ABSTRACT
ABT-492 is a novel quinolone with potent activity against gram-positive, gram-negative, and atypical pathogens, making this compound an ideal candidate for the treatment of community-acquired pneumonia. We therefore compared the in vitro pharmacodynamic activity of ABT-492 to that of levofloxacin, an antibiotic commonly used for the treatment of pneumonia, through MIC determination and time-kill kinetic analysis. ABT-492 demonstrated potent activity against penicillin-sensitive, penicillin-resistant, and levofloxacin-resistant Streptococcus pneumoniae strains (MICs ranging from 0.0078 to 0.125 μg/ml); β-lactamase-positive and β-lactamase-negative Haemophilus influenzae strains (MICs ranging from 0.000313 to 0.00125 μg/ml); and β-lactamase-positive and β-lactamase-negative Moraxella catarrhalis strains (MICs ranging from 0.001 to 0.0025 μg/ml), with MICs being much lower than those of levofloxacin. Both ABT-492 and levofloxacin demonstrated concentration-dependent bactericidal activities in time-kill kinetics studies at four and eight times the MIC with 10 of 12 bacterial isolates exposed to ABT-492 and with 12 of 12 bacterial isolates exposed to levofloxacin. Sigmoidal maximal-effect models support concentration-dependent bactericidal activity. The model predicts that 50% of maximal activity can be achieved with concentrations ranging from one to two times the MIC for both ABT-492 and levofloxacin and that near-maximal activity (90% effective concentration) can be achieved at concentrations ranging from two to five times the MIC for ABT-492 and one to six times the MIC for levofloxacin.
ABT-492, an investigational fluoroquinolone, targets both bacterial DNA gyrase and topoisomerase IV (L. L. Shen, Y. Cai, and A. M. Nilius, Abstr. 42nd Intersci. Conf. Antimicrob. Agents Chemother., abstr. F-545, 2002). Thus, the quinolone has potent activity against gram-positive and gram-negative pathogens. ABT-492 has also demonstrated in vitro activity against ciprofloxacin-resistant Streptococcus pneumoniae (H. J. Smith, K. A. Nichol, L. Palatnick, B. Weshnoweski, T. Bellyou, E. Rimmer, D. J. Hoban, and G. G. Zhanel, Abstr. 42nd Intersci. Conf. Antimicrob. Agents Chemother., abstr. F-549, 2002) and Legionella species (J. Dubois and C. St-Pierre, Abstr. 42nd Intersci. Conf. Antimicrob. Agents Chemother., abstr. F-553, 2002), making this antibiotic a prime candidate for the treatment of community-acquired respiratory tract infections.
In North America there has been a significant increase in the prevalence of quinolone-resistant S. pneumoniae in the community setting (1, 2; R. N. Jones and M. A. Pfaller, Letter, J. Clin. Microbiol. 38:4298-4299, 2000). More importantly, quinolone resistance has been linked to therapeutic failure in the clinical setting (3, 4, 7). It is therefore crucial for researchers to design and develop potent antibacterial compounds to target all community-acquired respiratory pathogens, including those pathogens that are resistant to other antibacterials. Because few data on ABT-492 have been published, we sought to compare its activity to that of levofloxacin, an antibiotic commonly used for the treatment of community-acquired pneumonia. Our comparison evaluates the pharmacodynamic activities of ABT-492 and levofloxacin against community-acquired respiratory pathogens by describing the concentration-effect relationship through time-kill kinetic studies and sigmoidal maximal-effect (Emax) models. Knowledge of this antibacterial's pharmacodynamics will provide a more rational basis for determining optimal dosing regimens during its development.
(This work was presented in part at the poster session of the 42nd Interscience Conference on Antimicrobial Agents and Chemotherapy, San Diego, Calif., 2002. This work was also presented in part at the poster session of the 40th Annual Meeting of the Infectious Disease Society of America, 2002.)
MATERIALS AND METHODS
Bacteria and antimicrobial agents.Two strains of penicillin-sensitive (MIC, <0.1 μg/ml) and two strains of penicillin-resistant (MIC, 2 μg/ml) S. pneumoniae, two strains of β-lactamase-positive and two strains of β-lactamase-negative Haemophilus influenzae, and two strains of β-lactamase-positive and one strain of β-lactamase-negative Moraxella catarrhalis were obtained from clinical respiratory specimens from the microbiology laboratories at the University of Illinois at Chicago Medical Center, Chicago. Isolates were stored at −70°C in double-strength skim milk (Remel, Lenexa, Kans.) and were subcultured three times prior to use. In addition, one quinolone-resistant S. pneumoniae strain was obtained from The Jones Group (JMI Laboratories, North Liberty, Iowa) and stored similarly. The ABT-492 and levofloxacin antibiotics for the MIC determination and time-kill kinetic studies were supplied as laboratory powders of known potency from Abbott Laboratories, Abbott Park, Ill. Antibiotic stock solutions were prepared according to the manufacturer's recommendations, diluted to concentrations ranging from 1,000 to 10,000 μg/ml, stored in unit-of-use vials, and frozen at −70°C until needed. The media used for MIC and time-kill kinetic studies included cation-adjusted Mueller-Hinton broth (Becton Dickinson, Sparks, Md.) with 5% lysed horse blood (Remel) (11) for the S. pneumoniae isolates and Haemophilus test medium (Dade Behring, West Sacramento, Calif.) for the H. influenzae and M. catarrhalis isolates.
MIC determination.The MIC for each isolate was determined by broth microdilution techniques as outlined by the National Committee for Clinical Laboratory Standards, and testing was performed in duplicate (10). Control strains (S. pneumoniae ATCC 49619 and H. influenzae ATCC 49247) were used to validate the MIC results. The inoculum was prepared by suspending S. pneumoniae organisms grown on blood agar plates or H. influenzae and M. catarrhalis organisms grown on chocolate agar plates, which had been incubated for a full 24 h, in 2 ml of sterile saline. Suspensions were adjusted to a 0.5 McFarland turbidity standard by using a spectrophotometer and diluted in broth to obtain a final inoculum of approximately 5 × 105 CFU/ml for each well. Inoculum checks were performed via colony counts. The microtiter plates were incubated overnight at 35°C in humidified air, and the results were read at 24 h. The lowest concentration of antibiotic in the wells showing no visible growth was defined as the MIC.
Time-kill assays.Prior to time-kill kinetic studies, a carryover study was performed for each isolate as outlined by the National Committee for Clinical Laboratory Standards guidelines (9). The time-kill studies were performed with antibiotic concentrations selected to comprise four doubling dilutions below and three doubling dilutions above the MIC for each test isolate, as previously described (3, 11, 14). A growth control (no antibiotic) was included for each isolate. Briefly, glass tubes containing 4.950 ml of the appropriate medium and antibiotic were inoculated with 50 μl of a diluted bacterial suspension with a pipette below the surface of the broth. S. pneumoniae was incubated in broth medium (described above) for 16 h prior to use, and H. influenzae and M. catarrhalis were taken directly from a 24-h culture plate. Inoculated tubes containing a concentration of 5 × 105 to 5 × 106 CFU/ml were incubated at 35°C on a rotator for 24 h.
Counts of viable cells in the antibiotic-containing suspensions and the growth controls were performed at 0, 2, 4, 6, 12, and 24 h by plating 10-fold dilutions of 0.1-ml aliquots from each tube on blood agar plates for S. pneumoniae and on chocolate agar plates for H. influenzae and M. catarrhalis. Recovery plates were incubated at 35°C for 24 to 48 h. Colony counts at 0 h were used to validate the starting inoculum. For undiluted inocula, carryover was addressed by immediately spreading the broth over the agar plate until it was dry (11, 14). The lower limit of sensitivity for colony counts was 300 CFU/ml.
Colony counts (log10 numbers of CFU per milliliter) from duplicate time-kill runs performed on different days were averaged and plotted versus time for each isolate. Bacteriostatic activity was defined as a 0- to <3-log10 reduction, and bactericidal activity was defined as a ≥99.9% (a ≥3-log10) reduction in the number of CFU per milliliter from that of the starting bacterial concentration. Composite concentration-effect graphs were constructed by plotting the change in log10 numbers of CFU per milliliter from the starting inoculum at each time point for all isolates versus the numbers in cultures with antibiotic concentrations standardized to multiples of the MIC.
Dose-effect response.To compare the concentration-effect relationships at 2, 4, 6, 12, and 24 h of ABT-492 and levofloxacin against S. pneumoniae, H. influenzae, and M. catarrhalis, the mean time-kill data at each time point for each bacterial isolate were combined according to bacterial species. The net change (log10 numbers of CFU per milliliter) in bacterial density at each time point for each concentration was fitted by multivariate nonlinear-regression analysis with a four-parameter sigmoidal Hill (Emax) model by using SigmaPlot 2000 for Windows (version 6.0; SPSS Inc.) (6, 8). In this model, effect (the net change in the log10 number of CFU per milliliter) is equal to E0 − (Emax × Cn)/(EC50n + Cn), where E0 is the baseline effect (bacterial growth for the control), Emax is the maximal bacterial-kill effect, C is the concentration of interest (a multiple of the MIC), EC50 is the antibacterial concentration that produced 50% of the maximal effect, and n is the sigmoidicity factor that gives flexibility to the shape of the curve. E0, EC50, Emax, and n were given in the regression output. The concentration that produced the EC90 between E0 and Emax was calculated.
RESULTS
Antibacterial susceptibility results.The MICs of each agent for each isolate are listed in Table 1. All 12 strains, except the quinolone-resistant S. pneumoniae strain (MIC = 32 μg/ml), were sensitive to levofloxacin. The breakpoints for ABT-492 have not been established; however, ABT-492 demonstrated potent activity against all bacterial isolates for which the MICs were ≤0.125 μg/ml. Against S. pneumoniae, the MICs of ABT-492 ranged from 0.0078 to 0.125 μg/ml compared to 1 to 32 μg/ml for levofloxacin. For H. influenzae and M. catarrhalis, the MICs of ABT-492 were much lower and ranged from 0.000625 to 0.0025 μg/ml compared to 0.0156 to 0.0625 μg/ml for levofloxacin.
MICs for all test isolates
Antibacterial carryover.A carryover effect was seen in all S. pneumoniae isolates at eight times the MIC, regardless of the tested quinolone. At four times the MIC, all carryover was eliminated, except that a slight carryover was noted at four times the MIC of ABT-492 for the quinolone-resistant S. pneumoniae isolate. A potential for carryover existed when direct sampling of the inoculated test tubes was needed at higher multiples of the MIC; however, the 0.1-ml aliquots were immediately spread over the agar plates, which contained 25 ml of medium. This reduced the carryover by causing a 1:250 dilution of the antibiotic (11, 14). Carryover was not observed with any of the H. influenzae and M. catarrhalis isolates.
Concentration-effect response.The mean starting inoculum for all concentrations in the time-kill curves was 5.813 log10 CFU/ml (standard deviation, 0.470). When all 12 isolates were compared based on the extent of the bacterial kill, ABT-492 and levofloxacin at eight times the MIC demonstrated bactericidal activity against all isolates. ABT-492 and levofloxacin at four times the MIC were bactericidal against 83% (10 of 12) and 100% (12 of 12) of the isolates, respectively. A bacteriostatic effect was observed between one and two times the MIC for many isolates. A delay in the bacterial growth compared to that of the control was observed at 0.25 and 0.5 times the MIC. When time-kill data were separated according to bacterial species, ABT-492 failed to produce bactericidal effects at four times the MIC for the quinolone-resistant isolate and one penicillin-resistant S. pneumoniae isolate. At two times the MIC, ABT-492 was bactericidal against 20% (one of five) of the S. pneumoniae isolates compared to 100% (five of five) with levofloxacin. Regrowth (defined as an increase in growth of ≥2 log10 CFU/ml after 6 h) (9) was not seen in the ABT-492-tested isolates but occurred in one levofloxacin-tested penicillin-sensitive S. pneumoniae. The time-kill kinetic data for H. influenzae and M. catarrhalis were similar for both ABT-492 and levofloxacin. Concentrations at four and eight times the MIC produced bactericidal activity in all strains. At two times the MIC, bactericidal activity was observed in 29% (two of seven) of the H. influenzae and M. catarrhalis isolates with ABT-492 compared to 71% (five of seven) with levofloxacin.
A time-kill plot of the activities of ABT-492 and levofloxacin against each bacterial species plus the quinolone-resistant S. pneumoniae isolate is shown in Fig. 1. The plots of the log10 numbers of CFU per milliliter versus time for all test isolates demonstrate the concentration-dependent bactericidal activity. Increases in the concentrations of both ABT-492 and levofloxacin consistently led to an increase in the rate and extent of bactericidal activity. The results of sigmoidal Emax models, plotted as the antibacterial activities (the net change in numbers of CFU per milliliter) of both quinolones relative to the concentrations used (multiples of the MIC) after 4, 6, and 12 h of exposure, for the quinolone-sensitive S. pneumoniae isolates are presented in Fig. 2. The quinolone-resistant S. pneumoniae isolate was not included in the S. pneumoniae analysis because the results from time-kill data significantly differed from the data for quinolone-sensitive isolates. The results of sigmoidal Emax models for H. influenzae and M. catarrhalis were similar to those for S. pneumoniae and are thus not depicted.
Time-kill curves for four clinical isolates. Curves on the left represent the activity of ABT-492, and the curves on the right represent the activity of levofloxacin against the same isolate. ▴, control; ⋄, 0.0625 times the MIC; ♦, 0.125 times the MIC; □, 0.25 times the MIC; ▪, 0.5 times the MIC; ▿, 1 times the MIC; ▾, 2 times the MIC; ○, 4 times the MIC; •, 8 times the MIC.
Composite concentration-response curves for penicillin-sensitive S. pneumoniae (SPS) and penicillin-resistant S. pneumoniae (SPR) isolates following antibiotic exposure to ABT-492 (left-hand side) and levofloxacin (right-hand side) at 4, 6, and 12 h.
The EC50, EC90, and Emax data obtained from all plots are presented in Table 2. The EC50s for all isolates ranged from 0.884 (standard error, 0.086) to 1.96 (standard error, 0.125) times the MIC. The EC90s for all isolates ranged from 1.270 to 6.383 times the MIC. With increased time exposure, the maximal bacterial-killing effect increased and the multiple of the MIC needed to achieve the EC90 decreased, except at 24 h. When the EC50s of ABT-492 were compared to those of levofloxacin, the values fell within the standard errors of each other at 2 and 4 h for S. pneumoniae and H. influenzae, respectively. At the 12- and 24-h antibacterial exposures, the EC50 of levofloxacin was lower than that of ABT-492, except in the case of H. influenzae.
Composite Emax model parameters
DISCUSSION
ABT-492 is a novel fluoroquinolone that is currently in development. This compound has enhanced antibacterial activity against a variety of pathogens, including fluoroquinolone-resistant isolates (10a). In this study, we characterized its pharmacodynamic activity and compared it to that of levofloxacin by using MIC determinations and in vitro time-kill kinetic studies. As stated earlier, ABT-492 had very low MICs when it was tested against gram-positive and gram-negative isolates. This potency has been validated by other researchers and extends to other bacterial species (A. M. Nilius, D. Hensey-Rudloff, L. Almer, J. Beyer, and R. K. Flamm, Abstr. 42nd Intersci. Conf. Antimicrob. Agents Chemother., abstr. F-546, 2002). Whether these low MICs will correlate to in vivo potency will depend on the achievable free antibiotic concentration at the site of infection. The achievable free concentration of an antimicrobial is dependent on its protein binding, tissue distribution, and therapeutic index (i.e., the efficacy-to-safety profile). While the tissue distribution and safety profiles for ABT-492 are still unknown, protein binding has recently been studied in an in vitro experimental model with both human and rat serum (10a). Those researchers found ABT-492 to be 77 and 89% protein bound in human and rat serum, respectively. This protein binding was similar to that of trovafloxacin and significantly greater than that of ciprofloxacin. The in vitro activities of both ABT-492 and trovafloxacin were reduced when they were tested in both heat-inactivated human serum and rat serum, although to a lesser extent with human serum.
In time-kill kinetic studies, ABT-492 and levofloxacin demonstrated similar pharmacodynamics. Both compounds produced potent concentration-dependent kills over a 24-h period. At higher antibiotic concentrations (four to eight times the MIC), both antibiotics were predominantly bactericidal. At the middle concentrations (one to two times the MIC), the activity ranged from bactericidal to bacteriostatic. At lower concentrations (0.25 to 0.5 times the MIC), bacterial growth was delayed compared to that of the growth control. Because the maximal activities of both antibiotics increased with increasing concentrations for all isolates, we concluded that both quinolones had concentration-dependent bactericidal activities regardless of the tested isolate. Additionally, because the slope of the lines (rate of activity) in the time-kill curve varied with concentration, we can again conclude concentration dependence. Differences in the bactericidal activities of the fluoroquinolones were seen at one and two times the MIC. The less-frequent bactericidal activity of ABT-492 at these concentrations may have resulted from the experimentally obtained MIC. The in vitro MIC is often not the true MIC because only doubling concentrations of antibiotic are tested. The MICs obtained for levofloxacin may have overestimated the true MIC more so than those obtained for ABT-492. At one times the MIC, levofloxacin resulted in bacteriostatic to bactericidal activity more frequently than ABT-492 resulted in bacteriostatic activity or growth. Because both antibiotics are concentration dependent, slight differences in antibiotic concentration exposures will affect the rate and extent of the bacterial kill.
The sigmoidal Emax model used in this study was designed to further support the time-kill kinetic analysis of concentration dependence. The sigmoidal Emax model depicts the relationship between the concentration (multiple of the MIC) and effect (net change in colony count) of multiple bacterial isolates at one point in time. By generating the model we can calculate parameters such as Emax, EC50, and EC90. Looking at the relationship between these parameters allows us to evaluate the effect of concentration over time by looking at how the concentration-response curve or slope changes. The slope of the curve becomes steeper over time because at lower concentrations, bacteria continue growing while colony count reductions occur at higher concentrations, as can be seen in Fig. 2. The larger change (effect) between lower and higher antibiotic concentrations over time is reflected by the increase in the Emax produced by the antibiotic. At 12 and 24 h, higher antibiotic concentrations not only prevented a 3-log10 increase in growth seen at the lower antibiotic concentrations but also produced a bactericidal effect with a >3-log10 reduction in colony count. We can also see that the slope of the curve changes around one times the MIC because at that concentration, growth becomes inhibited and colony reduction begins to occur. By examining how the EC50 and EC90 change relative to each other over time, we can appreciate the concentration-response curve at each time point. If the distance between the EC50 and EC90 narrows over time, the model suggests concentration dependence (5, 6, 8). In our model, the EC50 stayed relatively constant while the EC90 decreased until 24 h, hence the distance between EC50 and EC90 narrowed. Because of this phenomenon, higher antibiotic concentrations are needed to achieve 90% of the maximal activity at earlier time points.
The slight increase observed in the EC50s and EC90s of both quinolones at 24 h may represent a small amount of regrowth seen at bacteriostatic concentrations. While a full 1- to 2-log10 increase in colony count cannot be appreciated in Fig. 1, slight increases in colony counts at multiple bacteriostatic concentrations may be enough to impact the EC50 and EC90. Compared to the value at 12 h, there was a decrease in the Emax at 24 h that may have resulted from the combination of a slight regrowth that takes place at 24 h and a small reduction in colony counts for the control and isolates at lower antibacterial concentrations as bacteria achieve maximal growth.
When comparing the activity of ABT-492 to that of levofloxacin, we can look at EC50s to compare potencies. The compound with the highest potency will have the lowest EC50. We can see from Table 2 that, with standard error taken into account, levofloxacin has a lower EC50 against all of the tested species between 6 and 24 h after antibiotic exposure. Likewise, levofloxacin has a lower calculated EC90. For the purpose of this study, antibacterial concentrations were measured in multiples of the MIC. We calculated our data as such to make the results of the sigmoidal Emax models comparable for ABT-492 and levofloxacin, because the MICs were very different. Although levofloxacin may look more potent with respect to its EC50, when we multiply the EC50 by the obtained MIC, levofloxacin has a much higher EC50 in terms of antibacterial concentration (micrograms per milliliter) than ABT-492.
The information obtained in this study can be further applied to optimize future dosing regimens for this agent. ABT-492 demonstrates concentration-dependent activity in these models; however, future in vitro and in vivo dynamic modeling studies with this agent are needed to further validate the pharmacodynamics. The pharmacodynamic modeling of other quinolones, such as levofloxacin, have correlated better to the area under the concentration-time curve (AUC) over 24 h compared to the bacterial MIC (AUC-to-MIC ratio) (12, 13). Regardless of whether the antibiotic pharmacodynamics correlate better with concentration dependence (peak concentration-to-MIC ratio) or the AUC-to-MIC ratio, it will still be important to maximize free concentrations of antibiotic in serum or antibiotic concentrations at the site of infection to improve antibacterial activity. By looking at our Emax model, we predict that close-to-maximal activity (EC90) for ABT-492 can be achieved from free concentrations in serum ranging from two to four times the MIC (0.0012 to 0.5 μg/ml for all tested isolates) of the bacterial isolate over a 24-h period.
Conclusion.This study demonstrates that ABT-492 is a potent concentration-dependent antibiotic with in vitro activity similar to that of levofloxacin. The compound's very low MICs for gram-positive, gram-negative, and levofloxacin-resistant S. pneumoniae respiratory pathogens warrant further investigation. Sigmoidal Emax models were constructed to validate the time-kill pharmacodynamic analysis. The relationship between EC50 and EC90 became closer over time, supporting concentration-dependent pharmacodynamic activity. The models demonstrated that 50% maximal activity occurred between one and two times the MIC, and 90% maximal activity was observed at one to six times the MIC. Knowledge of the concentrations that are needed to achieve near-maximal activity of an antibiotic can help determine a goal for free antibiotic concentrations in serum in vivo.
ACKNOWLEDGMENTS
We thank Paul Schreckenberger from the University of Illinois at Chicago Medical Center and Ronald Jones from JMI Laboratories for providing the experimental isolates. We also thank Diane Styrczula for the laboratory support.
This study was supported by a grant from Abbott Laboratories, Abbott Park, Ill.
FOOTNOTES
- Received 17 March 2003.
- Returned for modification 7 July 2003.
- Accepted 7 October 2003.
- Copyright © 2004 American Society for Microbiology