, respectively).

Fig 2.

Simulated pharmacokinetics [(AUC₂₄/MIC)/(C_max/MIC)/T_>MIC] of doripenem and imipenem and time courses of susceptible (0× MIC) and resistant (2×, 4×, 8×, and 16× MIC) subpopulations of P. aeruginosa 8997 exposed to the antibiotics. Antibiotic dosing is indicated by arrows.

As seen in Fig. 2, with an increase in the simulated AUC₂₄/MIC, C_max/MIC, and T_>MIC, the rate of killing of doripenem-exposed P. aeruginosa increased weakly: at the highest AUC₂₄/MIC, the time required to kill 99% of the bacteria (T_99%) was only 2-fold shorter than that at the lowest AUC₂₄/MIC ratio. Unlike T_99%, the extent of bacterial killing that is reflected by the minimum number of bacteria (N_min) decreased systematically with an increase in the simulated AUC₂₄/MIC, C_max/MIC, and T_>MIC: at the highest AUC₂₄/MIC ratio, the N_min was more than 4 orders lower than that at the lowest AUC₂₄/MIC ratio. The concentration-dependent—more specifically, the AUC₂₄/MIC-dependent—lowering of the N_min was accompanied by later bacterial regrowth. At the highest AUC₂₄/MIC ratio, the regrowth was observed after the end of treatment, whereas at the lowest AUC₂₄/MIC ratio, it had already occurred on the third day of treatment.

Similar patterns were inherent in killing kinetics of P. aeruginosa 8997 exposed to imipenem. Like doripenem, bacterial regrowth followed the concentration-dependent reduction in bacterial counts over the entire range of the simulated AUC₂₄/MIC ratios of imipenem. Again, T_99% appeared to be only minimally sensitive to the AUC₂₄/MIC, C_max/MIC, and T_>MIC, whereas the N_min was much smaller and the time to regrowth was longer at the highest AUC₂₄/MIC ratio than at the lowest AUC₂₄/MIC ratio.

Time-kill curves obtained with P. aeruginosa 8996 and P. aeruginosa 14051 exposed to doripenem and imipenem (data not shown) were similar to those observed with P. aeruginosa 8997.

When expressed by the I_E parameter, the effects of doripenem and imipenem on P. aeruginosa were dependent on the AUC₂₄/MIC, C_max/MIC, and T_>MIC in a strain-independent manner (Fig. 3). There were strong correlations between the I_E and each of the three predictors of the antibacterial effect. For this reason, only the AUC₂₄/MIC relationships of the I_E were used in further comparisons of doripenem with imipenem.

Fig 3.

Antipseudomonal effects of carbapenems related to different predictors. Solid lines and empty symbols, doripenem; dotted lines and filled symbols, imipenem; triangles, P. aeruginosa 8996; squares, P. aeruginosa 8997; circles, P. aeruginosa 14051.

With each organism exposed to doripenem, the I_Es observed at comparable AUC₂₄/MIC ratios were lower than those for imipenem (Fig. 4, top). However, at the clinically achievable AUC₂₄/MIC ratios (130 h versus 65 h for P. aeruginosa 8996 exposed to doripenem and imipenem, respectively, and 130 h for P. aeruginosa 14051 exposed to both antibiotics), the predicted antipseudomonal effects were similar, whereas the effect of doripenem on P. aeruginosa 8997 was greater than that of imipenem (AUC₂₄/MIC ratios, 130 and 65 h, respectively).

Fig 4.

AUC₂₄/MIC-dependent antipseudomonal effects of carbapenems on susceptible P. aeruginosa (top) and P. aeruginosa subpopulations resistant to 4× MIC (bottom). The descriptions of the symbols are the same as those presented in the legend to Fig. 3.

Combined data from all the three strains also exhibit carbapenem-specific AUC₂₄/MIC-response plots (Fig. 5, left): the difference between the respective regressions was statistically significant (P = 0.05), in terms of both the slope and the intercept. Because the doripenem plot is under the imipenem plot over the studied AUC₂₄/MIC ranges, a higher AUC₂₄/MIC ratio of doripenem is needed to provide the same antipseudomonal effect as imipenem. However, at clinically achievable AUC₂₄s related to the geometric means of the MICs for the three studied strains (1 μg/ml of doripenem and 1.6 μg/ml of imipenem), doripenem appeared to be slightly more efficient than imipenem, although this difference should not be considered substantial due to the pronounced scattering of the points.

Fig 5.

AUC₂₄/MIC-dependent antipseudomonal effects of carbapenems on susceptible P. aeruginosa and P. aeruginosa subpopulations resistant to 4× MIC (combined data for three strains). The descriptions of the symbols are the same as those presented in the legend to Fig. 3.

Carbapenem pharmacodynamics with resistant subpopulations of P. aeruginosa.

As seen in Fig. 2, at the lowest and intermediate AUC₂₄/MIC ratios (50 and 90 h with doripenem and 30 and 70 h with imipenem), killing of carbapenem-susceptible subpopulations of P. aeruginosa 8997 was accompanied by the enrichment of resistant subpopulations. These subpopulations were enriched earlier at the lowest AUC₂₄/MIC ratios than the intermediate AUC₂₄/MIC ratios. However, by the end of treatment, subpopulations resistant to 2× and 4× MIC of doripenem (AUC₂₄/MIC, 50 h) and to 2×, 4×, and 8× MIC of doripenem (AUC₂₄/MIC, 90 h) and subpopulations resistant to 2×, 4×, 8×, and 16× MIC of imipenem (AUC₂₄/MIC ratios, 30 and 70 h) completely replaced carbapenem-susceptible organisms. In contrast, at the highest AUC₂₄/MIC ratios (170 and 140 h with doripenem and imipenem, respectively), resistant subpopulations were not amplified during treatment. Time courses observed with resistant subpopulations of P. aeruginosa 8996 and P. aeruginosa 14051 exposed to doripenem and imipenem (data not shown) were similar to those observed with P. aeruginosa 8997.

As seen in Fig. 6, a pronounced enrichment of resistant subpopulations occurred immediately after cessation of treatment of P. aeruginosa 8997 with imipenem (AUC₂₄/MIC, 140 h) but not doripenem (AUC₂₄/MIC, 170 h). The same difference was observed after the treatment of P. aeruginosa 8996 (Fig. 6): the size of subpopulations resistant to 2×, 4×, 8×, and 16× MIC of imipenem was 4 to 7 orders higher than that of doripenem-resistant subpopulations. The amplification of imipenem-resistant subpopulations was accompanied by 16-fold (P. aeruginosa 8996) and 8-fold (P. aeruginosa 8997) elevations in the MICs (Fig. 7). The observed elevations in the MICs were stable after five passages on antibiotic-free agar plates. In contrast to imipenem, there was no loss in susceptibility of P. aeruginosa 8996 and 8997 treated with doripenem. Unlike P. aeruginosa 8996 and 8997, an increase in the size of resistant P. aeruginosa 14051 subpopulations occurred after treatment with both doripenem and imipenem.

Fig 6.

Resistant subpopulations of P. aeruginosa after exposure to doripenem (AUC₂₄/MIC, 170 h) and imipenem (AUC₂₄/MIC, 140 h). Black bars, doripenem; gray bars, imipenem.

Fig 7.

Changes in susceptibility of P. aeruginosa exposed to doripenem (AUC₂₄/MIC, 170 h) and imipenem (AUC₂₄/MIC, 140 h): ratio of the MIC determined 96 h after the start of treatment (MIC_{96 h}) to the initial MIC (MIC₀). The descriptions of the symbols are the same as those in the legend to Fig. 6.

Based on time courses of carbapenem-resistant subpopulations, the AUBC_Ms were plotted against the AUC₂₄/MIC ratio for each organism (Fig. 4, bottom). As seen in the figure, at clinically achievable AUC₂₄/MIC ratios, doripenem exhibits lower AUBC_Ms, i.e., greater antimutant effects, than imipenem against P. aeruginosa 8996 and 8997, which is reversed with P. aeruginosa 14051. The AUC₂₄/MIC relationships of the AUBC_Ms determined with resistant subpopulations of the three studied strains exposed to doripenem and imipenem are shown on the right of Fig. 5. Although AUC₂₄/MIC plots of the AUBC_Ms were not superimposed for doripenem and imipenem, the difference between the respective regressions was statistically insignificant (P = 0.05). With doripenem, the predicted AUBC_M was slightly smaller (greater antimutant effect) than that with imipenem.

DISCUSSION

Designed to ascertain concentration-response relationships with two carbapenems, this study predicts similar or slightly greater effects of doripenem than imipenem on susceptible and resistant subpopulations of P. aeruginosa exposed to clinically achievable AUC₂₄/MIC ratios. These findings are consistent with clinical reports of the similar efficacies of doripenem and imipenem (5) and the somewhat higher efficacy of doripenem in ventilator-associated pneumonia (16, 17). A more pronounced enrichment of resistant subpopulations observed in our study after (but not during) treatment with imipenem can be attributed to its greater MPC (1,024 μg/ml) compared to that of doripenem (8 μg/ml).

Contrary to the widespread idea that carbapenem effects are dependent on T_>MIC but not AUC₂₄/MIC and C_max/MIC, all three predictors exhibited similar correlations with the antipseudomonal effect (Fig. 3). The same conclusion can be drawn from recently reported data (14) that were originally interpreted using T_>MIC only. Despite the fact that the authors avoided any mention of AUC₂₄/MIC or C_max/MIC, due to the similar pharmacokinetics of doripenem and imipenem (18, 22), the respective AUC₂₄/MIC relationships of the effect can be represented by its relationships to the MIC-related daily dose that corresponds to the clinical AUC₂₄/MIC for the unbound antibiotic (D/MIC). As seen in Fig. 8, killing of P. aeruginosa exposed to each carbapenem correlates with the D/MIC ratio (and, therefore, with AUC₂₄/MIC) even better than with T_>MIC. Furthermore, according to our analysis of reported in vivo data (1), the abilities of AUC₂₄/MIC, C_max/MIC, and T_>MIC to predict doripenem effects in a neutropenic murine thigh infection model are also similar (r² = 0.77, 0.77, and 0.67, respectively). This analysis once again raises questions about the advisability of searching for the “best” predictor of the antibacterial effect among highly covarying variables. Attempts to contrast equally good (or equally bad) predictors have been criticized previously (7, 10). Moreover, these attempts were often incorrect: erroneous conclusions about the advantages of one predictor over another resulted from the use of inconsistent functions relating the effect to its predictors, inappropriate combining of data obtained with pharmacokinetically different antibiotics, etc. Methodological pitfalls in analysis of the predictor-response relationships have been discussed in detail elsewhere (12).

Fig 8.

Antipseudomonal effects of doripenem (empty symbols) and imipenem (filled symbols) related to different predictors. Circles, P. aeruginosa PAO1; squares, stable depressed PAO1; triangles, OprD-downregulated PAO1 (reconstructed from reference 14).

Overall, at clinically achievable AUC₂₄/MIC ratios, doripenem was slightly more efficient than imipenem and prevented mutant selection somewhat better for two of the three tested strains.