Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

The therapeutic potential of antibacterial compounds is primarily determined by their in vitro activity and their in vivo pharmacokinetics. Among respiratory pathogens, Haemophilus influenzae and Moraxella catarrhalis are inhibited by all FQs in concentrations of less than 0.25 µg/ml (Hooper and Wolfson, 1991). FQs have large apparent volumes of distribution, and as a result, their peak serum concentrations are generally no higher than 6 mg/l (Neuman, 1988). On the other hand, the good diffusion of FQs into body tissues, which includes penetration into the extravascular and intracellular compartments (Nix et al., 1991), may be valuable in the treatment of bacterial pneumonia (pulmonary tissue-to-serum ratio >2) (Davies and Maesen, 1986). Based on their activity against Streptococcus pneumoniae, the organism responsible for about 50% of bacteriologically documented community-acquired pneumonias (Moine et al., 1994), FQs can be classified into three groups (Piddock, 1994): (1) compounds whose in vitro MICs are far greater than concentrations achievable in humans (8-16 µg/ml) and therefore inconsistent with therapeutic efficacy, e.g., pefloxacin, enoxacin, fleroxacin and lomefloxacin; (2) compounds with lower MICs (1- 4 µg/ml) that, however, remain close enough to achievable serum peak levels to carry a risk of therapeutic failure (Perrez-Tallero et al., 1990; Lee et al., 1991) and of selection of resistant mutants (Lafredo et al., 1993; Bernard et al., 1994) when these compounds are used in standard regimens, e.g., ciprofloxacin and ofloxacin; (3) newer compounds, most notably sparfloxacin, whose in vitro activity is improved (MICs, 0.125-0.5 µg/ml), such as sparfloxacin, trovafloxacin, clinafloxacin, grepafloxacin, DU-6895a and BAY 12-8039 (Stein, 1996). However, routine MIC determination provides information only on in vitro susceptibility of bacteria. We previously have showed the greater efficacy of sparfloxacin compared with ciprofloxacin and ofloxacin, in relation to its prolonged elimination half-life, in experimental mouse pneumonia models (Azoulay-Dupuis et al., 1992).

The specific relationships that link pharmacokinetic parameters (e.g., peak serum concentration, AUC, T_1/2) to parameters measuring pharmacodynamic interactions between the antimicrobial and the bacteria (i.e., the MIC) can be used to define Pk/Pd surrogate markers for bacterial killing or clinical cure endpoints (Hyatt et al., 1995). These Pk/Pd parameters depend on the antimicrobial class. A recently published abstract by Andes et al. (1995) reported that maximum efficacy for the -lactam amoxicillin against S. pneumoniae was achieved when serum levels exceeded the MIC of the strain during 40 to 50% of the dosing interval. In experimental animal pneumonia models, evidence has been found that the AUC/MIC ratio and the peak/MIC ratio may be the most useful Pk/Pd surrogates for bacterial killing of Klebsiella pneumoniae or Pseudomonas aeruginosa by FQs; this has been demonstrated, in particular, for CPFX in a model of neutropenic animals (Schentag et al., 1993).

We used a mouse model of severe pneumococcal pneumonia both to determine the Pk/Pd parameters that best reflect pulmonary killing curves and clinical cure endpoints and to evaluate the impact of dosing intervals on treatment efficacy during administration of two FQs, CPFX and SPFX. We chose these two marketed drugs because they belong to the same quinolone class but exhibit different profiles: CPFX, whose efficacy may be amenable to improvement and whose short half-life, unchanged in infected animals (Vallée et al., 1992), allows use of divided doses to avoid potential risks of accumulation, and SPFX, which has the lowest MICs against S. pneumoniae and a long half-life, in particular in infected animals (Azoulay-Dupuis et al., 1992). (This work was presented in part at the 33rd International Conference on Antimicrobial Agents and Chemotherapy, New Orleans, October 17-20, 1993, abstract 83).

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Animals. Female C57Bl/6 mice aged 6 to 7 weeks and weighing 18 to 20 g were obtained from Iffa-Credo Laboratories (L'Arbesle, France) and housed in our laboratory for 1 week before the experiment.

Bacterial strain. S. pneumoniae (Sp) strain P4241, a virulent serotype 3 originally isolated from blood and kindly provided by the Centre de Référence du Pneumocoque (Dr. P. Geslin, Créteil, France), was used throughout the study. The isolate was stored at 70°C in brain-heart infusion broth (Bio-Mérieux, Lyon, France) supplemented with 5% filtered horse serum (Diagnostic Pasteur, Ville D'Avray, France) and nutrient broth (10% glycerol, 4% sorbitol; Difco Laboratories, Paris).

Antibiotics. CPFX was provided by Bayer Laboratories (Wuppertal, Germany) and SPFX by Rhône-Poulenc-Rorer (Vitry-sur-Seine, France). An aqueous solution of the hydrochloride salt of CPFX was used for subcutaneous injections to mice. SPFX base was prepared for in vivo experiments by homogenizing the powder in a standard diluent containing carboxymethylcellulose (70 cP) in a concentration of 2 g % in isotonic saline solution (0.09%).

In vitro susceptibility tests. MICs and MBCs for the test strain were determined with microtiter broth dilution techniques with Mueller-Hinton infusion broth (Difco, Detroit, MI) as the test medium and final inocula ranging from 5 × 10⁵ to 10⁷ CFU/well. The MIC was defined as the lowest antimicrobial concentration preventing visible growth after aerobic incubation for 18h at 37°C. The MBC was determined by plating 0.01-ml samples from wells with no visible growth onto horse blood agar (Bio-Mérieux, Lyon, France), incubating the plates overnight at 37°C under aerobic conditions and recording the lowest antimicrobial concentration that killed >99.9% of the original inoculum.

Experimental pneumonia model. Pneumococcal pneumonia was induced in mice as described in detail elsewhere (Azoulay-Dupuis et al., 1991a). Mice anesthetized by intraperitoneal injection of sodium pentobarbital were inoculated endotracheally via the mouth with about 10⁵ log-phase CFU of the Sp test strain. The mice developed subacute pneumonia and died within 7 days, with the peak mortality rate occurring on the fifth day after inoculation. Bacteremia developed within 6 h in half the animals and within 24 h in all animals, and lung bacterial counts increased progressively. Bacterial counts were greater than 10⁸ CFU/lung and 10⁶ CFU/ml of blood at the time of death.

In vivo bacterial killing kinetics. The time course of bactericidal activity was recorded in the lungs during the 24-h period after single subcutaneous injections of CPFX (25, 50, 100 and 200 mg/kg) or SPFX (6.25, 12.5, 25 or 50 mg/kg). A group of untreated control mice also was studied. The FQs were administered 48 h after inoculation with S. pneumoniae, at which time there was histological evidence of advanced experimental pneumonia (Azoulay-Dupuis et al., 1991a). Mice were sacrificed by CO₂ asphyxiation. The lungs and blood of mice (n = 3) were collected 1, 3, 6, 9, 12 and 24 h after drug administration. Blood was collected by intracardiac puncture and was used for qualitative cultures (brain-heart infusion broth, Difco). Lungs were harvested from exsanguinated mice, washed and homogenized in 1 ml of sterile saline solution (Ultra-Turrax T25, Ika-Labortechnik, Staufen i. Br., Germany). Viable bacteria were counted in whole-lung homogenates by plating 0.1 ml of serial 10-fold dilutions of samples onto Columbia agar with 5% sheep blood and incubating the agar aerobically for 24 h at 37°C. Results are expressed as log₁₀ CFU/lungs (limit of detection: 1 log CFU). E_max is the maximal bactericidal effect observed (maximal log CFU reduction) with each tested dose of antimicrobial relative to untreated control mice. The %AUBC designates the area under the bactericidal curve observed during the 24-h period after drug administration, relative to that under the growth curve of untreated control animals.

Therapeutic trials. Treatment was initiated 48 h after inoculation, as mentioned above. Antibiotics were administered subcutaneously for 3 days as repeated doses every 24, 12, 8, 6 or 3 h. Fifteen animals per treatment group were used, and groups were randomized to treatment conditions. All animals in each experiment were infected simultaneously. Experiments were repeated at least twice. Cumulative survival rates were recorded daily for 10 days in each treatment group. Results are expressed as per cent survival rates 7 days after the end of therapy (no deaths occurred after this time interval).

Pharmacokinetic studies. The antimicrobial concentrations in serum were determined in infected mice. CPFX and SPFX were administered subcutaneously 48 h after inoculation with S. pneumoniae. Doses used were as follows: 25, 50, 75, 100, 150 and 200 mg/kg of CPFX and 6.25, 12.5, 25 and 50 mg/kg of SPFX. The mice were sacrificed with diethylether, and the blood from groups of three mice were collected 0.5, 1, 3, 5, 7, 9, 12 and 24 h after drug administration, as described previously (Vallée et al., 1992). Drug concentrations were determined by the agar well diffusion method with Escherichia coli ATCC 39118 as the indicator organism and Antibio-Medium 2 (Difco) as the test medium. Standard solutions of CPFX and SPFX in phosphate buffer, pH 6.8, were prepared for detection of the active free fraction of the drugs in the specimens. Concentrations were determined by averaging diameters from three replicate plates and comparing the results to a standard curve. Results were expressed as micrograms per milliliter of serum. The standard curves were linear from 0.06 to 32 µg/ml, and the lower limit of sensitivity was 0.06 µg/ml for both CPFX and SPFX. The coefficient of between- and within-day variation for replicates (n = 5) was 5% for both FQ assays. Concentration-time data were modeled, and pharmacokinetic parameters (peak level, T_1/2 and AUC) were calculated by nonlinear least-squares regression analysis (nonlinear Apis software, Mips, Marseille, France). Multiple models were evaluated. Best fit of experimental points was obtained with a one-compartment model with zero-order absorption and first-order elimination. Optimization was accomplished by use of the maximum likehood estimation criterion.

Statistical analysis. The observers involved in data collection and analysis were not completely blind to treatment conditions. However, the methodology used for sample identification prevented subjective bias in the experiments. On the other hand, doses and animals were randomized to treatment conditions. Data are expressed as mean ± S.D. Means were compared between groups by one-way or two-way (drug/dose) variance analysis. Correlations between parameters were sought by use of the linear regression coefficient r. Constant terms of the regression lines computed for the two FQs were compared when appropriate. To define the relationship between survival rate and Pk/Pd indexes at the steady state, the regression method was applied to the linear portion of the theoretically sigmoid curve (i.e., excluding 0 and 100% survival data). P values of .05 or less were considered significant. All statistical analyses were conducted with version 4.5 of Statview, Abacus Concepts, Inc., Berkeley, CA.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

MIC and MBC determinations. Against the challenge organism, MICs were 0.75 µg/ml for CPFX and 0.25 µg/ml for SPFX. MBC values were twice the MICs. Inoculum size, which ranged from 5 × 10⁵ to 10⁷ CFU/ml did not affect these values.

Pharmacokinetics in mouse serum. Single-dose pharmacokinetics of CPFX and SPFX administered at various doses were modeled from data from individual animals. Profiles were best described by a one-compartment model with monoexponential elimination. Parameter estimates and profiles were used to calculate the AUC above the MIC (AUC>MIC) and the time spent with serum concentrations greater than the MIC of the challenge strain (tMIC). The mean data for various doses of CPFX and SPFX are reported in table 1. For both FQs significant linear correlations were found between peak level, T_1/2 or AUC and doses, with P values <10⁴ in all cases (peak level vs. dose: r = 0.880 and 0.916; T_1/2 vs. dose: 0.851 and 0.932; AUC vs. dose: r = 0.865 and 0.956 for CPFX and SPFX, respectively). At similar doses, i.e., 25 and 50 mg/kg, peak levels were lower for SPFX than CPFX (P = .020), but half-lives were longer (P < 10⁴) and AUCs larger (P = .0016). This, added to lower MIC for SPFX, resulted in significantly higher values of pharmacodynamic parameters and greater potency of SPFX than CPFX, i.e., peak/MIC ratio (P < 10⁴), AUC/MIC ratio (P < 10⁴), AUC>MIC (P = .0005) and tMIC (P < 10⁴).

In vivo bacterial killing. Bacterial killing curves in lung for the various doses of CPFX and SPFX are shown in figure 1, and killing constants calculated from experimental data are given in table 2. At similar doses (i.e., 25 and 50 mg/kg) bactericidal effect was greater for SPFX than for CPFX, which indicated a difference in potency (P = .0074). Killing rate was not dose-dependent and was significantly slower with SPFX than with CPFX (0.34 ± 0.05 vs. 0.56 ± 0.06 log CFU/h, P = .0009), but the bactericidal effect of SPFX lasted much longer.

View larger version (25K): [in this window] [in a new window]   Fig. 1.   (a) Bacterial clearance from the lungs after single s.c. injections of various doses of CPFX (mg/kg) at various times after infection. (b) Bacterial clearance from the lungs after single s.c. injections of various doses of SPFX (mg/kg) at various times after infection.

4

View larger version (31K): [in this window] [in a new window]   Fig. 2.   Relationships between various Pk/Pd parameters and Emax for CPFX and SPFX, derived from pulmonary bacterial killing curves after single injections in our S. pneumoniae pneumonia model.

4

4

View larger version (36K): [in this window] [in a new window]   Fig. 3.   Relationships between various Pk/Pd parameters and %AUBC for CPFX and SPFX, derived from pulmonary bacterial killing curves after single injections in our S. pneumoniae pneumonia model.

Therapeutic trials. The goal of the therapeutic trials was to evaluate the impact of dose fractionation on treatment efficacy. We selected the minimal daily dosage that provided a 75 to 80% survival rate in infected animals when administered once daily for 3 days. This daily dosage was then fractionated into two, four or eight doses (given every 12, 6 or 3 h, respectively) (table 3). With CPFX (200 mg/kg/day), survival rate was significantly higher in the once-daily dosing group than the other three regimens which gave similar outcome (75 ± 9% vs. 29 ± 14%, P = .0004). With SPFX, survival rate was not modified significantly by fractionation of the daily dosage (50 mg/kg). The eight-time daily schedule was not used with SPFX because it resulted in major drug accumulation from dose to dose.

4

View larger version (33K): [in this window] [in a new window]   Fig. 4.   Relationships between various Pk/Pd parameters and survival rate for CPFX and SPFX, derived from dose fractionation experiments in our S. pneumoniae pneumonia model.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

S. pneumoniae is the most common cause of severe, community-acquired, bacterial pneumonia, of which many cases are fatal (Moine et al., 1995). The increasing emergence throughout the world of S. pneumoniae strains with resistance to -lactams and other antimicrobials (Baquero, 1995) is raising new therapeutic challenges. Demonstration of antipneumococcal efficacy of new classes of antimicrobials broadens the therapeutic armamentarium available for S. pneumoniae pneumonia. FQs are not only active on intracellular organisms responsible for community-acquired pneumonia (Thyset al., 1991) but also exhibit similar in vitro activity on S. pneumoniae strains with and without susceptibility to penicillin (Canton et al., 1992). Rational selection of an antimicrobial and of its administration route for the treatment of a focus of bacterial infection requires knowledge of the specific relationships that link the pharmacokinetic parameters of the drug to parameters reflecting pharmacodynamic interactions between the drug and the bacterium (e.g., the MIC) (Dudley, 1991; Nix and Schentag, 1988; Hyatt et al., 1995). Studies have found that the efficacy of -lactams is correlated with the time spent with serum antimicrobial levels above the MIC for the causative organisms, and that the efficacy of aminoglycosides is correlated with the peak/MIC ratio or the AUC/MIC ratio (Vogelman et al., 1988; Craig et al., 1991). In our experimental study, we used two in vivo pharmacodynamic endpoints, intrapulmonary bacterial killing after single antibiotic injections and survival after therapy with fractionated doses. We also used two parameters for evaluating bacterial killing, E_max (the maximal bactericidal effect) and %AUBC (the total bactericidal effect throughout time). In our model, intrapulmonary bacterial killing by CPFX or SPFX depended on the concentration of the antimicrobial, in keeping with previous studies (Vallée et al., 1991). All Pk/Pd study parameters were correlated strongly with E_max, based on their covariance (Schentag et al., 1993; Peloquin et al., 1989), but the strongest correlation was seen with AUC>MIC, a parameter that reflects both the concentration of the drug and the time spent with a concentration greater than the MIC for the causative pathogen. AUC>MIC represents the amount of drug above the inhibition threshold and depends both on the peak serum level and on the half-life of the drug. In our experiments, peak/MIC seems to have a greater contribution to achievement of the E_max than the half-life. For a given peak/MIC value, E_max values for CPFX and SPFX were similar; E_max was greater than 2 log₁₀ CFU/ml of lung homogenate when the peak/MIC ratio was >7. On the other hand, CPFX was characterized by a shorter time to E_max and a steeper killing slope after a single injection. These differences are probably ascribable to the more favorable AUC>MIC value of CPFX. The %AUBC incorporates a time component and therefore the possibility of bacterial regrowth. The close correlation between %AUBC and tMIC therefore was expected. For a given peak/MIC value, %AUBC was consistently greater for SPFX, whose half-life is longer than that of CPFX. The strong influence of tMIC was consistent with the absence of any postantibiotic effect in our in vivo model; studies done in vitro found a modest postantibiotic effect that ranged from 2 to 6 h in duration according to the bacterial species (Chin and Neu, 1987). It has been reported that a peak/MIC value greater than 8 prevented selection of resistant P. aeruginosa strains (Blaser et al., 1987; Leggett et al., 1991; Drusano et al., 1993). Because resistance rarely emerges in vivo after a single injection, we looked for resistant strains in animals that failed to respond to various multiple-dose therapeutic regimens. None of these animals harbored any resistant strains.

A few animal studies have focused on the Pk/Pd parameters predictive of clinical cure in S. pneumoniae pulmonary infections (Azoulay-Dupuis et al., 1991b, 1992). With a dose fractionation approach, which was the only means to identify differences among variables, we found that the AUC/MIC was the best predictor of survival, and that an AUC/MIC of 160 or more was associated with a 100% clinical cure rate independently from the dosage schedule. In contrast, AUC/MIC breakpoints probably vary across experimental models. Lower AUC/MIC values have provided similar efficacy in models of less severe S. pneumoniae infection with SPFX and levofloxacin (Vesga and Craig, 1995; Vesga et al., 1995). Also with the dose fractionation approach, Drusano et al. (1993) found that the significance of AUC/MIC in a neutropenic rat model of Pseudomonas sepsis varied with the peak/MIC ratio; if it is high (10-20), peak/MIC ratio was linked to survivorship, but at lower doses, producing peak/MIC ratios lower than 10, the AUC/MIC was linked closely to outcome. In a study of a mouse protection model, Sullivan et al. (1993) found that 100% protection by CPFX was achieved when the peak/MIC ratio reached 10.6. From an analysis of data from several clinical trials of patients with nosocomial pneumonia caused by Gram-negative bacilli, Schentag and colleagues determined that an AUC/MIC of 125 was required to achieve a clinical cure and that AUC/MIC values in the 250 to 500 range were associated with increased in vivo bacterial killing (Forrest et al., 1993; Hyatt et al., 1994). In our study of an S. pneumoniae model, 100% clinical cure was achieved with an AUC/MIC value of 160, which suggests that Pk/Pd parameters of FQ activity in severe pneumonia may be quite similar for Gram-negative bacilli and S. pneumoniae. The MIC of the bacterium to be eradicated should be confronted with kinetic data. For instance, when using CPFX in a high dose of 400 mg every 8 h intravenously, an AUC/MIC value in excess of 125 (the lowest effective value) was achieved only when the MIC of the causative organism (S. pneumoniae, Staphylococcus aureus, P. aeruginosa) was less than 0.5 mg/l (Hyatt et al., 1994). Concerning interspecies differences, comparisons of in vitro killing curves for CPFX at different concentrations (1.55-25 mg/l), demonstrated a concentration-effect relationship not only for P. aeruginosa but also for S. pneumoniae and S. aureus (all these organisms have equivalent MICs of 0.4 mg/l) with only subtle differences in killing rates (Hyatt et al., 1994).

This experimental approach shows clearly that optimization of FQ therapy for an S. pneumoniae infection requires knowledge of both the MIC of the organism and the serum pharmacokinetic parameters of the drug with the dosing schedule used. Only the relationships between these two types of parameters can provide information on the antipneumococcal potential of a FQ. The Pk/Pd data provided by our study could explain why CPFX therapy is inconsistently effective in severe pneumococcal pneumonia (MIC₉₀  2 mg/l, low peak/MIC and AUC/MIC) (Cooper and Lawlor, 1989; Scully, 1993). SPFX has several characteristics suggestive of better antipneumococcal activity, including a lower MIC (MIC 90 = 0.25 mg/l) (Chin et al., 1991) and a longer half-life (20 h), although it also has a low serum peak (1.7 mg/l after 400 mg per os), it has a greater AUC_0-24 of about 35 mg·h/l and then an AUC/MIC of 140 (Montay et al., 1994). Owing to these better Pk/Pd parameters, SPFX has been proven effective in pneumococcal pneumonia without criteria for severe disease (Aubier et al., 1996).

To be sufficiently active against S. pneumoniae and useful for the empirical treatment of community-acquired pneumonia, a FQ would have to exhibit a higher AUC/MIC ratio (i.e., lower MICs and/or higher AUCs via a higher serum peak and/or a longer half-life) and at the same time maintain a satisfactory safety-side effects profile.