Efficacy of Ceftolozane-Tazobactam in Combination with Colistin against Extensively Drug-Resistant Pseudomonas aeruginosa, Including High-Risk Clones, in an In Vitro Pharmacodynamic Model

Combination therapy is an attractive therapeutic option for extensively drug-resistant (XDR) Pseudomonas aeruginosa infections. Colistin has been the only treatment available for these infections for many years, but its results are suboptimal. Ceftolozane-tazobactam (C/T) is a newly available therapeutic option that has shown good antipseudomonal activity, even against a number of XDR P. aeruginosa strains. However, data about combinations containing C/T are scarce.

ABSTRACT

Combination therapy is an attractive therapeutic option for extensively drug-resistant (XDR) Pseudomonas aeruginosa infections. Colistin has been the only treatment available for these infections for many years, but its results are suboptimal. Ceftolozane-tazobactam (C/T) is a newly available therapeutic option that has shown good antipseudomonal activity, even against a number of XDR P. aeruginosa strains. However, data about combinations containing C/T are scarce. The aim of this study was to analyze the activity of C/T and colistin alone and in combination against a collection of XDR P. aeruginosa strains containing 24 representative clinical isolates from a multicentre Spanish study. Twenty-four time-kill experiments performed over 24 h were conducted in duplicate to determine the effects of colistin and C/T alone and combined. An in vitro pharmacodynamic chemostat model then was used to validate this combination against three selected XDR P. aeruginosa ST175 isolates with different susceptibility levels to C/T. Static time-kill assays demonstrated superior synergistic or additive effect for C/T plus colistin against 21 of the 24 isolates studied. In the in vitro dynamic pharmacokinetic/pharmacodynamic (PK/PD) model, the C/T regimen of 2/1 g every 8 h with a steady-state concentration of 2 mg/liter colistin effectively suppressed the bacterial growth at 24 h. Additive or synergistic interactions were observed for C/T plus colistin against XDR P. aeruginosa strains and particularly against C/T-resistant strains. C/T plus colistin may be a useful treatment for XDR P. aeruginosa infections, including those caused by high risk-clones resistant to C/T.

INTRODUCTION

Antibiotic resistance has existed since ancient times (1), but it is now a serious global health threat, responsible for over 0.7 million deaths each year (2). Bacteria can easily acquire new antibiotic resistance through chromosomal mutations and horizontal gene transfer (3). Pseudomonas aeruginosa has an outstanding capacity to develop antibiotic resistance and is a leading cause of morbidity and mortality worldwide. This microorganism has a nonclonal epidemic population structure (4), and its antibiotic resistance can be caused by several mechanisms.

The recently designated high-risk P. aeruginosa clones are widely distributed in hospitals around the world and have been directly linked to severe, complex, and difficult-to-treat infections. ST111, ST175, and ST235 appear to be the most prevalent of these clones (4). ST175 is particularly common in several European countries, including Spain and France (5). It has been associated with multidrug-resistant (MDR) isolates and is a recognized hospital contaminant.

Patients with extensively drug-resistant (XDR) P. aeruginosa infections are at an increased risk of receiving inadequate initial antimicrobial therapy because of the limited treatment options available. Colistin is, and for many years has been, the only option available (6). Its use, however, is limited by nephrotoxicity (7–10), difficulties achieving therapeutic levels, and heteroresistance (11). Data from pharmacokinetic studies have confirmed that colistin plasma concentrations following EMA and FDA dosage recommendations are low and inadequate for the treatment of MDR/XDR P. aeruginosa infections (12). Thus, there is clearly a need to investigate the performance of colistin combined with other antibiotics in P. aeruginosa infections (13).

Ceftolozane-tazobactam (C/T) has emerged as a promising solution to the lack of new antibiotics that are effective against P. aeruginosa (14). C/T is generally reserved for MDR/XDR P. aeruginosa infections, but there are clinical situations, including XDR P. aeruginosa infections with a C/T MIC of >4 mg/liter (C/T-nonsusceptible isolates), in which treatment needs to be optimized, probably through combination therapy (15).

Several studies have recommended antibiotic therapy combining an antipseudomonal β-lactam and an aminoglycoside or fluoroquinolone (16–20). Data on combinations containing C/T, however, are scarce (21–24). Combination regimens featuring colistin are more common and were recently recommended in the IDSA guidelines for the treatment of critically ill patients with ventilator-associated pneumonia due to the high prevalence of MDR/XDR microorganisms (25).

This study aimed to evaluate the antibacterial activity of C/T and colistin alone and in combination against a representative collection of clinical XDR P. aeruginosa strains in an in vitro pharmacodynamic model simulating the free drug concentration achieved with recommended dosing regimens for each antibiotic.

RESULTS

In vitro antimicrobial susceptibility studies.

All of the isolates were tested for antibiotic susceptibility. Twelve (ST111 [10-009], ST235 [06-042], ST175 [12-012], ST235 [07-004], ST111 [04-017], ST253 [01-008], ST175 [07-016], ST2221 [10-019], ST2533 [10-021], ST274 [09-011], ST175 [15-001], and ST175 [09-012]) were resistant to C/T. Seven of the 12 C/T-resistant isolates harbored carbapenemases. Only one isolate, ST111 (10-009), was resistant to colistin. The susceptibility profiles and the β-lactam resistance mechanisms of the XDR isolates studied were obtained from a previous Spanish multicenter study (26) and are shown in Table 1.

TABLE 1.

Susceptibility profiles and resistance mechanisms of the 24 XDR P. aeruginosa isolates studied

Isolate	ST	β-Lactamase(s)	AmpC hyperproduction	OprD deficiency	Polymyxin resistance mechanism	MIC (mg/liter) fora:
						C/T	CST
04-017	111	OXA-46	Yes	No		8/4	2
10-009	111	VIM-2	Yes	Yes	parR-nt621Δ4	>64/4	4
04-025	175		Yes	Yes		2/4	1
07-016	175	GES-5	No	Yes		16/4	2
10-023	175		Yes	Yes		2/4	2
12-012	175	VIM-20, OXA-2	No	Yes		>64/4	2
06-014	179	OXA-10	Yes	Yes		4/4	2
07-004	235	GES-19, OXA-2	No	Yes		>64/4	2
06-042	235	VIM-47	No	No		>64/4	2
12-003	244		Yes	Yes		4/4	2
01-008	253	VIM-1	No	Yes		>64/4	2
09-011	274		Yes	Yes		8/4	1
09-007	313		Yes	Yes		4/4	2
10-017	395		Yes	No		1/4	2
06-035	455		Yes	No		4/4	0.5
10-019	2221		Yes	Yes		8/4	2
10-021	2533		Yes	Yes		8/4	1
06-025	2534		Yes	Yes		4/4	2
06-027	2535		Yes	No		4/4	2
06-001	2536		Yes	Yes		4/4	2
04-024	175		Yes	Yes		4/4	2
12-017	175		Yes	Yes		4/4	2
15-001	175		Yes	Yes		8/4	2
09-012	175		Yes	Yes		8/4	2

^aAbbreviations: C/T, ceftolozane-tazobactam; CST, colistin.

Three of the previous isolates were selected to perform a one-compartment in vitro chemostat model: ST175 (10-023), ST175 (09-012), and ST175 (07-016).

ST175 (10-023) was C/T susceptible, with a MIC of 2 mg/liter, and showed the typical resistance profile associated with this clone (resistance to all β-lactams except C/T) due to OprD inactivation and AmpC hyperproduction. ST175 (09-12) showed intermediate resistance to C/T with a MIC of 8 mg/liter. In addition to OprD inactivation and AmpC hyperproduction, it showed a specific mutation in PBP3 (R504C) associated with increased β-lactam resistance. Finally, ST175 (07-016) was C/T resistant with a MIC of 16 mg/liter and produced the class A carbapenemase GES-5.

Time-kill studies.

In the time-kill assays, colistin and C/T alone were not bactericidal in 83% and 87.5% of the samples, respectively. However, the combination of C/T plus colistin was synergistic in 54% of isolates and additive in 87.5%. The combination regimen was efficacious for both colistin and C/T alone in 21 of the 24 isolates studied, and no antagonism was observed for any of the isolates. The results are shown in Table 2. The Kruskal-Wallis test showed statistically significant differences between the conditions studied (X² = 57.52, gl = 3, P <  0.001). Post hoc tests allow us to affirm that μA > (μB = μC) > μD; therefore, the in vitro combined treatment of C/T and colistin is more effective than monotherapy with any of them. In the time-kill studies, growth in the nontreatment controls reached 9 to 10 log₁₀ CFU/ml by the 24-h time point for all regimens.

TABLE 2.

Time-kill experiments performed against 24 XDR P. aeruginosa strains^a

^aBactericidal effect (≥3-log₁₀ reduction in CFU/ml after 24 h) is highlighted in blue. Synergistic and additive effects (≥2-log₁₀ or ≥1-log₁₀ reduction in CFU/ml at 24 h with the combination compared with the most active single drug) are highlighted in orange and green, respectively. Abbreviations: Atb, antibiotic; CST, colistin; C/T, ceftolozane-tazobactam. Mean bacterial concentration (log₁₀ CFU/ml) is shown for each strain and antibiotic treatment. Positive results for bactericidal effect (Bact.E) and synergy at 24 h (Syn.24h) are highlighted.

Time-kill curves with colistin alone showed a similar pattern in all the isolates, with an initial 3 to 5 log₁₀ reduction after 2 h, followed by regrowth in all cases. In 11 isolates (ST111 [10-009], ST175 [10-023], ST175 [12-012], ST313 [09-007], ST395 [10-017], ST455 [06-035], ST2534 [06-025], ST2536 [06-001], ST175 [04-024], ST175 [12-017], and ST175 [15-001]), bacterial regrowth occurred, reaching concentrations similar to those of the controls. A bactericidal effect was observed for colistin in four isolates (ST235 [06-042], ST274 [09-011], ST2535 [06-027], and ST175 [09-012]). Three different behaviors were observed for C/T: five isolates (ST111 [10-009], ST175 [12-012], ST235 [07-004], STS35 [06-042], and ST235 [01-008]) showed the same behavior as the untreated control; 16 isolates showed a 2- to 4-log₁₀ CFU/ml reduction at 8 h, followed by a plateau in 14 cases; and the remaining isolate, ST2534 (06-025), showed a 4-log₁₀ CFU/ml increase at 24 h. A bactericidal effect was detected in five isolates (ST175 [07-023], ST2533 [10-021], ST2535 [06-027], ST2536 [06-001], and ST175 [09-012]).

All isolates treated with the C/T plus colistin combination showed an initial 2- to 5-log₁₀ CFU/ml reduction followed by a plateau. The combination had bactericidal effect in all cases, a synergistic effect at 24 h in 13 isolates (ST111 [04-017], ST111 [10-009], ST175 [10-023], ST175 [12-012], ST179 [06-014], ST244 [12-003], ST313 [09-007], ST395 [10-017], ST2534 [06-025], ST2536 [06-001], ST175 [04-024], ST175 [12-017], and ST175 [15-001]), and an additive effect at 24 h in all but three isolates (ST 2533 [10-021], ST 2535 [06-027], and ST175 [09-012]). In this last case, the results were similar to those observed for C/T and colistin as monotherapy. No antagonism was detected.

Antibiotic exposures.

The observed concentrations and pharmacokinetic parameters calculated for all antibiotic regimens over the 24 h of the chemostat experiments are shown in Table 3. Overall, the observed versus predicted drug exposures of ceftolozane achieved in this model were considered satisfactory for all regimens based on observed r² values of 0.97, 0.96, and 0.94 for ST175 (10-023), ST175 (09-12), and ST175 (07-016), respectively (Fig. 1).

TABLE 3.

Observed concentrations and calculated pharmacokinetic parameters calculated for all antibiotic regimens over 24 h in the chemostat experiments^c

Isolate and regimen	Free peak concn (mg/liter)		Free trough concn (mg/liter)
	Target	Observed [means (SD)]	Target	Observed [means (SD)]
ST175 (10-023)
C/Ta	87.6	85.39 (14.71)	17.4	15 (2.42)
Colistinb	1	1.90 (1.15)	NC	NC
ST175 (09-012)
C/T	87.6	112.8 (8.36)	17.4	30.6 (5.29)
Colistin	1	1.08 (0.33)	NC	NC
ST175 (07-016)
C/T	87.6	121.7 (12.05)	17.4	25.5 ± 8.91
Colistin	1	1.72 (0.38)	NC	NC

^aConcentration and pharmacokinetic data were targeted only for the ceftolozane component of C/T.

^bColistin was administered as a continuous infusion; the free peak concentration is the mean steady-state concentration over the 24 h of the experiment.

^cData are presented as the means and standard deviations. Abbreviations: C/T, ceftolozane-tazobactam; NC, not calculated.

FIG 1.

Relationships between observed and targeted ceftolozane concentrations for the three selected XDR P. aeruginosa ST175 isolates: ST175 (10-023), ST175 (09-012), and ST175 (07-016).

Chemostat studies.

In the in vitro pharmacokinetics/pharmacodynamics model, the C/T regimen of 2/1 g every 8 h (q8h) combined with steady-state concentrations of 2 mg/liter colistin effectively suppressed the bacterial growth at 24 h. Additive and synergistic interactions were observed for C/T plus colistin against XDR P. aeruginosa strains and particularly against strains that were resistant to C/T. These results are shown in Fig. 2.

FIG 2.

In vitro chemostat experiments with three selected XDR P. aeruginosa ST175 isolates with different susceptibility levels to C/T: ST175 (10-023), with a MIC of 2 mg/liter, ST175 (09-012), with a MIC of 8 mg/liter, and ST175 (07-016), with a MIC of 16 mg/liter. Values shown are mean numbers of CFU over 24 h for each P. aeruginosa isolate and antibiotic. LLOD, lower limit of detection.

In the C/T-susceptible ST175 (10-023) isolate, the addition of colistin to C/T resulted in a significant reduction in number of CFU compared with that of either antibiotic alone. There was a 1.74-log₁₀ CFU/ml difference in the reduction achieved with the more active antibiotic (C/T); therefore, the combination was deemed to have an additive effect.

In the intermediate-resistant C/T ST175 (09-12) isolate, C/T monotherapy resulted in a 2.41-log₁₀ CFU/ml reduction at 24 h. Nevertheless, the addition of colistin resulted in a significant reduction in bacterial burden compared with that for either antibiotic alone. Colistin monotherapy resulted in strong regrowth at 24 h. The difference of 2.57 log₁₀ CFU/ml achieved with the combination compared with the more active antibiotic (C/T) meant that the combination was classified as synergistic.

ST175 (07-016) was C/T resistant. Nevertheless, C/T monotherapy resulted in a 3.46-log₁₀ CFU/ml reduction at 24 h, but its bactericidal effect was low compared with that of the combination regimen. C/T plus colistin resulted in a significant reduction of 1.44 log₁₀ CFU/ml compared with the most active single drug (C/T), qualifying this combination as additive.

As an alternative endpoint, for each regimen (including the growth control) for the duration of the study, we calculated the log ratio (LR) for the area under the curve for CFU (AUCFU) as the total bacterial exposure. Relative to the control (reference), all monotherapies achieved greater than 1-log reduction against the three isolates, including a 2-log reduction achieved with C/T compared with the reference in the ST175 (07-016) isolate. AUCFU reductions with the combination regimens were <1 log relative to the reduction achieved with each antibiotic alone in the three isolates studied.

Table 4 shows in vitro chemostat results as log difference at 24 h and LR of AUCFU for each antibiotic alone (test) compared with the control (reference) and for each antibiotic combination (test) compared with each antibiotic alone (reference).

TABLE 4.

In vitro chemostat results

^aLog difference at 24 h for each antibiotic alone compared with the control and for each antibiotic combination compared with each antibiotic alone. Synergy and additive effect (≥2-log₁₀ or ≥1-log₁₀ reduction in CFU/ml at 24 h with the combination compared with the most active single drug) are highlighted in orange and green, respectively.

^bThe log difference is presented as the log ratio (LR), which is used to compare any number of log₁₀ CFU of two regimens (test/reference). AUCFU, area under the curve for CFU.

Resistance studies.

In the chemostat model, C/T-resistant strains were not found with C/T alone or in combination with colistin over the 24 h of the experiments. The emergence of a colistin-resistant subpopulation was detected at the end of the experiment in the control and in the chemostat cultures receiving colistin alone. No colistin-resistant subpopulations were detected for the combination of C/T plus colistin. The frequency of the colistin-resistant subpopulation at concentrations of 2-fold the MIC was 1 CFU/ml in 3.26 × 10¹¹, 4.08 × 10⁸, and 3.84 × 10⁷ CFU/ml for ST175 (10-023), ST175 (09-12), and ST175 (07-016), respectively. At concentrations 4-fold the MIC, the respective frequencies were 1 CFU/ml in 3.8 × 10¹¹, 1.37 × 10⁸, and 2.67 × 10⁸ CFU/ml.

DISCUSSION

C/T emerged as a beacon of hope for the treatment for MDR/XDR Gram-negative bacteria, and it is one of the latest additions to the antibiotic armamentarium for treating severe infections caused by P. aeruginosa. In routine practice, however, there are many patients with particularly severe infections, or with a MIC for the MDR/XDR P. aeruginosa strains tested that is above the susceptibility breakpoint, who could benefit from combination therapy with C/T and colistin. Another reason for using this combination is to prevent the development of resistance, especially to β-lactams.

Our study evaluated combination therapy with C/T and colistin in a large collection of representative XDR P. aeruginosa isolates, including prevalent high-risk clones. The 24 isolates selected were resistant to all the β-lactams tested, and 12 of them (50%) were also resistant to C/T, with 7 of them harboring carbapenemases. Just one isolate (ST111) was resistant to colistin. We selected these highly resistant strains because they cause precisely the type of infections that could benefit from combination therapy. Few studies have examined combination therapy of C/T plus colistin. Monogue and Nicolau (27) published the first study assessing synergy of C/T against Gram-negative microorganisms and demonstrated synergistic effects for C/T combined with colistin or fosfomycin using time-kill curves. Rico Caballero et al. (24) further added to this body of knowledge in an in vitro pharmacodynamic study that showed greater overall reductions in bacterial burden and additive or synergistic effects for C/T combined with amikacin or colistin.

Time-kill curves were generated to evaluate the effects of C/T and colistin alone and in combination in all of the isolates. Colistin showed a similar pattern across the isolates (initial 3- to 5-log reduction in bacterial growth after 2 h, followed by regrowth in all cases), supporting previous reports (24).

The combination of colistin and C/T led to a rapid and sharp decrease in bacterial burden in all isolates but one, regardless of C/T resistance. A moderate decrease was also observed for C/T plus colistin for the isolate resistant to both antibiotics. Following the initial decrease in bacterial burden, there was a plateau in the curves with this antibiotic combination. In disagreement with other reports (21, 24, 27) and in contrast to when colistin was used as a monotherapy, no regrowth was observed for C/T plus colistin for any of the isolates. C/T combined with colistin had a bactericidal effect in all of the isolates. The combination was synergic at 24 h in 13 isolates and additive in an additional eight. Interestingly, synergy was observed in five of the non-C/T-resistant isolates, two of which had a MIC of ≥64 mg/liter. The aforementioned rapid killing caused by colistin probably contributed to the synergistic effect observed for this antibiotic combination, even in some of the C/T-resistant isolates. This effect is probably due to the different mechanisms of action of these two antibiotics (28), and their combined effect on bacterial cells would explain the mechanism of action of the combination. Colistin acts against the lipopolysaccharide in the outer membrane, causing local disturbance, permeability changes, osmotic imbalance, and death cell. The resulting increase in permeability would facilitate the uptake of C/T.

The above-described observations varied from the in vitro pharmacodynamics model. We chose three isolates from the most prevalent high-risk clone in our environment, ST175, which had C/T MIC values of between 2 and 16 mg/liter. Interestingly, the human-simulated free concentration of C/T and colistin had a synergistic effect in the ST175 (09-012) isolate, with a C/T MIC of 8 mg/liter. An additive effect was observed in the C/T-susceptible strain ST175 (10-023) and, interestingly, also in the resistant strain ST175 (07-016), with a C/T MIC of 16 mg/liter. It should be noted that the resistant isolate responded to C/T monotherapy, achieving a 3.5-log reduction, which is at odds with data from Rico Caballero et al. (24). Nevertheless, previous studies have demonstrated that adequate C/T concentrations confer a more favorable PTA profile for infections with higher MICs (15, 29, 30).

No antagonism for C/T and colistin against any of the P. aeruginosa isolates was observed in our study.

In agreement with previous reports, we have shown that the combined use of β-lactams and colistin led to increased activity against MRD/XDR P. aeruginosa compared with that of either agent used as monotherapy. One particularly relevant discovery was the role of this combination in C/T-resistant strains.

In our study, monotherapy with colistin resulted in the development of colistin-resistant subpopulations, and it was also present in the control without any exposure to antibiotic during the experiment. These data are helpful to understand the results, since heteroresistance already could be present, and therefore these isolates killed with the addition of C/T, or resistance could be caused by suboptimal colistin concentrations, and therefore C/T can prevent this resistance development from occurring.

Some limitations of this study should be noted. In vitro studies cannot examine toxicity, the contribution of the immune system, or the different PK/PD effects occurring at the specific site of an infection. Studies were only 24 h in length; longer-duration experiments are needed to represent the clinical administration guidelines and to assess the emergence of resistance.

To our knowledge, this is the first study to assess the synergy of C/T plus colistin against a large collection of representative XDR P. aeruginosa isolates, including prevalent high-risk clones, and the first to focus on ST175 C/T nonsusceptibility isolates. Our findings may help to identify novel strategies to improve the treatment of MDR/XDR P. aeruginosa infections using currently available drugs.

MATERIALS AND METHODS

Bacterial isolates.

Twenty-four clinical XDR P. aeruginosa isolates were used. These isolates previously had been collected from nine Spanish hospitals in the multicenter COLIMERO trial and characterized at a molecular level using pulsed-field gel electrophoresis, multilocus sequence typing, and whole-genome sequencing (26). The selected isolates are representative of all the clones and resistance mechanisms detected in the trial.

Antibiotics.

C/T (Zerbaxa) and colistin (colistin sulfate) were provided by Merck Sharp & Dohme (MSD) and from Sigma-Aldrich, respectively. Antibiotic solutions were prepared according to CLSI guidelines (31). Concentrations for time-kill experiments were based on AUC serum levels: for colistin, 4.5 MIU q12h, area under the concentration-time curve for 24 h (AUC₂₄), 50 μg·h/ml (32–34); for C/T, 2/1 g q8h, AUC₂₄, 912/150 μg·h/ml (35).

In the chemostat model, they were administered to simulate free plasma concentrations in critically ill patients under treatment for several infections. The simulated C/T dosing regimen was 2/1 g every 8 h by intravenous infusion over 1 h (current standard) to achieve a free maximum concentration of 90 mg/liter, with a simulated elimination half-life of 3 h and protein binding of 20% (23, 36, 37). It was assumed that tazobactam would be eliminated at the same rate as ceftolozane, since it has a limited role in this drug’s activity against P. aeruginosa (38). In the dynamic model, we simulated a continuous infusion of colistin to achieve concentrations of 2 mg/liter to mimic plasma colistin concentration-time profiles in critically ill patients (34). Due to protein binding for colistin being 50% (24), we simulated a free steady-state concentration of 1 mg/liter. C/T and colistin concentrations were validated by high-performance liquid chromatography (HPLC) (39, 40).

In vitro antimicrobial susceptibility testing.

Antimicrobial susceptibility testing was performed according to the CLSI guidelines (31) for broth microdilution using cation-adjusted Mueller-Hinton broth (CAMHB).

Time-kill experiments.

Time-kill analyses were performed on each of the 24 isolates to analyze the activity of C/T and colistin alone and in combination at clinically achievable free drug concentrations (when maximum indicated clinical doses were used). All experiments were performed in duplicate. The study flow is shown in Fig. 3.

FIG 3.

Study flow. We conducted 24 time-kill experiments with three selected antibiotics on our entire collection of P. aeruginosa high-risk clones to identify additive and synergistic effects. The combination of C/T plus colistin was validated in three selected ST175 isolates with different C/T susceptibility levels in a one-compartment chemostat model.

An overnight culture of each isolate was diluted with CAMHB and further incubated at 35°C to reach early-log-phase growth. The bacterial suspension was diluted with CAMHB according to absorbance at 630 nm; 30 ml CAMHB was placed in 50-ml sterile conical flasks with the corresponding antibiotics. The final concentration of the bacterial suspension in each flask was approximately 7 to 8 log₁₀ CFU/ml. Flasks were incubated at 35°C in a shaker water bath for 24 h. Bacterial growth was measured at 0, 2, 4, 8, 12, and 24 h. A 1-ml aliquot was obtained from each flask, centrifuged twice at 5,000 rpm for 5 min, and then reconstituted with sterile saline solution to its original volume to minimize drug carryover. Serial decimal dilutions in CAMHB were performed, and 200 μl was plated on Muller-Hinton E (MHE) agar plates to quantify the total bacterial count for each sample. The inoculated plates were incubated in a humidified incubator (35°C) for 18 to 24 h. Bacterial colonies were counted after the overnight incubation, and the bacterial density from the original sample was calculated based on the dilution factor.

Pharmacodynamic time-kill parameters.

Bactericidal effect was defined as a ≥3-log₁₀ CFU/ml reduction at 24 h from the starting point of the curve; synergy was defined as a ≥2-log₁₀ CFU/ml reduction in the culture at 24 h for the combination compared with the most active single drug; indifference was defined as a <2-log₁₀ CFU/ml change at 24 h; antagonism was defined as ≥1-log₁₀ CFU/ml regrowth achieved with the combination compared with the least active component; and additivity was defined as a 1- to 2-log₁₀ CFU/ml reduction based on the final count of colonies in the antibiotic combination compared with the count for the more effective of the two components (21, 41).

In vitro pharmacodynamic model.

A one-compartment in vitro chemostat model (42) was used to validate the C/T plus colistin combination against three isolates of XDR P. aeruginosa ST175 with different susceptibility levels to C/T, ST175 (10-023), ST175 (09-012), and ST175 (07-016), which have MICs between 2 and 16 mg/liter. These isolates were selected because ST175 is the most prevalent high-risk clone in our environment. The chemostat model consisted of four independent glassware reactor models studied simultaneously: one contained an antibiotic-free growth control, one contained C/T, one contained colistin, and one contained C/T plus colistin. The experiment was placed in an incubator at 37°C. All reactors were filled with 300 ml of CAMHB broth under constant stirring. Several colonies were inoculated in the reactors to achieve 10⁷ to 10⁸ log₁₀ CFU/ml of each isolate. They were supplemented with the corresponding concentration of the selected antibiotics, which were infused into the reactors via antibiotic pumps. Antibiotics were added as boluses into the treatment models to achieve target peak concentrations. Fresh broth was supplied via a peristaltic pump (Masterflex L/S model 7524-40; Cole-Parmer Instrument Company, Vernon Hills, IL) programmed to achieve the human-simulated half-life of the antimicrobial being tested. Samples were obtained from each of the models at specific time points (0, 2, 4, 8, 12, and 24 h) throughout the experiment and were serially diluted in normal saline to assess changes in bacterial density over time. Aliquots from each diluted sample were plated onto TSA II plates and incubated at 37°C for 18 to 24 h for quantitative cultures. Numbers of CFU were counted after the overnight incubation. The lower limit of detection was 1.3 log₁₀ CFU/ml. All experiments were conducted in duplicate over 24 h (24).

In the chemostat model, a portion of the bacterial suspension was quantitatively cultured onto agar supplemented with C/T at 2-fold and 4-fold the baseline MIC or with colistin at 2-fold and 4-fold the MIC to assess the effect of each regimen on the less susceptible bacterial population.

Pharmacokinetic studies.

Antibiotic concentrations were collected from the reactors at the predetermined time points and stored at –80°C until analysis. Samples were taken to validate antibiotic concentrations. All exposures to simulate the steady-state human pharmacokinetics of unbound drugs were based on half-lives of 3 h for ceftolozane and colistin (43). All treatment regimens were compared with a no-treatment control. Over the first 24 h of the study, all pharmacokinetic samples were determined by HPLC.

Statistical analysis.

We performed the nonparametric Kruskal-Wallis test to determine if these differences were statistically significant between the conditions (A, control; B, colistin; C, C/T; D, C/T plus colisitn). This test is preferred to the ANOVA (analysis of variance) test when a normal distribution is not assumed. We corrected multiple comparisons between pairs of variables with Bonferroni correction. A P value of ≤0.05 was considered statistically significant.

For antibiotic exposures, we performed a linear regression and assessed the global fit with the coefficient of determination (R²), which represents the proportion of variability of the dependent variable (Y) that can be attributed to X.

For each regimen (including the growth control) for the duration of the study, the LR difference in area under the curve for CFU (AUCFU) was calculated to compare any two regimens (test and reference). We compared each antibiotic alone with the control isolate and each antibiotic combination with the antibiotic alone. We calculated the log ratio of AUCFU (LR) as LR = log₁₀ (AUCFU_test/AUCFU_reference), where the reference regimen is the growth control. An LR value of −1 or −2 means that the test regimen (compared to the reference) reduced exposure by 90% (10-fold reduction) or 99% (100-fold reduction), respectively. No definitions for synergy, additivity, etc., have been established for the LR of the AUCFU method (24, 44).