ABSTRACT
The antibacterial effect of amoxicillin-clavulanate in two formulations, pharmacokinetically enhanced 16:1 amoxicillin-clavulanate twice a day (b.i.d.) and standard 7:1 amoxicillin-clavulanate b.i.d., were studied in an in vitro pharmacokinetic model of infection. Five strains of Streptococcus pneumoniae and two of Haemophilus influenzae, all associated with raised MICs (2 to 8 mg/liter), were used. The antibacterial effect was measured over 24 h by the area under the bacterial kill curve (AUBKC) and the log change in viable count at 24 h (Δ24). A high 108 CFU/ml and low 106 CFU/ml initial inocula were used. Employing the Δ24 effect measure, the time above MIC (T>MIC) 50% maximum effect (EC50) for S. pneumoniae was in the range 21 to 28% with an 80% maximal response of 41 to 51%, for the AUBKC measure, the value was 26 to 39%, irrespective of inoculum. For H. influenzae, the T>MIC EC50 was 28 to 37%, and the 80% maximum response was 32 to 48% for the Δ24 measure and 20 to 48% for AUBKC. The maximum response occurred at a T>MIC of 50 to 60% for both species and inocula. The S. pneumoniae data were analyzed by analysis of variance to assess the effect of inoculum, formulation, and MIC on antibacterial effect. Standard and enhanced formulations had different effects depending on MIC, with the standard formulation less effective at higher amoxicillin-clavulanate MICs. This is explained by the greater T>MICs of the enhanced formulation. Although resistant to amoxicillin-clavulanate by conventional breakpoints, S. pneumoniae and H. influenzae strains for which MICs are 2 or 4 mg/liter may well respond to therapy with pharmacokinetically enhanced formulation amoxicillin-clavulanate.
The dominant pharmacodynamic index determining the effect of penicillins against Streptococcus pneumoniae is the time the drug concentration remains over the pathogen MIC (T>MIC) (15). The magnitude of the parameter for maximal effect in animal models varies from 20 to 75% depending on the definition of maximal effect used, the host immune status, the site of infection, and perhaps also the bacterial strain used (1, 7, 8, 14, 16). It has been suggested that the appropriate pharmacodynamic target for β-lactams is a T>MIC of 40 to 50% (8) and that in humans an efficacy of 80 to 85% in clinical trials of treatment of acute otitis media is achieved by a T>MIC of 40 to 50% (5). In one animal model, amoxicillin efficacy was demonstrated at a T>MIC of ≥35%. In others, enhanced bacterial activity occurred up to 80% (1, 7, 16). It follows, therefore, that dosing strategies which reliably produce an amoxicillin T>MIC of 40 to 50% against the target S. pneumoniae should be effective in in vitro and animal pharmacokinetic models of infection. T>MIC can be increased by use of larger drug doses, increased frequency of administration, and use of slow release formulations. Pharmacokinetically enhanced amoxicillin-clavulanate employs a larger dose of amoxicillin (1,125 mg of amoxicillin trihydrate and 125 mg of clavulanate potassium) as well as a sustained release formulation resulting in a T>MIC of almost >50% for S. pneumoniae strains for which amoxicillin MICs were 4 mg/liter, if mean healthy volunteer plasma concentration time profiles are used (10). The formulation is a bi-layer tablet with 437.5 mg of sustained release sodium amoxicillin in one layer plus 562.5 mg of immediate release amoxicillin trihydrate and 62.5 mg clavulanate potassium in the second layer. Two tablets are given for each dose (9). Data on the pharmacodynamics of amoxicillin-clavulanate against Haemophilus influenzae in animal and in vitro models is less extensive than that for S. pneumoniae (V. Berry, C. Singley, J. Satterfield, and G. Woodnutt, Abstr. 41st Intersci. Conf. Antimicrob. Agents Chemother., abstr. 988, 2001), and the effect of inoculum on bacterial clearance of both species is not studied at all.
In this series of experiments, we assessed the antibacterial effect of pharmacokinetically enhanced amoxicillin-clavulanate against S. pneumoniae strains for which amoxicillin MICs were >3 mg/liter and H. influenzae strains for which MICs were >2 mg/liter. A standard formulation was also used for comparison, and the impact of high 108 CFU/ml and low 106 CFU/ml inocula on the magnitude of the pharmacodynamic index measured for both S. pneumoniae and H. influenzae was assessed.
MATERIALS AND METHODS
Model.A New Brunswick (Hatfield, Hertfordshire, England) Bioflow 1000 in vitro model was used to simulate the total drug concentrations in serum associated with oral administration of amoxicillin-clavulanate at various doses. The apparatus, which has been described previously, consists of a single central chamber (volume, 360 ml) connected to a dosing chamber which is in turn attached to a reservoir containing broth. It is a single compartment open dilutional system. The central chamber is connected to a collecting vessel for overflow (12). The contents of the dosing chamber and central chamber were diluted with brain heart infusion broth (Oxoid, Basingstoke, England) by using a peristatic pump (Ismatec, Bennett & Co., Weston-super-Mare, England) at a flow rate of 200 ml/h. The temperature was maintained at 37°C, and the broth in the dosing and central chambers was agitated by a magnetic stirrer.
Media.Brain heart infusion broth (75%) was used in all experiments, and H. influenzae experiments were supplemented with hematin (16.7 mg/liter) and NAD (16.7 mg/liter). Nutrient agar plates supplemented with 5% whole horse blood (S. pneumoniae) and chocolated nutrient agar plates (H. influenzae) were used for determination of viable counts.
Strains.Six S. pneumoniae clinical strains for which amoxicillin-clavulanate MICs were >2 mg/liter were provided by V. Berry, GlaxoSmithKline Pharmaceuticals, Collegeville, Pa. The strains were given the following designations: 05010S, 16001S, 3005S, 47003S, 5003S, and 404053. The amoxicillin-clavulanate MICs determined using NCCLS methodology (13) were 2, 4, 4, 8, 8, and 8 mg/liter, respectively. Two H. influenzae clinical strains for which amoxicillin-clavulanate MICs were >2 mg/liter (designations 103-255 and 643-119) were also used. The MICs determined using NCCLS techniques were 2 and 4 mg/liter.
Antimicrobials.Amoxicillin and clavulanic acid were obtained from GlaxoSmithKline Pharmaceuticals (Uxbridge, Middlesex, United Kingdom). Stock solutions were prepared according to British Society of Antimicrobial Chemotherapy (BSAC) guidelines (3) and stored at −70°C.
MICs.MICs were also determined using BSAC methodology, except that sequential 1-mg/liter steps of amoxicillin-clavulanate were used rather than doubling dilutions. This allows for a more accurate determination of the MIC for subsequent analysis.
Pharmacokinetics and bacterial killing curves.The in vitro activity of amoxicillin-clavulanate at several dose simulations against the S. pneumoniae and H. influenzae strains described was tested in the model. For the amoxicillin-clavulanate pharmacokinetically enhanced formulation dose of 2,000 mg of amoxicillin plus 125 mg of sodium clavulanate (16:1 ratio), the target maximum concentration was 17.0 mg/liter at 1.5 h, and the half-life was 1.25 h. For the amoxicillin-clavulanate standard formulation dose of 875 mg of amoxicillin plus 125 mg of sodium clavulanate (7:1 ratio), the target maximum concentration was 12 mg/liter at 1.5 h, and the half-life was 1.25 h. In addition, the dose-ranging simulations used with S. pneumoniae strains 05003S and 16001S were 0.5, 0.25, and 0.33 times the maximum targets for the pharmacokinetically enhanced formulation, and for both H. influenzae strains, 2, 0.4, 0.17, and 0.1 times the maximum concentration targets for the pharmacokinetically enhanced formulation were used. This was to produce a range of T>MICs. Total drug concentrations were modeled for each drug, and in all simulations two doses were used. The model was run for 24 h.
For high inocula experiments, 200 μl of McFarland 4 standard suspension of the test strain was inoculated into the central culture chamber (volume, 360 ml) and grown for 18 h to give a final bacterial inoculum of 108 CFU/ml. For low inocula experiments, 720 ml of McFarland 4 standard was inoculated into the central culture chamber and left for 45 min before dosing to give an inoculum of 106 CFU/ml. Amoxicillin-clavulanate was then added to the dosing chamber, and single samples were taken from the central chamber over 24 h at 0, 1, 2, 3, 4, 6, 7, 12, and 24 h for assessment of viable bacterial counts. Approximately 0.5 ml of broth was withdrawn from the central chamber, and β-lactamase (supplied by T. Walsh, BCARE, Bristol, United Kingdom) was added to neutralize the amoxicillin and immediately placed on antibiotic-free media. The bacteria were quantified by using a Spiral plater (Don Whitley Spiral Systems, Shipley, West Yorkshire, England); the minimum detection level was 2 × 102 CFU/ml.
Additional aliquots were stored at −70°C for the measurement of amoxicillin-clavulanate with a bioassay (4). Samples were assayed with Bacillus subtilis NCTC 10400 as the indicator. All standards and samples were prepared as needed in the same concentration of brain heart infusion broth as was used in the simulations. Stock solutions were made on the day of use, and aliquots from the models were assayed within 7 days. The detection limit was 1 mg/liter and the coefficiency of variation was 6.49% for the 7:1 formulation and 6.64% for the 16:1 formulation. All pharmacokinetic simulation and killing curves were performed at least in triplicate.
Measurement of antibacterial effect.The antibacterial effect was assessed by calculating the log change in viable counts between time zero and 3 (Δ3), 6 (Δ6), 12 (Δ12), and 24 h (Δ24). The times taken for the inoculum to fall by 99 and 99.9% from its value at time zero are defined as T99 and T99.9. In addition, the area under the bacterial killing curve (AUBKC; measured as log CFU per milliliter · h) was calculated by using the log linear trapezoidal rule for the period 0 to 24 h (AUBKC24). The AUBKC was also calculated for the period 0 to 12 h (AUBKC12). For pharmacodynamic analysis, the T>MIC for the 24-h observation period was compared to AUBKC24 or Δ24 by using an inhibitory Emax model (where Emax represents the maximum effect) with Win Nonlin software (Scientific Consulting). For S. pneumoniae only, the effect of MIC, drug formulation, and initial bacterial inoculum on AUBKC24 was investigated by fitting a model to the data exploiting the factorial structure of the study design (analysis of variance).
RESULTS
MICs.The amoxicillin-clavulanate MICs for the S. pneumoniae strains were as follows: strain 5010S, 3 mg/liter; strain 16001S, 4 mg/liter; strain 30005S, 5 mg/liter; strain 47003S, 6 mg/liter; strain 5003S, 8 mg/liter; and strain 404053, 8 mg/liter. The MICs for the H. influenzae strains were as follows: strain 103-255, 2 mg/liter; and strain 643-119, 5 mg/liter.
Pharmacokinetic curves and pharmacodynamic parameters.There was good agreement between target and achieved amoxicillin-clavulanate concentrations in the model (percent error, 8.2% for the 7:1 formulation and 1.6% for the 16:1 formulation). The target T>MIC for S. pneumoniae for the pharmacokinetically enhanced formulations was 39 to 60% and for the standard formulation was 20 to 41%. Equivalent values for H. influenzae were 51 and 64% and 29 and 48%, respectively. Use of the dose-ranging simulation gave T>MICs of 0 to 60% for S. pneumoniae and 0 to 92% for H. influenzae, and the AUBKC24 and Δ24 associated with experiments producing these ranges of T>MIC were used in the inhibitory Emax models.
Antibacterial effects.The antibacterial effects of pharmacokinetically enhanced and standard formulation amoxicillin-clavulanate on S. pneumoniae at inoculums of 106 and 108 CFU/ml are shown in Tables 1 and 2. The antibacterial effects of the two formulations on H. influenzae are shown in Table 3. At an S. pneumoniae inoculum of 106 CFU/ml, the enhanced formulation of amoxicillin-clavulanate resulted in a >4-log drop in viable count in the model for five of six strains by 24 h; however, simulations of the standard formulation only reduced counts by >4 log for the two strains for which MICs were <4 mg/liter. At the high inoculum (108 CFU/ml), the enhanced formulation produced a >6-log reduction in count at 24 h for four of six strains and a >5-log reduction in five of six, but with one strain (404053; MIC, 8 mg/liter), grow-back was noted. With the standard formulation simulations, a >6-log reduction in counts occurred with the two strains for which the MIC was <4 mg/liter, and grow-back occurred in three of the four strains for which MICs were >5 mg/liter. The enhanced amoxicillin-clavulanate simulations against H. influenzae resulted in a >4-log reduction in count for strain 103-255 (MIC, 2 mg/liter) and a >3-log drop for strain 643-119 (MIC, 5 mg/liter). The standard formulation produced similar results, except with the more resistant strain at an inoculum of 108 CFU/ml, where the reduction in count at 24 h was 2.5 logs.
Antibacterial effect of pharmacokinetically enhanced amoxicillin-clavulanate (2,000 mg of amoxicillin) on S. pneumoniae
Antibacterial effect of standard formulation amoxicillin-clavulanate (875 mg of amoxicillin) on S. pneumoniae
Antibacterial effect of pharmacokinetically enhanced (2,000 mg of amoxicillin) and standard formulation (875 mg of amoxicillin) on H. influenzae
For S. pneumoniae and H. influenzae, both AUBKC24 and Δ24 could be related to T>MIC by using an inhibitory Emax model, where EC50 represents the T>MIC needed to obtain 50% of the Emax. For both antibacterial effect measures, the relationship could be adequately described by the model as judged by Akaike criteria, plots of fitted values, and the correlation between observed and predicted values (r) (Table 4). The EC50s for S. pneumoniae were in the range 21 to 28%, and the T>MIC for 80% maximal response was 41 to 51% for Δ24 and 26 to 34% for AUBKC24. The maximum response occurred with a T>MIC of 50 to 60% for both the high and low inoculum experiments. The EC50s for H. influenzae were similar, being in the range 28 to 37, and the T>MIC for 80% maximal response was 32 to 48% (Δ24) and 20 to 48% (AUBKC24).
Parameters fitted by Emax model of T > MIC to AUBKC24 and Δ24 for S. pneumoniae and H. influenzae at inocula of 106 and 108 CFU/ml (mean ± SD)
The S. pneumoniae data were analyzed exploiting the factorial structure by analysis of variance to assess the effect of inoculum, amoxicillin-clavulanate formulation, and MIC on AUBKC24. There was insufficient data for a similar analysis for H. influenzae. All three factors had affected ln AUBKC24, and in addition to the interaction between formulation and inoculum, there was also a significant interaction between formulation and MIC, suggesting that the effect of formulation on ln (AUBKC24) was different for different MICs. An interaction between inoculum and MIC was not indicated (P = 0.27). The assumptions underpinning the analysis were assessed graphically and were satisfied (data not shown). The least squares mean ln (AUBKC24) and standard errors estimated from the model for each formulation-inoculum-MIC combination is illustrated in Fig. 1. As can be seen in the figure, the mean ln (AUBKC24) for the two inocula are parallel to each other (no interaction with MIC), while the two formulations act differently at different MICs. The standard and enhanced formulations diverge as the MIC increases, with the standard formulation having a larger mean ln (AUBKC24) value, implying less bacterial clearance.
Relationship between antibacterial effect and MIC for amoxicillin-clavulanate pharmacologically enhanced formulation and standard formulation.
DISCUSSION
Determining the size of the pharmacodynamic index for optimal antibacterial effect is critical to an antibiotic pharmacodynamic and clinical evaluation, as it helps determine dosing regimens and also may be of use in defining breakpoints, and hence clinically relevant resistance.
A key difference between animal or in vitro pharmacokinetic models and human chest infection is that the bacterial load or inoculum is often lower in pharmacokinetic models than in human infection (6). It is therefore important to assess the impact of inoculum on the size of the pharmacodynamic index. In this study, we used two inocula, 106 CFU/ml for comparison to previous work in pharmacokinetic-pharmacodynamic models (1, 11) and a higher inoculum (108CFU/ml) to more closely mimic human infection. By using a neutropenic rat pneumonia model, an amoxicillin T>MIC of about 40% was associated with a maximum −4-log10 kill of S. pneumoniae (16). In a neutropenic mouse thigh model, the amoxicillin ± clavulanic acid T>MIC for a −3-log10 kill was about 60% (1). The maximum bacterial clearance in this model occurred at a T>MIC of 50 to 60% with an S. pneumoniae inoculum of 106 CFU/ml, which is in close agreement with these values. The initial S. pneumoniae inoculum used had little or no effect on the drug exposure needed to achieve the EC50, 80% of the Emax, or Emax.
Recent work using an in vitro pharmacokinetic model has suggested that a amoxicillin-clavulanate T>MIC of 73 to 79% was required for clearance of H. influenzae over a period of 24 h with an initial inoculum of about 106 CFU/ml (11). Our data imply that the T>MIC required for an antibacterial effect against H. influenzae is similar to that for S. pneumoniae and also not affected by bacterial inoculum, therefore, the target T>MIC reported here is somewhat lower than that suggested by Lowdin et al. (11). However, this may be accounted for by the different strains used in the two studies as well as different analytical approaches.
In the rat respiratory tract infection model, pharmacokinetically enhanced amoxicillin-clavulanate (16:1) is effective in reducing the bacterial counts of three S. pneumoniae strains for which MICs were 8 mg/liter (Berry et al., 41st ICAAC). Two of the three strains used in the study by Berry et al. were also used in this in vitro model. At a low inoculum similar to that used in the rat model, clearance was observed, but this did not occur at the higher inoculum. However, the pharmacokinetically enhanced formulation resulted in a 99.9% kill of the inoculum by 12 h for the three strains for which the MICs were ≥6 mg/liter; this did not occur in the standard dose simulations. Both rat and in vitro models were also in agreement as to the activity of pharmacokinetically enhanced amoxicillin-clavulanate against S. pneumoniae and H. influenzae strains for which amoxicillin-clavulanate MICs were <4 mg/liter (2). Standard formulation amoxicillin-clavulanate (7:1) was in general equipotent to the enhanced formulation against S. pneumoniae strains for which MICs were <5 mg/liter. However, for strains associated with higher MICs, pathogen clearance was reduced with the standard 7:1 formulation. This is in keeping with the lower T>MICs which occur with this dosing regimen against strains for which amoxicillin-clavulanate MICs are higher.
In conclusion, these data further illustrate that the definitions of resistance to antimicrobials in terms of clinical or antibacterial responses to therapy are not absolute. Most of the strains tested here would be regarded as amoxicillin-clavulanate-resistant using existing North American or European breakpoints (3, 13). However, when using an amoxicillin dose and formulation which maximizes T>MIC, strains for which amoxicillin-clavulanate MICs were up to 6 mg/liter would be expected to respond clinically to pharmacokinetically enhanced amoxicillin-clavulanate (16:1). It is therefore important to define clinical breakpoints in relationship to the drug formulation and dosing regimen used.
ACKNOWLEDGMENTS
We thank GlaxoSmithKline for financial support.
We thank A. White, D. Payne, and G. Woodnutt of GlaxoSmithKline for nonfinancial support.
FOOTNOTES
- Received 18 December 2003.
- Returned for modification 17 February 2004.
- Accepted 17 March 2004.
- Copyright © 2004 American Society for Microbiology
