Ceftobiprole Efficacy In Vitro against Panton-Valentine Leukocidin Production and In Vivo against Community-Associated Methicillin-Resistant Staphylococcus aureus Osteomyelitis in Rabbits

Abstract

Community-associated methicillin-resistant Staphylococcus aureus (CA-MRSA) can cause osteomyelitis with severe sepsis and/or local complications in which a Panton-Valentine leukocidin (PVL) role is suspected. In vitro sub-MIC antibiotic effects on growth and PVL production by 11 PVL⁺ MRSA strains, including the major CA-MRSA clones (USA300, including the LAC strain; USA400; and USA1000), and 11 PVL⁺ methicillin-susceptible S. aureus (MSSA) strains were tested in microplate culture. Time-kill analyses with ceftobiprole at its MIC were also run with LAC. Efficacies of ceftobiprole (40 mg/kg of body weight subcutaneously [s.c.] four times a day [q.i.d.]) or vancomycin (60 mg/kg intramuscularly [i.m.] twice a day [b.i.d.]) alone or combined with rifampin (10 mg/kg b.i.d.) against rabbit CA-MRSA osteomyelitis, induced by tibial injection of 3.4 × 10⁷ CFU of LAC, were compared. Treatment, started 14 days postinoculation, lasted 14 days. In vitro, 6/11 strains cultured with sub-MICs of ceftobiprole produced 1.6- to 4.8-fold more PVL than did the controls, with no link to specific clones. Rifampin decreased PVL production by all tested strains. In time-kill analyses at the LAC MIC (0.75 mg/liter), PVL production rose transiently at 6 and 8 h and then declined 2-fold at 16 h, concomitant with a 2-log₁₀-CFU-count decrease. In vivo, the mean log₁₀ CFU/g of bone for ceftobiprole (1.44 ± 0.40) was significantly lower than that for vancomycin (2.37 ± 1.22) (P = 0.034), with 7/10 versus 5/11 bones sterilized, respectively. Combination with rifampin enhanced ceftobiprole (1.16 ± 0.04 CFU/g of bone [P = 0.056], 11/11 sterile bones) and vancomycin (1.23 ± 0.06 CFU/g [P = 0.011], 11/11 sterile bones) efficacies. Ceftobiprole bactericidal activity and the rifampin anti-PVL effect could play a role in these findings, which should be of interest for treating CA-MRSA osteomyelitis.

INTRODUCTION

Methicillin-resistant Staphylococcus aureus (MRSA) infections are increasingly being detected in the community worldwide (4). In the United States, the community-associated (CA) MRSA genotype USA300 has emerged as the major circulating strain type. CA-MRSA strains are generally more virulent than hospital-acquired (HA) MRSA, a finding consistent with the ability of CA-MRSA to cause disease in children and adults without predisposing factors (8). At present, skin and soft tissue infections represent the majority of the CA-MRSA disease burden, but severe acute diseases, such as necrotizing pneumonia, sepsis, and osteomyelitis, have also been described (3, 15, 16). Concomitantly with the recent worrying emergence of CA-MRSA strains, the S. aureus osteomyelitis incidence in hospitalized children doubled between 2002 and 2007 in the United States, and that increase was due exclusively to MRSA (14). Furthermore, pediatricians reported more unusual cases of staphylococcal osteomyelitis, characterized by severe sepsis (16) and/or local extension with concomitant myositis or pyomyositis (2).

Most CA-MRSA strains produce Panton-Valentine leukocidin (PVL). The PVL contribution to the course of osteomyelitis, suspected very early by Panton and Valentine themselves (31), was highlighted by the observation that S. aureus bone-and-joint infections were more severe and required prolonged treatment when the strain produced PVL (2, 9). In experimental Los Angeles County clone (LAC USA300) CA-MRSA rabbit osteomyelitis, PVL was shown to play a role in the local extension of the infection to muscle (6), a result consistent with the ability of PVL to cause muscle damage in a mouse model of necrotizing soft tissue infection (38).

At present, little is known about the therapeutic options for bone infections caused by PVL⁺ CA-MRSA. Therapeutic recommendations have been derived from those of HA-MRSA infections (25), and the optimal regimen remains to be defined. Unlike HA-MRSA, CA-MRSA strains are often susceptible in vitro to several oral antimicrobials (4). However, some of those compounds can be ineffective in vivo (32). In addition to the bacteriostatic/bactericidal activity, the effect of antibiotics on PVL release by the strains could also play a role in their in vivo efficacy. Indeed, it was shown in vitro that subinhibitory antibiotic concentrations influenced PVL release (10). In osteomyelitis, the bacteria may encounter such low concentrations because of poor antibiotic penetration into cortical bone (34).

Vancomycin alone or combined with rifampin remains the first-line parenteral therapeutic option for MRSA osteomyelitis in the Infectious Diseases Society of America (IDSA) guidelines (25). However, vancomycin efficacy might not be optimal and some data suggested that it could be less effective than beta-lactams against methicillin-susceptible S. aureus (MSSA) bacteremia (36).

Ceftobiprole is the first broad-spectrum cephalosporin with bactericidal activity against MRSA, including CA-MRSA isolates (23, 43). It was shown to be as effective as vancomycin against complicated skin and skin structure infections caused by Gram-positive bacteria, including PVL⁺ isolates (29). Its efficacy alone against MRSA experimental endocarditis and localized osteomyelitis models has already been demonstrated (37, 42). However, its in vivo efficacy against PVL-producing CA-MRSA osteomyelitis has never been tested.

The goals of this study were (i) to evaluate in vitro ceftobiprole activity against the isolates belonging to major PVL⁺ CA-MRSA clones circulating worldwide and its impact on their capacity to produce PVL and (ii) to compare the in vivo efficacies of ceftobiprole and vancomycin alone or combined with rifampin against experimental CA-MRSA osteomyelitis in rabbits.

(This work was presented, in part, at the 21st European Congress of Clinical Microbiology and Infectious Diseases-27th International Congress of Chemotherapy, ECCMID-ICC, Milan, Italy, 7 to 10 May 2011 [35a].)

MATERIALS AND METHODS

In vitro study design. (i) Bacterial strains.

The PVL⁺ S. aureus LUG855 strain obtained by lysogenization of the reference strain RN6390 with phage phiSLT as control (28); 11 S. aureus isolates, including the major PVL-producing CA-MRSA clones; and 10 PVL⁻ MSSA strains were used (Table 1). Strains were characterized by multilocus sequence typing (MLST), with spa typing and mecA and pvl gene detection performed as previously described (12, 17, 20). Six mecA-negative strains showed agr-1 and ST8 by MLST and t008, t024, or t1911 (spa type) and belonged to the USA300 lineage. The mecA⁺ strains belonged to the major CA-MRSA clones spreading throughout the United States: 5 strains showed agr-1 and ST8 (MLST) and t008 (spa type) and belonged to the USA300 major lineage, including the LAC strain (kindly provided by Franck DeLeo, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Hamilton, MT) (41); the 5 agr-3 and ST1 (MLST) and t125, t128, or t175 spa-type strains belonged to the USA400 lineage; the 1 agr-1, ST59, and t216 strain belonged to the USA1000 lineage.

Table 1.

Characterization of clinical strains and MICs of selected antibiotics^a

Strain	Sequence type	spa type	mecA status	agr allele	lukSF-PV	Source	MIC (mg/liter) in CCY mediumb
							OXA	RIF	CEF	VAN
ST20070572	8	t1911	−	1	+	This study	0.25	0.007	0.5	1
ST20070573	8	t008	−	1	+	This study	0.25	0.015	0.5	1
ST20070575	8	t008	−	1	+	This study	0.25	0.015	0.5	1
ST20070576	8	t008	−	1	+	This study	0.25	0.03	0.5	1
ST20070577	88	t2742	−	3	+	This study	0.25	0.03	0.5	1
ST20070579	121	t645	−	4	+	This study	0.25	0.03	0.5	1
ST20070580	8	t024	−	1	+	This study	0.25	0.007	0.5	1
ST20070581	121	t940	−	4	+	This study	0.25	0.015	0.5	1
ST20070582	8	t024	−	1	+	This study	0.12	0.03	0.5	1
ST20070583	25	t528	−	1	+	This study	0.12	0.03	0.5	1
LUG855	8	ND	−	1	+	10	0.12	0.015	0.5	1
HT20010251	8	t008	+	1	+	This study	4	0.03	1	1
HT20030203	8	t008	+	1	+	This study	16	0.03	1	1
HT20030206	8	t008	+	1	+	This study	4	0.03	1	1
HT20030207	8	t008	+	1	+	This study	8	0.015	1	1
LAC USA300	8	t008	+	1	+	41	8	0.015	1	2
HT20010751	59	t216	+	1	+	This study	8	0.007	1	1
HT20010253	1	t125	+	3	+	This study	16	0.03	1	1
HT20010734	1	t128	+	3	+	10	16	0.007	1	1
HT20010252	1	t175	+	3	+	This study	8	0.03	1	1
HT20010255	1	t175	+	3	+	This study	8	0.015	1	2
HT20020338	1	t128	+	3	+	This study	8	0.007	1	1
ATCC 29213	ND	ND	−	ND	ND	ATCC	0.12	0.007	0.25	0.5

^aAbbreviations: ND, not determined; OXA, oxacillin; CEF, ceftobiprole; RIF, rifampin; VAN, vancomycin.

^bMICs were determined according to CLSI recommendations for 22 strains cultured in CCY broth.

(ii) Antibiotics.

Antibiotic MICs were determined for each isolate using the broth microdilution method recommended by the Clinical and Laboratory Standards Institute (CLSI) in Mueller-Hinton medium (bioMérieux, Marcy l'Etoile, France) (5) and adapted to CCY medium containing casein hydrolysate and yeast extract (bioMérieux) because optimal PVL yield is obtained in this medium. Ceftobiprole was provided by Basilea Pharmaceutica (Basel, Switzerland), whereas oxacillin was purchased from Biochemika Fluka (Buchs, Switzerland), and rifampin and vancomycin were purchased from Sigma-Aldrich (L'Isle d'Abeau, France). S. aureus strain ATCC 29213 served as the control for MIC determinations.

(iii) Culture conditions for PVL production in the presence of antibiotics.

Isolates were precultured aerobically on blood agar at 37°C overnight before being tested for PVL production using the broth microdilution method in modified CCY medium, without or with antibiotics (at 0.50×, 0.25×, and 0.12× MIC) according to the CLSI standard procedure. After 24 h at 37°C without shaking, samples were taken for counting of bacterial colonies in diluted broth and PVL quantification by enzyme-linked immunosorbent assay (ELISA), as previously described (1, 10). Experiments were run in triplicate. Results are expressed as the mean ratios of μg of PVL/log₁₀ CFU of bacteria grown with each antibiotic concentration to that of bacteria grown without antibiotic (control).

Time-kill experiments cultured the LAC strain in CCY medium without or with ceftobiprole at its MIC starting with an inoculum of 10⁷ CFU/ml. Aliquots of each culture (without or with ceftobiprole) were sampled at 6, 8, and 16 h for PVL measurements and bacterial counts. To measure PVL, cultures were adjusted to an optical density at 600 nm (OD₆₀₀) of 1 and PVL was assessed with a specific ELISA at all times. Experiments were run in triplicate and results are expressed as μg of PVL/ml of adjusted culture.

In vivo study design. (i) Test strain.

The ceftobiprole-susceptible LAC strain used in this study was isolated from a patient with a skin abscess.

(ii) Preparation of bacterial inocula.

The organisms were stored at −80°C until used. Before experiments, they were cultured in CCY medium at 37°C for 18 h with shaking. After centrifugation, the pellets were washed twice and resuspended in phosphate-buffered saline to the desired bacterial density immediately before inoculation. All inocula were quantified by plating serial dilutions on tryptic soy agar (bioMérieux).

(iii) Experimental model.

Norden's method (30) was used to induce osteomyelitis in female New Zealand White rabbits, each weighing 2 to 3 kg. The rabbits were housed in individual cages and received food and water ad libitum. The experimental protocol complied with French legislation on animal experimentation and was approved by the Animal Use Committee of Maison Alfort Veterinary School. The animals were anesthetized by intramuscular injection of 25 mg of ketamine/kg of body weight (Vibrac, Carros, France) and 25 mg/kg of 2% xylazine (Rompum; Bayer Santé, Division Santé Animal, Puteaux, France). An 18-gauge needle was inserted percutaneously through the lateral right tibial metaphysis into the medullary cavity. Infection was induced by direct injection of a sclerosing agent (0.1 ml of 3% sodium tetradecyl sulfate (Trombovar; Kreussler Pharma, La Chaussée-Saint-Victor, France), followed by 0.2 ml of CA-MRSA LAC inoculum (3.4 × 10⁷ CFU) and 0.1 ml of saline. Patch analgesia (Durogesic; Janssen-Cilag, Issy-les Moulineaux, France) was given for 7 days following surgery. A 3.4 × 10⁷-CFU inoculum was selected because previous experiments (data not shown) had shown that it induced persistent osteomyelitis in 90% of animals with low early mortality due to severe sepsis (<10%).

(iv) Treatment and its evaluation.

Fourteen days postinfection, rabbits were randomly assigned to the untreated (control) or treated group. The latter received ceftobiprole (40 mg/kg subcutaneously [s.c.] four times a day [q.i.d.], n = 10) or vancomycin (60 mg/kg intramuscularly [i.m.] twice a day [b.i.d.]) alone (n = 11) or combined with rifampin (10 mg/kg i.m. b.i.d., n = 11). Vancomycin and rifampin doses were selected based on previous experiments (33) showing that they achieved concentrations close to those recommended for humans (trough vancomycin concentrations of 15 to 20 μg/ml and rifampin dose equivalent to 900 mg/day) (25). The ceftobiprole dose was selected based on the pharmacokinetic study described below. Each regimen was administered for 14 days. Control rabbits (n = 9) were left untreated. Animals were killed by intravenous (i.v.) injection of pentobarbital 7 days after the end of therapy (day 35) to allow for bacterial regrowth after ending treatment while avoiding the persistence of residual antibiotic in the bone. Control rabbits were also killed on day 35. At the time of death, the right hind limb was dissected, and the tibia and femur were separated from the surrounding soft tissues. For quantitative bacterial counts, the upper third of the tibia (3 cm long), including compact bone and marrow, was isolated, split with a bone crusher, weighed, cut into small pieces, frozen in liquid nitrogen, and crushed in an autopulverizer (Spex 6700; Freezer/Mill Industries Inc., Metuchen, NJ). The pulverized bone was suspended in 10 ml of sterile saline; serial dilutions were made and plated on tryptic soy agar. After 24 h of incubation at 37°C, the number of viable microorganisms was determined. Results are expressed as means ± standard deviations (SD) log₁₀ CFU/g of bone and as the percentage of animals with sterile bone. Bone was considered sterile when the culture showed no growth after incubation for 48 h at 37°C and the number of CFU was recorded as the lowest detectable bacterial count (1.10 to 1.30 CFU/g of bone, depending on the weight of the sample).

(v) In vivo selection of mutants.

Aliquots (0.1 ml) of each undiluted bone homogenate were also plated onto Mueller-Hinton agar (Becton, Dickinson, Rungis, France) containing ceftobiprole, rifampin, or vancomycin, at 2 and 4 times their MICs, to detect potentially emerging resistant mutants after 24, 48, and 72 h of incubation at 37°C. When bacterial growth was observed, MICs were determined using the Etest method (AB Biodisk, Solna, Sweden). Mutants were defined as having 3-fold-higher MICs than the initial strain.

(vi) Serum ceftobiprole levels.

Serum antibiotic levels were determined in uninfected rabbits. Dose regimens were selected on the basis of previous experimental studies in rabbits (42) and then adjusted to achieve serum levels equivalent to those obtained in humans. Four rabbits received an s.c. injection of ceftobiprole, and blood was drawn 15 and 30 min and 1, 3, 6, and 8 h later. Ceftobiprole concentrations were measured by high-performance liquid chromatography with Waters Alliance e2695 and Waters 2487 UV detectors. The column was a Nova-Pack C₁₈, 4.0 μm (150 by 3.9 mm) (Waters, St Quentin-en-Yvelines, France). The mobile phase was composed of a mixture of 0.1 M KH₂PO₄ buffer (pH 6.1) and acetonitrile, 95/5 (vol/vol). The flow rate was 1 ml/min, and ceftobiprole was detected by UV absorbance at 300 nm. The internal standard was ceftazidime. The detection limit was 0.5 mg/liter. The coefficient of variation was <7.5 mg/liter.

Statistical analyses.

Bacterial densities in bones from the experimental groups were compared by analysis of variance (ANOVA), followed by Scheffe's test for multiple comparisons. Results are expressed as means ± SD. For PVL measurements, statistical analyses were based on one-way ANOVA followed, when necessary (samples of unequal variance), by Dunnett's a posteriori test. The impact of subinhibitory concentrations of oxacillin, ceftobiprole, vancomycin, and rifampin on bacterial growth was analyzed with Student's t test. A P value of <0.05 was considered significant. Analyses were computed with SPSS v19.0 software.

RESULTS

In vitro study. (i) MIC determinations.

As expected, all mecA⁺ strains were resistant to oxacillin, and all strains were susceptible to vancomycin, rifampin, and ceftobiprole (Table 1).

(ii) Antibiotic effect on bacterial growth.

CFU counts from 3 different experiments with the LAC strain after incubation of microplate cultures without or with subinhibitory antibiotic concentrations (0.125×, 0.25×, and 0.5× MIC) are shown in Table 2. As expected from previous experiments, growth was inhibited (bacterial inoculum loss exceeded 1 log₁₀ CFU) when oxacillin or ceftobiprole was used at 0.5× or 0.25× MIC, but also when cultures were incubated with vancomycin or rifampin at 0.5× MIC. Bacterial growth in the presence of oxacillin was significantly lower than that with vancomycin at 0.125× MIC (P = 0.045) or rifampin at 0.125× MIC (P = 0.017) and 0.25× MIC (P = 0.007). It was also lower in the presence of ceftobiprole than in the presence of rifampin at 0.25× MIC (P = 0.028). For all other antibiotic concentrations, bacterial counts were comparable to control values.

Table 2.

Antibiotic effects on bacterial growth^a

Antibiotic	Bacterial count, log10 CFU/ml
	No antibiotic	0.125× MIC	0.25× MIC	0.50× MIC
Oxacillin	9 ± 0.3	8 ± 0.2	7.5 ± 0.3	6.5 ± 0.3
Ceftobiprole	9 ± 0.4	8.5 ± 0.4	7.5 ± 0.6	6.5 ± 0.6
Vancomycin	9 ± 0.5	8.5 ± 0.3 A	8 ± 0.4	7 ± 0.4
Rifampin	9 ± 0.3	9 ± 0.4 A	8.7 ± 0.3 A, B	6.8 ± 0.3

^aThe S. aureus LAC strain was incubated in CCY medium without or with oxacillin, ceftobiprole, vancomycin, or rifampin (at 0.50×, 0.25×, and 0.125× MIC) according to the CLSI standard procedures for 24 h at 37°C without shaking. Results of 3 experiments are expressed as mean ± SD log₁₀ CFU of bacteria. Statistical significance is shown by capital letters: A, P < 0.05 versus oxacillin; B, P < 0.05 versus ceftobiprole.

(iii) Antibiotic effect on PVL production in microplate cultures (Fig. 1).

Fig 1.

Antibiotic effect on PVL production by 22 S. aureus strains. Strains were incubated on CCY-containing microplates without (●) or with antibiotics (oxacillin [♢], vancomycin [△], rifampin [○], and ceftobiprole [□]) at 0.50×, 0.25×, and 0.125× MIC according to CLSI standard procedures for 24 h at 37°C without shaking. Samples were taken for bacterial counting and PVL quantification by ELISA. Results are expressed as the ratio of mean μg of PVL/log₁₀ CFU of bacteria grown at the indicated antibiotic concentration to that of bacteria cultured without antibiotic (control). Results are the means of 3 independent experiments for each strain, and horizontal bars indicate the median values. *, P < 0.05 (ANOVA), antibiotic versus control. atb, antibiotic; OXA, oxacillin; VAN, vancomycin; RIF, rifampin; CEF, ceftobiprole.

PVL production by all strains increased when grown with 0.125 or 0.25 times the oxacillin MIC. The amplitude of the enhancement was strain dependent and ranged from 1.2 to 7.2 times, compared to control PVL production. Thus, the medians increased significantly by 1.86-fold for 0.125× MIC and 2.15-fold for 0.25× MIC (P < 0.05). When cultured with 0.5× MIC, overall PVL production was not modified.

PVL production increased for only 15/22 strains cultured with 0.25 or 0.125 times the ceftobiprole MIC. The amplitude increase was strain dependent and ranged from 1.7 to 4.8 times, compared to the control PVL production. Thus, the medians rose significantly, 1.43-fold for 0.125× MIC and 1.18-fold for 0.25× MIC (P < 0.05). At 0.125× or 0.25× ceftobiprole MIC, induced PVL production levels were significantly lower than those observed with oxacillin (P < 0.05). As for oxacillin at 0.5× MIC, the overall PVL production was not modified in the presence of 0.5× ceftobiprole MIC.

PVL production decreased in the presence of 0.5 or 0.25 times the rifampin MIC for all strains and for most of the strains at 0.125 times its MIC. The reduction amplitude was strain dependent and ranged from 2 to 20 times below control growth. Thus, the medians decreased significantly by 1.45-fold for 0.125× MIC, 5-fold for 0.25× MIC, and 16.67-fold for 0.5× MIC (P < 0.05).

Vancomycin, at any concentration tested, did not significantly affect PVL production, which was significantly lower with vancomycin than with ceftobiprole at 0.125×, 0.25×, and 0.5× MIC (P < 0.006).

(iv) Ceftobiprole effect on LAC PVL production in time-kill cultures.

Figure 2 shows the transient PVL production rise of 1.91 and 1.84 times control growth after 6 and 8 h of incubation with ceftobiprole at its MIC. After 16 h, S. aureus cultured with ceftobiprole showed a 2-fold PVL production decline concomitant with a 2-log₁₀-CFU decrease of bacterial counts in time-kill analyses.

Fig 2.

Ceftobiprole effect on PVL production by S. aureus LAC. The S. aureus LAC strain was cultured in CCY broth under time-kill conditions, without (black bars) or with ceftobiprole at its MIC (white bars). Aliquots were taken at 6, 8, and 16 h for PVL measurement after being adjusted to an OD₆₀₀ of 1. Experiments were run in triplicate, and results are expressed in μg of PVL/ml of adjusted culture (mean ± SD). *, P < 0.05 (ANOVA), with antibiotic versus without antibiotic.

Experimental CA-MRSA (LAC) osteomyelitis study.

Following a single s.c. injection of ceftobiprole (40 mg/kg), its mean peak plasma concentration was 100.1 ± 24.7 μg/ml, the mean trough concentration was 4.53 ± 2.07 μg/ml, and area under the curve from 0 to 8 h was 283 mg · h/liter (n = 4). Those parameters are close to those obtained in humans given the 1,000-mg dose three times a day (t.i.d.) (27).

Control animals infected with LAC had a mean bacterial count of 4.57 ± 1.09 log₁₀ CFU/g of bone (Fig. 3). In the vancomycin-treated group, 5/11 animals had sterile bone and the mean bacterial count in bone (2.37 ± 1.22 log₁₀ CFU/g) differed significantly from that of controls. In the ceftobiprole-treated group, 7/10 animals had sterile bone and the mean bone bacterial density (1.44 ± 0.40 log₁₀ CFU/g) was significantly lower than those of controls and vancomycin-treated rabbits. Compared to each primary antibiotic alone, the addition of rifampin significantly enhanced the efficacies of ceftobiprole (1.16 ± 0.04 log₁₀ CFU/g of bone, P = 0.056) and vancomycin (1.23 ± 0.06 log₁₀ CFU/g of bone, P = 0.011), and all animals had sterile bone. No ceftobiprole- or vancomycin-resistant strain emerged in the bones of treated animals.

Fig 3.

Antibiotic effect against a rabbit model of CA-MRSA LAC osteomyelitis. Rabbits were injected for 14 days with ceftobiprole (Cef; 40 mg/kg, s.c., q.i.d.) or vancomycin (Van; 60 mg/kg, i.m., b.i.d.) alone or combined with rifampin (Rif; 10 mg/kg, i.m., b.i.d.). Bone was considered sterile when the culture showed no growth after incubation for 48 h at 37°C and the number of CFU was recorded as the lowest detectable bacterial count (1.10 to 1.30 CFU/g of bone, depending on the weight of the sample).

DISCUSSION

In this study, we tested in vitro ceftobiprole activity against a large representative sample of the major clones of PVL⁺ CA-MRSA. We also evaluated its in vivo efficacy in a PVL⁺ CA-MRSA experimental osteomyelitis model as monotherapy and combined with rifampin.

In vitro, all tested strains were susceptible to vancomycin, rifampin, and ceftobiprole, in accordance with other in vitro studies that observed ceftobiprole MICs between 0.5 and 2 μg/ml against 100 CA-MRSA strains (23).

Here, PVL production by strains cultured with 0.12× or 0.25× MIC of ceftobiprole or oxacillin increased, while it was not modified by vancomycin. However, the ceftobiprole-induced PVL rise was smaller than that induced by oxacillin. We recently showed that PVL induction by beta-lactams was triggered by penicillin-binding protein 1 (PBP1) inhibition. Therefore, only beta-lactams inhibiting PBP1 were able to increase PVL expression (11). Ceftobiprole is a broad-spectrum cephalosporin with high affinity for PBP2a with a 50% inhibitory concentration (IC₅₀) of about 0.5 mg/liter for MRSA (13). Ceftobiprole has good affinity for MSSA PBP1 to PBP4, with an IC₅₀ of about 1 mg/liter (7). However, oxacillin affinity for PBP1 is about 20 times higher than that of ceftobiprole (26), thereby explaining their different PVL-inducing abilities.

The LAC strain selected for the in vivo study belongs to the USA300 lineage, the major PVL⁺ CA-MRSA clone in the United States. To mimic growth conditions similar to the experimental setting, we determined PVL production by the LAC strain, during exposure to ceftobiprole at its MIC. PVL production rose transiently after 6 and 8 h of culture but declined thereafter, probably linked to the antibiotic's bacterial growth-inhibitory effect. These in vitro data support using ceftobiprole as an appropriate beta-lactam to treat CA-MRSA infection, as its overall observed effect combined suppression of bacterial growth and that of PVL expression.

Among the antibiotics tested, only rifampin, at sub-MICs, decreased PVL production for all tested strains, including LAC, while, as shown previously, vancomycin had no effect on PVL expression (10).

In vivo, ceftobiprole was significantly more effective than vancomycin at decreasing bone LAC counts. However, the clinical significance of that observation merits being discussed in its context. First, the vancomycin MIC of the tested LAC strain (2 μg/ml) might have favored ceftobiprole. Indeed, it was previously shown that elevated MICs (1.5 to 2 μg/ml) enhanced the likelihood of vancomycin therapeutic failure (25). Second, none of the monotherapies was able to sterilize 100% of the animals. Moreover, the good ceftobiprole efficacy found in our model agrees with other experimental studies performed with different staphylococcal strains.

In experimental endocarditis due to MRSA strain COL (vancomycin MIC, 1 μg/ml) (37), ceftobiprole led to significantly lower bacterial counts in vegetations than did vancomycin. Also, in a rabbit MRSA osteomyelitis model (vancomycin MIC, 0.78 μg/ml) (42), ceftobiprole was able to sterilize 100% of the animals compared to 73% of vancomycin- or linezolid-treated animals. Notably, ceftaroline, another cephalosporin with anti-MRSA bactericidal activity, performed better than vancomycin against MRSA rabbit endocarditis and acute osteomyelitis (18, 19). The previously reported good in vivo activity of ceftobiprole monotherapy, noted early in experimental endocarditis (13), remains only partially understood; its bactericidal effect against MRSA might play a role. In our in vivo study, plasma ceftobiprole dosage yielded trough concentrations twice the MIC. Authors of previous in vitro time-kill studies reported ceftobiprole to be bactericidal against MRSA at 2× MIC at 24 h (24). Another explanation could be good antibiotic diffusion at the infection site. However, to our knowledge, ceftobiprole bone penetration has not been studied. The absence of ceftobiprole-resistant strains after monotherapy in our model is also pertinent in the perspective of its use in clinical situations.

Combination with rifampin is well known to enhance the efficacy of treatment of experimental bone-and-joint infections because of its bactericidal activity against slow-growing bacteria (44). However, this remarkable in vivo efficacy could be hampered by the emergence of rifampin-resistant strains, even when rifampin is used in combination (21, 40). In our study, combination with rifampin improved the efficacy of ceftobiprole or vancomycin and sterilized 100% of the rabbits.

In addition to its bactericidal effect on bone-enclosed bacteria, the antibiotic impact on the capacity of S. aureus to release PVL could interfere with its in vivo efficacy against PVL⁺ CA-MRSA osteomyelitis. Indeed, we previously showed that PVL plays a role in the persistence and rapid local extension of CA-MRSA (LAC) rabbit osteomyelitis (6). In our in vitro model, ceftobiprole sub-MICs slightly increased LAC PVL production. Rifampin sub-MICs significantly lowered PVL production by LAC, by as much as 20 times less than the controls. Early effective suppression of PVL expression by rifampin combined with ceftobiprole or vancomycin might explain the therapeutic successes of these regimens in our in vivo model.

The experimental model used here, induced by 10⁷ CFU of LAC, reproduces CA-MRSA hematogenous osteomyelitis in children with early cortical bone involvement and local extension to muscles and joints (6). However, also described in children (22, 39) is a more severe systemic disease with potential lethal sepsis that can be experimentally reproduced by using a higher inoculum (35). The results obtained here cannot be extrapolated to those cases that warrant further therapeutic studies.

Unlike vancomycin, which must be closely monitored to avoid treatment failures and toxicity (25), beta-lactam antibiotics, with their extensive history of use, are generally considered safe and easy to administer. Taken together, our in vitro and in vivo findings suggest that beta-lactams active against MRSA, like ceftobiprole, should be of interest in treating CA-MRSA osteomyelitis.