Activities of Ceftobiprole, a Novel Broad-Spectrum Cephalosporin, against Haemophilus influenzae and Moraxella catarrhalis

ABSTRACT Ceftobiprole, a broad-spectrum pyrrolidinone-3-ylidenemethyl cephem currently in phase III clinical trials, had MICs between 0.008 μg/ml and 8.0 μg/ml for 321 clinical isolates of Haemophilus influenzae and between ≤0.004 μg/ml and 1.0 μg/ml for 49 clinical isolates of Moraxella catarrhalis. Ceftobiprole MIC50 and MIC90 values for H. influenzae were 0.06 μg/ml and 0.25 μg/ml for β-lactamase-positive strains (n = 262), 0.03 μg/ml and 0.25 μg/ml for β-lactamase-negative strains (n = 40), and 0.5 μg/ml and 2.0 μg/ml for β-lactamase-negative ampicillin-resistant strains (n = 19), respectively. Ceftobiprole MIC50 and MIC90 values for β-lactamase-positive M. catarrhalis strains (n = 40) were 0.12 μg/ml and 0.5 μg/ml, respectively, whereas the ceftobiprole MIC range for β-lactamase-negative M. catarrhalis strains (n = 9) was ≤0.004 to 0.03 μg/ml. Ceftriaxone MICs usually were generally at least twofold lower than those of ceftobiprole, whereas amoxicillin-clavulanate MICs usually were higher than those of ceftobiprole. Azithromycin and telithromycin had unimodal MIC distributions against H. influenzae, with MIC90 values of azithromycin and telithromycin of 2 μg/ml and 4 μg/ml, respectively. Except for selected quinolone-nonsusceptible H. influenzae strains, moxifloxacin proved highly active, with MIC90 values of 0.12 μg/ml. Time-kill analyses showed that ceftobiprole, ceftriaxone, cefpodoxime, amoxicillin-clavulanate, azithromycin, telithromycin, and moxifloxacin were bactericidal at 2× MIC by 24 h against all 10 H. influenzae strains surveyed. Only modest increases in MICs were found for H. influenzae or M. catarrhalis clones after 50 serial passages in the presence of subinhibitory concentrations of ceftobiprole, and single-passage selection showed that the selection frequency of H. influenzae or M. catarrhalis clones with elevated ceftobiprole MICs is quite low.

Haemophilus influenzae, a bacterium with demanding nutritional requirements, is a major cause (together with Streptococcus pneumoniae and Moraxella catarrhalis) of communityacquired respiratory infections such as sinusitis, otitis media, pneumonia, acute exacerbations of chronic bronchitis, and chronic obstructive pulmonary disease (11,23,26). In countries such as the United States, where the H. influenzae type b vaccine is widely used, H. influenzae type b strains have been replaced by untypeable H. influenzae strains.
Synthesis of ␤-lactamases (TEM-1 and rarely ROB-1) is the principal antibiotic resistance trait expressed in H. influenzae (9,17,24,33); lack of susceptibility to ␤-lactams due to alternative mechanisms is rare in most parts of the world (20). A study performed during 1997 reported the incidence of ␤-lactamase production among 1,676 untypeable H. influenzae strains isolated throughout the United States to be 41.6% (17). While the incidence of ␤-lactamase-negative ampicillin-resistant (BLNAR) strains in the United States is Ͻ1% (17,25), the incidence of BLNAR strains in Japan and in some parts of France (9,24) approaches 30%.
Almost all clinical strains of H. influenzae are fluoroquinolone susceptible (17), but nonsusceptibility towards this antibiotic class has been described previously (12), with one fatality recorded for a patient whose infecting H. influenzae strain had not been tested initially for quinolone susceptibility (2). Colo-nization with a quinolone-nonsusceptible clone of H. influenzae was recently reported in a long-term care facility in New York City, N.Y. (29). The prevalence of H. influenzae with reduced susceptibility to quinolones is very low (3) but will undoubtedly increase as quinolones are more frequently prescribed for respiratory tract infections.
Cefixime and cefpodoxime are the oral ␤-lactam antibiotics considered most efficacious against H. influenzae in terms of both MICs and pharmacokinetic/pharmacodynamic properties, followed by amoxicillin-clavulanate and cefuroxime (17). Among the macrolides azithromycin has the lowest MICs against H. influenzae, followed by erythromycin and clarithromycin (8,21,30). However, the presence of a macrolide efflux pump in H. influenzae strains classified as susceptible according to CLSI (formerly NCCLS) breakpoints (31,32), the pharmacokinetic/pharmacodynamic properties of macrolides (16), and results of double-tap otitis media studies (10) all raise questions about the clinical efficacy of this antibiotic class against H. influenzae infections. MICs of telithromycin approximate those of azithromycin (21), but the pharmacological properties of telithromycin against H. influenzae infections are yet to be fully explored (W. A. Craig, personal communication).
M. catarrhalis is an aerobic diplococcus frequently found as a commensal of the upper respiratory tract. It is an important cause of upper respiratory tract infections in otherwise healthy children and elderly people and of lower respiratory tract infections in adults with chronic obstructive pulmonary disease (19,27). At least 85% of M. catarrhalis isolates are ␤-lactamase producers (38).
Ceftobiprole (previously known as BAL9141) is an experi-mental broad-spectrum intravenous cephalosporin. Its prodrug form ceftobiprole medocaril, previously known as BAL5788, is currently in phase III clinical trials. Ceftobiprole has demonstrated antibacterial activity against gram-positive cocci (including penicillin-resistant pneumococci, methicillin-and vancomycin-resistant staphylococci, and ampicillin-susceptible enterococci) and many gram-negative bacteria (4,14,15,18,22). In this study we have compared the activities of ceftobiprole with those of amoxicillin, amoxicillin-clavulanate, ceftriaxone, cefpodoxime, azithromycin, telithromycin, moxifloxacin, and tetracycline by (i) MIC testing of 321 H. influenzae and 49 M. catarrhalis clinical isolates by the microdilution broth method; (ii) macrodilution broth and time-kill studies of the aforementioned drugs against 10 H. influenzae and two M. catarrhalis strains with differing ␤-lactam, macrolide, and quinolone susceptibilities; and (iii) multi-and single-passage studies of the proclivities of ceftobiprole, amoxicillin, ceftriaxone, moxifloxacin, azithromycin, and telithromycin to select for clones with elevated antibiotic MICs against eight H. influenzae and two M. catarrhalis strains.

MIC determinations.
A total of 321 strains of H. influenzae were surveyed. These were isolated between 1999 and 2002, predominantly from sputum, bronchial aspirate, blood, and cerebrospinal fluid (the last two sources from countries which do not use the H. influenzae type b vaccine), and were comprised of 262 ␤-lactamase-positive strains (including two ␤-lactamase-positive amoxicillin-clavulanate-resistant [BLPACR] strains), 40 ␤-lactamase-negative strains, and 19 BLNAR strains. The H. influenzae panel also included nine quinolone-nonsusceptible strains, two macrolide-hypersusceptible strains, and one macrolide-hyperresistant strain (31,32). The 49 M. catarrhalis strains were comprised of 40 ␤-lactamase producers and nine ␤-lactamase nonproducers. Strains were stored at Ϫ70°C in double-strength skim milk (Difco Laboratories, Detroit, Mich.) before MIC testing and periodically examined for purity by cultivation and Gram staining.
Susceptibility testing was performed by the microdilution broth method according to CLSI guidelines (28)  Time-kill studies. For time-kill profiling 10 H. influenzae strains with the following phenotypes were selected: four ␤-lactamase positive, two ␤-lactamase negative, two BLPACR, and two BLNAR. Of these, six strains with amoxicillin MICs of Ͼ32 g/ml were not tested with amoxicillin. Two ␤-lactamase-positive M. catarrhalis strains also were examined by time-kill assays.
Glass tubes containing 5 ml of freshly prepared HTM containing doubling antibiotic concentrations were inoculated with 5 ϫ 10 5 to 5 ϫ 10 6 CFU/ml and incubated at 35°C in a shaking water bath. Viability counts of antibiotic-containing suspensions and controls lacking antibiotic were obtained at 0, 3, 6, 12, and 24 h by plating 10-fold dilutions (HTM) of 0.1-ml aliquots from each tube onto chocolate agar plates, which were incubated for up to 48 h in 5% CO 2 at 35°C. Colony counts were performed on plates yielding 30 to 300 colonies; the lower limit of sensitivity of colony counts was 300 CFU/ml. The number of strains yielding a ⌬(log 10 CFU/ml) of Ϫ1 (corresponding to 90% killing), Ϫ2 (99% killing), and Ϫ3 (99.9% killing) at 3, 6, 12, and 24 h relative to the inoculum size was determined. A given concentration of antibiotic (expressed as a multiplicity of MIC) was considered bactericidal if it reduced the inoculum viable count by Ն3 log 10 CFU/ml or bacteriostatic if it reduced the inoculum viable count by Ͻ3 log 10 CFU/ml during a specified time period.
Multistep selection studies. For resistance selection studies eight H. influenzae isolates (six ␤-lactamase producers including a BLPACR strain and two ␤-lac-  tamase nonproducers including a BLNAR strain) and two M. catarrhalis isolates (both ␤-lactamase positive) were chosen. Serial passage of each strain (initial inoculum size, ϳ10 8 CFU/ml) was performed daily using freshly prepared HTM supplemented with increasing subinhibitory concentrations of a given antibiotic. For each subsequent daily passage the inoculum was taken from the tube nearest the MIC, usually 1 or 2 log 2 dilution steps below the MIC, which had approximately the same turbidity as antibiotic-free controls. Strains were passaged until MICs of Ͼ64 g/ml were obtained, up to a maximum of 50 serial passages, at which time subculturing in the presence of antibiotic was discontinued, and selected resistant/nonsusceptible clones underwent 10 daily passages on chocolate agar without antibiotics. The identity of parental strains and their derived clones was confirmed by pulsed-field gel electrophoresis using a CHEF DR III apparatus (Bio-Rad, Hercules, CA). MICs for each resistant clone for each drug were confirmed by microdilution broth assay. Resistance mechanisms for some parental strains and derived clones were determined as described below.
Mechanisms of resistance. The presence of mutations in ribosomal L4 and L22 proteins and in 23S rRNA (for macrolide-resistant clones) and in quinolone resistance-determining regions (for moxifloxacin-resistant clones) was examined using primers and conditions as described previously (6,31). After PCR amplification products were purified using a QIAquick PCR purification kit (QIAGEN, Valencia, Calif.). Nucleotide sequences were obtained using a CEQ8000 Genetic Analysis System (Beckman Coulter, Fullerton, CA).
Single-step selection studies. Selection frequencies of clones resistant to 2ϫ, 4ϫ, and 8ϫ the MIC of antibiotics were determined by spreading 10 9 to 10 11 CFU (in 100 l) onto plates of either HTM (H. influenzae) or Mueller-Hinton agar supplemented with 5% sheep blood (M. catarrhalis). After incubation at 35°C in 5% CO 2 for 48 h, resistant colonies were confirmed by replica plating onto medium containing antibiotics. Resistance frequencies were calculated as the number of resistant colonies per inoculum (21).

RESULTS
The MIC range for ceftobiprole against the 321 H. influenzae strains surveyed was 0.008 to 8 g/ml; MIC 50 and MIC 90 values were 0.06 and 0.25 g/ml for ␤-lactamase-positive strains, 0.03 and 0.25 g/ml for ␤-lactamase-negative strains, and 0.5 and 2.0 g/ml for BLNAR strains, respectively ( Table 1). The MIC range for ceftobiprole towards the 49 M.catarrhalis strains surveyed was Յ0.004 to 1 g/ml; MIC 50 and MIC 90 values were 0.12 and 0.5 g/ml, respectively, for ␤-lactamase-positive strains, whereas the MIC range for the nine ␤-lactamase-negative strains was Յ0.004 to 0.03 g/ml (Table 1). Of a total of 370 H. influenzae and M. catarrhalis strains examined, only four (1.1%), all ␤-lactamase-positive H. influenzae isolates susceptible to azithromycin, moxifloxacin, and tetracycline, had ceftobiprole MICs exceeding 2.0 g/ml (MIC of 4 g/ml, one strain; MIC of 8 g/ml, three strains). ␤-Lactamase levels in enzyme-positive strains with raised ceftobiprole MICs were not quantified.
Ceftriaxone MICs for H. influenzae were usually at least twofold lower than those of ceftobiprole, whereas MICs of amoxicillin (with or without clavulanate) were usually two-to fourfold higher than those of ceftobiprole. Cefpodoxime MICs were similar to or slightly higher than those of ceftobiprole. Azithromycin and telithromycin yielded unimodal MIC distributions for both H. influenzae and M. catarrhalis and were active against M. catarrhalis at MICs of Յ1 g/ml. Moxifloxacin was active against all strains except those H. influenzae isolates chosen for their fluoroquinolone nonsusceptibility, while tetracycline yielded MICs between 2 and 8 g/ml for H. influenzae and between Յ1 and Ͼ8 g/ml for M. catarrhalis.
Time-kill studies. Macrodilution broth MICs for the 12 strains chosen for time-kill studies are listed in Table 2, and summaries of time-kill results for the H. influenzae strains surveyed are presented in Table 3. Ceftobiprole was bactericidal against 7 of 10 H. influenzae strains at 1ϫ MIC and against all 10 strains at 2ϫ MIC by 24 h. Amoxicillin-clavulanate, ceftriaxone, cefpodoxime, azithromycin, telithromycin, and moxifloxacin produced similar time-kill profiles relative to their MICs. Bactericidal activity at 2ϫ MIC after 24 h was observed for tetracycline against 6 of 10 strains and for amoxicillin against two of four strains tested.
Time-kill analyses of two ␤-lactamase-positive M. catarrhalis strains (data not shown) found that ceftobiprole and amoxicillin were bactericidal against one strain at 4ϫ MIC by 24 h. Amoxicillin-clavulanate, ceftriaxone, cefpodoxime, azithromycin, telithromycin, and moxifloxacin were all bactericidal against both M. catarrhalis strains at 2ϫ MIC after 24 h. Tetracycline was bactericidal against one strain at 2ϫ MIC and against both M. catarrhalis strains at 4ϫ MIC after 24 h.
After 50 serial passages in the presence of ceftobiprole MICs for the eight H. influenzae strains surveyed remained constant for five strains and increased 1 to 2 log 2 dilution steps for the other three strains; the highest MIC recorded was 1 g/ml. Ceftriaxone selection likewise led to MIC increases of 0 to 2 log 2 dilution steps, the highest ceftriaxone MIC recorded being 0.25 g/ml. No increase in MICs was observed for strains subcultured in amoxicillin-clavulanate. During serial passage seven of eight H. influenzae strains gave rise to azithromycinnonsusceptible clones according to CLSI breakpoint criteria (7); two of these clones, recovered after 34 to 37 passages, had undergone MIC increases of 1 to Ͼ64 g/ml. In contrast, only three of eight strains evolved resistance to telithromycin (2 to 16 g/ml, 1 to 32 g/ml, and 0.5 to Ͼ64 g/ml) according to CLSI breakpoint criteria. Selection with moxifloxacin produced only one clone nonsusceptible to this fluoroquinolone (0.06 to 32 g/ml), though MIC increases as high as 32-fold were found for the remaining seven H. influenzae clones passaged in the presence of moxifloxacin. Mutations in 23S rRNA and/or in ribosomal proteins L4 and L22 were identified in three of six azithromycin-nonsusceptible and telithromycinresistant clones ( Table 4). Sequencing of five moxifloxacinnonsusceptible clones (MIC, 0.5 to 32 g/ml) revealed mutations in GyrA, GyrB, and/or ParC in three clones ( Table 4). The MIC for azithromycin of one azithromycin-nonsusceptible clone, derived from H. influenzae parental strain 153-008, was somewhat unstable, dropping from Ͼ64 g/ml to 32 g/ml after 10 passages in drug-free medium (Table 5), though MICs for other antibiotics in the drug panel either remained unchanged or rose by 1 to 2 log 2 dilution steps. After 10 passages on drug-free medium, the azithromycin-nonsusceptible clone   derived from H. influenzae parental strain 110-061 showed no MIC changes for azithromycin, ceftobiprole, ceftriaxone, or amoxicillin-clavulanate outside a single log 2 dilution step, though its moxifloxacin MIC dropped from 1 g/ml to 0.125 g/ml (Table 5). During passage on drug-free medium the MIC for telithromycin of this clone rose from 4 g/ml to Ͼ32 g/ml. We have no explanation for this phenomenon, which is currently under investigation. After 50 serial passages in the presence of antibiotic, MICs for the two M. catarrhalis strains either remained stable or increased by 1 log 2 dilution step (Table 4). Selection by moxifloxacin and azithromycin did not lead to increases in MICs compared to the parental strains (0.06 g/ml for moxifloxacin and 0.03 g/ml for azithromycin).

DISCUSSION
Ceftobiprole, the active component of the prodrug ceftobiprole medocaril (formerly BAL5788), is a broad-spectrum pyrrolidinone-3-ylidenemethyl cephem with well-documented in vitro and in vivo activities against most clinically relevant pathogens, including penicillin-resistant pneumococci, ␤-lactamase-positive H. influenzae, and Enterobacteriaceae devoid of extended-spectrum ␤-lactamases (1,13,14,15,18). The MICs for H. influenzae and M. catarrhalis found in this study generally mirror those reported by Hebeisen at al (14) and by Jones et al. (18). A small number of H. influenzae strains with high MICs, up to 8 g/ml, was identified by another group (N. C. Issa, M. S. Rouse, K. E. Piper, J. M. Steckelberg, and R. Patel, Abstr. 43rd. Intersci. Conf. Antimicrob. Agents Chemother., abstr. F-540, 2003). These discrepancies may reflect methodological differences or differences in the sources from which clinical isolates were collected. In our survey, which included BLNAR and BLPACR strains, ceftobiprole had MICs of Յ2 g/ml towards 98.8% of the 321 H. influenzae clinical isolates examined. The highest ceftobiprole MIC, 8 g/ml, was encountered in three strains (0.9%), all of which were ␤-lactamase positive and susceptible to azithromycin, moxifloxacin, and tetracycline.
Kill kinetics for ceftobiprole resembled those of other ␤-lactams examined, with bactericidal activity at 2ϫ MIC by 24 h against all 10 H. influenzae strains profiled. Other ␤-lactams had similar kill kinetics relative to their different MICs, which were usually higher than those for ceftobiprole. The significance of time-kill results in the two M. catarrhalis strains studied awaits confirmation by examination of more strains. M. catarrhalis is a difficult organism to test by time-kill assay because of clumping.
Our results indicate that ceftobiprole has a low proclivity for emergence of resistance in H. influenzae and M. catarrhalis. Under intensive selective pressure (50 serial passages) cephalosporins did not promote emergence of clones with MICs exceeding four times that of parental strains; the largest observed change in ceftobiprole MIC was from 0.06 g/ml to 0.25 g/ml. Likewise, the single-step selection procedure failed to detect H. influenzae or M. catarrhalis clones with elevated MICs for ceftobiprole. In contrast, prolonged serial passage of H. influenzae in the presence of a macrolide (azithromycin or telithromycin) led to a Ͼ4-fold increase in MIC for most strains. While mutations in ribosomal proteins L4 and/or L22, as well as in domains II and/or V of 23S rRNA, can be responsible for macrolide resistance in H. influenzae (5,31), we were unable to detect any such mutations in three azithromycin-nonsusceptible clones with MICs of Ն32 g/  The results of this study, combined with the excellent antipneumococcal activity of ceftobiprole (1,22) and favorable pharmacokinetic and safety profiles of the prodrug ceftobiprole medocaril (21,(34)(35)(36)(37), make ceftobiprole a very promising drug for treatment of patients with community-acquired respiratory tract infections who require hospitalization.