INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The emergence of resistance in clinical pneumococcal isolates to the -lactam antibiotics, sulfonamides, tetracycline, quinolones, and macrolide antibiotics leaves the glycopeptides as the only class of antibiotics against which no resistance mechanism so far exists among pneumococci (4, 13, 18). This situation has led to a renewed interest in the pharmacodynamics of glycopeptides.

The glycopeptide antibiotics have been used in the treatment of infections in humans since the introduction of vancomycin in 1956 (6, 10). Beside vancomycin, other glycopeptides have been tested, but only teicoplanin, introduced in 1984, has been widely used until now for the treatment of humans (26, 27). In most parts of the world, the glycopeptides have been reserved as second-line antibiotics for infections caused by -lactam-resistant gram-positive bacteria. The use of glycopeptides has also been restricted to intravenous (i.v.) or intramuscular (i.m.) administration and installation in cavities, such as the peritoneum, cerebral ventricles, and the lumen of the gut (6, 26).

The correlations between doses and pharmacokinetic profiles of glycopeptides in humans are well described, but the correlation between the pharmacokinetic profile and the effect of treatment of patients has not been fully elucidated (1, 12, 16, 26, 28). Because of fear of toxic effects, it was previously recommended that serum drug concentrations be monitored regularly during glycopeptide therapy, at least as trough concentrations but eventually also as peak concentrations (1, 6, 7, 12, 16, 26, 28, 29). The effect of treatment has been correlated to dose and serum drug concentrations, but most studies have been retrospective and have focused on treatment failures and the correlation to serum drug concentrations (1, 6, 7, 12, 16, 26, 28, 29). It has not been possible to conclude which one or more of the pharmacokinetic or pharmacodynamic (PK/PD) parameters are the most important and best predictors for the effects of treatment in humans (1, 6, 12, 16, 26, 28).

In animal studies, it is possible to overcome many of the variables seen in human studies; however, infections in animals are not easily extrapolated to clinical situations. The effect of glycopeptide antibiotics has been studied in animal models (2, 3, 11, 19, 21-25), including the relationship to dosing regimens (3, 22, 25), but more often one glycopeptide regimen has been compared to regimens with other drugs (2, 11, 19, 21, 23, 24). Until now, it has not been possible to clearly identify which one or more of the PK/PD parameters are the most important in treatment with glycopeptides in animal studies.

The aim of this study was to clarify the importance of the different PK/PD parameters with respect to treatments with glycopeptides. We used the mouse peritonitis model with immunocompetent mice and with Staphylococcus aureus and Streptococcus pneumoniae as infective organisms. We used a wide spectrum of different treatment regimens with vancomycin and teicoplanin.

(Parts of this work have been presented at the following international congresses: 2nd International Symposium on Infection Models in Antimicrobial Chemotherapy, Reykjavík, Iceland, 17 to 20 July 1996, abstr. 14; 36th International Conference on Antimicrobial Agents and Chemotherapy, New Orleans, La., 15 to 18 September 1996, abstr. A39; 20th International Congress of Chemotherapy, Sydney, Australia, 29 June to 3 July 1997, abstr. 4076.)

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacteria and media. A strain of S. pneumoniae (I-1320, serotype 6B) and a strain of S. aureus (E-2371) were used throughout this study; these strains have been used in animal infection models before (9, 14, 15). The pneumococcal suspension for animal inoculation was prepared immediately before use by suspending colonies from a fresh overnight culture on 5% blood agar plates in sterile beef broth, adjusted to an optical density at 540 nm of 0.5, giving a concentration of approximately 10⁸ CFU/ml. The suspension was further diluted in beef broth and finally diluted 1:1 in 10% (wt/vol) mucin, giving a final concentration of 1 × 10⁶ to 5 × 10⁶ CFU/ml, and 5% (wt/vol) mucin in beef broth. The staphylococcal inoculum was prepared by suspending colonies from an overnight culture on 5% blood agar plates in Mueller-Hinton broth in tubes. The tubes were incubated for approximately 20 h at 35°C in a shake stand, followed by centrifugation at 2,000 × g for 10 min at 4°C. The bacterial pellet was resuspended twice in saline and centrifuged at 2,000 × g. The bacteria were dissolved in saline to an optical density at 540 nm of 1.0, giving a concentration of approximately 5 × 10⁸ CFU/ml. This suspension was centrifuged, half the supernatant was removed, and the suspension was diluted 1:1 in human plasma, giving a final concentration of 4 × 10⁸ to 5 × 10⁸ CFU/ml, and 50% human plasma in saline. For each experiment, the size of the inoculum was determined after 10-fold dilutions in saline were made; 20 µl was plated in spots in duplicate on 5% blood agar plates, with subsequent counting of colonies after incubation overnight at 35°C.

Antimicrobial agents. The antibiotics used were products for i.v. treatment of humans, vancomycin (Vancocin; Eli Lilly and Company, Indianapolis, Ind.) and teicoplanin (Targocid; Astra, Södertälje, Sweden). The purities for vancomycin and teicoplanin, as given by the manufacturers, were 92 and 82% (wt/wt), respectively; the drugs were used without any weight compensation for the difference in purity to mimic the clinical situations.

In vitro experiments. MICs of vancomycin and teicoplanin were determined by the microdilution method using Mueller-Hinton broth supplemented with 5% sheep blood (20) and using the E-test, a gradient diffusion method, according to the instructions of the manufacturer (AB Biodisk, Solne, Sweden). Protein binding of the glycopeptides in mouse serum was determined earlier using the ultrafiltration method (5, 15). In brief, after 2 h of incubation of the media (human serum, mouse serum, or beef broth) in ambient air at 35°C, the pH was adjusted to 7.0 to 7.5 by bubbling with CO₂. The glycopeptides at concentrations of 150 and 50 µg/ml were incubated for 2 h at 35°C in the media. The samples were divided into two parts; one part was centrifuged in tubes with filters with a cutoff of approximately 30 kDa (Centricon 30, no. 4208; Amicon, Beverly, Mass.) in a fixed-angle rotor at 3,000 × g for 20 min. The antibiotic concentrations in both parts were determined by the agar cup method (strain: nonhemolytic Streptococcus EB-68; Statens Serum Institut). A standard curve was produced using antibiotic concentrations dissolved in the same media (variation coefficients, <5%; for r² for standard curves, >0.90). Thus, the protein-bound fraction could be calculated.

In vivo models. In the mouse peritonitis model, outbred female ssc:CF1 mice were used. The models have been described earlier (9, 14). In the model for pneumococci, the mice weighed 28 to 32 g and were approximately 8 weeks old. Mice were inoculated intraperitoneally with 0.5 ml of the pneumococcus suspension (1 × 10⁶ to 5 × 10⁶ CFU/ml) and 5% (wt/vol) mucin in beef broth. Treatments of the pneumococcal infections were always started 1 h after challenge. In the model for staphylococci, originally designed to be performed as a foreign-body model (9), the mice weighed 35 to 40 g and were 10 to 12 weeks old. Mice were inoculated intraperitoneally with 1 ml of the staphylococcal suspension (4 × 10⁸ to 5 × 10⁸ CFU/ml) and 50% human plasma in saline. Treatments were started immediately thereafter. The mice were kept in cages at five to a cage, and they had free access to food and water.

In vivo investigations. Four different treatment trials were carried out.

(i) Trial 1. The single doses that protected 50% of the mice (ED₅₀s) for teicoplanin and vancomycin given subcutaneously for both the staphylococcus and the pneumococcus were calculated using the Hill equation. Doses from 0.1 to 100 mg/kg of body weight were used as follows: for the pneumococcus and vancomycin, 0.10, 0.25, 0.50, 0.67, 1.0, 2.5, 10.0, and 100 mg/kg; for the pneumococcus and teicoplanin, 0.10, 0.25, 0.39, 0.50, 0.60, 0.75, 1.0, 10.0, and 100 mg/kg; for the staphylococcus and vancomycin, 0.27, 1.4, 2.7, 4.7, 6.8, 13.5, 27.0, and 54.1 mg/kg; and for the staphylococcus and teicoplanin, 0.68, 1.4, 2.7, 5.4, 6.8, 8.1, 9.5, 13.5, 21.6, and 86.5 mg/kg. Each dose was given to mice in groups of at least 5 (range, 5 to 30). In each trial, a group of mice treated with saline was included as a control for the lethality of the infection. Mice were observed for 6 days, and deaths were recorded daily.

(ii) Trial 2. Treatments with total doses close to the ED₅₀s for single doses were given subcutaneously as one or two doses for both drugs and both strains to mice in groups of 24 or 30. When two doses were given, the intervals between treatments were 2 h for vancomycin and 12 h for teicoplanin; these intervals were 3.8 and 4.8 times the t_1/2 values, respectively. These intervals were chosen so that the second doses were given approximately at the time when the serum drug concentrations became lower than the MICs. Mice were observed daily for 6 days.

(iii) Trial 3. Multidosing regimens given subcutaneously in the pneumococcal peritonitis model were carried out for vancomycin and teicoplanin with a total treatment time of 48 h, followed by observation of the mice for 6 days. Mice were treated twice (with 24 h between treatments), with 4 doses (every 12 h), with 8 doses (every 6 h), with 12 doses (every 4 h), or with 24 doses (every 2 h). The total doses for both drugs were from 0.001 to 1,000 mg/kg in the different dosing regimens as follows: 0.001 and 0.01 mg/kg in 2, 4, and 8 doses; 0.1, 1.0, 5.0, and 10 mg/kg in 2, 4, 8, 12, and 24 doses; and 20, 100, and 1,000 mg/kg in 12 and 24 doses. There were at least 5 mice (range, 5 to 15) in each treatment group.

(iv) Trial 4. Because of differences in the serum protein binding of the drugs and therefore possible differences in penetration to the focus of infection in the peritoneal cavity, the following experiments were performed. Mice were challenged with pneumococci and treated in groups of five for 48 h with vancomycin or teicoplanin as a 12-dose regimen (every 4 h) or as a 24-dose regimen (every 2 h), for total doses of 1, 5, 10, and 20 mg/kg, as in the multidosing regimens (trial 3); however, here the treatments were given intraperitoneally.

Statistical methods. The Hill equation, a nonlinear regression method with a variable slope (GraphPad Prism; GraphPad Software Inc., San Diego, Calif.), was used to calculate the ED₅₀s: E = E_min + [(E_max  E_min)/(1 + 10^{(logED₅₀ logX)H}), where X is the concentration of the drug and H is the Hill slope of the sigmoid curve going from E_min to E_max (trials 1 and 2). This method was also used to analyze the correlation between C_max-free/MIC (as the X value) and survival in groups of mice after 6 days (as the E value) (trial 3). The correlation between T_>MIC-free, (X values) and survival in groups of mice after 6 days (E value) was analyzed using the E_max model: E = E_max [X/(ED₅₀ + X)] (trial 3).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

The serum drug elimination in mice and the protein binding in mouse serum were determined previously (15). For vancomycin a t_1/2 of 32 min was used, and for teicoplanin, a t_1/2 of 151 min was used. For protein binding in mouse serum, values of 25 and 90% were used for vancomycin and teicoplanin, respectively.

The MICs and single-dose ED₅₀s for the strains and the glycopeptides (trial 1) are shown in Table 1. No major difference was found between the MICs determined by the microdilution method and the E-test. The differences in ED₅₀s reflected the differences in MICs, as the staphylococcus was less susceptible to both glycopeptides than the pneumococcus, and higher doses were needed to achieve the same effect in the mouse model.

In Table 2, the results of the one- and two-dose experiments (trial 2) are shown. The different regimens with the same total dose were compared using the survival curves and the Mantel-Haenszels test. In all five trials, the one-dose group showed better survival than the two-dose group, a result which was statistically significant in four of the five trials. The PK/PD parameters corresponding to the dosing regimens also are listed in Table 2. The only significant correlations between the percent survival of the mice after 6 days and each of the PK/PD parameters were the C_max/MIC for S. aureus (Spearman rho, 0.97; P, 0.02) and the C_max-free/MIC for S. pneumoniae (Spearman rho, 0.81; P, 0.03) when data for vancomycin and teicoplanin trials were combined. When the data for both bacteria were combined, only the C_max/MIC for vancomycin was significantly correlated to survival (Spearman rho, 0.87; P, 0.03). No other significant correlations were found.

In Table 3, the ED₅₀s for the different multidosing regimens in the pneumococcal peritonitis model (trial 3) are given, as calculated using the Hill equation. The ED₅₀s increased with decreasing dosing intervals. For three of the values, it was not possible to calculate the 95% confidence interval (CI) because few values were located between 100% survival and 100% death of mice in groups receiving different total doses; however, in all calculations, the fitting of the values to the nonlinear regression curves was high (R² for all, 0.99) (Table 3).

The multidosing regimens for pneumococcal peritonitis in the 48-h trials (including the one-dose ED₅₀ trials) included 39 different vancomycin regimens and 40 different teicoplanin regimens (trials 1 and 3). For all of these regimens, different PK/PD parameters were calculated and correlated to the effect, i.e., survival of groups of mice after 6 days with different methods. The parameters analyzed are shown in Tables 4 and 5. Highly significant correlations were found for all parameters and effect, with a Spearman rho of >0.75 and P < 0.001 (Table 4). The parameters were also correlated to each other for all 79 regimens, especially between T_>MIC-free and AUC/MIC (Spearman rho, 0.87; P, <0.001), between C_max-free/MIC and AUC/MIC (Spearman rho, 0.77; P, <0.001), and between T_>MIC-free and C_max-free/MIC (Spearman rho, 0.79; P, <0.001).

Logistic regression analysis was performed to evaluate the importance of the parameters T_>MIC-free, AUC/MIC, and C_max-free/MIC. The effect was expressed as a binomial distribution of survival of mice in different treatment groups in a logit transformation [logit (E_BN)]. For both drugs, the log₁₀(C_max-free/MIC) was related to the logit(E_BN) in a threshold relationship, and the log₁₀(T_>MIC-free) and the log₁₀(AUC/MIC) were related to the logit(E_BN) in linear relationships. For vancomycin, the logit(E_BN) could be explained by the parameters log₁₀(C_max-free/MIC) and log₁₀(AUC/MIC), but log₁₀(T_>MIC-free) seemed to be without importance, expressed as arbitrary distances from the intercept, the gain in 2log(likelihood) (Table 5). For teicoplanin, the logit(E_BN) could be explained by the parameters log₁₀(C_max-free/MIC) and log₁₀(T_>MIC-free), but log₁₀(AUC/MIC) seemed to be without importance, again expressed as arbitrary distances from the intercept, the gain in 2log(likelihood) (Table 5).

Figure 1 shows the nonlinear regression curves for effect (survival after 6 days) as a function of increasing T_>MIC-free and log₁₀(C_max-free/MIC) for 48-h vancomycin and teicoplanin treatment regimens. For these four evaluations, statistically significant correlations were found for T_>MIC-free and effect for vancomycin and teicoplanin. The E_max model showed R² values of 0.65 and 0.82, respectively; the Hill equation correlated the data from log₁₀(C_max-free/MIC) and effect for vancomycin and teicoplanin, with R² values of 0.76 and 0.96, respectively. The change from no effect to full effect took place within approximately one log₁₀ increase in C_max-free/MIC for both teicoplanin and vancomycin. The thresholds for these shifts in C_max-free/MIC were five to six for vancomycin and two to three for teicoplanin. Because these thresholds for shifts in effect were different for the two drugs, logistic regression was not performed for all data together for the two drugs. When the different regimens with the same total doses were analyzed, a highly significant difference in survival of mice over the 6 days of observation was found for both drugs at a total dose of 1 mg/kg, showing the one-dose regimens to be the best (log rank test: P, <0.001). No statistically significant differences between regimens with the same total doses could be shown for doses other than a total dose of 1 mg/kg.

View larger version (20K): [in this window] [in a new window]   FIG. 1.   Correlations between pharmacodynamic parameters and effect in the multidosing experiment with the pneumococcus (trial 3), shown as nonlinear regression curves. Effect is the survival of groups of mice after 6 days. (a) For T>MIC-free in hours, we used the Emax model. (b) For the Cmax-free/MIC ratio, we used the Hill equation. Vancomycin data are given as solid lines and circles; teicoplanin data are given as broken lines and crosses. The goodness of fit of values for the curves were as follows: for effect and vancomycin, T>MIC-free R2 was 0.65 and Cmax-free/MIC R2 was 0.76; for effect and teicoplanin, T>MIC-free R2 was 0.82 and Cmax-free/MIC R2 was 0.96.

To illustrate the different regimens, Figure 2 shows the extrapolated serum drug values, both as the total and as the free serum drug concentrations for vancomycin and teicoplanin, at the same total dose of 1 mg/kg. The MICs of both drugs are indicated.

View larger version (23K): [in this window] [in a new window]   FIG. 2.   Examples of serum drug profiles for different regimens with the same total dose of 1 mg/kg over 48 h for vancomycin (a) and teicoplanin (b). Values were calculated by extrapolation from PK/PD data. The first 12 h of the regimens are shown. The left y axes are the total serum drug concentrations, and the right y axes are the free drug concentrations in serum. The t1/2 values for vancomycin and teicoplanin were 32 and 151 min, respectively; the levels of protein binding of vancomycin and teicoplanin in mouse serum were 90 and 25%, respectively. The MICs for the pneumococcus are indicated.

To examine whether one or more PK/PD parameters were constant for the ED₅₀s in the different regimens, the T_>MIC-free, AUC/MIC, and C_max-free/MIC were studied for the ratios of ED₅₀/number of doses. The values for T_>MIC-free and AUC/MIC were multiplied by number of doses, but no statistically significant correlations could be found.

A comparison of the effects of the same regimens given as subcutaneous or intraperitoneal injections of vancomycin and teicoplanin, according to the total dose given (trials 3 and 4), did reveal a difference for vancomycin. For vancomycin, all eight regimens of intraperitoneally administered treatments were better than or equal to subcutaneously administered treatments (sign test: P, 0.04). For teicoplanin, no statistically significant differences were found for the eight regimens (sign test: P, 0.72) (Table 6). The overall rates of survival for vancomycin given intraperitoneally versus subcutaneously were 73 versus 49%, respectively; corresponding rates for teicoplanin were 75 versus 56% (Table 6).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Among animal studies, two studies in particular have previously provided hints of the importance of the various PK/PD parameters (3, 22). In a study of teicoplanin in a rabbit endocarditis model with S. aureus, doses were given as i.v. or i.m. treatments (3). The same dose given i.v. and i.m. showed differences in serum drug profile; the i.v. regimen resulted in a higher peak and more rapid elimination of the drug than in the i.m. regimen. A statistically better effect, i.e., a smaller number of CFU in the valve vegetations, was found for the i.m. regimen. A nonstatistically higher drug concentration was found in valves from animals in the i.m. regimen than in those from animals in the i.v. regimen after 24 h of therapy. The mean of the free-drug concentration was not above the MIC in the i.v. regimen, as it was in the i.m. regimen. The authors argue for regimens providing trough free-drug concentrations above the MIC (3). That is, the authors argue for regimens where the serum free-drug concentrations never go below the MIC during the treatment period.

However, in another study, comparing i.v. and i.m. treatments for Staphylococcus haemolyticus in the thigh model of normal or neutropenic mice with vancomycin or teicoplanin, the authors did not find any difference between i.v. and i.m. treatments but did find differences between treatments of normal and neutropenic mice (25). The ED₅₀s of both vancomycin and teicoplanin in neutropenic mice were approximately four times higher than those in normal mice (25). In a study of the effect of vancomycin and teicoplanin against S. aureus in the thigh model with neutropenic mice, the effect in vivo correlated with the effect predicted from in vitro results (22). The killing effect of the drugs in vitro as the difference in CFU in the thighs between treated and control mice was correlated to the effect seen in time-kill curves, according to concentration, and with respect to serum protein binding. The killing effect in vivo could be predicted for vancomycin and, to a certain degree, also for teicoplanin; i.e., a maximum killing effect was found for the time when the serum drug concentration was above the MIC, and a concentration-dependent killing effect was found during the time when the drug concentration was below the MIC (22). Both of these previous studies (3, 22) focus on T_>MIC (or T_>MIC-free) as the most important parameter for effect.

In the one- and two-dose trials with staphylococci and pneumococci in the present study, it seemed obvious that the glycopeptides should be given in few high doses and that C_max was the most important parameter for achieving effect. In the one- and two-dose regimens with teicoplanin against staphylococci, significant effects were observed; 15 of 24 mice and 11 of 24 mice, respectively, survived, although the free-drug concentration did not reach the MIC and the T_>MIC-free was zero. Since the ED₅₀s increased with number of doses given in the multidosing experiment, the results seemed to confirm the importance of C_max (Table 3). The significant log-rank test for both drugs showing a better effect with fewer doses than with many doses in the different regimens with total doses of 1 mg/kg in the multidosing trials is also in concordance with C_max being the most important parameter for effect.

All the PK/PD parameters were significantly correlated with effect by the Spearman rank test for the multidosing trials, and T_>MIC-free and C_max-free/MIC could be correlated with effect in different models, the E_max model and the Hill equation, respectively. Logistic regression did not show C_max-free/MIC alone to be the best predictor for effect but, in combination with AUC/MIC for vancomycin and with T_>MIC-free for teicoplanin, it explained the survival of mice in the multidosing trials.

The effects observed in the one- or two-dose trial (trial 2), where the free concentrations of teicoplanin in serum were below the MICs (Table 2) possibly can be explained by the bactericidal effect of glycopeptides at concentrations lower than the MIC; teicoplanin has a bactericidal effect against Staphylococcus at concentrations one-half the MIC (data not shown and reference 22). Another important factor could be that the mice used were immunocompetent and leukocytes might enhance the killing of bacteria by concentrations below the MIC.

In 1950, Eagle et al. performed a study of different dosing regimens for penicillin against intraperitoneally inoculated pneumococci in mice, where the 50% protective dose diminished with increasing number of doses given (8). This result has also been shown for other combinations of -lactam antibiotics and bacteria in animal models (17). It is well known that T_>MIC is increased for a certain total dose if the dose is divided into a number of smaller doses; therefore, a decrease in the 50% protective dose with increasing number of doses given is indirect proof that T_>MIC is the most important parameter for effect. This connection between further division of doses and increasing T_>MIC is abolished when the C_max/MIC ratio is below four. The C_max/MIC values achieved at the ED₅₀s for -lactams are usually essentially higher than four (8, 14, 17); therefore, T_>MIC increases and C_max/MIC decreases with number of doses. The effect of the glycopeptides seems to be achieved at relatively low drug concentrations in comparison with the effect of the -lactams; a division of doses when the C_max/MIC ratio is below four will result in decreases in the parameters C_max/MIC and T_>MIC or zero values for T_>MIC.

Figure 2 shows examples of how the regimens with total doses of 1 mg/kg are related to the MICs in this model of pneumococci and glycopeptides. The example of 1 mg/kg was chosen because, in the multidosing trial (trial 3), the regimens had ED₅₀s of below 1 mg/kg for both vancomycin and teicoplanin when given in one or two doses; however, when doses were given in 4, 8, 12, or 24 injections, the ED₅₀s were above 1 mg/kg. Figure 2 shows why the 50% protective doses (ED₅₀s in our experiments) do not decrease with the number of doses given, whatever the parameter of importance is, T_>MIC or C_max. The ratios between C_max-free and MIC are less than four, so that a division of the doses results in both very short or no T_>MIC values and lower C_max values.

Because of the difference in the serum protein binding properties of the two drugs, penetration to the focus in our model, the peritoneal cavity, could be different; therefore, the effect of the two drugs could be difficult to compare. It could have been a disadvantage for teicoplanin, with a relatively high level of protein binding in mouse serum, to be compared with vancomycin in a mouse model with a peritoneal focus for sepsis. The peritoneal fluid following an intraperitoneal challenge with bacteria is an exudate and therefore is presumed to contain at least as much protein as serum. Instead of trying to determine the antibiotic concentration and the protein content in the peritoneum, we treated the mice with either subcutaneous or intraperitoneal injections in equivalent regimens (Table 5). Only for vancomycin was a slightly better effect achieved with the intraperitoneal dosing. No difference was found for teicoplanin, a result which could be explained by higher inactivation of the drug because of the high protein content in the peritoneal fluid. We concluded that this difference did not have any major influence on our results, and we focused on the importance of working with the free fractions of the drugs.

In all these calculations, simple extrapolations from relatively few pharmacokinetic data are done, and the fact that the elimination of drugs is highly dependent on the illness of the animals is ignored. We start to treat the animals 1 h after challenge, when the mice have bacteremia, approximately 10⁴ CFU/ml (14), and the serum elimination of drugs is not yet impaired. If the animals become severely ill, the elimination rates used in the calculations are probably not correct; the T_>MIC values are longer and the C_max values are higher than the values that we use in the calculations. It would be ideal to use individualized pharmacokinetics, but that is very complicated.

From these observations, it seems possible to conclude that C_max-free is of major importance, but this parameter alone cannot explain the effects achieved with either of the two glycopeptides studied here, vancomycin and teicoplanin.