Bactericidal Effect and Pharmacodynamics of Cethromycin (ABT-773) in a Murine Pneumococcal Pneumonia Model

Abstract

Cethromycin (ABT-773), a new ketolide, possesses potent in vitro activity against Streptococcus pneumoniae. The objective of this study was to investigate the in vivo bactericidal activity of cethromycin against macrolide-susceptible and -resistant S. pneumoniae in a murine pneumonia model and to describe the pharmacodynamic (PD) profile of cethromycin. Eight (two macrolide susceptible, six macrolide resistant) clinical isolates of S. pneumoniae were investigated. Cyclophosphamide administration rendered ICR mice transiently neutropenic prior to intratracheal inoculation with 0.05 ml of an S. pneumoniae suspension containing 10⁷ to 10⁸ CFU/ml. Oral cethromycin was initiated 12 to 14 h postinoculation over a dosage range of 0.1 to 800 mg/kg of body weight/day. Lungs from seven to eight mice per treatment and control groups were collected at 0 and 24 h posttherapy to assess bacterial density. The cumulative mortality (n = 12 to 13) was assessed at 120 h (end of therapy) and at 192 h (3 days posttherapy). Recovery of pneumococci from the lungs of infected animals prior to the initiation of therapy ranged from 4.6 to 7.2 log₁₀ CFU. Growth in untreated control animals over a 24-h study period increased 0.3 to 2.7 log₁₀ CFU. Cethromycin demonstrated a substantial bactericidal effect, regardless of macrolide susceptibility. Correlation between changes in bacterial density (24 h) and survival over both 120 and 192 h were statistically significant. All three PD parameters demonstrated a significant correlation with changes in log₁₀ CFU/lung (Spearman's correlation coefficient, P < 0.001); however, the goodness of fit as assessed with the maximum effect (E_max) model revealed that the maximum concentration of free drug in serum (C_{max free})/MIC and the area under the free drug concentration-time curve (AUC_free)/MIC best explained the relationship between drug exposure and reductions in viable bacterial counts. These data reveal that an approximate cethromycin AUC_free/MIC of 50 or C_{max free}/MIC of 1 results in bacteriostatic effects, while higher values (twofold) maximize survival.

Ketolides are a new chemical subclass of 14-member-ring macrolides with a 3-keto functional group, which is semisynthesized by oxidation of the 3-OH group. Through this chemical structure transformation, ketolides achieve a high degree of stability in acidic environments and potent in vitro bactericidal effects against gram-positive cocci, including macrolide-susceptible and -resistant Streptococcus pneumoniae (2). Recently, interest in ketolides has been prompted by the rising incidence of macrolide-resistant S. pneumoniae, which is estimated to be 15 to 20% (5, 13). The mechanism of resistance to macrolides is mainly due to expression of the mef(A) and/or erm(B) genes (12). Fortunately, ketolides have in vitro bactericidal activity against S. pneumoniae with both the mef(A) and erm(B) genes (M. M. Neuhauser, J. L. Prause, R. Jung, N. Boyea, J. M. Hackleman, L. H. Danziger, and S. L. Pendland, Abstr. 39th Intersci. Conf. Antimicrob. Agents Chemother., abstr. 2139, 1999).

Cethromycin (ABT-773) is an investigational ketolide with an in vitro bactericidal effect against macrolide-susceptible and -resistant S. pneumoniae (Neuhauser et al., Abstr. 39th Intersci. Conf. Antimicrob. Agents Chemother., 1999; G. G. Zhanel, J. A. Karlowsky, A. Wierzbowski, and D. J. Hoban, Abstr. 40th Intersci. Conf. Antimicrob. Agents Chemother., abstr. 2169, 2000; B. M. Willey, L. Trpeski, S. Pong-Porter, J. de Azavedo, K. Weiss, R. Davidson, A. McGeer, D. E. Low, and Canadian Bacterial Surveillance Network, Abstr. 40th Intersci. Conf. Antimicrob. Agents Chemother., abstr. 2161, 2000). However, the in vivo bactericidal effect has not been established in the pneumonia model. Moreover, the pharmacodynamics (PD) of cethromycin in the pneumonia model has not been studied. Before an optimal pneumonia dose regimen for cethromycin can be established, it is of critical importance to determine the optimal PD profile for the maximization of antibiotic efficacy in such a model.

The first objective of this study is to investigate the in vivo bactericidal activity of cethromycin against macrolide-susceptible and -resistant S. pneumoniae. The second objective is to describe the pharmacokinetic (PK) and PD profiles of cethromycin against macrolide-susceptible and -resistant S. pneumoniae in a murine pneumonia model.

MATERIALS AND METHODS

Antibiotics and microorganisms.

Laboratory standard powders of cethromycin (Abbott Laboratories, Chicago, Ill.), azithromycin (Pfizer, Groton, Conn.), and clindamycin (Sigma Chemical Co., St. Louis, Mo.) were used for all in vitro and in vivo testing.

Eight clinical isolates of S. pneumoniae were utilized for all in vitro and in vivo experiments.

Clinical isolates included two macrolide-susceptible isolates, four mef(A) isolates, one mef(A) erm(B) isolate, and one erm(B) isolate (Table 1). All strains were maintained in skim milk medium (Becton Dickinson, Cockeysville, Md.) at −80°C and subcultured twice onto Trypticase soy agar with 5% sheep blood (Becton Dickinson) before use in all experiments.

TABLE 1.

Median MICs for the S. pneumoniae test isolates

Isolate no.a	MIC (μg/ml)			Phenotype or genotype
	Azithromycin	Clindamycin	Cethromycin
21	0.06	0.03	0.008	Non-mef(A) or erm(B)
75	0.19	0.125	0.008	Non-mef(A) or erm(B)
87	4	0.09	0.125	mef(A)
100	2	0.06	0.03	mef(A)b
109	16	0.06	0.03	mef(A)
102	8	0.06	0.06	mef(A)b
95	>128	>16	0.03	erm(B)
106	>128	>16	0.5	mef(A) and erm(B)b

^aInternal strain designation.

^bGenotype (according to PCR testing results provided by Abbott Laboratories).

Susceptibility testing.

The MICs were determined by a minimum of triplicate independent tests by the broth microdilution method according to the National Committee for Clinical Laboratory Standards (10).

Lung infection model.

Specific-pathogen-free outbred female ICR mice (body weight, approximately 25 g) were obtained from Harlan Sprague Dawley, Inc. (Indianapolis, Ind.). Animals were maintained and utilized in accordance with National Research Council recommendations (9) and were provided food and water ad libitum. The general study design for the lung infection model is summarized in Fig. 1. Briefly, cyclophosphamide (Cytoxan; Bristol-Myers Squibb, Princeton, N.J.), given as intraperitoneal injections (150 mg/kg of body weight/0.2 ml on days −4 and −1 before inoculation) rendered ICR mice transiently neutropenic (1, 8). A suspension of S. pneumoniae cells was prepared from the second subculture that had been incubated for less than 20 h and was adjusted to a 3.0 McFarland turbidity standard in a 5% dextrose saline solution, approximating 10⁸ CFU/ml. The bacterial density of each run was confirmed by serial dilution and culture of an aliquot from each inoculum suspension for quality control (QC).

FIG. 1.

Cethromycin murine pneumonia model schematic. IP, intraperitoneally; Tx, treatment.

Isoflurane (2% [vol/vol] in 100% oxygen carrier) was used to induce a semianesthetized state. Intratracheal inoculation was implemented to induce pneumonia by instilling a 0.05-ml bacterial suspension into the mouth and completely blocking the nares of the animal, thus resulting in bacterial inhalation through the mouth to the lungs. After allowing mice to fully recover from anesthesia in an oxygen-enriched chamber, they were randomized into control, bacterial density, or survival groups. Oral administration of cethromycin was initiated via oral gavage 12 to 14 h postinoculation (0 h dosing time). Nontreated control animals orally received drug-free vehicle (10% 95% ethanol and 90% 0.1 M [pH 6.5] phosphate buffer solution, 80:20 KH₂PO₄-Na₂HPO₄) in the same volume and on the same schedule as cethromycin. A preliminary study in which the bacterial densities of the lungs from nontreated control animals were measured at 6, 12, and 24 h postinoculation showed an increasing bacterial density starting at 12 h postinoculation, at which time dosing was initiated.

Murine PK study.

Mice were prepared as described in the section on the lung infection model. Concentrations of cethromycin oral solution in serum were determined after single doses of 25, 50, 100, or 200 mg/kg of body weight. The vehicle for dosing solutions was 10% 95% ethanol and 90% (vol/vol) 0.1 M (pH 6.5) buffer solution. Blood samples from 3 to 10 mice (mean of 6 mice) per time point were collected at 0.25, 0.5, 1, 2, 4, 6, 8, 12, 24, 48, and 72 h post single dose via cardiac puncture after euthanasia through CO₂ inhalation and cervical dislocation. After clotting, each sample was centrifuged for 10 min at 6,900 × g. The serum was collected and frozen at −80°C until analysis. Serum drug concentrations were determined as described below in the drug concentration assay section. Protein binding data were supplied by Abbott Laboratories (Drug Metabolism Department), where radiolabeled cethromycin was determined in ICR female mice by using a dialysis system and measured by radioassay. These data were used to calculate serum free drug concentrations.

Efficacy as assessed by bacterial density in lung.

Thirty-six groups of 21 to 64 mice per test isolate were prepared as described in the section on the murine pneumonia model and treated with various oral dose regimens of cethromycin ranging from 0.1 to 800 mg/kg/day administered in one to four doses per day, encompassing regimens of 0.1 mg/kg once a day to 200 mg/kg four times a day. Ten groups of nontreated control animals (5 to 10 mice per test isolate) received oral drug-free vehicle in the same volume and on the same schedule as cethromycin. Just prior to dosing (0 h), lung tissue from control mice (five per group) was harvested and cultured. After 24 h postinitiation of cethromycin dosing, lung tissue was cultured from nontreated mice (five mice per group) and treated mice (seven to eight mice per group). Following euthanasia through CO₂ inhalation and cervical dislocation, the pleural cavity was opened to aseptically harvest all five lobes of the lung. They were rinsed with sterile water, blotted on a surgical sterile towel, and placed in a sterile tube containing 0.5 ml of normal saline. A tissue homogenizer (PRO Scientific, Monroe, Conn.) was used to homogenize lung tissues. Tissue suspensions were diluted to appropriate 10-fold dilutions, plated onto Trypticase soy agar with 5% sheep blood, and incubated at 35°C for approximately 24 h in 5% CO₂ for CFU determinations. The detection limit of the bacterial density was 50 CFU/lung.

When the antibiotic carryover was suspected, the 0.1-ml aliquot of homogenized tissue suspension was repeatedly washed by mixing tissue suspension with 10 ml of Mueller-Hinton broth by using a VSM-3 mixer (Shelton Scientific Manufacturing, Inc., Shelton, Conn.), followed by centrifugation of broth cultures at 2,540 × g and 25°C for 25 min and removal of the supernatant. The 0.05- and 0.01-ml aliquots of the washed cultures were plated onto Trypticase soy agar with 5% sheep blood and incubated at 35°C for approximately 24 h in 5% CO₂ for CFU determinations.

Efficacy (change in bacterial density) was calculated by subtracting the mean log₁₀ CFU per lung of the control mice (just prior to cethromycin administration) from the log₁₀ CFU per lung of cethromycin-treated or untreated control mice at the end of therapy (24 h). The mean value of changes in bacterial density for each group was used in analyses.

Survival analysis.

Twenty-five groups of 36 to 48 mice per test isolate (12 to 13 mice per drug regimen) were prepared as described in the section describing the murine pneumonia model and treated with various oral dose regimens of cethromycin ranging from 3 to 180 mg/kg/day administered in one to four doses per day for 120 h. Ten animals served as a nontreated control for each isolate. The cumulative percent survival (%S) was assessed at 120 h (end of therapy) and at 192 h (3 days post end of therapy). The contemporary standards of mortality observation were implemented for our methodology in order to minimize the pain and suffering of animals (7). Specifically, animals were monitored three times daily by researchers who had been trained to recognize the signs of illness and abnormal behavior, such as substantial alterations in posture (e.g., abnormal posture or head tucked into abdomen), coat, exudate around eyes and/or nose, and breathing or movement. Animals that were judged to display signs of severe illness and abnormal behavior were removed from the group housing and were euthanized. Death caused by either the natural infection process or euthanasia was considered as one end point for experimental and statistical purposes.

Drug concentration assay.

The murine concentrations of cethromycin in serum were analyzed by a validated high-performance liquid chromatography (HPLC) assay with fluorescence detection (excitation, 324 nm; emission, 364 nm). The system consisted of the following components: a model 515 HPLC pump (Waters, Milford, Mass.), a WISP 717 Plus autosampler (Waters, Milford, Mass), an Alltech Nucleosil 100 C₁₈ column (10-μm pore size, 4.6 mm by 25 cm; Phenomenex Co.), a model 980 fluorescence detector (Applied Biosystems, Foster City, Calif.), and an EZChrome Elite chromatography data system (version 2.2; Scientific Software, San Ramon, Calif.). The mobile phase was a mixture of aqueous buffers (0.01 M tetramethylammonium hydroxide in 0.05 M KH₂PO₄ [pH 6.0]) and acetonitrile-methanol (100/90/10 [vol/vol/vol]). The flow rate was 1.0 ml/min under ambient temperature.

Cethromycin standard samples and QC samples (0.1 and 1.6 μg/ml) were prepared by spiking control female ICR mouse serum with an appropriate volume of stock solution. Abbott-267257 (lot no. A-267257; Abbott Laboratories, Chicago, Ill.) was used as an internal standard. Aliquots of 200 μl of standard, QC, or unknown mouse serum samples were pipetted into labeled 1.7-ml snap-cap tubes, followed by addition of 100 μl of 0.5 M sodium carbonate and 50 μl of internal standard. Each sample was then extracted by vortexing with 1.2 ml of hexane-ethyl acetate (1:1 [vol/vol]) for 30 s, followed by centrifugation at 3,600 × g for 6 min. The supernatants were transferred into labeled clean tubes and evaporated to dryness under a stream of nitrogen at 40°C. The residuals were then mixed with 100 μl of a reconstituted solution (hydrochloric acid added to 2:1 [vol/vol] 0.05 M KH₂PO₄ [pH 6.0]-acetonitrile to obtain 0.01 N HCl). Then 1.0 ml of hexane was added to each tube, and the mixture was vortexed for 30 s and centrifuged at 3,600 × g for 6 min. The organic layer was discarded, aqueous solution was transferred to micro vials, and a typical injection of 30 μl was injected onto the HPLC system. The standard curve ranged from 0.05 to 2.0 μg/ml with good linearity (r > 0.999). The coefficients of variation (CVs) for interday assay were 4.99 and 2.15% for low- and high-check samples, respectively. The CVs for intraday assay were 2.69 and 2.01% for low- and high-check samples, respectively.

Data analysis.

Animal PK data were analyzed with the double-precision NONMEM program (version IV, level 1.0; NONMEM Project Group, San Francisco, Calif.) to obtain population PK parameter estimates. The PK profile of each dosage regimen was calculated based on the NONMEM results obtained with the WinNonlin 3.0 program (Pharsight Corporation, Mountain View, Calif.) as described elsewhere (14). Protein binding data (Abbott Laboratories) were analyzed with the WinNonlin program, and the results were then used to calculate serum free drug concentrations.

Spearman's correlation coefficient test evaluated the association between antibiotic efficacy (mortality and changes in CFU) of cethromycin and three PD parameters: percentage of time that serum drug levels are above the MIC (%T > MIC) of free drug (MIC_free), maximum concentration of drug in serum (C_max)/MIC_free, and area under the concentration-time curve (AUC)/MIC_free. Likewise, the relationships between changes in CFU and mortality (at both the 120- and 192-h observation periods) were compared by the Spearman's correlation coefficient test. A P value of <0.01 was considered significant. The sigmoid maximum effect (E_max) dose-effect model was used to characterize the relationship between antibiotic efficacy and three PD parameters.

RESULTS

Susceptibility of tested isolates.

Median MICs of cethromycin, azithromycin, and clindamycin are listed in Table 1. The median MIC for each isolate was used to calculate the PD profile.

Murine PK study.

Cethromycin demonstrated nonlinear PK characteristics. NONMEM analysis indicated that a two-compartment open model with Michalis-Menten elimination kinetics characterized the PK profile of cethromycin (14). Based on the NONMEM results, serum drug concentration-time curves were simulated for each dosage regimen. The mean observed serum concentration-time curves and estimated PK parameters of the four PK dosing regimens are displayed in Fig. 2 and Table 2, respectively.

FIG. 2.

Mean serum drug concentration-time profiles of cethromycin following oral administration of 25 (⧫), 50 (▪), 100 (▴), or 200 (•) mg/kg in mice with pneumococcal lung infection. Reprinted with permission from reference 14.

TABLE 2.

PK parameters^a of cethromycin in mice after oral single-dose administration

Cethromycin dose (mg/kg)	Cmax (μg/ml)	Tmax (h)b	AUC (h · μg/ml)	Half-life (h)	Protein binding (%)c
12.5	0.86	2.0	14.4	5.4	94.0
25	1.77	2.2	40.7	6.7	93.9
50	3.63	2.6	130.3	15.9	93.7
100	7.40	2.8	474.8	38.1	93.3
200	14.59	2.2	1,520.0	67.7	92.5

^aData were calculated based on the population PK results obtained from NONMEM (14).

^bT_max, time to maximum concentration of drug in serum (C_max).

^cData provided by Abbott Laboratories.

Efficacy as assessed by bacterial density.

Recovery of pneumococci from the lungs of infected animals immediately prior to cethromycin therapy (0 h) ranged from 4.62 to 7.19 log₁₀ CFU/lung (mean ± standard deviation [SD], 6.00 ± 0.76 log₁₀ CFU/lung). Growth in untreated control animals over a 24-h study period increased 0.30 to 2.65 log₁₀ CFU/lung (mean ± SD, 1.17 ± 0.81 log₁₀ CFU/lung). Cethromycin reduced bacterial density by as much as 3.92 log₁₀ CFU/lung when a total daily dose of 100 mg was administered.

The sigmoid E_max model-fitted graphs of changes in bacterial density at 24 h versus %T > MIC_free, maximum concentration of free drug in serum (C_{max free})/MIC, and AUC_free/MIC are shown in Fig. 3, 4, and 5, respectively. All three PD parameters demonstrated a significant correlation with the change in log₁₀ CFU per lung (Spearman's correlation coefficient, P < 0.001); however, the goodness of fit as assessed with the E_max model revealed that the C_{max free}/MIC and area under the free drug concentration-time curve (AUC_free)/MIC best explained (r² = 0.81) the relationship between drug exposure and reductions in viable bacterial counts. These data reveal that an approximate AUC_free/MIC of 50 or C_{max free}/MIC of 1 resulted in bacteriostatic effects, whereas higher values (AUC_free/MIC of 1,000 or C_{max free}/MIC of 100) were required to achieve maximal bactericidal activity in this neutropenic pneumonia model.

FIG. 3.

%T > MIC of unbound cethromycin versus changes in log CFU plotted for individual bacterial isolates (n = 46, with each point representing seven to eight mice). SP, S. pneumoniae.

FIG. 4.

C_maxfree/MIC of unbound cethromycin versus changes in log CFU plotted for individual bacterial isolates (n = 46, with each point representing seven to eight mice). SP, S. pneumoniae.

FIG. 5.

AUC/MIC of unbound cethromycin versus changes in log CFU plotted for individual bacterial isolates (n = 46, with each point representing seven to eight mice). SP, S. pneumoniae.

Efficacy as assessed by survival.

Mortality rates of 97.5 and 98.8% were observed in nontreated groups over 120 and 192 h, respectively, with the exclusion of the erm(B) isolate, which did not result in >80% mortality in untreated animals. Thus, this isolate was only used in the bacterial density study. The correlation between bacterial density (24 h) and %S over both 120 h (P = 0.002) and 192 h (P = 0.001) was statistically significant (Fig. 6). The sigmoid E_max model-fitted graphs of %S over 120 h versus three PD parameters is shown in Fig. 7, 8, and 9. The sigmoid E_max model-fitted graphs of %S over 192 h versus three PD parameters displayed a profile similar to that observed with the 120-h survival data (data not shown). All three PD parameters demonstrated a significant correlation with %S over 120 h (P < 0.001) and 192 h (P < 0.001). Maximal survival at 120 h (Fig. 9) and 192 h (data not shown) appears to be achieved with an AUC_free/MIC or C_{max free}/MIC that is approximately twice that required to achieve bacteriostatic effects in this model.

FIG. 6.

Relationship between changes in CFU and %S.

FIG. 7.

Relationship between %T > MIC of unbound cethromycin and %S over 120 h (n = 33, with each point representing 10 to 13 mice). SP, S. pneumoniae.

FIG. 8.

Relationship between C_{max free}/MIC of unbound cethromycin and %S over 120 h (n = 33, with each point representing 10 to 13 mice). SP, S. pneumoniae.

FIG. 9.

Relationship between AUC_free/MIC of unbound cethromycin and %S over 120 h (n = 33, with each point representing 10 to 13 mice). SP, S. pneumoniae.

DISCUSSION

“Pharmacodynamics” is a hybrid term that incorporates PK and antibiotic efficacy in order to optimize antibiotic usage. The thigh infection model is the most commonly used in vivo model for assessment of PD, since this model produces relatively less data variability than other infection models (3, 4). However, the thigh model has been criticized for its lack of relevance in practical clinical applications. For instance, in the case of lung infection, the drug distribution of lung tissue differs from that of skeletal muscle. Therefore, we adopted the pneumonia model, since cethromycin targets respiratory tract infections. In our study, by performing intratracheal inoculation, we could recover a sufficient bacterial load (6.00 ± 0.76 log₁₀ CFU/lung) from the lung at the start of drug administration. Moreover, the growth phase of isolates in this adopted pneumonia model was confirmed by our preliminary study, in which the bacterial densities of lungs from nontreated control animals were measured up to 24 h postinoculation. In addition, high mortality rates were observed in nontreated groups over 120 and 192 h, respectively. Finally, the correlations between bacterial density (24 h) and survival over both 120 and 192 h were statistically significant. Despite the inherent potential for introducing data variability, the murine pneumonia model provides more relevant results to predict the outcome of pneumonia treatment.

Our study demonstrated that cethromycin follows nonlinear kinetics, which is indicated by best fitting with a two-compartment open model with Michalis-Menten elimination kinetics in NONMEM analysis. Andes and Craig also found, in their murine thigh model study, nonlinear kinetics with increases in half-life (0.8 to 2.6 h) and AUC/dose (0.5 to 1) with doses from 1.5 to 24 mg/kg (D. R. Andes and W. A. Craig, Abstr. 40th Intersci. Conf. Antimicrob. Agents Chemother., abstr. 2139, 2000). Our study showed greater changes with increases in half-life (2.4 to 67.7 h) and AUC/dose (0.3 to 7.6), since we used a broader dose regimen ranging from 0.1 to 200 mg/kg. The slight difference in half-life between their study and that in our study may have resulted from the use of a different drug solution vehicle, route of administration, and/or infection model.

For the purpose of determining serum free drug concentrations, we used the results of a protein binding study with ICR mice conducted by Abbott Laboratories. As a result of the wide range of drug exposures (0.1 to 200 mg/kg) in the model, protein binding decreased from 94.0 to 92.5% over the dosage range studied. However, Andes and Craig found that protein binding in ICR/Swiss mice was fixed at 90% (Andes and Craig, Abstr. 40th Intersci. Conf. Antimicrob. Agents Chemother., abstr. 2139, 2000). This discrepancy may have resulted from the use of different methods for determining protein binding. Andes and Craig adopted an ultrafiltration method without labeling cethromycin, whereas Abbott Laboratories used an equilibrium dialysis system with [¹⁴C] labeling.

Since only free drug is expected to be distributed to lung tissues and exert antibiotic effects (6), we studied the relationship between antibiotic efficacy and three PD parameters by using free drug concentrations. By study design, we achieved wide ranges of C_{max free}/MIC (0.01 to 137) and AUC_free/MIC (0.07 to 56,999). Moreover, we could also obtain a wide range of %T > MIC_free despite less dispersed data of between 1 and 99%, which resulted from the nonlinear kinetics of the compound and the observation that slight dose increases reduced the elimination of cethromycin, resulting in the achievement of a %T > MIC_free of 100% for many of the regimens studied.

Since the manufacturer of cethromycin has not established a dose regimen for humans at the time of our study initiation, and thus human PK profiles had not been fully investigated, we used a wider range of doses in an attempt to further delineate the PD profile of the compound. In our study, we found that all three commonly utilized PD parameters demonstrated a significant correlation with changes in log₁₀ CFU/lung (Spearman's correlation coefficient, P < 0.001); however, when the goodness of fit was assessed with the E_max model, both the C_{max free}/MIC and AUC_free/MIC appeared to best explain the relationship between drug exposure and reductions in bacteria from harvested lungs. These data reveal that an approximate AUC_free/MIC of 50 or C_{max free}/MIC of 1 resulted in a bacteriostatic effect in this neutropenic pneumonia model. This finding of AUC_free/MIC or C_{max free}/MIC as the predictive parameter related to outcome is not unexpected with the ketolides, since these agents have been reported to produce concentration-dependent bactericidal activity, and this observation relating the PD importance of AUC or C_max has been observed both with cethromycin and a structurally related compound of this class (11; W. A. Craig and D. R. Andes, Abstr. 40th Intersci. Conf. Antimicrob. Agents Chemother., abstr. 2139 and 2141, 2000). Moreover Andes and Craig (Abstr. 40th Intersci. Conf. Antimicrob. Agents Chemother., abstr. 2139, 2000) further suggested that AUC/MIC is the predominant parameter, since it was the parameter most highly correlated (r² = 0.88) with changes in bacterial density of infected thighs as compared to the other correlation values for C_max/MIC of r² = 0.78 and %T > MIC of r² = 0.64. Additionally, they found that bacteriostatic effects in the mouse thigh model were achieved when AUC_free/MIC ranged from 25 to 40 for susceptible isolates and 10 to 108 for resistant isolates. Despite the use of different infection models among the investigations noted, similar PD parameters and magnitudes were linked to outcome as assessed by bacterial killing in infected tissues.

In the present study, we also noted a correlation with %S at both 120 and 192 h for all three parameters, although the data around the fitted line were considerably more variable than those observed when relating change in bacterial density to these pharmacodynamic parameters. In this study, maximal survival at 120 and 192 h appears to be achieved with an AUC_free/MIC or C_{max free}/MIC, which is approximately twice that required to achieve bacteriostatic effects in this model. Moreover, in this study, the changes in the bacterial density were significantly correlated with %S over 120 h (end of therapy) and 192 h (3 days post-end of therapy). Of note, Andes and Craig did not perform a %S study with their thigh infection model, so comparisons with this outcome parameter between infection models are unable to be made.

There was no significant difference between the %S over 120 h and the %S over 192 h. Several reasons could account for this observation. First, bacterial load reduction was almost complete during treatment, leaving an insufficient bacterial load for inducing death after treatment. Second, accumulated drug in lung tissue, sub-MIC effect, and/or postantibiotic effect may have also contributed to this observation.

Finally, we found that an approximate AUC_free/MIC of 50 or C_{max free}/MIC of 1 results in bacteriostatic effects, while higher values (twofold) maximize survival when investigated over a 120- or 192-h observation period. In addition, the S. pneumoniae resistance pattern did not alter the antibiotic efficacy of cethromycin. Our results are consistent with observations of Craig and Andes (Abstr. 40th Intersci. Conf. Antimicrob. Agents Chemother., abstr. 2140, 2000), because the observed bacteriostatic effects in the mouse thigh model were achieved with similar exposures to cethromycin.