Comparative Efficacy of Continuous Ceftazidime Infusion vs. Intermittent Bolus against In Vitro Ceftazidime-Susceptible and -Resistant Pseudomonas aeruginosa Biofilm

Background: As the anti-biofilm pharmacokinetic/pharmacodynamic (PK/PD) properties of antibiotics are not well-defined, we have evaluated the PK/PD indices for different regimens of ceftazidime (CAZ; with/without colistin) against Pseudomonas aeruginosa biofilm. Methods: We have used the Center for Disease Control and Prevention Biofilm Reactor with two susceptible (PAO1 and HUB-PAS) and one resistant (HUB-XDR) strains of P. aeruginosa. The regimens were CAZ monotherapies (mimicking a human dose of 2 g/8 h, CAZ-IB; 6 g/daily as continuous infusion at 50 mg/L, CAZ-CI50; and 9 g/daily at 70 mg/L, CAZ-CI70) and CAZ-colistin combinations. Efficacy was correlated with the CAZ PK/PD parameters. Results: CAZ-CI70 was the most effective monotherapy against CAZ-susceptible strains (Δlog CFU/mL 54–0 h = −4.15 ± 0.59 and −3.05 ± 0.5 for HUB-PAS and PAO1, respectively; p ≤ 0.007 vs. other monotherapies), and adding colistin improved the efficacy over CAZ monotherapy. CAZ monotherapies were ineffective against the HUB-XDR strain, and CAZ-CI50 plus colistin achieved higher efficacy than CAZ-IB with colistin. The PK/PD index that correlated best with anti-biofilm efficacy was fAUC0–24h/MIC (r2 = 0.78). Conclusions: CAZ exhibited dose-dependent anti-biofilm killing against P. aeruginosa, which was better explained by the fAUC0–24h/MIC index. CAZ-CI provided benefits compared to CAZ-IB, particularly when using higher doses and together with colistin. CAZ monotherapies were ineffective against the CAZ-resistant strain, independently of the optimized strategy and only CAZ-CI plus colistin appeared useful for clinical practice.


Introduction
Antibiotic therapy for biofilm-related infections represents a challenge for clinicians [1,2]; thus, optimizing the use of antimicrobials in this setting is recommended.However, the conventional pharmacokinetic/pharmacodynamic (PK/PD) indices of antibiotic efficacy were defined by using planktonic bacteria and have not proved useful for guiding the treatment of biofilm-related infections [3].
The proportion of osteoarticular and orthopedic device-related infections caused by Gram-negative bacilli (GNB) has increased progressively over recent years, with infections caused by Pseudomonas aeruginosa being of great concern due to their ability to form biofilms and acquire resistance mechanisms [4].Beta-lactams still remain a viable firstline therapy for osteoarticular infections, including those caused by multidrug-resistant GNB [5].Locally, ceftazidime (CAZ) is still a valuable option against most P. aeruginosa strains; however, almost 15-17% of these isolates are multi-drug or extremely-drug resistant, mainly belonging to high-risk clones [6].Beta-lactams have traditionally been considered to have poor activity against biofilms as they are principally active against bacteria in the growth phase (planktonic) [7].Particularly, the time-dependent killing of beta-lactams has been questioned against bacterial biofilms based on results from pre-clinical works that mainly evaluated the efficacy of ceftazidime and meropenem [8][9][10].Overall, few previous works have focused on the evaluation of specific PK/PD indices of beta-lactams used in biofilm-related infections or have suggested the most appropriate dosage and type of administration in clinical settings to improve their efficacy in the treatment of device-related infections [11,12].
In vitro dynamic models, such as the Center for Disease Control and Prevention Biofilm Reactor (CBR), allow the PK/PD parameters of antimicrobials against bacterial biofilms to be mimicked over time.It has already been validated and used to evaluate the efficacy of different antimicrobials against biofilm-embedded bacteria, including several P. aeruginosa strains [10,13].In our previous works, we evaluated the PK/PD indices of meropenem, predicting anti-biofilm efficacy against several meropenem-susceptible and -resistant P. aeruginosa strains [14].Also, we performed a partial evaluation of the efficacy of CAZ administered in continuous infusion at a range of drug concentrations equivalent to clinical doses lower than 6 g/daily (4 mg/L to 40 mg/L) against CAZsusceptible P. aeruginosa strains [10].With this in mind, in the present work we used the same CBR model to further explore the anti-biofilm PK/PD indices of CAZ against P. aeruginosa isolates.For this aim, we evaluated the efficacy of (i) two regimens of CAZ administration (intermittent bolus and continuous infusion), (ii) doses equivalent to the highest dosages used in clinical practice (6 g daily and 9 g daily), and (iii) the efficacy against two CAZ-susceptible P. aeruginosa strains (HUB-PAS and PAO1) and one CAZresistant P. aeruginosa strain (HUB-XDR).We also studied the anti-biofilm effect of CAZ plus colistin.

Results
All the minimum inhibitory concentrations (MICs), minimum biofilm inhibitory concentrations (MBICs), and minimum biofilm eradication concentrations (MBECs) for CAZ and colistin are shown in Table 1.The MIC values of CAZ were 1, 2, and 32 mg/L for HUB-PAS, PAO1, and HUB-XDR, respectively.
Against CAZ-susceptible strains (HUB-PAS and PAO1), all CAZ monotherapies (CAZ-IB, CAZ-CI50, and CAZ-CI70) presented greater efficacy than the control, with CAZ-CI70 being the most active treatment against each strain, achieving a bactericidal activity (Figure 1a,b).Against the CAZ-resistant strain (HUB-XDR), all CAZ monotherapies had reduced efficacy with only CAZ-CI70 showing a significantly higher decrease in bacterial counts than the control (Figure 1c).Against the CAZ-resistant strain (HUB-XDR), all CAZ monotherapies had reduced efficacy with only CAZ-CI 70 showing a significantly higher decrease in bacterial counts than the control (Figure 1c).
Overall, only the CAZ-CI 70 regimen could completely avoid the appearance of CAZresistant strains.
Table 2 shows the main PK/PD indices obtained for the different CAZ regimens (IB, IC 50 , and IC 70 ) against each strain.All treatments maintained CAZ concentrations above the MIC throughout treatment (f %T > MIC = 100%) except for the CAZ-IB regimen for HUB-XDR (f %T > MIC = 48.5%),which has the highest MIC.The f AUC 0-24 , and therefore f AUC 0-24h /MIC values, were higher for the CAZ-CI 70 regimen than for CAZ-IB and CAZ-CI 50 , but the greatest f C max was achieved by CAZ-IB.The PK/PD index that best correlated with the anti-biofilm efficacy of CAZ was the f AUC 0-24h /MIC (r 2 = 0.78).Figure 2 shows the correlation between the main PK/PD indices and the efficacy for each regimen.Overall, only the CAZ-CI70 regimen could completely avoid the appearance of CAZresistant strains.
Table 2 shows the main PK/PD indices obtained for the different CAZ regimens (IB, IC50, and IC70) against each strain.All treatments maintained CAZ concentrations above the MIC throughout treatment (f%T > MIC = 100%) except for the CAZ-IB regimen for HUB-XDR (f%T > MIC = 48.5%),which has the highest MIC.The fAUC0-24, and therefore fAUC0-24h/MIC values, were higher for the CAZ-CI70 regimen than for CAZ-IB and CAZ-CI50, but the greatest fCmax was achieved by CAZ-IB.The PK/PD index that best correlated with the anti-biofilm efficacy of CAZ was the fAUC0-24h/MIC (r 2 = 0.78).Figure 2 shows the correlation between the main PK/PD indices and the efficacy for each regimen.
The addition of colistin improved the efficacy of all CAZ monotherapies, and achieved with this combined therapy a bactericidal effect against all strains (except for CAZ-IB plus colistin against HUB-XDR).Overall, the combination CAZ-CI 50 plus colistin demonstrated bactericidal activity higher than that of CAZ-IB plus colistin against CAZ-susceptible and CAZ-resistant P. aeruginosa strains.In all cases, combination therapy also prevented the emergence of colistin-resistant and CAZ-resistant strains.

Discussion
There is still a need to evaluate the PK/PD indices most useful for predicting the efficacy of beta-lactams against biofilm-related infections, because traditional parameters were defined to achieve an f %T > MIC above 40-60% and a drug concentration approximately 3-4 times the MIC value for planktonic bacteria and not taking the particularities of biofilm-embedded bacteria into account [3,15].In the present work, we evaluated the PK/PD parameters of several CAZ regimens equivalent to the maximum dosages used in humans (6 g and 9 g daily) administered by IB and CI, against CAZ-susceptible and CAZresistant P. aeruginosa strains with different MIC values.The f %T > MIC index, considered the most important PK/PD index for beta-lactams, was optimized to its maximum (100%) in all cases but the CAZ-IB regimen for the HUB-XDR strain (48.5%;Table 2).However, under these conditions, we observed notable differences between the efficacy of CAZ, with and without colistin, which depended on the strain susceptibility, the dosage, and the form of administration.Overall, CAZ monotherapies achieved poor efficacy against the HUB-XDR strain (Figure 1c), whereas relevant and sometimes bactericidal activity was observed for CAZ-susceptible strains, although benefits were noted when using CAZ-CI at usual (CAZ-CI 50 ) and at high doses (CAZ-CI 70 ) compared to CAZ-IB (Figure 1a,b).Specifically, CAZ-CI 70 monotherapy showed the greatest efficacy, being significantly higher than CAZ-CI 50 or CAZ-IB against the HUB-PAS strain, which had the lowest MIC value (Figure 1a).When comparing the activity of CAZ-IB and CAZ-CI 50 , which were equivalent to a human dose of 6 g daily, the benefits of CI were mainly observed in combination with colistin for all strains.
To our knowledge, few studies have evaluated the specific PK parameters of CAZ against biofilms.Hengzhuang et al. published three works using in vitro methods with planktonic and biofilm bacteria and an in vivo model of biofilm lung infection formed by the PAO1 strain and some derived mutants.In these works, the authors noted a change in beta-lactam (imipenem and CAZ) PD values in biofilms compared to the time-dependent killing in planktonic bacteria, and therefore, the role of dose in these infections [8,16].They observed that CAZ exhibited concentration-dependent killing that was more evident against a beta-lactamase-overproducing mutant strain [9].Our group used the CBR model to evaluate the efficacy of CAZ using CI at concentrations of 4 to 40 mg/L against two CAZ-susceptible P. aeruginosa strains, mimicking concentrations allowed in clinical practice with doses lower than 6 g/d, revealing an improved anti-biofilm activity when higher concentrations of CAZ were used for longer periods [10].In the present work, we performed additional and complementary experiments to evaluate in depth the anti-biofilm PK/PD indices of CAZ; concisely, compared to our previous works, herein we evaluated (i) the comparative efficacy of ceftazidime administered by intermittent bolus or in continuous infusion, (ii) the efficacy of ceftazidime administered at the highest dosages used in clinical practice (6 g daily and 9 g daily), and (iii) the ceftazidime efficacy against ceftazidimesusceptible and -resistant P. aeruginosa strains.Overall, the present work highlights how to optimize the PD parameters of CAZ therapy.
The efficacy of CAZ against P. aeruginosa, when administered by IB and CI, has been compared in a few experimental works, but none has evaluated activity on biofilm cells.By using a two-compartment in vitro PD model, Cappelletty et al. compared the activity of CAZ-IB (equivalent to 2 g/8 h) and CAZ-CI at 5, 10, and 20 mg/L against one CAZ-susceptible and one CAZ-resistant strain.This showed that CAZ at 20 mg/L had a significantly greater efficacy than the other CI regimens and with similar activity to CAZ-IB against the CAZ-susceptible strain, despite the CI regimen (AUC 0-24h values = 677) mimicking a lower dose than the IB regimen (AUC 0-24h = 1780).Moreover, all monotherapies were ineffective against the resistant strain [17].In another work, Mouton et al. used a similar in vitro model to compare the activity of the same dose of CAZ (300 mg/L/24 h), administered by CI or IB against three P. aeruginosa strains.They noted that CI exposed the strains to high and sustained CAZ concentrations and thus confirmed that this regimen was the most effective [18].
Our results mostly agree with these previous studies by Cappelletty et al. and Mouton et al., though we emphasize that the f AUC/MIC ratio, and not the f %T > MIC, was the PK/PD index that best correlated with CAZ efficacy in biofilm cells.We found that the f AUC/MIC ratio could be optimized by using high doses and CI administration.In fact, this index has been related to the efficacy of most antibiotics, considered either timedependent killing or concentration-dependent killing agents, and an AUIC value of 125 (which represents near 80% of the AUC above the MIC) is a key point associated with efficacy and resistance being overcome [19][20][21].Our findings of ineffective CAZ monotherapies against the HUB-XDR strain are of particular interest here.Despite CAZ-IB, CAZ-CI 50 , and CAZ-CI 70 having f %T > MIC values of 48.5%, 100%, and 100%, respectively (Figure 2a), the f AUC/MIC values were clearly below the cut-off in all cases (Figure 2b).While awaiting further studies, these results support the position that CAZ monotherapy, independent of the optimal dosage and form of administration, is not a valid therapeutic alternative to treat biofilms formed by CAZ-resistant P. aeruginosa strains.Given the recent appearance of other beta-lactams that can be used alone or in combination with beta-lactamase inhibitors, their efficacy against biofilm-related infections should be evaluated [22][23][24].
Colistin has been reported to have a good activity against biofilm cells, and in particular, the combination of beta-lactams plus colistin appears to be highly effective against biofilms formed by P. aeruginosa [8,25].Consistent with previous works, our results again show the synergistic effects of CAZ with colistin, with this combination resulting in the best therapy.Overall, the intravenous administration of beta-lactams to allow high drug serum concentrations, as well as their use in combination with colistin, represent two therapeutic strategies with proven therapeutic efficacy in clinical settings with osteoarticular and orthopedic device-related infections caused by GNB.
There are some limitations in the present study, mainly concerning the use of the in vitro biofilm model.Although the CBR allows the evaluation of antibiotic PD parameters against biofilms, this complex structure may differ from in vivo biofilm infections.Furthermore, the model did not mimic other environmental variables, such as host-pathogen interactions.The use of only three strains may preclude study generalization to all P. aeruginosa; however, the present work is still of interest due to its evaluation of CAZ PD parameters against susceptible and resistant strains.The synergistic effects provided by the combination of colistin plus ceftazidime at the highest concentration (70 mg/L) could also be beneficial and should be further explored.Finally, we note that in vitro models can induce the emergence of resistant strains, which may also affect the efficacy of antibiotics, particularly over a prolonged therapy.
In conclusion, CAZ exhibited dose-dependent anti-biofilm killing of P. aeruginosa strains, which was best explained by the f AUC/MIC index.The administration of CAZ by CI provided benefits compared to CAZ by IB, particularly when using high doses alone and in combination with colistin.However, CAZ monotherapy was ineffective against biofilms formed by CAZ-resistant P. aeruginosa strains, independent of dosage optimization; in this case, only colistin plus CAZ administered by CI appeared useful.These findings not only show the potential to optimize CAZ therapy against biofilm-related infections due to P. aeruginosa but also demonstrate its limited activity against CAZ-resistant strains.

Bacterial Isolates and Antimicrobial Agents
Three P. aeruginosa strains were used.The first was the reference PAO1 strain (American Type Culture Collection, Rockville, MD, USA).The second was the clinical isolate HUB-PAS (from a case of prosthetic joint infection at Hospital Universitari Bellvitge).Both strains were susceptible to CAZ, but with different minimum inhibitory concentrations (MICs).The third strain, HUB-XDR, belonged to the high-risk clone ST-175 that has shown extensive drug resistance (AmpC ß-lactamase hyperproduction plus OprD porin deletion), including resistance to CAZ.All strains were susceptible to colistin.
Ceftazidime and colistin were purchased from manufacturers and the purified powder for each antibiotic was resuspended following laboratory recommendations.

Determination of Minimum Inhibitory Concentrations, Minimum Biofilm Inhibitory Concentrations, and Eradication Concentrations
The MICs of CAZ and colistin were determined by broth microdilution, following standard recommendations [26].The MBICs and the MBECs were determined using an MBEC device (Innovotech Inc., Edmonton, AB, Canada) [27] with a range concentration from 0.5 mg/L to 512 mg/L of each antibiotic.All experiments were performed at least in triplicate.

In Vitro Pharmacokinetic/Pharmacodynamic Biofilm Model
A CDC Biofilm Reactor system (CBR; BioSurface Technologies Corp., Bozeman, MT, USA) was used based on a previously described protocol [10,13], which consisted of a biofilm conditioning phase for 48 h followed by a therapeutic phase for 54 h.After the biofilm conditioning phase, where biofilm was formed in Teflon coupons, the therapeutic phase started (baseline, 0 h) with fresh medium (free or with CAZ) was pumped at a flow rate of 2 mL/min to reproduce the half-life of CAZ (t ½ = 2 h).
The following therapeutic regimens were evaluated, mimicking human dosages: CAZ by intermittent bolus (CAZ-IB; free C max = 134 mg/L), equivalent to 2 g every 8 h in humans; CAZ by continuous infusion of 50 mg/L (CAZ-CI 50 ), equivalent to 6 g/24 h; and CAZ by continuous infusion of 70 mg/L (CAZ-CI 70 ), equivalent to 9 g/24 h [28,29].We also evaluated colistin at 3.50 mg/L by CI, simulating the unbound plasma steady-state concentration observed in patients receiving 6-9 MU/day of colistin methanesulfonate [30].Flow rates were calibrated before each experiment and monitored throughout to ensure optimal system operation.We evaluated control therapy (no antibiotic), monotherapy (CAZ-IB, CAZ-CI 50 , CAZ-CI 70 and colistin), and combination therapy (colistin plus CAZ-CI 50 and colistin plus CAZ-IB).All experiments were performed at least in duplicate.

Pharmacokinetic/Pharmacodynamic Analysis
For the PD analyses, at least three coupon samples (biofilm-embedded bacteria) were collected at 0, 6, 24, 30, 48, and 54 h.The removed coupons were then processed as described elsewhere [10], with the coupon samples serially diluted (10-fold), plated on agar plates (Beckton Dickinson, Barcelona, Spain), and incubated at 37 • C for 24-48 h.Bacterial counts were expressed as log 10 CFU/mL and the efficacy was evaluated as a decrease in bacterial counts (∆log 10 CFU/mL X-0 h).The activity of monotherapy or combination regimens was defined as bactericidal or bacteriostatic, synergistic or antagonistic based on standard definitions [31].
The emergence of resistance to CAZ and colistin was screened during the treatment.Samples recovered from coupons were plated into agar plates containing 8 mg/L of CAZ and 4 mg/L of colistin and incubated over 24-48 h.After incubation, MICs were determined following standard protocol [26].
To measure the concentrations of CAZ and colistin, 1 mL of medium containing antibiotic was collected from the CBR system at different times, placed in 1.5 mL microcentrifuge tubes, and stored immediately at −80 • C. We used high-performance liquid chromatography to determine antibiotic levels, as previously described [10].

Confocal Laser Scanning Microscopy
Coupons were evaluated by confocal laser scanning microscopy to confirm biofilm infection (0 h) and treatment activity (54 h), according to a previously reported protocol [10].

Statistical Analysis
Data were analyzed using IBM SPSS v.20.0 (IBM Corp., Armonk, NY, USA).The efficacy of each therapy was evaluated using the log change method from 0 h to each time-point (∆log CFU/mL X-0 h).Therapeutic groups were compared by analysis of variance and Tukey's post-hoc test per treatment.Differences were considered statistically significant at a p-value of <0.05.
The PK parameters, PK/PD indices, and their correlation with the efficacy of each therapeutic regimen were determined according to previously reported methodology [14].Briefly, the included PK parameters, considering free-drug concentrations, were as follows: elimination half-life (t ½), peak concentration (f C max ) for IB, steady-state concentration (f Css) for CI, and area under the concentration-time curve (f AUC 0-t ).For PK/PD integration of CAZ, we calculated the area under the curve for 24 h versus MIC (f AUC 0-24h /MIC), the free-drug peak concentration versus MIC (f C max /MIC), and the percentage of time during which the free-drug level exceeded the MIC (f %T > MIC).The efficacy of the therapeutic regimens at the end of experiments (∆log CFU/mL 54-0 h) was correlated with these PK/PD indices for each CAZ regimen and used to generate fit curves.

Figure 2 .
Figure 2. Correlation between PK/PD indices of therapeutic regimens of CAZ (f %T > MIC (a), f AUC/MIC (b), and f C max /MIC (c)) and anti-biofilm efficacy against each strain.Abbreviations: f AUC, area under the concentration-time curve; CFU, colony-forming unit; MIC, minimum inhibitory concentration; f C max , peak concentration; f %T > MIC, time the free antimicrobial concentration remains above the MIC.

Table 2 .
Pharmacokinetic parameters of each CAZ regimen for each strain used.

Table 2 .
Pharmacokinetic parameters of each CAZ regimen for each strain used.