|Home | About | Journals | Submit | Contact Us | Français|
We compared drugs (imipenem and doripenem), doses (500 mg and 1 g), and infusion times (0.5 and 1.0 [imipenem], 1.0 and 4.0 h [doripenem]) in our hollow-fiber model, examining cell kill and resistance suppression for three isogenic strains of Pseudomonas aeruginosa PAO1. The experiments ran for 10 days. Serial samples were taken for total organism and resistant subpopulation counts. Drug concentrations were determined by high-pressure liquid chromatography-tandem mass spectrometry (LC/MS/MS). Free time above the MIC (time > MIC) was calculated using ADAPT II. Time to resistance emergence was examined with Cox modeling. Cell kill and resistance emergence differences were explained, in the main, by differences in potency (MIC) between doripenem and imipenem. Prolonged infusion increased free drug time > MIC and improved cell kill. For resistance suppression, the 1-g, 4-h infusion was able to completely suppress resistance for the full period of observation for the wild-type isolate. For the mutants, control was ultimately lost, but in all cases, this was the best regimen. Doripenem gave longer free time > MIC than imipenem and, therefore, better cell kill and resistance suppression. For the wild-type organism, the 1-g, 4-h infusion regimen is preferred. For organisms with resistance mutations, larger doses or addition of a second drug should be studied.
Pseudomonas aeruginosa continues to be a major problem in the nosocomial setting. Increasing rates of resistance make the development of effective therapeutic regimens problematic.
Doripenem is a new carbapenem antibiotic with potent activity against Pseudomonas aeruginosa. Preclinical studies have indicated that it is highly stable to the AmpC enzyme seen in this pathogen and that it interacts differently with the pathogen regarding oprD downregulation, resulting in lower MIC shifts, in at least 50% of instances (6). Clinically, the use of the prolonged infusion has been shown to have a salutary impact on Pseudomonas resistance emergence during therapy, relative to the impact of imipenem (1). We chose to study imipenem because the doripenem clinical trial program employed imipenem as a comparator, as meropenem does not have the breadth of FDA indications present for imipenem (e.g., nosocomial pneumonia).
Previous work from our group has shown that the use of prolonged infusion optimizes time above the MIC (time > MIC) target attainment and may have an impact on resistance emergence (3, 10). This leads to four major factors requiring exploration: (i) drug (potency), (ii) dose, (iii) infusion schedule, and (iv) differences in mechanism of resistance between drugs.
In order to ascertain the contribution of each, we decided to study three different isogenic isolates: a wild-type isolate (PAO1), an isolate with a stably derepressed chromosomal AmpC enzyme (AmpC β-lactamase production is markedly increased when a mutation in the repressor system occurs, and the increase is stable and not dependent upon the presence or absence of drug), and an isolate with a defined downregulation of OprD (OprD is a carbapenem-specific transport porin; when it is downregulated, less drug is available per unit time in the periplasmic space). In addition, we decided to examine both doripenem and imipenem to ascertain the impact of differing potencies and interactions with OprD downregulation. Finally, we hypothesized that infusion time would have an impact. Therefore, we studied doripenem at a 500-mg dose with a 1-h infusion, a 500-mg dose with a 4-h infusion, and a 1-g dose with a 4-h infusion. Imipenem's stability is such that a 4-h infusion cannot be recommended clinically. We therefore decided to examine two regimens: 500 mg every 6 h with a half-hour infusion and 1 g every 8 h with a 1-h infusion. Both regimens are consistent with the package insert for imipenem. The endpoints were cell kill at 24 h (before emergence of resistant clones would obfuscate the endpoint) and emergence of resistance (both the initial time when the number of resistant clones exceeded that at baseline and the time to near-maximal number of resistant clones). “Near maximal” is defined as being within 1 standard deviation of true maximal. This is approximately 0.3 log10 CFU/ml.
The Pseudomonas aeruginosa strain PAO1, its oprD-downregulated isogenic mutant (selected by exposure to imipenem; ampC RNA levels were checked by reverse transcription-PCR [RT-PCR] to document that the oprD selection did not stably derepress ampC; expression of mexAB, mexCD, mexEF, and mexXY were also checked and shown not to differ from expression of wild-type genes), and a stably derepressed (selected by exposure to ceftazidime) isogenic mutant were the kind gift of Karen Bush and Brian Morrow. MIC values were determined by CLSI broth macrodilution methodology for both doripenem and imipenem (7). This was done on at least three occasions. Imipenem was employed as the selecting agent. We had done an extensive stability evaluation of the drugs in Mueller-Hinton II (MH II) agar at an incubator temperature that drove the choice of selecting agent (data not shown).
A schematic diagram of the system is shown in reference 4. Doripenem or imipenem was directly injected into the central reservoir over a period of 0.5 to 4 h on an appropriate schedule (every 6 or every 8 h) to achieve the peak concentration desired. The experiments were carried out for 10 days or more. Methods specific to the operation of this model have been published previously (4).
Each isolate was stored at −70°C in skim milk. Fresh isolates were grown on blood agar plates (BBL Microbiology Systems, Cockeysville, MD) for 24 h at 35°C before each experiment. The bacterial inoculum was prepared by employing three medium-sized colonies grown overnight in Mueller-Hinton II broth at 35°C. Hollow-fiber systems were maintained at 35°C in a humidified incubator. Bacterial culture (15 ml) in late-log-phase growth (circa 1 × 108 CFU/ml) was infused into six hollow-fiber cartridges, one each for (i) nominal drug exposure (area under the curve at 24 h [AUC24]/MIC) of 0 (control), (ii) a doripenem dose (1-h infusion) of 500 mg every 8 h, (iii) a doripenem dose (4-h infusion) of 500 mg every 8 h, (iv) a doripenem dose (4-h infusion) of 1 g every 8 h, (v) an imipenem dose (half-hour infusion) of 500 mg every 6 h, and (vi) an imipenem dose (1-h infusion) of 1 g every 8 h. These regimens simulated steady-state human pharmacokinetics of unbound doripenem and imipenem. Experimentally attained doripenem and imipenem exposures (central compartment) were based upon drug concentrations quantified by a validated high-pressure liquid chromatography-tandem mass spectrometry (LC/MS/MS) method (see below). Samples were obtained 20 to 22 times (depending on regimen) over 47 h. At 0, 0.17, 1, 2, 3, 4, 6, 8, and 10 days, samples of the bacterial cultures were obtained, centrifuged at 3,200 × g for 15 min, resuspended in normal saline in order to minimize drug carryover effect, and diluted 10-fold serially. The serially diluted samples were quantitatively cultured onto drug-free Mueller-Hinton II agar plates to enumerate the total bacterial population. The resistant bacterial subpopulation was quantified by culturing on medium plates supplemented with imipenem at a concentration 2.5 times the baseline MIC, except for the oprD isogenic mutant, where 3 times the baseline MIC was employed because of the shift in imipenem MIC. The medium plates were incubated at 35°C for 24 and 48 h to evaluate them for the impacts on the total and antibiotic-resistant subpopulations, respectively. MIC values of the resistant subpopulations to doripenem and imipenem at experiment end were determined to confirm resistance.
Mueller-Hinton II broth pharmacokinetic simulation samples were diluted with high-pressure liquid chromatography (HPLC) water (0.050-ml sample into 1.00 ml water) and were analyzed by HPLC-tandem mass spectrometry (LC/MS/MS) for doripenem concentrations. The LC/MS/MS system was comprised of a Shimadzu Prominence HPLC system and an Applied Biosystems/MDS Sciex API5000 LC/MS/MS.
Chromatographic separation was performed using a Waters Novapak C18 column (5 μm, 150 by 3.9 mm) and a mobile phase consisting of 85% 0.1% formic acid in water and 15% 0.1% formic acid in acetonitrile, at a flow rate of 1.0 ml/min. Doripenem concentrations were obtained using LC/MS/MS, monitoring the MS/MS transition m/z 421 → m/z 274. Analysis run time was 2.5 min. The assay was linear over a range of 0.100 to 25.0 μg/ml (r2 > 0.996). The interday coefficients of variation (CVs) for the quality control samples analyzed in replicates of three at three concentrations on each analysis day (0.500, 4.00, and 20.0 μg/ml) ranged from 2.04 to 6.38%, with accuracies (%REC) ranging between 98.2% and 102%.
Mueller-Hinton II broth pharmacokinetic simulation samples were diluted with HPLC water (0.050-ml sample into 1.00 ml water) and were analyzed by high-pressure liquid chromatography-tandem mass spectrometry (LC/MS/MS) for imipenem concentrations. The LC/MS/MS system was comprised of a Shimadzu Prominence HPLC system and an Applied Biosystems/MDS Sciex API5000 LC/MS/MS.
Chromatographic separation was performed using a Waters Novapak C18 column (5 μm, 150 by 3.9 mm) and a mobile phase consisting of 90% 0.1% formic acid in water and 10% 0.1% formic acid in acetonitrile, at a flow rate of 1.0 ml/min. Imipenem concentrations were obtained using LC/MS/MS, monitoring the MS/MS transition m/z 300 → m/z 170. Analysis run time was 3.0 min. The assay was linear over a range of 0.100 to 50.0 μg/ml (r2 > 0.996). The interday CVs for the quality control samples analyzed in replicates of three at three concentrations on each analysis day (0.500, 4.00, and 25.0 μg/ml) ranged from 1.50 to 7.45%, with accuracies (%REC) ranging between 98.2% and 108%.
The working solution of nitrocefin (500 mg/liter) was diluted 10-fold in buffer (0.1 M phosphate; 1 mM EDTA, pH 7.0). Changes in absorbance were measured in the spectrophotometer (SpectraMax M5 plate reader, running SoftMax Pro-5 software; Molecular Devices, Sunnyvale, CA) at 486 nm. The molar extinction coefficient of hydrolyzed nitrocefin at 486 nM is 20,500 M−1 cm−1. The wild-type isolate and its isogenic, stably derepressed mutant were examined with this method.
Concentration-time profiles for each drug and each regimen for each isolate were analyzed employing maximum likelihood estimation. The identification module of the ADAPT II package of programs of D'Argenio and Schumitzky (2) was employed. As computer-controlled infusion pumps drove the profile, a one-compartment open model with zero-order input and first-order elimination was employed.
Free drug time > MIC was estimated by integrating the following differential equation, which was a system output:
where X(1) is the amount of drug in the central compartment, Vol is the volume of the central compartment, MIC is the MIC of the appropriate drug for the pathogen being studied, and GE is “greater than or equal to.”
This measure of the independent variable (free drug time > MIC) was linked to cell kill at 24 h using either a sigmoid Emax model or an Emax model, depending upon whether Hill's constant was statistically supportable (i.e., model expansion tested with Akaike's information criterion).
Additionally, we employed an inhibitory sigmoid Emax model:
and an inhibitory Emax model:
where Effect is the bacterial burden at 24 h in treated animals (log10 CFU/g), Econ is the bacterial burden at 24 h in the no-treatment control animals (log10 CFU/g), Emax (log10 CFU/g) is the largest bacterial kill obtainable with the time frame employed, E50 is the measure of the independent variable (fraction of the dosing interval that drug concentration exceeds the MIC) that drives half of the maximal effect, and H is Hill's constant (also called the sigmoidicity constant).
These models recognize that exposure responses are relatively flat at low values of the independent variable, are steep and relatively linear in the midrange (between approximately 20% of maximal effect and 80% of maximal effect), and again flatten out at higher values of the independent variable.
Time to resistance and time to near-maximal resistance were analyzed using Cox proportional hazards regression (SYSTAT for Windows version 11.0).
The MIC values of doripenem and imipenem were 1.0 and 2.0 mg/liter, respectively, for the wild-type PAO1 isolate. For the stably derepressed isogenic mutant, the values were 2.0 and 2.0 mg/liter. For the ΔoprD isolate, the MIC values were 2.0 mg/liter for doripenem and 16.0 mg/liter for imipenem.
The hydrolysis of nitrocefin was statistically different for the wild-type PAO1 isolate relative to the stably derepressed mutant, as would be expected (data not shown).
The total CFU/ml and resistant colony counts for the control and different regimens of doripenem and imipenem are displayed in Fig. 1A to F. The high (1-g)-dose, 4-h infusion arm is the only one to completely suppress resistance emergence.
The total CFU/ml and resistant colony counts for the control and different regimens of doripenem and imipenem are displayed in Fig. 2A to F. The high (1-g)-dose, 4-h infusion arm is the only one to hold the total population under some control and also limit, but not suppress completely, the amplification of resistant mutants. The doripenem 500-mg, 4-h infusion slightly underperformed at this dose, in contrast to the wild-type isolate.
This resistance mechanism had a greater impact on cell kill than stable derepression of the AmpC β-lactamase for both drugs and all the regimens (Fig. 3A to F). As noted previously, doripenem had overall greater log kill than imipenem, which is likely related to the MIC of this mutant for the drugs (2 mg/liter for doripenem and 16 mg/liter for imipenem). Only the doripenem 1-g, 4-h-infusion regimen administered every 8 h achieved a 3-log10-CFU/ml cell kill.
The time > MIC was calculated for each regimen and for each organism and is displayed in Table Table1.1. Because of the previous observations of Tam et al. (9), the time > 6.2× MIC was also calculated (data not shown).
We examined time to the first time that the resistant mutant isolates exceeded their time zero baseline values and also the time that they achieved near-maximal colony counts. Tam and colleagues had indicated for the carbapenem meropenem that trough concentrations greater than 6.2 times the baseline MIC were necessary to suppress resistance in Pseudomonas aeruginosa (9). Consequently, we calculated time > this value and employed Cox proportional hazards regression to ascertain whether the time to resistance for these agents and organisms was influenced by time > 6.2× MIC. We also used organisms and drug as stratification variables to determine if the time to resistance (first resistance and amplification to near-maximal numbers) was influenced by these factors. Neither drug nor organism significantly influenced the outcome as a stratification variable (Mantel test). Time > 6.2× MIC significantly influenced time to first resistance (estimate, −6.95; 95% confidence interval, −12.93 to −0.976; P = 0.023) and time to amplification to near-maximal number of colony counts (estimate, −11.76; 95% confidence interval, −19.979 to −3.533; P = 0.005). These survival curves are displayed in Fig. 5A and B.
A new, more potent weapon in the war against serious Pseudomonas aeruginosa infection would be most welcome. In order to determine whether doripenem had advantageous microbiological properties, we tested it in our hollow-fiber infection model over a period of time (10 to 13 days) that is consistent with clinical usage. In addition, we also wished to examine the impact of differing mechanisms of resistance on the microbiological effect of doripenem. Finally, we also wished to examine the impact of prolonged (4-h) infusion on the observed microbiological effect. In order to place these findings into proper perspective, we also examined two regimens of imipenem against these isolates.
Doripenem was more potent (1 dilution) than imipenem in the wild-type isolate and was equivalent to imipenem in the stably derepressed mutant. In the oprD mutant, however, the MIC was 2 mg/liter for doripenem and 16 mg/liter for imipenem, validating the previous observations of Mushtaq et al. (6).
We examined both cell kill activity of the regimens and emergence of resistance. For cell kill, we first simply examined the colony counts at the end of 24 h of drug exposure, as this time point will be least confounded by amplification of resistant mutant subpopulations.
For the wild-type PAO1, all three doripenem regimens produced cell kill greater than either imipenem regimen, with the 1-g doripenem dose administered as a 4-h infusion every 8 h producing the largest cell kill of 5.4 log10 CFU/ml. The 500-mg doripenem regimens demonstrated an advantage for the 4-h infusion, with the 1-h-infusion regimen having a kill of 3.71 log10 CFU/ml and the 4-h-infusion regimen of 500 mg producing a kill of 4.07 log10 CFU/ml.
For the stably derepressed isogenic mutant of PAO1, the hierarchy of effect was the same, with all three doripenem regimens producing more 24-h cell kill than either of the imipenem regimens. The advantage of the high (1-g)-dose doripenem arm with the 4-h infusion was also manifest here.
With the ΔoprD isogenic mutant, essentially the same findings were seen. The only exception was that the 1-g, 1-h infusion imipenem regimen was slightly superior to the doripenem 500-mg, 1-h-infusion regimen, with cell kill of 3.47 versus 3.12 log10 CFU/ml, respectively. As in the other experiments, the best cell kill was seen with the doripenem 1-g, 4-h infusion regimen, at 4.24 log10 CFU/ml.
When one examines all the regimens, it becomes apparent that the differences in cell kill activity seen are mediated through the MIC and through the mode of administration (prolonged infusion). In Fig. 2A to C, the cell kill activities for the three different regimens are displayed as a function of time > MIC. Examination of the curves demonstrates that the fit of the model to the data was precise and unbiased. The 95% confidence intervals around the point estimates of the parameters were quite acceptable. The fit of the model to the data explained a significant amount of the variance (i.e., the fit was significant). In all instances, once the MIC is accounted for and the dose and mode of administration are accounted for by calculating time > MIC, there is little bias between drugs and modes of administration. As noted above, doripenem has greater cell kill because it has lower MIC values and the mode of administration generates longer time > MIC. The largest Emax is seen with the wild-type isolate, at 6.06 log10 CFU/ml. The stably derepressed mutant and the ΔoprD mutant had Emax values of 5.05 and 4.56 log10 CFU/ml, respectively. This suggests that the mutations lead to a lower log kill rate than was seen with the wild-type isolate and is not accounted for in the time > MIC calculation. One regimen in the oprD experiment was an outlier, where imipenem at 500 mg every 6 h as a half-hour infusion caused substantial cell kill but with a calculated time > MIC of zero hours. This is likely because the imipenem MIC for this isolate was 16 mg/liter. We ran an arithmetic dilution MIC, and the imipenem MIC was 12 mg/liter. For this value, the fraction of the dosing interval with time > MIC (free drug) was 0.3736. This makes the fit of the model to the data quite acceptable (see Fig. Fig.4C;4C; r2 = 0.949; imipenem at 500 mg every 6 h is the open triangle).
With respect to resistance emergence, Tam et al. (9) have previously demonstrated that for the carbapenem meropenem, a trough concentration 6.2 times the baseline MIC is required to suppress resistance when that agent is administered alone. Consequently, we calculated the time that drug concentrations remained above 6.2 times the MIC. As an endpoint we looked at the time until the regimen had resistance emergence, defined as having the resistant population increase to concentrations above the baseline number. We also looked at the time until the resistant subpopulation was amplified until near-maximal values were obtained. Because we were looking at a time-to-event, we employed Cox proportional hazards regression as the analytical tool. For both endpoints, time > 6.2× MIC was a statistically significant covariate in the analyses. For time till first emergence of resistance, the time > 6.2× MIC had a P value of 0.023 (Fig. (Fig.5A).5A). The endpoint of time till near-maximal resistance amplification was even more strongly influenced by time > 6.2× MIC (P = 0.005) (Fig. (Fig.5B5B).
It is important to note that of all the regimens for all the isogenic isolates, only the doripenem 1-g, 4-h infusion every 8 h against wild-type PAO1 suppressed amplification of the resistant subpopulation for the full 10 days of the experiment. The concentration-time profile here remained above 6.2 times the MIC for a fraction of 0.516 of the dosing interval. This is different from the data obtained by Tam et al. (9) but is consistent with the idea that carbapenem regimens need to be above a multiple of the MIC for a considerable period of time to suppress resistance.
Given this, we wished to see if a larger doripenem dose would suppress resistance. We calculated that 1,500 mg of doripenem every 8 h as a 4-h infusion would give a free drug time > 6.2× MIC similar to that seen with the wild-type isolate. When free drug time > 6.2× MIC was directly measured, we achieved a fraction of the dosing interval of free drug time > 6.2× MIC of 0.497 (versus 0.516 for the wild-type isolate). This did not suppress resistance for the full 10 days of the experiment (data not shown). We speculate that oprD mutants also have some Mex pump overexpression, which may make them more difficult to suppress (5, 8).
Other defined resistance mechanisms changed the MIC and shortened the time that concentrations remained above 6.2× the MIC and, therefore, allowed ultimate amplification of resistant subpopulations. It should be noted, however, that the doripenem 1-g, 4-h infusion performed best with each of the isogenic mutants.
In summary, cell kill and suppression of resistance for these two carbapenem antibiotics were optimized by increasing the time > threshold. For resistance suppression, this threshold was substantially higher. This principle held, irrespective of the specific resistance mechanism. Doripenem, particularly at the 1-g dose with the 4-h infusion, performed best in all the evaluations. This is because of the increased potency of doripenem and because the prolonged infusion combined to maximize the time > MIC and time > 6.2× MIC. Given these findings, the high-dose, prolonged infusion of doripenem should be evaluated in clinical circumstances, such as treatment of patients with ventilator-associated pneumonia, where risk of emergence of resistance is highest. Because resistance was suppressed for the full 10 days by the high-dose 4-h infusion only in the wild-type isolate, it is likely that Pseudomonas strains carrying a resistance mechanism should be treated by this regimen in combination with a second drug.
This work was supported in part by a grant from Ortho-McNeil Pharmaceutical. This work was also supported, in part, by R01AI079578, a grant from the National Institute of Allergy and Infectious Diseases (NIAID) to the Emerging Infections and Pharmacodynamics Laboratory.
The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIAID or the National Institutes of Health.
Karen Bush, Anne-Marie Queenan, Brian Morrow, Mohammed Khashab, James B. Kahn, and Susan Nicholson are or were employees of Johnson and Johnson. George Drusano accepted a research grant from Johnson and Johnson that partially paid for this investigation. No other author has any conflicts.
Published ahead of print on 22 March 2010.