JAC Comparison of in-vitro pharmacodynamics of once and twice daily ciprofloxacin

Transcription

1 Journal of Antimicrobial Chemotherapy (1999) 44, JAC Comparison of in-vitro pharmacodynamics of once and twice daily ciprofloxacin K. E. Bowker 1 *, M. Wootton 1, C. A. Rogers 1, R. Lewis 2, H. A. Holt 1 and A. P. MacGowan 1 1 Bristol Centre for Antimicrobial Research and Evaluation, Southmead Health Services NHS Trust and University of Bristol, Department of Medical Microbiology, Southmead Hospital, Westbury-on-Trym, Bristol BS10 5NB, UK; 2 Faculty of Applied Sciences, University of the West of England, Frenchay, Bristol BS16 1QY, UK The pharmacodynamics of ciprofloxacin were explored in an in-vitro continuous bacterial culture model of infection, by simulating two oral dosing regimens; 0.5 g 12-hourly (bd) and 1 g 24-hourly (od). Three strains of Escherichia coli (ciprofloxacin MICs 0.03, 0.5 and 2 mg/l); two strains of Pseudomonas aeruginosa (MICs 0.09 and 1.5 mg/l), two strains of Staphylococcus aureus (MICs 0.12 and 1 mg/l) and two strains of Streptococcus pneumoniae (MICs 0.5 and 2 mg/l) were used. Three pharmacodynamic parameters, T > MIC, C max /MIC and AUC/MIC (T = time, C max = peak serum concentration, AUC = area under the curve), were compared with area under the bacterial-kill curve (AUBKC) (after transformation of the AUBKC) using a simple E max or sigmoidal E max model. AUBKC was taken to be the main antibacterial effect measure. The models were compared by inspection of residuals and Akaike information criterion. E max models adequately described the relationship between AUC/MIC and AUBKC and between C max /MIC and AUBKC, but not between T > MIC and AUBKC. All three pharmacodynamic parameters are related to each other but multiple regression analysis indicated that AUC/MIC was the best individual predictor of AUBKC. Despite this, comparison of od and bd regimens indicates some advantage to od in terms of early antibacterial effect. Serum concentration time curve shape has some importance in determining antibacterial effect. These data indicate that for ciprofloxacin AUC/MIC ratio is not the sole determinant of antibacterial effect. Introduction The pharmacodynamics and dosing regimens for fluoroquinolones have been extensively studied using in-vitro animal and human model systems. 1 These drugs have the properties of concentration-dependent killing in the therapeutic range and a post-antibiotic effect of h for both Gram-positive and Gram-negative bacteria. 2 Some animal models indicate that the area under the curve (AUC)/MIC ratio gives the best correlation with outcome, 3,4 while others have suggested that the peak serum concentration (C max )/MIC ratio relates to survival. 5 7 Similarly, using in-vitro models, AUC/MIC ratio has been related to antibacterial effect; 8 as has C max /MIC and also time above MIC (T MIC). 9,10 The inter-relationship between AUC/MIC, C max /MIC and T MIC features constantly in investigations. 6,11 In some studies only one parameter has been compared with antibacterial effect, for example C max /MIC for enoxacin 9 and AUC/MIC for ciprofloxacin and ofloxacin. 12 Alternative predictors, e.g. T MIC, are not always included. 8 If AUC/MIC is the sole determinant of outcome in quinolone therapy, then the shape of the serum concentration time curve will have no impact on antibacterial effect, while if C max /MIC is dominant large infrequent doses would give the best therapeutic results. In this study some of these issues have been explored using a dilutional in-vitro model and a dose fractionation design to compare od and bd regimens using multiple parameters to assess antimicrobial efficacy. Materials and methods Pharmacokinetics Two oral dosing regimens of ciprofloxacin (0.5 g 12-hourly bd and 1 g 24-hourly od) were simulated. The target AUC for each simulation was 24.2 mg/l.h. The time to peak *Corresponding author. Tel: ; Fax: ; bowker_k@southmead.swest.nhs.uk 1999 The British Society for Antimicrobial Chemotherapy 661

2 K. E. Bowker et al. concentration (T max ) was 1 h for both dosing simulations, the target C m a x for bd dosing was 1.7 mg/l and for od dosing 3.4 mg/l. For C min these values were 0.3 and 0.1 mg/l, respectively. Bacterial strains The following clinical isolates, held at the Bristol Centre for Antimicrobial Research and Evaluation, were used: Escherichia coli SMH 5773, SMH 5774 and SMH 5311 (ciprofloxacin MICs 0.03, 0.5 and 2 mg/l, respectively); Pseudomonas aeruginosa SMH 8545 and SMH 5761 (ciprofloxacin MICs 0.09 and 1.5 mg/l), Staphylococcus aureus SMH 8546 and SMH 8548 (ciprofloxacin MICs 0.12 and 1 mg/l) plus Streptococcus pneumoniae SMH and SMH (ciprofloxacin MICs 0.5 and 2 mg/l). Antibiotic and media Ciprofloxacin hydrochloride was supplied by Bayer plc., 1 g/l stock solutions, dissolved in water were stored at 70 C, and a fresh aliquot used for each simulation. Stock solution was added to 32 ml Isosensitest broth (ISB) or Brain Heart Infusion broth (BHI) (Oxoid, Basingstoke, UK) containing nicotine adenine dinucleotide (NAD), haemin and histidine (10 mg/l) for bd and od simulations. Viable counts were performed on nutrient agar plates (Merck, Dorset, UK), containing 1% magnesium chloride (BDH, Newbury, Berks, UK) to neutralize the ciprofloxacin, using a spiral plater (Don Whitley, Shipley, West Yorkshire, UK). For S. pneumoniae strains nutrient agar plates containing 1% magnesium chloride supplemented with 8% whole horse blood (TCS Microbiology, Buckingham, UK) were used. Plates were incubated at 37 C in air for 18 h for E. coli, P. aeruginosa and S. aureus. S. pneu - moniae plates were incubated at 37 C in 5% CO 2 for 18 h. In-vitro model description A New Brunswick Biostat C-30 (New Brunswick, Hatfield, Hertfordshire, UK) simulating a one compartment open model for oral administration was used. The apparatus consisted of a reservoir containing diluted ISB (2% for E. coli and P. aeruginosa; 6% for S. aureus and 75% BHI supplemented with 10 mg/l haemin, NAD and L-histidine for S. pneumoniae) connected via silicon and aluminium tubing to a dosing chamber and on to a culture chamber. Medium was pumped into the chambers via a peristaltic pump (Ismatec, Bennett and Co., Weston-Super-Mare, UK) at a flow rate of 1.1 ml/min to give an elimination half-life of 230 min. The chamber was agitated at 300 rpm. The temperature of the culture chamber was maintained at 37 C. A 100- L aliquot of an overnight broth of the test organism was inoculated into the culture chamber and the model then run for 18 h to enable the organism to reach a steady state concentration of approximately c f u / m L. Ciprofloxacin was then inoculated into the dosing chamber according to the dosing regimen being simulated. Aliquots were collected via the outflow tube for viable count and and assay by high pressure liquid chromatography (HPLC) 13 at 0, 1, 2, 3, 4, 5, 6, 7, 8, 10, 12, 14, 22 and 24 h after the dose. Each simulation was performed once. A subculture of the test organism from the beginning and the end of each simulation was stored on Dorset Egg slopes or at 70 C for MIC determination by agar incorporation technique. 14 Pharmacodynamics and statistics The parameters used to measure antibacterial effect were maximum change in viable count ( max, log cfu/ml), change in viable count at 6 h ( 6, log cfu/ml), change in viable count at 12 h ( 12, log cfu/ml), change in viable count at 24 h ( 24, log cfu/ml), slope of the bacterial time kill curve between 0 and 6 h (S), time to kill 99.9% of the initial inoculum (T99.9) and the area under the bacterial-kill curve (AUBKC, log cfu/ml.h). The AUBKC was calculated after the inoculum was standarized, by the log linear trapezoidal rule (Graph Pad Prizm TM, San Diego, CA, USA). For pharmacodynamic analysis the percentage of time the concentration exceeded the MIC (T MIC), and C max /MIC and AUC/MIC ratios were calculated. Linear regression had previously shown that the AUBKC correlated best with bacterial killing (data not shown), and has been used by others, 8 therefore this parameter was used to compare AUC/MIC, C max /MIC and T MIC using a simple E max and a sigmoidal E max model. AUBKC was transformed into ln (1/AUBKC) 6, before model fitting using WinNonlin Software (Pharsight, Mountain View, CA, USA) as described previously. 8 The two different models were compared by inspection of plots of residuals versus predicted values and Akaike information criterion. The correlation between the pharmacodynamic variables was assessed using Spearman rank correlation and subsequently multiple regression analysis was used to examine the combined effect of AUC/MIC, C max /MIC and T MIC on log (AUBKC) and log ( max 7). Both forward and backward variable selection methods were used. The antibacterial effect parameters were compared between the od and bd regimens using the statistical Sign test for related samples. Results Ciprofloxacin concentrations in the in-vitro model The target concentrations associated with each simulation (bd and od) are shown in Figure 1. The concentration achieved in each simulation and mean values are also shown. T MIC (% of 24 h) ranged from 0 to 100% for bd and 18 to 100% for od, depending on the pathogen MIC; C max /MIC ranged from 0.9 to 58 bd and 1.8 to 117 od, while 662

3 Fluoroquinolone pharmacodynamics shown in Figure 2 (a i). In general and as expected, bacterial killing was greater the lower the MIC for the pathogen irrespective of the simulation used. S. pneumoniae was different from the other three species tested, in that bacterial killing was noticeably poorer in comparison with E. coli strains, even for strains with the same MICs (0.5 mg/l and 2 mg/l). Figure 1. Target and actual ciprofloxacin concentrations with (a) 500 mg bd (bd) (, target concentration (mg/l);, mean concentration (mg/l)); and (b) 1 g od (od) simulations (, target concentration (mg/l);., mean concentration (mg/l)). AUC/MIC ratios ranged from 12 to 808 for both simulations (Table I). Bacterial time kill responses to bd and od simulations Bacterial time kill curves for the bd and od simulations, for E. coli, P. aeruginosa, S. aureus and S. pneumoniae are Use of E max models to fit the data and relationship between pharmacodynamic variables The simple E m a x model best described the relationship between AUC/MIC and ln (1/AUBKC) 6 (Figure 3), while the relationship between C m a x /MIC and ln (1/AUBKC) 6 was best described by a sigmoid E m a x model (Figure 4). The relationship between T MIC and ln (1/AUBKC) 6 could only be described by a sigmoid E m a x model as the simple model failed to converge. However, inspection of the residuals and model r 2 indicated the data did not fit well for T MIC, suggesting that the relationship is not sigmoidal. Despite the dose fractionation design there was a strong correlation between the three pharmacodynamic variables (Tables I and II). Multiple regression analysis Both backward and forward variable selection methods gave concordant results. AUC/MIC was the best single predictor of log (AUBKC) and of log ( max 7) and no s i g n i ficant improvement in fit was observed when C m a x / M I C or T MIC were added. There was some evidence to Table I. Relationships of pharmacodynamic parameters Pharmacodynamic parameter Simulation Pathogen MIC (mg/l) AUC/MIC C max /MIC T MIC E. coli, bd E. coli, od E. coli, bd E. coli, od E. coli, bd E. coli, od P. aeruginosa, bd P. aeruginosa, od P. aeruginosa, bd P. aeruginosa, od S. aureus, bd S. aureus, od S. aureus, bd S. aureus, od S. pneumoniae, bd S. pneumoniae, od S. pneumoniae, bd S. pneumoniae od

4 K. E. Bowker et al. Figure 2. Bacterial time kill curves comparing bd (solid squares) and od (open squares) simulations with ciprofloxacin MICs for E. coli: (a) 0.03 mg/l;, bd simulation;, od simulation; (b) 0.5 mg/l;, bd simulation;, od simulation; (c) 2.0 mg/l;, bd simulation;, od simulation; P. aeruginosa: (d) 0.09 mg/l;, bd simulation;, od simulation; (e) 1.5 mg/l;, bd simulation;, od simulation; S. aureus: (f) 0.12 mg/l;, bd simulation;, od simulation; (g) 1.0 mg/l,, bd simulation;, od simulation; S. pneumoniae: (h) 0.5 mg/l;, bd stimulation;, od simulation; and (i) 2.0 mg/l;, bd simulation;, od stimulation. 664

5 Fluoroquinolone pharmacodynamics Table III. Comparison of bactericidal activities of od and bd simulations Occasions when od more bactericidal Sign test for related Parameter than bd samples Figure 3. Relationship between AUC/MIC ratio and ln (1/AUBKC) 6 (Simple E max model);, observed;, predicted. max 8/9 P /8 P /9 P /9 P 1.0 S 9/9 P 0.01 AUBKC 6/9 P 0.5 T99.9 5/6 P 0.2 max, maximum reduction in viable count; 6, 12, 24, reduction in viable count at 6, 12 and 24 h; S, slope; AUBKC, area under the bacterial kill curve; T99.9, time to 99.9% kill. terms of max, 12 and S (P 0.05) but not 6, 24, AUBKC or T Figure 4. Relationship between C max /MIC ratio and ln (1/AUBKC) 6 (Sigmoid E max model);, observed;, predicted. suggest that the inclusion of the quadratic term (AUC/ MIC 2 ) improved the fit of the model describing log (AUBKC) (P 0.07) but not log ( max 7), although examination of plots of residuals versus predicted values suggested that the variation in log (AUBKC) and log ( max 7) could not be described wholly in terms of AUC/MIC. Comparison of od and bd regimens using different measures of antibacterial effect The data on od and bd regimens was assessed by comparing the occasions when od was more bactericidal than bd (Table III). The Sign test indicated that od was superior in Variables Table II. Correlation between variables Spearman rank correlation (95% CI) AUC/MIC and C max /MIC 0.95 ( ) AUC/MIC and T MIC 0.97 ( ) T MIC and C max /MIC 0.93 ( ) Susceptibility of isolates after ciprofloxacin MICs performed on isolates before and after ciprofloxacin dosing showed no changes with either simulation. Discussion Rational drug dosing depends on a knowledge of the pharmacodynamic properties of the agent. If the time the drug concentration remains higher than the pathogen MIC (T MIC) is the dominant factor in deciding clinical or bacteriological outcomes, then small frequent doses will give the best therapeutic response; in contrast, if maximum serum concentrations (C max )/MIC ratios are the more important then large infrequent doses are best. If the AUC/MIC is most important, then the frequency of dosing is not vital and multiple dosing will be as effective as single doses, provided that the total AUC is the same. In the case of the fluoroquinolones there are many data supporting the concept that AUC/MIC ratios are important in determining outcome. 11,12,15 Animal models indicate that log (AUC) is the most important indicator of efficacy. 3 This has been confirmed in in-vitro models, when AUC/MIC and C max / MIC were related to antibacterial effect. 8 (T MIC was not used in that particular analysis.) 8 When the antibacterial effect was defined by the 24 h AUBKC and various regimens of ofloxacin and ciprofloxacin studied, AUC/MIC was related to the antibacterial effect against P. aerugi - nosa. 12 In this study, a similar technique to that used by the authors in a study quoted above 12 was used to relate AUC/ MIC and C max /MIC to antibacterial effect (AUBKC). Sigmoid models were satisfactory for the explanation of these 665

6 K. E. Bowker et al. relationships, as either a sigmoid E max model or a simple E max model could describe the relationship between ln (1/AUBKC) 6 and AUC/MIC or C max /MIC (Figures 3 and 4). This is not true of T MIC which cannot easily be related to AUBKC using sigmoidal models. Our study incorporated a dose fractionation design, examining a wide range of pathogens with different ciprofloxacin MICs. Analysis of the three major pharmacodynamic parameters (AUC/MIC, C max /MIC and T MIC) indicated, with the use of multiple regression analysis, that AUC/MIC ratio was the main determinant of outcome. Others have used linear relationships to relate antibacterial effect of ciprofloxacin or trovafloxacin to AUC/MIC or T MIC, without performing multivariate analysis. 10 In these experiments the antibacterial effect was calculated taking into account regrowth up to the point at which counts recover to near levels of the controls not exposed to the drug. This end point is significantly different from that used by ourselves and others. 6,12 This may also explain why other authors feel that the best predictor of quinolone antibacterial effect is T MIC. 10 The importance of AUC/MIC as a determinant of outcome is supported by clinical studies in which ciprofloxacin has been used to treat ITU-acquired pneumonia and grepafloxacin employed in the therapy of exacerbation of chronic obstructive pulmonary disease. 11,15 Animal models provide additional information to support the concept that, given equivalent AUC/MIC ratios, regimens using infrequent dosing of ciprofloxacin are superior to continuous infusion, when time to death is used as the outcome measure. 5 These observations were explored further by Drusano et al., 6 using a rat P. aeruginosa infection model. Treating with high-dose lomefloxacin, survival was found to correlate best with the C max /MIC ratio and T MIC was not an important factor. These authors also showed, in a further series of experiments, that smaller daily doses of lomefloxacin administered either once daily or twice daily were equivalent in terms of survival; that is, AUC/MIC was also related to outcome. 6 In in-vitro models, for some bacterial strains, once a day dosing appeared to be superior to twice a day, as defined by the time to kill 99.9% of the initial inoculum. 16 Our data also indicate once a day dosing is superior to twice a day, in terms of initial speed of killing and maximum bactericidal effect, but there was no difference in the effect over 24 h as indicated by AUBKC and 24 antibacterial effect measures. It is possible that initial increased rate of kill observed in our model with od dosing is translated into survival in animal models, but it was not possible to show that C max /MIC was the best predictor of log ( max 7) using multivariate analysis. Recently, based on animal and human data, it has been suggested that if the C max /MIC ratio is 10 then this parameter is the most significant in determining outcome, but at lower C max /MIC ratios AUC/MIC is important. 17 In this study, in 12 of 18 experiments the C max /MIC was 10, which may explain why AUC/MIC was found to be a better predictor of outcome. The comparative data for od and bd dosing indicates that, at least by some parameters of antibacterial effect, od dosing is superior. MIC has previously been demonstrated to correlate with bacterial killing. This was also apparent in this study. S. pneumoniae is exceptional in this respect, little or no bacterial killing being seen at MICs at which this occurs in other species. Hyatt et al. 18 had previously demonstrated that at the same MICs bacterial killing was greater for P. aeruginosa than for S. aureus or S. pneumoniae. It was suggested that this could be due to the higher MBCs of the latter. In conclusion, these data support the concept that AUC/MIC is important in determining the efficacy of quinolones, but that if AUC/MIC ratios are the same, then od produces more rapid and greater bacterial killing. While od dosing is superior to bd using some measures of antibacterial effect, bd is never superior to od. Clearly the precise relationship between AUC/MIC, C max /MIC and outcome remains to be finally established. Acknowledgement We thank Professor A. Dalhoff (Bayer AG, Germany) for his support. References 1. MacGowan, A. P. & Bowker, K. E. (1997). Pharmacodynamics of antimicrobial agents and rationale for their dosing. Journal of Chemotherapy 9, Suppl. 1, MacKenzie, F. M. & Gould, I. M. (1993). The post-antibiotic effect. Journal of Antimicrobial Chemotherapy 32, Leggett, J. E., Ebert, S., Fantin, B. & Craig, W. (1990). Comparative dose-effect relations at several dosing intervals for lactam, aminoglycoside and quinolone antibiotics against Gram-negative bacilli in murine thigh-infection and pneumonitis models. Scandinavian Journal of Infectious Diseases 74, Suppl., Fantin, B., Legett, J., Ebert, S. & Craig, W. A. (1991). Correlation between in vitro and in vivo activity of antimicrobial agents against Gram-negative bacilli in a murine infection model. Antimicrobial Agents and Chemotherapy 35, Roosendaal, R., Bakker-Woudenberg, I. A., van den Berghe-van Raffe, M., Vink-van den Berg, J. C. & Michel, B. M. F. (1989). Impact of the dosage schedule on the efficacy of ceftazidime, gentamicin and ciprofloxacin in Klebsiella pneumoniae, pneumonia and septicemia in leukopenic rats. European Journal of Clinical Microbiology and Infectious Diseases 8, Drusano, G. L., Johnson, D. E., Rosen, M. & Standiford, H. C. (1993). Pharmacodynamics of a fluoroquinolone antimicrobial agent in a neutropenic rat model of Pseudomonas sepsis. Antimicrobial Agents and Chemotherapy 37, Sullivan, M. C., Cooper, B. W., Nightingale, C. H., Quintiliani, R. & Lawlor, M. T. (1993). Evaluation of the efficacy of ciprofloxacin against Streptococcus pneumoniae by using a mouse protection model. Antimicrobial Agents and Chemotherapy 37, Madaras-Kelly, K. J., Ostergaard, B. E., Horde, L. B. & Rotshafer, J. C. (1996). Twenty-four-hour area under the concentration-time- 666

7 Fluoroquinolone pharmacodynamics curve/mic ratio as a generic predictor of fluroquinolone antimicrobial effect using three strains of Pseudomonas aeruginosa in an in vitro pharmacodynamic model. Antimicrobial Agents and Chemotherapy 40, Blaser, J., Stone, B. B., Groner M. C. & Zinner, S. H. (1987). Comparative study with enoxacin and netilmicin in a pharmacodynamic model to determine importance of ratio of antibiotic peak concentration to MIC for bactericidal activity and emergence of resistance. Antimicrobial Agents and Chemotherapy 31, Firsov, A. A., Shevchenko, A. A., Vostrov, S. N. & Zinner, S. H. (1998). Inter and intra quinolone predictors of antimicrobial effect in an in vitro dynamic model: new insight into a widely used concept. Antimicrobial Agents and Chemotherapy 42, Forrest, A., Nix, D. E., Ballow, C. H., Gross, T. F., Birmingham, M. C. & Schentag, J. J. (1993). Pharmacodynamics of intravenous ciprofloxacin in seriously ill patients. Antimicrobial Agents and Chemotherapy 37, Madaras-Kelly, K. J., Larsson, A. J. & Rotschafer, J. C. (1996). A pharmacodynamic evaluation of ciprofloxacin and ofloxacin against two strains of Pseudomonas aeruginosa. Journal of Antimicrobial Chemotherapy 37, Gau, W., Ploschke, H. J., Schmidt, K. & Weber, B. (1985). Determination of ciprofloxacin in biological fluids by high performance liquid chromatography. Journal of Liquid Chromatography 8, Holt, H. A., Bywater, M. J. & Reeves, D. S. (1990). In-vitro activity of cefpodoxime against 1834 isolates from domicilary infections in 20 UK centres. Journal of Antimicrobial Chemotherapy 26, Suppl. E, Forrest, A., Chodosh, S., Amantea, M. A., Collins, D. A. & Schentag, J. J. (1997). Pharmacokinetics and pharmacodynamics of oral grepafloxacin in patients with acute bacterial exacerbations of chronic bronchitis. Journal of Antimicrobial Chemotherapy 40, Suppl. A, Kang, S. A., Rybak, M. J., McGrath, B., Kaatz, G. W. & Seo, S. M. (1994). Pharmacodynamics of levofloxacin, ofloxacin and ciprofloxacin alone and in combination with rifampin against methicillin susceptible and resistant Staphylococcus aureus in an in vitro infection model. Antimicrobial Agents and Chemotherapy 38, Preston, S. L., Drusano, G. L., Berman, A. L., Fowler, C. L., Chow, A. T., Dornseif, B. et al. (1998). Pharmacodynamics of levofloxacin: A new paradigm for early clinical trials. JAMA 279, Hyatt, J. M., Nix, D. E. & Schentag, J. J. (1994). Pharmacokinetic and pharmacodynamic activities of ciprofloxacin against strains of Streptococcus pneumoniae, Staphylococcus aureus and Pseudomonas aeruginosa for which MICs are similar. Antimicrobial Agents and Chemotherapy 38, Received 3 December 1998; returned 1 March 1999; revised 16 April 1999; accepted 14 July

8