Meropenem-Tobramycin Combination Regimens Combat Carbapenem-Resistant Pseudomonas aeruginosa in the Hollow-Fiber Infection Model Simulating Augmented Renal Clearance in Critically Ill Patients

Augmented renal clearance (ARC) is common in critically ill patients and is associated with subtherapeutic concentrations of renally eliminated antibiotics. We investigated the impact of ARC on bacterial killing and resistance amplification for meropenem and tobramycin regimens in monotherapy and combination. Two carbapenem-resistant Pseudomonas aeruginosa isolates were studied in static-concentration time-kill studies. One isolate was examined comprehensively in a 7-day hollow-fiber infection model (HFIM).

ABSTRACT

Augmented renal clearance (ARC) is common in critically ill patients and is associated with subtherapeutic concentrations of renally eliminated antibiotics. We investigated the impact of ARC on bacterial killing and resistance amplification for meropenem and tobramycin regimens in monotherapy and combination. Two carbapenem-resistant Pseudomonas aeruginosa isolates were studied in static-concentration time-kill studies. One isolate was examined comprehensively in a 7-day hollow-fiber infection model (HFIM). Pharmacokinetic profiles representing substantial ARC (creatinine clearance of 250 ml/min) were generated in the HFIM for meropenem (1 g or 2 g administered every 8 h as 30-min infusion and 3 g/day or 6 g/day as continuous infusion [CI]) and tobramycin (7 mg/kg of body weight every 24 h as 30-min infusion) regimens. The time courses of total and less-susceptible bacterial populations and MICs were determined for the monotherapies and all four combination regimens. Mechanism-based mathematical modeling (MBM) was performed. In the HFIM, maximum bacterial killing with any meropenem monotherapy was ∼3 log₁₀ CFU/ml at 7 h, followed by rapid regrowth with increases in resistant populations by 24 h (meropenem MIC of up to 128 mg/liter). Tobramycin monotherapy produced extensive initial killing (∼7 log₁₀ at 4 h) with rapid regrowth by 24 h, including substantial increases in resistant populations (tobramycin MIC of 32 mg/liter). Combination regimens containing meropenem administered intermittently or as a 3-g/day CI suppressed regrowth for ∼1 to 3 days, with rapid regrowth of resistant bacteria. Only a 6-g/day CI of meropenem combined with tobramycin suppressed regrowth and resistance over 7 days. MBM described bacterial killing and regrowth for all regimens well. The mode of meropenem administration was critical for the combination to be maximally effective against carbapenem-resistant P. aeruginosa.

TEXT

Pseudomonas aeruginosa is a common cause of serious infections in hospitalized patients, which are increasing in prevalence and result in substantial morbidity and mortality (1, 2). In the United States, it is estimated that 51,000 health care-associated infections each year are due to P. aeruginosa (3). Patients suffering from P. aeruginosa were found to have a higher mortality rate than those infected by other Gram-negative pathogens (4). P. aeruginosa is intrinsically resistant to many antibiotics and, in addition, has a particularly high propensity to develop resistance to all available antipseudomonals during monotherapy (5). The carbapenems are key antibiotics in the treatment of serious P. aeruginosa infections, but unfortunately, carbapenem resistance has become prevalent. In the United States, 19% of isolates from selected health care-associated infections were nonsusceptible to carbapenems, and some European countries reported resistance rates of 50 to 60% (6, 7).

Infections caused by carbapenem-resistant P. aeruginosa strains are particularly worrisome for critically ill patients, who not only commonly develop serious infections but also frequently display substantially altered pharmacokinetics (PK) compared with those of noncritically ill patients (8, 9). Of major concern is that these patients (particularly if suffering from burns, sepsis, or trauma) commonly have augmented renal clearance (ARC) (9, 10). ARC places the patient at risk of subtherapeutic concentrations of renally eliminated drugs, including most antibiotics in the three major classes used against P. aeruginosa (β-lactams, aminoglycosides, and fluoroquinolones) (10–12). Subtherapeutic concentrations commonly lead to the emergence of resistance (13, 14), which presents serious clinical problems for the patients being treated and others subsequently infected with the newly resistant strain. We have previously shown that when the commonly used β-lactam antibiotics meropenem and piperacillin are administered as monotherapy and subjected to the pharmacokinetics associated with ARC, bacterial killing is substantially decreased and resistance emergence increased, even at the maximum approved doses (15, 16). Data from clinical studies suggest the potential benefit of combination therapy with a carbapenem as a backbone against infections caused by carbapenem-resistant Gram-negative pathogens (17–19). However, there have been no studies that have examined carbapenem-based combinations against carbapenem-resistant P. aeruginosa in the context of ARC.

Therefore, the current study evaluated the combination of meropenem and tobramycin in static-concentration time-kill studies (SCTK) against two carbapenem-resistant P. aeruginosa isolates from critically ill patients with burns. The hollow-fiber infection model (HFIM) is the state-of-the-art in vitro pharmacodynamic (PD) model used to accurately represent antibiotic concentration-time profiles as observed in patients (20, 21). HFIM results correlate well with clinical endpoints for bacterial killing and the time course of emergence of resistance (22). Our current study simulated in the HFIM the pharmacokinetic profiles of patients with ARC for four meropenem regimens (two dose levels in short-term intermittent or continuous infusion [CI]) and a tobramycin regimen, in monotherapies and combinations, to characterize their effects on bacterial killing and resistance amplification of carbapenem-resistant P. aeruginosa. Furthermore, we developed a mechanism-based mathematical model (MBM) to quantitatively describe the relationship between meropenem and tobramycin concentration-time profiles (alone and in combination) and bacterial killing and regrowth over time under the ARC condition.

RESULTS

The initial inoculum for each isolate in the SCTK was ∼10⁷ CFU/ml (Fig. 1). For P. aeruginosa isolate CR382 (meropenem MIC of 8 mg/liter) in monotherapy, 25 mg/liter meropenem (reflecting the peak concentration for 1 g given as a 30-min infusion, equivalent to ∼3× MIC) was required to achieve ∼2 log₁₀ CFU/ml bacterial killing and suppress regrowth over 48 h. Meropenem at 4 mg/liter and 8 mg/liter (reflecting the average concentrations at steady state for daily doses of 3 g and 6 g meropenem, respectively) resulted in <1 log₁₀ CFU/ml bacterial killing at 5 h, followed by regrowth. In contrast, for P. aeruginosa isolate CR380 (meropenem MIC of 32 mg/liter), even 25 mg/liter meropenem as monotherapy achieved <2 log₁₀ CFU/ml bacterial killing at 5 h, followed by regrowth. Tobramycin monotherapy, even at 8 mg/liter (8× MIC, reflecting ∼35% of the unbound maximum concentration in a typical patient with substantial ARC given 7 mg/kg of body weight tobramycin as a 30-min infusion every 24 h [q24h]), did not suppress regrowth for either isolate. For isolate CR382, the combination of 8 mg/liter meropenem and 2 mg/liter tobramycin achieved >3.5 log₁₀ CFU/ml killing and suppression of regrowth over 48 h. In contrast, for isolate CR380, 25 mg/liter meropenem was required in combination with 1 or 2 mg/liter tobramycin to achieve suppression of regrowth over 48 h. Overall, clear synergy (≥2 log₁₀ CFU/ml killing compared to that of the most active monotherapy) was demonstrated for the meropenem and tobramycin combination for both isolates. The treatments that were conducted in replicates demonstrated good reproducibility.

FIG 1.

Static-concentration time-kill studies examining various meropenem (MEM) and tobramycin (TOB) concentrations, alone and in combination, against isolates CR382 and CR380. Counts below the limit of counting (1.0 log₁₀ CFU/ml) are plotted at 1.0 log₁₀ CFU/ml. Data for MEM monotherapies, TOB 8 mg/liter, and MEM combinations with TOB 8 mg/liter are presented as average values and standard deviations of three biological replicates.

Representative observed, targeted, and fitted concentration-time profiles for meropenem and tobramycin in the HFIM are shown in Fig. S1 in the supplemental material. Measured concentrations were well matched to the target profiles produced in the in silico design simulations with fitted concentration-time profiles used in the MBM. The time courses of total and less-susceptible bacterial populations for isolate CR382 in the HFIM are shown in Fig. 2 and 3, respectively. The proportions of mutants (mutant frequencies) and MICs are presented in Tables S1 and S2.

FIG 2.

Time courses of total bacterial populations of isolate CR382 simulated in the HFIM under conditions of augmented renal clearance (ARC) (CL_CR of 250 ml/min, corresponding to meropenem and tobramycin clearances of 30.3 and 12.9 liters/h, respectively). The study comprised four clinically relevant dosing regimens of meropenem (MEM) (1 g and 2 g i.v. given every 8 h [q8h] as 0.5-h infusion or 3 g/day and 6 g/day as continuous infusion [CI]), a clinically relevant dosing regimen of tobramycin (TOB) (7 mg/kg given q24h as 0.5-h infusion), and their combinations. The combination of MEM as 6-g/day CI with TOB was evaluated in two biological replicates, and the average values are shown in the graph. A growth control is also included. Bacterial counts below the limit of counting (1.0 log₁₀ CFU/ml) are plotted at 1.0 log₁₀ CFU/ml.

FIG 3.

Effects of each dosing regimen of meropenem and tobramycin, alone or in combination, on less-susceptible bacterial populations of CR382 in the HFIM. Less-susceptible populations were those able to grow in the presence of 3× or 5× the meropenem MIC or 5× or 10× the tobramycin MIC. The total bacterial population at each time point, from the experiments whose results are shown in Fig. 2, is included for ease of comparison. Panel labeled “MEM 6 g/day CI + TOB” shows the average values from two biological replicates.

The initial inocula for CR382 in the HFIM were 6.80 to 6.97 log₁₀ CFU/ml for the different treatment arms. No colonies grew on antibiotic-containing agar prior to treatment (Fig. 3). The log₁₀ mutant frequency (log₁₀MF) at 0 h was <−8.95 on agar containing 3× or 5× the meropenem MIC and also on agar containing 5× or 10× the tobramycin MIC (Table S1). For the growth control, bacteria resistant to meropenem (growing at 3× MIC, i.e., 24 mg/liter meropenem) and less susceptible to tobramycin (growing at 5× MIC, i.e., 5 mg/liter) plateaued at ∼1 to 3 log₁₀ CFU/ml (Fig. 3). No colonies grew on agar containing meropenem at 5× MIC (40 mg/liter), while there was 1 to 2 log₁₀ CFU/ml of growth on agar containing 10× the MIC (10 mg/liter) of tobramycin at 119 and 167 h. The MIC of colonies isolated from agar containing 10 mg/liter tobramycin at 167 h was 16 mg/liter (Table S2).

For meropenem monotherapies, bacterial killing ranged from virtually no killing with the 1 g every 8 h (q8h) regimen to ∼3 log₁₀ CFU/ml at 7 h for the 6-g/day continuous infusion (CI) regimen (Fig. 2). However, rapid regrowth occurred with all meropenem monotherapy regimens such that by 23 h, growth exceeded that of the initial inoculum by at least 1 log₁₀ CFU/ml. By day 7, for the two highest-dose regimens (2 g q8h via a 30-min infusion and 6 g/day via CI), ∼30 to 70% of the total populations were replaced by bacteria growing on agar containing meropenem at 5× MIC (Fig. 3; Table S1); for the lower-dose meropenem regimens, bacteria growing at 5× MIC increased to ∼7 log₁₀ CFU/ml. The MICs of colonies isolated from agar containing 5× the meropenem MIC at 47 and 167 h were 64 to 128 mg/liter (Table S2). Tobramycin monotherapy achieved rapid and extensive bacterial killing such that no viable bacteria were detected at 4 and 7 h (Fig. 2). However, rapid and extensive regrowth to within 0.5 log₁₀ CFU/ml of the growth control count had occurred by 23 h. Bacteria growing on agar containing tobramycin at 5× MIC plateaued at ∼4.5 log₁₀ CFU/ml from 71 h onwards, and the MIC was 32 mg/liter on day 7 (Fig. 3, Table S2).

Combination regimens containing intermittently administered meropenem suppressed regrowth of the total bacterial population over that of bacteria treated with tobramycin alone by ∼1 extra day, followed by rapid regrowth that surpassed the initial inoculum by 71 h (Fig. 2). By day 7, there was 5.0 to 5.7 log₁₀ CFU/ml of growth measured on agar containing 5× the meropenem MIC, and 6.0 to 6.6 log₁₀ CFU/ml grew on agar containing 10× the tobramycin MIC; the meropenem and tobramycin MICs of the colonies collected from the plates were 64 and 32 mg/liter, respectively (Fig. 3, Table S2). The combination containing 3 g/day CI meropenem achieved suppression of regrowth over 3 to 4 days. After that, regrowth occurred, and the total population was within 2.0 to 2.5 log₁₀ CFU/ml of the control count on days 6 and 7 (Fig. 2). By 167 h, 4.6 log₁₀ CFU/ml of growth was measured on both agar containing 5× the meropenem MIC and agar containing 10× the tobramycin MIC; the meropenem and tobramycin MICs of the colonies collected from the plates were 128 and 32 mg/liter, respectively (Fig. 3, Table S2). Only the combination containing 6 g/day CI meropenem achieved suppression of regrowth of the total population across 7 days (Fig. 2), and this was confirmed by conducting two biological replicates on different days, which showed excellent reproducibility. There were only two less-susceptible colonies observed (both growing on agar containing 5× the tobramycin MIC at 167 h for one of the two biological replicates) at any time (Fig. 3).

To explore the impact of an even higher meropenem MIC on the antibacterial effect of the combination, we evaluated a second carbapenem-resistant clinical P. aeruginosa isolate (CR380) in the HFIM (Fig. S3). For CR380 (initial inoculum of 6.8 log₁₀ CFU/ml), the combination of 6 g/day CI meropenem plus tobramycin (7 mg/kg q24h) achieved extensive reductions in bacterial counts following each of the first three tobramycin doses. Regrowth was observed in between these doses and after day 3. No colonies were observed on antibiotic-containing plates, except for 6.8 log₁₀ CFU/ml on agar containing 5 mg/liter tobramycin at 167 h.

The MBM that was developed simultaneously described the effects of meropenem (as short-term infusions and CI), tobramycin, and the combinations against isolate CR382 in the HFIM and yielded unbiased and sufficiently precise curve fits of the viable counts (Fig. 4; Fig. S2A and B; see the text in the supplemental material for further details on MBM). The parameter estimates are in Table S3. Three bacterial subpopulations with different susceptibilities to meropenem and tobramycin (i.e., different antibiotic concentrations are required for half-maximal bacterial killing) were required to describe the observed bacterial counts. Direct bacterial killing (as opposed to inhibition of successful replication or inhibition of growth) by both meropenem and tobramycin best described the antibacterial effects. Amplification of preexisting resistant subpopulations explained the bacterial regrowth for the failed regimens in the model. Subpopulation synergy was able to describe the time course of bacterial killing and regrowth for the combination dosage regimens. The effect of mechanistic synergy due to tobramycin enhancing the target site concentration of meropenem (i.e., via the permeabilizing effect of tobramycin on the bacterial outer membrane [OM]) was estimated to be relatively small. The parameter I_max,OM,TOB (i.e., maximum fractional decrease of KC_50,MEM by tobramycin via outermembrane disruption) could not be estimated with good precision and had a large uncertainty. In contrast, all standard errors in the final model (without an outer membrane effect) were <25%. Also, the outer membrane effect did not significantly enhance the model’s performance for this carbapenem-resistant isolate. Thus, there was no statistical justification for the inclusion of this effect in the model, and the most parsimonious model was chosen as the final model. The model developed for CR382 was able to predict the overall trend in the observed viable-count profile for CR380 with a change in only one parameter estimate, an increase in KC_50,IR,MEM (i.e., meropenem concentration required for half-maximal killing of the IR population) to 6.9 mg/liter, in agreement with isolate CR380 being more meropenem resistant than isolate CR382 (Fig. S3).

FIG 4.

Mechanism-based model (MBM) fits (lines) for observed (symbols) viable-count profiles of the total populations for the indicated meropenem (MEM) and tobramycin (TOB) regimens, in monotherapies and combinations, against CR382 in the HFIM. Observed and fitted viable counts below the limit of counting (i.e., below 1.0 log₁₀ CFU/ml) are plotted at 1.0 log₁₀ CFU/ml.

DISCUSSION

In this study, we examined meropenem and tobramycin regimens against carbapenem-resistant P. aeruginosa isolates. Carbapenems and aminoglycosides are often used in critically ill patients (23, 24). The most important pharmacokinetic/pharmacodynamic (PK/PD) index for the activity of β-lactam antibiotics is the duration of time (T) over which unbound (free [f]) concentrations remain above a multiple of the MIC of the infecting pathogen (%fT_>MIC) (25). Short-term infusions (0.5 h) remain the standard mode of administration for β-lactams, including meropenem, in many countries (23, 26), although this dosing approach commonly produces suboptimal antibiotic concentrations in critically ill patients with conserved renal function (26–28). PK/PD simulation studies have suggested greatly improved PK/PD target attainment for continuous or extended (2- to 4-h) infusions (27–30), and consequently, this mode of administration is increasingly used in critically ill patients (11, 23, 26). Patients with ARC have substantially increased meropenem clearances and, thus, decreased meropenem exposures (28, 31–33), placing them at significant risk of subtherapeutic antibiotic concentrations. Similarly, tobramycin clearances are affected by ARC, which impacts on PK/PD target attainment, especially for the area under the unbound concentration-time curve over 24 h divided by the MIC (fAUC₂₄/MIC), an important PK/PD index for aminoglycosides (10, 31, 34).

Using isolate CR382, we simulated the PK of ARC in the HFIM for meropenem at the standard and maximum FDA-recommended daily doses and for tobramycin at the most common daily dose for sepsis and septic shock (24). The results of meropenem monotherapy at 1 or 2 g q8h (0.5-h infusion) in the HFIM, namely, early treatment failure with regrowth and increases in less-susceptible populations, agreed with the SCTK results for CR382. In the SCTK, monotherapy of 25 mg/liter meropenem (achieved only briefly with intermittent short-term infusions) was required for suppression of regrowth over 48 h. The results of the meropenem intermittent monotherapy in the HFIM also closely match those of our previous HFIM study where these intermittent regimens were simulated under the ARC condition against a meropenem-susceptible (MIC of 0.25 mg/liter) P. aeruginosa isolate (15). For carbapenems, in vitro and animal in vivo studies traditionally reported that a %fT_>MIC of >40% is required for optimal bactericidal activity (35, 36). However, in the current study, rapid regrowth accompanied by increases of less-susceptible populations occurred even with 6 g/day meropenem as CI, where a %fT_>MIC of 100% was achieved (Table 1). Similar failure of meropenem monotherapy with emergence of resistance has been reported elsewhere, including in clinical studies, despite target attainment at or close to 100% fT_>MIC (11, 37, 38). Regrowth with β-lactams is often observed when concentrations fall below the MIC (39, 40). Such findings from in vitro and clinical studies support a more aggressive target in critically ill patients, challenging the traditional PK/PD targets (11, 31, 41–44). Various studies on β-lactams that have examined resistance suppression have shown a range of values from 100% fT_{>2× MIC} to 100% fT_{>6× MIC} required to prevent regrowth and suppress resistance (15, 37, 38, 45), and a target of 100% fT_{>4–5× MIC} has been suggested to prevent resistance emergence (46, 47). In our previous HFIM studies, all meropenem and piperacillin regimens that achieved 100% fT_{>5× MIC} successfully suppressed regrowth and resistance (15, 16). However, against P. aeruginosa in the presence of ARC, such targets are unachievable with short-term infusions and would be achievable only for susceptible isolates with CI. In a recent clinical study, Carrié et al. (11) found that 12% of 79 patients did not achieve the target of 100% fT_>4×MIC, and this was significantly associated with creatinine clearance (CL_CR) of ≥170 ml/min, i.e., ARC. Patients with CL_CR of ≥170 ml/min also had significantly higher rates of therapeutic failure. Overall, resistance emergence during treatment was reported in 2/13 patients (15%) with therapeutic failure. Thus, investigation of alternative treatment strategies for critically ill patients with ARC, including combination antibiotic regimens, is essential.

TABLE 1.

Concentrations, exposures, and pharmacokinetic/pharmacodynamic indices for the meropenem and tobramycin dosing regimens evaluated in the HFIM against P. aeruginosa CR382 and CR380^a

Isolate, antibiotic	Regimen	fCmax/fCmin or fCSS (mg/liter)	fAUC24 (mg · h/liter)	fCmax/MIC or fCSS/MIC	fCmin/MIC	fAUC24/MIC	%fT>MIC	%fT>5×MIC
CR382
Meropenem	1 g q8h	26.5/0.01	97.3	3.31	0.001	12.2	19.7	0
	2 g q8h	52.9/0.02	195	6.61	0.002	24.3	28.9	5.9
	3 g/day CI	4.06	97.3	0.51		12.2	0	0
	6 g/day CI	8.11	195	1.01		24.3	100	0
Tobramycin	7 mg/kg q24h	23.9/0.00	43.8	23.9	0.000	43.8
CR380
Meropenem	6 g/day CI	8.11	195	0.253		6.08	0	0
Tobramycin	7 mg/kg q24h	23.9/0.00	43.8	23.9	0.000	43.8

^aCR382 has a meropenem MIC of 8 mg/liter and a tobramycin MIC of 1 mg/liter. CR380 has a meropenem MIC of 32 mg/liter and a tobramycin MIC of 1 mg/liter. All values presented relate to the PK at steady state for patients with augmented renal clearance (ARC [CL_CR of 250 ml/min]). CI, continuous infusion; fC_max, unbound maximal concentration; fC_min, unbound minimal concentration; fC_SS, unbound concentration at steady state; fAUC₂₄, area under the unbound concentration-time curve over 24 h; fC_max/MIC, ratio of fC_max to MIC; fC_SS/MIC, ratio of fC_SS to MIC; fC_min/MIC, ratio of fC_min to MIC; fAUC₂₄/MIC, ratio of fAUC₂₄ to MIC; %fT_>MIC, percentage of time that unbound concentration exceeded MIC; %fT_>5×MIC, percentage of time that unbound concentration exceeded 5× MIC.

To the best of our knowledge, this is the first HFIM study to investigate the meropenem-tobramycin combination in the context of ARC. Combinations containing intermittent meropenem failed ∼1 day later than monotherapies, whereas the combination with meropenem at 3 g/day CI suppressed regrowth of CR382 until the end of day 3. Only the combination containing meropenem as a 6-g/day CI suppressed regrowth over 7 days. The success or failure of the regimens is interesting in light of the PK/PD targets discussed above. All meropenem regimens were ineffective as monotherapies and failed to achieve targets for resistance suppression, such as 100% fT_{>2× MIC} or higher (15, 37, 38, 45). The 6-g/day CI meropenem regimen, however, exceeded (if only just) the MIC at all times, including in combination when tobramycin concentrations had declined. Thus, while no meropenem regimen achieved the PK/PD targets for resistance suppression, this was offset for the 6-g/day CI when in combination with tobramycin. That the 2-g q8h regimen (same daily dose as a 6-g/day CI) in combination with tobramycin failed early highlights that the shape of the meropenem concentration-time profile is important for efficacy, not only for carbapenem monotherapy (48–50) but also as part of a synergistic combination. This may be particularly important for difficult isolates, e.g., those that are carbapenem resistant, as in this study, or hypermutators (51, 52).

For aminoglycosides, clinical PK/PD targets of unbound maximal concentration (fC_max)/MIC of 8 to 10 and fAUC/MIC of >70 have been proposed (31). In our present study, the fC_max/MIC for tobramycin at 7 mg/kg q24h was 24 and, thus, achieved its target, whereas the fAUC/MIC was 44 and below the target for this PK/PD index (Table 1). These values correspond well with the performance of the tobramycin monotherapy, i.e., rapid and extensive (∼7 log₁₀ CFU/ml) initial bacterial killing followed by regrowth and emergence of less-susceptible populations (Fig. 2 and 3). Therefore, the primary benefit of combination therapy with 6-g/day CI meropenem was not additional bacterial killing but suppression of the regrowth and emergence of meropenem- or tobramycin-resistant populations. Carbapenems enter the periplasmic space of P. aeruginosa through the OprD outer membrane porin, with meropenem resistance most commonly occurring via downregulation of the OprD gene (53, 54). Enzymatic inactivation via carbapenemases is also a major determinant of resistance (55). For tobramycin, upregulation of the MexXY-OprM efflux pump, which reduces the intracellular accumulation of antibiotics, including aminoglycosides, is the most common mechanism of aminoglycoside resistance in P. aeruginosa (56–58), although inactivation by aminoglycoside-modifying enzymes (59) and modification of the binding site by 16S rRNA methylases (60) also contribute. The success of the combination containing meropenem administered as a 6-g/day CI against CR382 may be due to the combination of maintaining the meropenem concentration above the MIC at all times and the different resistance mechanisms operative for each antibiotic. On the latter point, the requirement for separate and independent mutations for resistance development has been shown to help minimize the selection of mutants resistant to multiple drugs (61). Our MBM, which reflected this scenario through subpopulation synergy, successfully described and predicted the viable-count profiles for all treatments studied. Model predictions for the estimated subpopulations agreed with the observed bacterial counts from antibiotic-containing plates (Fig. 3), which suggested that regrowth with the suboptimal combination regimens was mainly due to the emergence of resistance to tobramycin. Overall, counts on tobramycin-containing agar were present earlier and at higher numbers than those on meropenem-containing agar. In the MBM, growth of the tobramycin-resistant population mainly drove the regrowth for the suboptimal combination regimens; meropenem—when present at sufficiently high and sustained concentrations—killed the tobramycin-resistant population, resulting in subpopulation synergy. The model-estimated tobramycin-resistant population had, to some extent, decreased susceptibility to meropenem; this agreed with the MIC results (Table S2).

The current study also investigated isolate CR380 (meropenem MIC of 32 mg/liter). In the HFIM, meropenem as a 6-g/day CI (unbound concentration at steady state [fC_SS] of 8.11 mg/liter, ∼0.25× MIC) plus tobramycin achieved extensive bacterial killing following the first three tobramycin doses; however, regrowth occurred in between these doses and thereafter. This agreed with the SCTK, where 25 mg/liter meropenem was required with 2 mg/liter tobramycin to suppress regrowth of CR380, whereas 8 mg/liter meropenem plus 2 mg/liter tobramycin was sufficient for regrowth suppression of CR382. Comparison of the results from isolates CR382 and CR380 suggests that a meropenem concentration at or close to the MIC may be required in combination with tobramycin to suppress regrowth and resistance of these carbapenem-resistant isolates.

The present study has important strengths. We studied two carbapenem-resistant isolates in both SCTK and HFIM studies. It is the first HFIM study to assess the meropenem-tobramycin combination against P. aeruginosa in the context of ARC and the first to examine carbapenem-resistant isolates while simulating ARC. Two HFIM studies (37, 52) and one in vivo animal study (62) also examined this combination against P. aeruginosa; however, those studies involved susceptible isolates and did not simulate ARC. In addition, our HFIM study was performed over 7 days, examined total as well as less-susceptible bacterial populations via drug-containing plates, and included the development of an MBM that described the data observed in the dynamic model well. Importantly, our current study demonstrated a substantial impact of the mode of meropenem administration (even for the same daily dose in the combination) on the overall synergistic effect against isolate CR382. Furthermore, we examined in the HFIM a second carbapenem-resistant clinical P. aeruginosa isolate (CR380) to explore the impact of an even higher meropenem MIC on the antibacterial effect of the combination. Future studies may use additional isolates to assess the combination. A limitation of our study is that genotypic data on the isolates were not available. The study focused on optimizing the meropenem regimen, and thus, one dosage regimen (recommended by the Infectious Diseases Society of America [IDSA]) was studied for tobramycin. This may have impacted the ability of the MBM to identify a tobramycin effect on the bacterial outer membrane as significantly contributing to the observed synergy. A general limitation of the HFIM is that it lacks an immune system. Future animal studies may be directed at evaluating immune effects on residual bacteria after the initial killing by the antibiotics. However, accurate simulation of humanized antibiotic concentration-time profiles is very challenging in animals due to the substantial differences in clearance and elimination half-life. Furthermore, as we showed in this HFIM study, evaluation of resistance suppression requires long study durations, which cannot be employed in the typical animal models for ethical reasons.

In conclusion, this study has demonstrated the great difficulties involved in effectively treating carbapenem-resistant P. aeruginosa infections under the pharmacokinetic conditions imposed by ARC. As monotherapy against a meropenem-resistant and tobramycin-susceptible isolate, even high doses of meropenem or tobramycin were unsuccessful: bacterial killing was limited, and early regrowth and amplification of less-susceptible populations occurred. In combination with tobramycin, even 6 g/day of intermittently administered meropenem or 3 g/day meropenem as a continuous infusion failed to provide a sustained antibacterial effect against an isolate with a meropenem MIC at the resistance breakpoint. Only the combination that included the high daily dose of meropenem administered as a continuous infusion successfully suppressed the regrowth and amplification of less-susceptible populations across 7 days. This combination regimen also achieved extensive bacterial killing over 3 days for an isolate that was even more resistant to meropenem. Further investigations of combination therapies, including rationally designed and optimized regimens supported by MBM and Monte Carlo simulations, are warranted in the context of ARC.

MATERIALS AND METHODS

Antibiotics, media, bacterial isolates, and susceptibility testing.

Meropenem (lot number MAUS1016; Fresenius Kabi, NSW, Australia) and tobramycin (lot number 70625C; AK Scientific, Union City, MD, USA) stock solutions were prepared as previously described (15, 63). Viable counting was performed on cation-adjusted Mueller-Hinton agar (CAMHA) (25 mg/liter Ca²⁺ and 12.5 mg/liter Mg²⁺; University of Melbourne, Australia). Drug-containing agar plates were prepared by adding meropenem or tobramycin stock solution to CAMHA (lot number 7076682; BD, Sparks, MD, USA). Cation-adjusted Mueller-Hinton broth (CAMHB) (lot number 6258541; BD, Sparks, MD, USA) containing 20 to 25 mg/liter Ca²⁺ and 10 to 12.5 mg/liter Mg²⁺ was used for all studies. The clinical P. aeruginosa isolates were obtained from a critically ill patient with burns at the Royal Brisbane and Women’s Hospital, Australia. Isolate CR382 (meropenem MIC of 8 mg/liter, tobramycin MIC of 1 mg/liter) was from a wound swab, and isolate CR380 (meropenem MIC of 32 mg/liter, tobramycin of MIC 1 mg/liter, multidrug resistant) from a tracheal aspirate. The CLSI agar dilution method was used to determine MICs prior to and following drug exposure (64). Susceptibility and resistance were defined, according to CLSI guidelines, as MICs of ≤2 mg/liter and ≥8 mg/liter for meropenem and ≤4 mg/liter and ≥16 mg/liter for tobramycin (64).

SCTK experiments.

The static-concentration time-kill (SCTK) experiments were conducted as described previously (65, 66). Clinically relevant meropenem (4, 8, and 25 mg/liter) and tobramycin (1, 2, and 8 mg/liter) concentrations were studied as monotherapies and combination therapies against both isolates (CR382 and CR380). These concentrations represented the range of unbound concentrations achieved in patients with ARC. The meropenem monotherapies, 8 mg/liter tobramycin monotherapy, and the meropenem combination therapies with 8 mg/liter tobramycin were studied in three biological replicates. The initial inoculum for each isolate was ∼10⁷ CFU/ml. Serial samples were collected at 5 to 10 min predose (0 h) and at 1.5, 5, 24, 29, and 48 h. The bacterial suspensions were centrifuged at 24 h, the supernatant carefully removed, and bacteria resuspended in fresh, prewarmed medium that contained the targeted antibiotic concentration(s). As previously described, supplemental meropenem doses corresponding to 30% of the original dose were added at 6 and 30 h to further offset the known thermal degradation of meropenem (52, 67). The samples for viable counting were washed twice in sterile saline, and counts determined by manually plating 100 μl of undiluted or appropriately diluted suspension in saline onto agar plates.

HFIM, simulated regimens, and assessment of microbiological response.

The hollow-fiber in vitro infection model (HFIM) employed has been described previously (15, 68, 69). Experiments with isolates CR382 and CR380 were conducted over 7 days (initial inoculum of ∼10⁷ CFU/ml) at 37°C in a humidified incubator and using cellulosic cartridges (batch B720170715b; FiberCell Systems, Inc., Frederick, MD, USA). The unbound meropenem and tobramycin plasma concentration-time profiles for patients with ARC (creatinine clearance [CL_CR] of 250 ml/min) (70) were simulated in silico using Berkeley Madonna (version 8.3.18), based on population pharmacokinetic models for critically ill patients (28, 34). For meropenem, the simulated clearance was 30.3 liters/h and the half-life (t_1/2) was 0.66 h, and for tobramycin, the clearance was 12.9 liters/h and the t_1/2 was 1.1 h. The dosing regimens studied against isolate CR382 were as follows: meropenem at 1 g and 2 g administered via a 30-min intravenous (i.v.) infusion q8h; meropenem at 3 g/day and 6 g/day as a continuous infusion (CI); and tobramycin at 7 mg/kg as a 30-min i.v. infusion q24h. The meropenem and tobramycin concentration-time profiles resulting from the in silico simulations were reproduced experimentally in the HFIM as monotherapies and combinations (Table 1; Fig. S1). The combination of 6-g/day CI meropenem with tobramycin was performed in two biological replicates on different days. No loading doses were required for intermittent regimens. For CI, meropenem was delivered as a constant concentration by spiking the medium flowing through the system, and a loading dose was given as a bolus to attain steady-state concentrations immediately. The diluent reservoirs and syringes delivering meropenem were stored in the refrigerator and changed every 24 h to prevent thermal degradation (71). An untreated growth control was included. The dosing regimen studied against isolate CR380 was 6-g/day CI meropenem plus tobramycin (7 mg/kg q24h).

At 0, 1.5, 4, 7, 13, 23, 28, 47, 55, 71, 95, 119, 143, and 167 h, samples (1.0 to 1.5 ml) were collected for viable counting, centrifuged (4,000 × g for 5 min), and resuspended in saline twice to reduce antibiotic carryover. Appropriately diluted samples were manually plated on CAMHA containing no drug (100 μl plated; limit of counting, 1.0 log₁₀ CFU/ml). Less-susceptible subpopulations were quantified at 0, 23, 47, 71, 119, and 167 h. Samples (200 μl; limit of counting, 0.7 log₁₀ CFU/ml) were manually plated onto CAMHA containing meropenem at 3× or 5× MIC (24 or 40 mg/liter meropenem) for isolate CR382 and at 3× MIC (96 mg/liter) for isolate CR380 or tobramycin at 5× or 10× MIC (5 or 10 mg/liter) for both isolates. Log₁₀ mutant frequencies (log₁₀MF) were calculated as follows: log₁₀MF = log₁₀(CFU/ml on antibiotic-containing agar) − log₁₀(CFU/ml on antibiotic-free agar). Although some viable counts were too low to observe colonies on antibiotic-containing agar, these samples still provided information on the upper limit of the log₁₀MF. For CR382, changes in MICs from the baseline were determined at 47 and 167 h using a subset of at least three colonies from plates containing meropenem at 5× MIC (40 mg/liter) or tobramycin at 10× MIC (10 mg/liter).

Quantification of meropenem and tobramycin.

Samples for pharmacokinetic determinations were collected periodically from the HFIM. Meropenem concentrations were determined by a validated ultrahigh-performance liquid chromatography–photodiode array (UHPLC-PDA) method. Tobramycin concentrations were determined by a validated UHPLC-tandem mass spectrometry (UHPLC-MS/MS) method. Detailed descriptions of the assays are provided in the supplemental material.

Mechanism-based modeling.

MBM of the HFIM data for isolate CR382 was conducted by using the S-ADAPT software (version 1.57) with an importance sampling algorithm (pmethod = 4) and SADAPT-TRAN for pre- and postprocessing (72, 73). Bacterial growth and replication of P. aeruginosa were modeled using a life cycle growth model (74–76). The extent and time course of synergy in bacterial killing and suppression of regrowth were evaluated. Models for subpopulation synergy (i.e., meropenem killing the tobramycin-resistant bacteria and vice versa) and/or mechanistic synergy (via an outer membrane [OM]-permeabilizing effect of tobramycin, i.e., tobramycin enhancing the bacterial killing by meropenem) were assessed (65, 75, 77–79). Competing models were distinguished by use of the S-ADAPT objective function value (−1 · log likelihood), standard diagnostic plots, the coefficient of correlation, visual predictive checks, and consideration of the biological plausibility of the parameter estimates (63, 80–82). All effects included in the final model were required to describe the bacterial-count profiles.

Data availability.

The supplemental material includes the differential equations for the final model.