Simulated Intravenous versus Inhaled Tobramycin with or without Intravenous Ceftazidime Evaluated against Hypermutable Pseudomonas aeruginosa via a Dynamic Biofilm Model and Mechanism-Based Modeling

ABSTRACT

Acute exacerbations of chronic respiratory infections in patients with cystic fibrosis are highly challenging due to hypermutable Pseudomonas aeruginosa, biofilm formation and resistance emergence. We aimed to systematically evaluate the effects of intravenous versus inhaled tobramycin (TOB) with and without intravenous ceftazidime (CAZ). Two hypermutable P. aeruginosa isolates, CW30 (MIC_CAZ, 0.5 mg/liter; MIC_TOB, 2 mg/liter) and CW8 (MIC_CAZ, 2 mg/liter; MIC_TOB, 8 mg/liter), were investigated for 120 h in dynamic in vitro biofilm studies. Treatments were intravenous ceftazidime, 9 g/day (33% lung fluid penetration); intravenous tobramycin, 10 mg/kg of body every 24 h (50% lung fluid penetration); inhaled tobramycin, 300 mg every 12 h; and both ceftazidime-tobramycin combinations. Total and less susceptible planktonic and biofilm bacteria were quantified over 120 h. Mechanism-based modeling was performed. All monotherapies were ineffective for both isolates, with regrowth of planktonic (≥4.7 log₁₀ CFU/ml) and biofilm (>3.8 log₁₀ CFU/cm²) bacteria and resistance amplification by 120 h. Both combination treatments demonstrated synergistic or enhanced bacterial killing of planktonic and biofilm bacteria. With the combination simulating tobramycin inhalation, planktonic bacterial counts of the two isolates at 120 h were 0.47% and 36% of those for the combination with intravenous tobramycin; for biofilm bacteria the corresponding values were 8.2% and 13%. Combination regimens achieved substantial suppression of resistance of planktonic and biofilm bacteria compared to each antibiotic in monotherapy for both isolates. Mechanism-based modeling well described all planktonic and biofilm counts and indicated synergy of the combination regimens despite reduced activity of tobramycin in biofilm. Combination regimens of inhaled tobramycin with ceftazidime hold promise to treat acute exacerbations caused by hypermutable P. aeruginosa strains and warrant further investigation.

TEXT

Respiratory infections caused by Pseudomonas aeruginosa in patients with cystic fibrosis (CF) are a significant challenge (1). In CF, acute infective exacerbations (AIEs) with this pathogen and progressive pulmonary insufficiency are responsible for high morbidity and mortality (2). Over time, extensive use of antibiotics and significant pulmonary environmental pressures promote a phenotypic shift of P. aeruginosa (3). The hypermutator phenotype (with an up to 1,000-fold increased mutation rate) occurs in up to ∼65% of P. aeruginosa isolates from patients with CF (4, 5). These strains are associated with the transition of early (planktonic-phase) infection to a respiratory infection involving biofilm formation (6, 7). The biofilm harbors phenotypic diversity, restricts access of antibiotics to the infecting pathogen, and diminishes their antibacterial activity (2, 8). Furthermore, hypermutable P. aeruginosa strains develop resistance to antibiotics much more rapidly than nonhypermutable strains, are linked to a significant increase in multidrug resistance, and consequently present a substantial clinical challenge (9).

Intravenous ceftazidime and tobramycin are considered first-line treatment options for P. aeruginosa exacerbations, but their use as monotherapy is not recommended and information about rational dosing of this antibiotic combination in CF is limited (10–12). P. aeruginosa has a large armamentarium of resistance mechanisms and can rapidly develop resistance following suboptimal antibiotic exposures, increasing the risk of therapeutic failure (13). In addition, for most antibiotics, including ceftazidime and tobramycin, the minimal biofilm inhibitory concentration is usually more than one 2-fold dilution higher than the MIC for planktonic bacteria, further complicating the treatment of AIEs in patients with CF (14).

While inhaled tobramycin is widely used in attempts to eradicate early acquisition of P. aeruginosa and control chronic infections, it is not usually employed in the management of P. aeruginosa AIEs (15–17). This mode of administration ensures delivery of high concentrations of the antibiotic into lung fluid (18–21). The effect of tobramycin concentration-time profiles that can be achieved with inhalation against hypermutable P. aeruginosa embedded in biofilms and in planktonic form has never been explored. Thus, we utilized a dynamic biofilm model to investigate the effect of clinically relevant pharmacokinetic (PK) profiles of ceftazidime and tobramycin (intravenous versus inhalation), as monotherapy and in combination, against clinical isolates of hypermutable P. aeruginosa. We developed the first mechanism-based mathematical model for the time course of viable counts from the dynamic biofilm model.

RESULTS

Pharmacokinetic validation and microbiological response.

The observed ceftazidime and tobramycin concentrations in the Centers for Disease Control and Prevention biofilm reactor (CBR) were, on average, within 5% of the targeted exposures that are shown in Table 1. Viable cell count profiles for total populations of planktonic and biofilm bacteria are shown in Fig. 1 for both isolates, and the corresponding profiles for less susceptible populations of CW30 and CW8 are in Fig. 2 and 3, respectively. Log changes in viable counts of total bacteria, mutant frequencies, and MICs of colonies from antibiotic-containing agar are in Tables 2, 3, and 4, respectively.

TABLE 1.

Clinically representative lung fluid concentrations, exposures, and PK/PD indices for CAZ and TOB against CW30 and CW8^a

Isolate and treatment	fCss, fCmax; fCmin (mg/liter)	fAUC24 (mg·h/liter)	fCss/MIC, fCmax/MIC; fCmin/MIC	fT>MIC (%)	fAUC24/MIC
CW30
CAZ 9 g/day CI	18	432	36	100	864
TOB 10 mg/kg q24h i.v.	12.3; 0.1	64.4	6.1; 0		32.2
TOB 300 mg q12h INH	100; 9.4	928	50; 4.7		464
CW8
CAZ 9 g/day CI	18	432	9	100	216
TOB 10 mg/kg q24h i.v.	12.3; 0.1	64.4	1.5; 0		8.1
TOB 300 mg q12h INH	100; 9.4	928	12.5; 1.2		116

^aCAZ, ceftazidime; TOB, tobramycin; CI, continuous intravenous infusion; q24h, every 24 h; q12h, every 12 h; i.v., intermittent intravenous infusion; INH, inhalation; fC_ss, unbound steady-state concentration; fC_max, unbound maximum concentration; fC_min, unbound minimum concentration before next dose; fAUC₂₄, the area under the unbound concentration-time curve over 24 h; fC_ss/MIC, the ratio of fC_ss to MIC; fC_max/MIC, the ratio of fC_max to MIC; fC_min/MIC, the ratio of fC_min to MIC; fT_>MIC, the cumulative percentage of a 24 h period that unbound concentrations exceeded the MIC; fAUC₂₄/MIC, the ratio of fAUC₂₄ to MIC. The simulated half-life in lung fluid was 3.5 h for tobramycin, as the tobramycin half-life is slightly longer in lung fluid than in plasma (66, 73). The simulated lung fluid penetration was 33% for ceftazidime and 50% for tobramycin (based on area under the concentration-time curve ratios) (66, 73–77).

FIG 1.

Total viable counts for growth control and treatments with ceftazidime (CAZ) and/or tobramycin (TOB) with clinically relevant lung fluid concentration-time profiles. CAZ was administered by continuous infusion (CI) and TOB by intermittent intravenous or inhalational (INH) administration. Samples were from the media within the reactor (i.e., planktonic bacteria) and from coupons (i.e., biofilm bacteria). The y axis starts at the limit of counting. The results from treatment C for CW30 and A plus C for both isolates are presented as averages ± SE from two replicates.

FIG 2.

Effect of each dosage regimen on the counts of CW30 able to grow on agar plates containing 2.5 or 10 mg/liter ceftazidime or 5 or 10 mg/liter tobramycin. CAZ was administered by continuous infusion (CI), and TOB administration reflected intermittent intravenous infusion or inhalation (INH). The results for treatment C and treatment A plus C are presented as averages ± SE from of two replicates. Results below the limit of counting are plotted at zero. To differentiate less-susceptible subpopulations from the predominant population, the antibiotic concentrations in agar were based upon Etest MICs, which were 0.064 mg/liter for ceftazidime and 0.75 mg/liter for tobramycin (35).

FIG 3.

Effect of each dosage regimen on the counts of CW8 able to grow on agar plates containing 6 or 10 mg/liter of ceftazidime or 10 or 20 mg/liter of tobramycin. CAZ was administered by continuous infusion (CI), and TOB administration reflected intermittent intravenous infusion or inhalation (INH). The results from treatment A plus C are presented as averages ± SE from two replicates. Results below the limit of counting are plotted at zero. To differentiate less susceptible subpopulations from the predominant population, the antibiotic concentrations in agar were based upon Etest MICs, which were 0.75 mg/liter for ceftazidime and tobramycin (35).

TABLE 2.

Log changes in viable cell counts of total bacteria at various time points with clinically relevant lung fluid concentration exposures of ceftazidime and/or tobramycin^a

^aCAZ, ceftazidime; TOB, tobramycin; CI, continuous intravenous infusion; Q24 h, every 24 h; Q12 h, every 12 h; IV, intermittent intravenous infusion; INH, inhalation; log₁₀(CFUt), log₁₀ CFU/ml or log₁₀ CFU/cm² at each sample collection time; log₁₀(CFU0), log₁₀ CFU/ml or log₁₀ CFU/cm² at time zero. The green background indicates synergy (a ≥2-log₁₀ decrease in the CFU/ml or CFU/cm² with the combination compared to its most active component and, for planktonic bacteria, a ≥2-log₁₀ decrease in the CFU/ml compared to the initial inoculum); the blue background indicates a 1.0- to <2-log₁₀ decrease in the number of CFU/ml or CFU/cm² with the combination compared to its most active component.

TABLE 3.

Log₁₀ mutant frequencies at 2.5 or 6 mg/liter and 10 mg/liter ceftazidime and 5 or 10 mg/liter and 10 or 20 mg/liter tobramycin^a

^aCAZ, ceftazidime; TOB, tobramycin; CI, continuous intravenous infusion; Q24 h, every 24 h; Q12 h, every 12 h; IV, intermittent intravenous infusion; INH, inhalation. The red background indicates a high mutant frequency, i.e., a large proportion of less susceptible bacteria being present in the total population; the green background indicates a low mutant frequency, i.e., a small proportion of less susceptible bacteria being present in the total population. When no colonies were present on antibiotic-containing plates, mutant frequencies reported represent an upper limit based on the total viable count.

TABLE 4.

MIC values of colonies isolated from antibiotic-containing agar plates (10 mg/liter ceftazidime and 10 or 20 mg/liter tobramycin) at 0 and 120 h for each dosage regimen^a

Isolate and arm	Time (h)	CAZ, 10 mg/liter		TOB, 10 or 20 mg/liter
		Planktonic bacteria	Biofilm bacteria	Planktonic bacteria	Biofilm bacteria
CW30
Control	0	NC	NC	NC	NC
	120	8	4	4	4
A: CAZ 9 g/day CI	120	64	32	—	—
B: TOB 10 mg/kg q24h i.v.	120	—	—	32	16
C: TOB 300 mg q12h INH	120	—	—	16	8
A + B	120	NC	NC	NC	NC
A + C	120	NC	NC	NC	NC
CW8
Control	0	8	8	16	8
	120	32	16	32	16
A: CAZ 9 g/day CI	120	128	128	—	—
B: TOB 10 mg/kg q24h i.v.	120	—	—	64	32
C: TOB 300 mg q12h INH	120	—	—	32	16
A + B	120	16	NC	NC	NC
A + C	120	8	NC	NC	NC

^aCAZ, ceftazidime; TOB, tobramycin; CI, continuous intravenous infusion; q24h, every 24 h; q12h, every 12 h; i.v., intermittent intravenous infusion; INH, inhalation; NC, no colonies grew on the antibiotic-containing agar plates; —, not tested.

Planktonic bacteria.

Tobramycin monotherapy simulating intravenous dosing produced initial killing of ≥2-log₁₀ CFU/ml at 7 h followed by regrowth close to the growth control by 48 h for both isolates (Fig. 1, Table 2). With this treatment, amplification of resistance was observed at 72 and 120 h, with an increase of tobramycin-resistant populations compared to the growth control (Fig. 2 and 3, Table 3). The MIC of colonies recovered from tobramycin-containing agar at 120 h was 32 mg/liter for CW30 and 64 mg/liter for CW8 (Table 4).

For CW30, tobramycin inhalation produced extensive initial killing of ∼4.2 log₁₀ CFU/ml at 7 h and viable counts remained suppressed at ∼4.7 log₁₀ CFU/ml across 72 to 120 h, which was ∼4 log below that with intravenous tobramycin. For CW8, inhaled tobramycin achieved ∼2.7 log₁₀ CFU/ml initial killing at 7 h, followed by slow regrowth to ∼5.9 log₁₀ CFU/ml at 120 h, which was ∼2.5 log below the count for intravenous tobramycin (Fig. 1, Table 2). Less susceptible tobramycin populations at 120 h were slightly lower (CW30) (Fig. 2) or substantially higher (CW8) (Fig. 3) than for the growth control. For both isolates and inhaled tobramycin, MICs of colonies recovered from drug-containing agar were one 2-fold dilution lower than those from the intravenous tobramycin treatment (Table 4).

Ceftazidime monotherapy achieved ≥2.2-log₁₀ CFU/ml initial killing at 7 h for both isolates, followed by regrowth (Fig. 1, Table 2). Less susceptible ceftazidime populations at 120 h had counts ∼2 to 5 log higher than those for the growth control (Fig. 2 and 3, Table 3); the MIC was 64 mg/liter for CW30 and 128 mg/liter for CW8 (Table 4).

For CW30, the combination of intravenous tobramycin with ceftazidime showed enhanced bacterial killing at 72 h and synergy at 96 and 120 h, whereas for CW8 enhanced activity was observed as early as 5 h and synergy was achieved at 48 h and beyond (Fig. 1, Table 2). No colonies were detected at 120 h on tobramycin-containing agar plates for both isolates and on ceftazidime-containing plates for CW30 (Fig. 2 and 3). With CW8, ≤3.2 log₁₀ CFU/ml was observed on ceftazidime-containing plates at 120 h, i.e., ∼1 to 2.5 log₁₀ CFU/ml below the growth control values; the MIC was 16 mg/liter (Fig. 3, Table 4).

The combination of tobramycin inhalation plus ceftazidime resulted in greater and more sustained killing than the combination with intravenous tobramycin; for CW30 and CW8, the counts (in CFU/ml) at 120 h of the combination simulating inhaled tobramycin were 0.47% and 36%, respectively, of those for the combination containing intravenous tobramycin. For CW30, enhanced bacterial killing was observed as early as 5 h, and synergy or enhanced activity was evident at all sampling times from 7 to 120 h. Viable counts of the total population were below the limit of counting at 28, 96, and 120 h (Fig. 1, Table 2), and no colonies were observed on agar containing either antibiotic (Fig. 2). For CW8, the combination simulating tobramycin inhalation produced enhanced activity at 5 h with ∼4 log₁₀ CFU/ml bacterial killing. Enhancement or synergy was evident at all sampling times from 28 to 120 h (Fig. 1, Table 2). No colonies were detected on tobramycin-containing plates for either isolate and on ceftazidime-containing plates for CW30 (Fig. 2 and 3). For CW8, ∼1.2 log₁₀ CFU/ml of bacteria less susceptible to ceftazidime were observed at 120 h; these counts were ∼3 to 4 log₁₀ CFU/ml below those from the growth control and ∼1.5 log below the counts for the combination with intravenous tobramycin (Fig. 3). The MIC at 120 h was 8 mg/liter (Table 4).

Biofilm bacteria.

The intravenous tobramycin and ceftazidime monotherapies were ineffective against biofilm bacteria and produced only modest killing of both isolates. Tobramycin inhalation monotherapy achieved maximal killing of ∼3.3 log₁₀ CFU/cm² for CW30 and ∼0.9 log₁₀ CFU/cm² for CW8, i.e., viable counts ∼4.3 and ∼2.6 log₁₀ below those for intravenous tobramycin (Fig. 1, Table 2). Amplification of ceftazidime- and tobramycin-resistant bacteria, compared to corresponding control values, was observed with all intravenous monotherapies (Fig. 2 and 3, Table 3). The MICs at 120 h are presented in Table 4. The combination simulating intravenous tobramycin produced synergy from 72 to 120 h for both isolates; ∼3 log₁₀ CFU/cm² on 5 mg/liter tobramycin-containing agar was observed at 120 h only for CW30, which was ∼1 log lower than that of the growth control (Fig. 2). The combination with tobramycin inhalation produced more pronounced killing for both isolates, with enhanced or synergistic effects evident beyond 24 to 48 h (Fig. 1, Table 2); for CW30 and CW8, the counts (in CFU/cm²) at 120 h for the combination simulating inhaled tobramycin were 8.2% and 13%, respectively, of those for the combination containing intravenous tobramycin. For CW30, ∼1.8 log₁₀ CFU/cm² of colonies was retrieved at 96 and 120 h; the viable count at this time was ∼1 log₁₀ CFU/cm² lower than that observed with the intravenous tobramycin combination. For CW8, the viable count from 48 to 120 h remained below ∼3 log₁₀ CFU/cm², and bacterial killing was ∼1 to ∼1.2 log₁₀ CFU/cm² greater than what occurred with the intravenous tobramycin combination. During treatment with the combination containing inhaled tobramycin, no colonies were observed on antibiotic-containing agar at any time (Fig. 2 and 3, Table 3).

Mechanism-based modeling.

The mechanism-based model (MBM) well described the planktonic and biofilm bacterial counts and yielded unbiased and precise curve fits for both isolates (Fig. 4; see also Fig. S1 in the supplemental material). For each mode of growth, the data from all dosage regimens and growth controls across the two different isolates could be described simultaneously by the same model and generally the same parameter estimates. Different parameter estimates were only required to describe differences in the initial biofilm inocula, abundance of preexisting resistant subpopulations, and tobramycin susceptibilities of the planktonic bacteria (Table 5). The lower estimate of KC_50,SS (tobramycin concentration causing 50% of the maximum killing rate constant for the susceptible population) for planktonic bacteria of CW8 compared to CW30 was needed to describe the greater extent of initial bacterial killing by intravenous tobramycin of CW8 compared to CW30. The parameters were estimated with sufficient precision, as the standard errors (SE) were below 30% coefficient of variation for the majority of parameters.

FIG 4.

Total viable counts of planktonic and biofilm bacteria of isolates CW30 and CW8 for growth control and treatments with ceftazidime (CAZ) and/or tobramycin (TOB) with mechanism-based model fitted lines. Samples below the limit of counting (1.0 log₁₀ CFU/ml, 1.6 log₁₀ CFU/cm²) are plotted at zero. The results from treatment C for CW30 and A plus C for both isolates (open and filled symbols) are biological replicates that were modeled separately.

TABLE 5.

Parameter estimates for the MBM for CAZ and TOB in monotherapy and combination regimens against planktonic and biofilm bacteria of CW8 and CW30

Parameter	Population mean estimate (SE%)
	Symbol (unit)	Planktonic	Biofilm
Bacterial growth and subpopulations
Initial inoculum	Log10CFU0	7.05 (1.8)	5.44a (2.2), 6.28b (2.2)
Maximum population size	Log10CFUmax	8.97 (1.8)	7.62 (2.1)
Mean generation time (MGT)
CAZs-TOBs	MGTSS (min)	90.3 (13)	457 (15)
CAZR-TOBI	MGTRI (min)	126 (9.2)	700 (13)
CAZI-TOBR	MGTIR (min)	240 (5.8)	2,377 (14)
Log10 mutation frequency
CAZ	LogMUT,CAZ	−4.64a (8.3), −4.52b (6.6)	−1.74a (19), −3.19b (8.7)
TOB	LogMUT,TOB	−3.36a (6.4), −3.75b (6.0)	−1.57a (11), −3.33b (8.1)
Inhibition of successful replication by CAZ
Maximum inhibition of successful replication	I max,REP	0.958 (11)	0.881 (14)
CAZ concn causing 50% of Imax,REP
CAZs-TOBs	IC50,REP,SS (mg/liter)	0.622 (36)	0.841 (35)
CAZR-TOBI	IC50,REP,RI (mg/liter)	128c (16)	102 (37)
CAZI-TOBR	IC50,REP,IR (mg/liter)	128c (16)	5.75 (26)
Bacterial killing by TOB
Maximum killing rate constant	Kmax,TOB (h−1)	7.37 (16)	0.283 (13)
TOB concn causing 50% of Kmax,TOB
CAZs-TOBs	KC50,SS (mg/liter)	7.32a (35), 40.1b (24)	7.70c (22)
CAZR-TOBI	KC50,RI (mg/liter)	118a (49), 75.0b (19)	7.70c (22)
CAZI-TOBR	KC50,IR (mg/liter)	15,647 (31)	1,044 (19)
Hill coefficient	HillTOB	2.15a (39), 1b (fixed)	1 (fixed)
Mechanistic synergy
Maximum fractional decrease of IC50,REP via mechanistic synergy	I max,OM,TOB	0.795 (22)	0.661 (35)
Concn of TOB causing 50% of Imax,OM,TOB	IC50,OM,TOB (mg/liter)	1.74 (40)	16.1 (21)
Hill coefficient	HillOM	3.50a (50), 1b (fixed)	1 (fixed)
Residual variability
SD of residual error on log10 scale	SDCFU	0.442 (6.3)	0.306 (7.6)

^aCW8.

^bCW30.

^cThe RI and IR subpopulations, or SS and RI subpopulations, were described by the same population mean estimate.

Comparison of the parameter estimates between planktonic and biofilm bacteria revealed a few trends across both isolates. The mean generation times (MGTs) for the three preexisting bacterial populations (i.e., double susceptible, ceftazidime resistant and tobramycin intermediate, and ceftazidime intermediate and tobramycin resistant) in the biofilm states were considerably (5- to 10-fold) higher than those estimated for the planktonic states. The parameter estimates representing inhibition of successful replication by ceftazidime were similar for biofilm and planktonic bacteria, except for a lower IC_50,REP,IR (ceftazidime concentration causing 50% of maximum inhibition of successful replication of the ceftazidime-intermediate, tobramycin-resistant population) in biofilm (Table 5). In contrast, the maximum killing rate constant of tobramycin (K_max,TOB) was estimated to be 26 times lower for biofilm than planktonic bacteria. Despite some KC₅₀ estimates being higher for planktonic than biofilm bacteria, the estimated overall killing effect of tobramycin was greater on planktonic than biofilm bacteria for all three subpopulations. The very high estimates of KC_50,IR indicate that the tobramycin-resistant planktonic and biofilm subpopulations were not killed by tobramycin in the model. In addition, the concentration of tobramycin required for 50% of the maximum extent of mechanistic synergy was estimated to be 9-fold higher for biofilm than planktonic bacteria, while the maximum extent of mechanistic synergy was similar for both modes of growth.

DISCUSSION

For β-lactams, including ceftazidime, the PK/pharmacodynamic (PD) index generally considered most predictive of antibacterial activity is the duration of the dosing interval over which the unbound concentration remains above a multiple of the MIC of the infecting pathogen (fT_>MIC) (22–26). For ceftazidime, favorable microbiological and clinical outcomes were more likely when fT_>MIC was >45% (27). Since the maximum rate of bacterial killing was found to occur at an unbound ceftazidime concentration of ∼4× MIC, it was suggested that administration of ceftazidime as a continuous infusion to maintain unbound concentrations 4-fold higher than the MIC should maximize efficacy (25, 28). In the present CBR study, 9 g/day ceftazidime was simulated as a continuous infusion. For isolate CW30, the ceftazidime concentration was 36× MIC across the entire study duration; the corresponding value for CW8 was 9× MIC. Even with these high levels of exposure, ceftazidime monotherapy was unable to suppress regrowth and resistance emergence for both planktonic and biofilm bacteria of each hypermutable isolate. In contrast, an in vitro study using clinical strains of P. aeruginosa with preexisting resistant mutants suggested that an unbound ceftazidime concentration >3.8-fold higher than the MIC was needed to suppress the amplification of resistant subpopulations (29). That study examined nonhypermutable planktonic bacteria in a hollow-fiber infection model, whereas in the present study the interaction and cycling of biofilm and planktonic hypermutable bacteria may have decreased the ability to suppress amplification of resistant subpopulations (13, 30).

For aminoglycosides, antibacterial activity has been correlated with the ratio of unbound exposure across a 24-h period to MIC (fAUC/MIC) and the ratio of unbound maximum concentration to MIC (fC_max/MIC) (31); an fAUC/MIC of >70 and fC_max/MIC of 8 to 10 have been proposed as targets to increase likelihood of clinical success (32). In the present study, tobramycin was administered representing two modes of delivery, intravenous and inhalation. For the regimen simulating intravenous administration (10 mg/kg of body weight every 24 h [q24h]), the fAUC/MIC and fC_max/MIC of CW30 were 32.2 and 6.1, respectively; the corresponding values for CW8 were 8.1 and 1.5. The exposure values for inhaled delivery (300 mg every 12 h) were 464 and 50 for fAUC/MIC and fC_max/MIC for CW30 and 116 and 12.5 for fAUC/MIC and fC_max/MIC for CW8. The fAUC/MIC and fC_max/MIC values achieved with the intravenous regimen did not reach the above-mentioned PK/PD targets for either isolate, but the targets were exceeded with inhaled delivery against both isolates. However, neither of the tobramycin regimens in monotherapy was able to suppress the regrowth of both planktonic and biofilm bacteria. This result agreed with our previous in vitro studies where tobramycin monotherapy failed to inhibit regrowth of planktonic hypermutable P. aeruginosa even with fAUC/MIC exposures substantially higher than 70 (33, 34).

In the current study, the monotherapy regimens of ceftazidime and tobramycin were less effective against biofilm than planktonic bacteria. There are multiple mechanisms that may contribute to the relative resistance of P. aeruginosa biofilm bacteria, some of which are general in nature while others are specific to particular classes of antibiotics, such as β-lactams and aminoglycosides (8). The failure of each of ceftazidime at 9 g/day as a continuous infusion and tobramycin, even when administered by inhalation, to effectively suppress both bacterial growth forms strongly argues against the use of monotherapy for treatment of infections caused by hypermutable and biofilm-forming P. aeruginosa strains.

The combination regimen containing intravenous tobramycin increased bacterial killing and decreased emergence of resistant subpopulations relative to the monotherapy treatments. The increased antibacterial effect culminated in enhanced or synergistic killing of planktonic and biofilm bacteria of the isolates across the last 2 to 4 days of the 5-day study period. The extent of the increased killing of biofilm bacteria, relative to either antibiotic administered as monotherapy, was especially marked. However, the largest antibacterial effect occurred with the combination containing tobramycin inhalation. Against both isolates, that combination regimen produced enhanced or synergistic bacterial killing of planktonic bacteria on each of the 5 days of the study and of biofilm bacteria across the last 3 to 4 days, over which time at least a 1- to 2-log₁₀ reduction in bacterial counts was observed relative to inhaled tobramycin alone. The combination regimen including inhaled tobramycin decreased the planktonic bacteria of both isolates to ≤2 log₁₀ CFU/ml at 28 h (and at 96 to 120 h for CW30). The immune system might eliminate the remaining planktonic cells, which may or may not be the case for remaining biofilm bacteria. An additional notable finding of both combination regimens was resistance suppression, given the isolates were strong hypermutators and at baseline had mutations in genes associated with resistance to β-lactams and aminoglycosides (35).

Previous studies investigated ceftazidime in combination with tobramycin in static and dynamic in vitro studies and reported synergistic outcomes against P. aeruginosa (36–39). However, all studies had one or more limitations, including that they were of short duration (24 to 48 h), examined only planktonic bacteria, and did not include hypermutable strains or report whether an isolate was hypermutable (36–39). Ceftazidime in combination with tobramycin achieved significant bacterial reductions compared to the components in monotherapy against nonhypermutable P. aeruginosa in vivo (40, 41).

The enhanced and synergistic bacterial killing in the current study may be the result of the difference between the two antibiotics in mechanisms of action and resistance. Ceftazidime inhibits cell wall synthesis via binding to penicillin-binding proteins (PBPs), and major mechanisms of resistance in P. aeruginosa isolates involve chromosomally mediated AmpC β-lactamase overexpression, enzymatic inactivation via β-lactamases, and reduced affinity of PBPs (13, 42). Tobramycin blocks protein synthesis but also disrupts the outer bacterial membrane, very likely resulting in increased ceftazidime concentrations in the periplasmic space where the PBPs are located (43–45). Resistance mechanisms of P. aeruginosa against aminoglycosides include increased expression of MexXY-OprM, target site modification, enzymatic cleavage, and reduced outer membrane permeability (46–50). The mechanism by which the combination regimens achieved such remarkable activity toward biofilm bacteria in the current study is unknown but may be related to the ability of tobramycin to inhibit adhesion and microcolony formation during cycling of biofilm and planktonic cells (51).

There are a number of potential advantages of administering antibiotics via inhalation for the treatment of lung infections (52). These advantages include the ability to achieve a substantially higher lung-to-plasma concentration ratio than can be achieved with intravenous administration (53). This means that for antibiotics with narrow therapeutic windows (e.g., tobramycin), it is possible to achieve concentrations in lung much higher than can be safely achieved with intravenous administration; for tobramycin, this implies reduced risk of nephrotoxicity. For the last few decades antibiotics have been administered by inhalation for the management of chronic lung infections, including those occurring in patients with CF (54, 55). A recent Cochrane systematic review considered the randomized controlled trials that have been conducted in people with CF experiencing a pulmonary exacerbation in whom treatment with inhaled antibiotics was compared to placebo, standard treatment, or another inhaled antibiotic for between 1 and 4 weeks (15). Four trials with 167 participants were included in the review. Unfortunately, due to heterogeneity in trial design, high risk from lack of blinding, difficulty in assessing risk of bias, and lack of statistical power, the systematic review was unable to demonstrate whether or not one treatment was superior to the other. The authors concluded that further research is needed to establish whether inhaled tobramycin can be used as an alternative to intravenous tobramycin for pulmonary exacerbations (15). The results of the study reported here lend strong support for the conduct of well-designed clinical trials to evaluate the delivery of tobramycin by inhalation in patients having pulmonary exacerbations. In designing clinical trials, it should be recognized that P. aeruginosa can distribute across different airway niches in the lung of patients with cystic fibrosis, some of which can be better reached by intravenous administration (56). Thus, intravenous plus inhaled tobramycin in combination with ceftazidime should be considered for inclusion as a treatment arm in clinical trials. The limited systemic bioavailability of inhaled tobramycin (∼9 to 17%) generates relatively low plasma exposure (53), and limited clinical experience suggests generally good safety of coadministration of intravenous and inhaled tobramycin as a component of antibiotic treatment for pulmonary exacerbations in patients with CF (15, 57).

Our developed MBM well described the antibacterial effects of all monotherapy and combination regimens against planktonic and biofilm bacteria for both isolates. Longer estimated mean generation times for biofilm compared to planktonic bacteria reflected the slower growth of biofilm bacteria (8). The parameter estimates related to inhibition of successful replication by ceftazidime were similar between biofilm and planktonic bacteria, except for the IC_50,REP of the tobramycin-resistant subpopulation. However, the slower growth of the biofilm bacteria resulted in an attenuated effect of ceftazidime monotherapy on biofilm compared to planktonic bacteria in the MBM, leading to a lower rate of bacterial killing. This is in agreement with the mechanism of action of ceftazidime affecting replicating bacteria (8). Similarly, the substantially lower maximum killing rate constant of tobramycin and higher tobramycin concentration required for 50% of the maximum mechanistic synergy for biofilm compared to planktonic bacteria reflect a lower activity of aminoglycosides in a biofilm matrix (8).

The strengths of this study include the following: it is the first study to examine the activity of clinically relevant regimens of ceftazidime and tobramycin in monotherapy and in combination against hypermutable isolates of P. aeruginosa in the CBR model; the impact of delivering tobramycin by inhalation versus intermittent intravenous infusions was investigated; the effects on both planktonic and biofilm bacteria were examined in a dynamic model over 5 days; and the time courses of both total and resistant subpopulations were characterized. The study also has some limitations. As with most nonclinical models, the planktonic and biofilm growth modes in the CBR model may not fully recapitulate those in the lungs of patients with CF, but a major advantage of the model we employed is that it enables accurate mimicking of the PK of antibiotics occurring in patients receiving clinical dosage regimens over 5 days. Like other in vitro infection models, the CBR also lacks an immune system and, therefore, the responses observed reflect the effects of the antibiotics only. While our previous studies on these isolates provided data on mutations in resistance genes present prior to treatment (35), we did not undertake genomic studies on emergent resistant populations of failed monotherapy regimens. It may also have been helpful to explore possible changes in biofilm structure in response to the different regimens. As is the case for most published MBM describing dynamic in vitro infection studies (58–62), the total viable counts were modeled and bacterial counts on drug-containing agar were not incorporated in the MBM. However, our study is the first to develop an MBM based on data from the dynamic biofilm model; therefore, future studies may investigate an extension of the current MBM to characterize the resistance emergence.

In conclusion, this study has provided evidence that ceftazidime and tobramycin, when administered in monotherapy against hypermutable P. aeruginosa, are unable to provide sustained reduction in total and resistant subpopulations of both planktonic and biofilm bacteria. This supports guideline recommendations that these antibiotics should not be used in monotherapy regimens for treatment of pulmonary exacerbations. Indeed, the study demonstrated the enhanced and synergistic activity of combination therapy, especially with tobramycin administered by inhalation. The latter finding strongly supports the need for further studies to investigate the administration of antibiotics by inhalation in the treatment of infective exacerbations.

MATERIALS AND METHODS

Antibiotics, bacterial isolates, and MIC testing.

Solutions of ceftazidime (lot WS16174; Waterstone Technology, Carmel, IN, USA) and tobramycin (lot LC24138; AK Scientific, Union City, MD, USA) were prepared in sterile Milli-Q water immediately before each experiment. Cation-adjusted Mueller-Hinton agar (CAMHA) and broth (CAMHB) supplemented with 1% tryptic soy broth (TSB; BD, Sparks, MD, USA) were used in all experiments.

Two previously described clinical hypermutable P. aeruginosa isolates (CW30 and CW8) were examined; hypermutability is defined in the supplemental material (35). Agar dilution MICs determined for each isolate in triplicate on three separate days using Clinical and Laboratory Standards Institute guidelines (63) were 0.5 mg/liter and 2 mg/liter for ceftazidime and 2 mg/liter and 8 mg/liter for tobramycin for CW30 and CW8, respectively.

Dynamic biofilm model and antibiotic dosing.

The Centers for Disease Control and Prevention biofilm reactor (CBR), as previously described (64, 65), was used over 120 h to investigate the effects of ceftazidime and tobramycin alone and in combinations. Before each experiment, isolates were subcultured onto CAMHA and incubated at 36°C for 48 h. Colonies (2 to 3) were randomly selected and grown overnight in 10 ml of TSB, from which early-log-phase growth was obtained. Early-log-phase bacterial suspension (1 ml) was inoculated into each reactor containing 350 ml TSB. The 28-h conditioning phase was performed as described previously (64, 65). At 0 h (before starting treatment), the flow of medium (CAMHB–1% TSB) through each CBR was set to 1.16 ml/min to mimic a tobramycin half-life (t_1/2) of 3.5 h in lung fluid (65–68).

For ceftazidime, the highest recommended daily dose of 9 g/day for intravenous administration, and for tobramycin, doses of 10 mg/kg/day for intravenous administration and 300 mg q12h for inhalation, were simulated in the CBR (12). The PK profiles were based on those expected in lung fluid of patients with CF given the respective antibiotic regimens. These expected concentration-time profiles were simulated in silico using Berkeley Madonna (version 8.3.18) (69) based on clinical studies and population PK models for patients with CF (67, 70–72). Information on lung fluid penetration of ceftazidime (33%) and tobramycin (50%) after intravenous administration was from published studies (66, 73–77). The lung fluid concentration-time profiles of tobramycin following simulated inhalation were informed by clinical studies that quantified tobramycin by bioassay and high-performance liquid chromatography in epithelial lining fluid (20, 78) or sputum (21).

To simulate intravenous and inhalational dosing in the CBR, tobramycin was delivered via syringe driver over 30 min every 24 h and over 15 min every 12 h, respectively. As tobramycin does not accumulate following multiple intravenous administrations, no loading dose was required to achieve steady-state concentrations; for tobramycin inhalation, a small loading dose to attain fC_min (Table 1) was delivered at 0 h. Continuous infusion dosing of ceftazidime was achieved by administering a loading dose at 0 h directly into the reactor to immediately attain the required steady-state concentration (fC_ss), which was subsequently maintained by delivering medium containing the same concentration (fC_ss) of ceftazidime to the reactor (Table 1). Ceftazidime-containing medium was kept in the fridge to avoid thermal degradation and changed every 24 h.

Growth controls were included, and some monotherapy and combination treatments were studied in two biological replicates. Serial samples of planktonic and biofilm bacteria were collected aseptically at the times indicated in Table 2; total viable planktonic and biofilm bacteria were counted as previously described (64, 65). Less susceptible subpopulations were quantified at baseline (0 h) and during treatment at 24, 72, and 120 h on CAMHA supplemented with ceftazidime at 2.5 or 6 mg/liter and 10 mg/liter or tobramycin at 5 or 10 mg/liter and 10 or 20 mg/liter for CW30 and CW8, respectively. MICs of colonies isolated from antibiotic-containing agar at 0 and 120 h were determined to verify phenotypically the presence of stable resistance. Ceftazidime and tobramycin in PK samples were measured using validated liquid chromatography tandem mass spectrometry (LC-MS/MS) assays, as described in the supplemental material.

Pharmacodynamic analysis.

Microbiological responses to monotherapy and combination regimens were examined using the log change method, as described previously (65, 79). Synergy was considered ≥2 log₁₀ CFU/ml or log₁₀ CFU/cm² more killing for the combination relative to the most active corresponding monotherapy at a specified time and, for planktonic cells, a viable cell count of ≥2 log₁₀ CFU/ml below the initial inoculum (80). Combination regimens achieving bacterial killing of ≥1 to <2 log₁₀ CFU/ml or log₁₀ CFU/cm² compared to the most active corresponding monotherapy at the same time were considered to represent enhanced antibacterial activity. Mutant frequencies were calculated as previously described (65).

Mechanism-based modeling methods are described in the supplemental material, and all parameters are explained in Table 5.

Data availability.

The figures and tables include data from the reported studies.