Effect of achieving bone sterilisation on bone architecture and bone marrow, in an experimental rabbit model of osteomyelitis caused by carbapenemase-producing Enterobacterales

OBJECTIVES

Natural history and treatment of bone infections caused by carbapenemase-producing Enterobacterales (CPE) are poorly defined. We evaluated the effect of treatment on the progression of subacute osteomyelitis in a rabbit model.

METHODS

Two isolates were used: a KPC-producing Klebsiella pneumoniae and an Escherichia coli harbouring blaOXA-48 and blaCTX-M15 inserts, both susceptible to gentamicin, colistin, fosfomycin, and ceftazidime-avibactam. Osteomyelitis was induced in rabbits by tibial injection of 2 × 108 colony-forming units/mL. Antibiotics were started 14 d later, for 7 d, in 6 groups of 12 rabbits. Three days after treatment completion (D24), rabbits were euthanised and bones were cultured. Bone marrow and bone architecture macroscopic changes were evaluated through analysis of pictures by investigators unaware of the rabbit treatment group and microbiological outcome, using scales ranging from 0 (normal) to 3 (severe lesions) depending on modifications.

RESULTS

Bone marrow modifications induced by local infection were similar between prematurely deceased animals and non-sterilised animals (P = 0.14) but differed significantly from animals that achieved bone sterilisation after treatment (P = 0.04). Conversely, when comparing bone deformity, rabbits who died early (n = 13) had similar bone architecture as those achieving bone sterilisation (P = 0.12), as opposed to those not sterilised after treatment (P = 0.04). After a multivariate logistic regression, bone marrow scale ≤2 was associated with bone sterilisation (P < 0.001), and bone architecture scale ≤2 was associated with bone sterilisation (adjusted odds ratio = 2.7; 95% confidence interval 1.14-6.37) and KPC infection (adjusted odds ratio = 5.1; 95% confidence interval 2.17-12.13).

CONCLUSION

Effective antibacterial treatment reduces bone architecture distortion and bone marrow changes. These variables may be used as proxy for bone sterilisation.

Hip joint infections caused by multidrug-resistant Enterobacterales among patients with spinal cord injury: experience of a reference center in the Greater Paris area

Introduction

We aimed to describe the management and treatment of hip joint infections caused by multidrug-resistant Enterobacterales among patients with spinal cord injury (SCI).

Methods

We included all hip joint infections associated with grade IV decubitus ulcers caused by Extended-Spectrum Beta-Lactamase producing Enterobacterales (ESBL-PE) and Carbapenemase-Producing Enterobacterales (CPE) treated in a reference center for bone and joint infections (BJIs) over 9 years in a retrospective study.

Results

Seventeen SCI patients with ischial pressure ulcers breaching the hip capsule (mean age 52 ± 15 years) were analyzed. In 16 patients, paraplegia was secondary to trauma and one was secondary to multiple sclerosis. Infections were mostly polymicrobial (n=15; 88.2%), notably caused by Klebsiella pneumoniae (n=10) and S. aureus (n=10). The carbapenemases identified were exclusively OXA-48-type (n=3) including 2 isolates co-expressed with ESBL-PE within the same bacterial host.

Multidrug-resistant Enterobacterales were commonly resistant to fluoroquinolones (n=12; 70.6%). Most therapies were based on carbapenems (n=10) and combination therapies (n=13). Median duration of treatment was 45 (6-60) days. Of 17 cases of hip joint infections, 94.1% (n=16) benefited from a femoral head and neck resection.

Infection control was initially achieved in 58.8% (n=10) of cases and up to 88.2% after revision surgeries, after a median follow-up of 3 (1-36) months.

Conclusion

Hip infections among SCI patients caused by multidrug-resistant Enterobacterales are often polymicrobial and fluoroquinolones-resistant infections caused by K. pneumoniae and S. aureus, highlighting the need for expert centers with pluridisciplinary meetings associating experienced surgeons, clinical microbiologists and infectious disease specialists.

Native bone and joint infections caused by extended-spectrum β-lactamase-producing Enterobacterales: experience of a reference centre in the Greater Paris area

Antibiotic treatment of native osteomyelitis caused by extended-spectrum β-lactamase-producing Enterobacterales (ESBL-PE) is a challenge. Limited epidemiological and outcome data are available. This retrospective cohort study included osteomyelitis patients with ESBL-PE infections treated in a reference centre for bone and joint infections (BJIs) between 2011-2019. Twenty-nine patients with native BJI (mean age, 44.4 ± 15.7 years) were analysed. Fifteen cases were paraplegic patients with ischial pressure sores breaching the hip capsule. Other cases included eight other hip infections, four tibial infections and two foot infections. Infections were mostly polymicrobial (n = 23; 79.3%), including Staphylococcus aureus (n = 13; 8 methicillin-resistant). Klebsiella pneumoniae (n = 13) was the most frequent ESBL-producing species identified, followed by Escherichia coli (n = 10), including 3 E. coli/K. pneumoniae co-infections, and Enterobacter spp. (n = 9). ESBL-PE were rarely susceptible to fluoroquinolones (n = 4; 13.8%). Most therapies were based on carbapenems (n = 22) and combination therapies (n = 19). The median duration of treatment was 41 (5-60) days. Primary control of the infection was achieved in 62.1% (18/29) of cases and up to 86.2% after second look surgeries, after a median follow-up of 6 (1-36) months. Infection with ESBL-producing K. pneumoniae was associated with failure (P = 0.001), whereas age, infection location, prior colonisation and antimicrobial therapy were not found to be predictors of outcome. ESBL-PE native BJIs are often polymicrobial and fluoroquinolone-resistant infections caused by K. pneumoniae, highlighting the need for expert centres with pluridisciplinary meetings with experienced surgeons.

Efficacy of generic meropenem products in combination with colistin in carbapenemase-producing Klebsiella pneumoniae experimental osteomyelitis

Guidelines for the management of carbapenemase-producing Enterobacterales (CPE) infections recommend a combination of two active agents, including meropenem if the minimum inhibitory concentration (MIC) is ≤8 mg/L. The therapeutic equivalence of meropenem generics has been challenged. We compared the bactericidal activity of meropenem innovator (AstraZeneca) and four generic products (Actavis, Kabi, Mylan and Panpharma), both in vitro and in vivo, in association with colistin. In vitro time-kill studies were performed at 4 × MIC. An experimental model of KPC-producing Klebsiella pneumoniae osteomyelitis was induced in rabbits by tibial injection of a sclerosing agent followed by 2 × 10⁸ CFU of K. pneumoniae KPC-99YC (meropenem MIC = 4 mg/L; colistin MIC = 1 mg/L). At 14 days after inoculation, treatment for 7 days started in seven groups of ≥10 rabbits, including a control group, a colistin group, and one group for each meropenem product (i.e. the innovator and four generics), in combination with colistin. In vitro, meropenem + colistin was bactericidal with no viable bacteria after 6 h, and this effect was similar with all meropenem products. In the osteomyelitis model, there was no significant difference between meropenem generics and the innovator when combined with colistin. Colistin-resistant strains were detected after treatment with colistin + meropenem innovator (n = 3) and generics (n = 3). The efficacy of four meropenem generics did not differ from the innovator in vitro and in an experimental rabbit model of KPC-producing K. pneumoniae osteomyelitis in terms of bactericidal activity and the emergence of resistance.

Case series of carbapenemase-producing Enterobacteriaceae osteomyelitis: Feel it in your bones

OBJECTIVES

Limited data have been reported regarding osteomyelitis due to carbapenemase-producing Enterobacteriaceae (CPE), including co-infections with extended-spectrum β-lactamase (ESBL)-producing micro-organisms.

METHODS

We conducted a retrospective study in a reference centre for bone and joint infections from 2011 to 2019 among patients infected with CPE.

RESULTS

Nine patients (mean age 46.8 ± 16.6 years), including three with infected implants, were identified. Infections were mostly polymicrobial (n = 8/9), including Staphylococcus aureus (n = 6/9). CPE were mainly OXA-48-type, associated with ESBL-producing Enterobacteriaceae (n = 8/9), of which 5/9 isolates were Klebsiella pneumoniae. Control of the infection was achieved in seven cases.

CONCLUSIONS

CPE osteomyelitides are essentially polymicrobial and fluoroquinolone-resistant infections, highlighting the need for efficient surgery with implant removal.

Critical analysis of experimental models of periprosthetic joint infection

INTRODUCTION

Because the extreme diversity of clinical situations makes formal clinical trials difficult to carry out, animal models of periprosthetic infection in orthopaedics are needed to understand the aetiology and pathology of these infections, and to test new treatment methods. These experimental models must reproduce the features of the infections encountered in clinical practice. One of the model variables is the method of inoculation: local (intra-articular), intravenous or intra-arterial. Another is the timing of the inoculation: intra-operative or postoperative. Together, these options simulate the different contamination methods: direct, by proximity or blood-borne. However, the chosen inoculation route can also affect the infection rate and severity in the various models, and in some cases do not accurately reproduce the postoperative infections encountered clinically.

HYPOTHESIS

The direct inoculation method is the most effective for inducing a local infection on a foreign body in a joint, and the least iatrogenic.

METHODS

A critical analysis of published studies was carried out to evaluate each model against three endpoints, according to the type of inoculation. The primary endpoint was the infection rate, which should be as close as possible to 100%. The secondary endpoints were the mortality rate and rate of spontaneous healing, both of which should be as low as possible. Twenty-one articles were reviewed.

RESULTS

Intra-articular and intra-medullary inoculations had induction rates between 70 and 100%; intra-arterial inoculations had an induction rate of 100%, while intravenous inoculation had a rate of 47 to 77%. The mortality rates were lower with the intra-articular and intramedullary inoculations (5 to 23%) than for the intra-arterial inoculations (37%) and intravenous inoculations (28 to 56%). The spontaneous healing rate was 0 to 30% for intra-articular and intramedullary inoculations, 30 to 53% for intravenous inoculations and 0% for intra-arterial inoculations.

CONCLUSION

Direct inoculation methods are most effective at reproducing chronic periprosthetic joints infections, without putting the animal's life at risk or allowing for spontaneous healing. The simulation of blood-borne infections is more random.

Activity of LY333328 combined with gentamicin in vitro and in rabbit experimental endocarditis due to vancomycin-susceptible or -resistant Enterococcus faecalis

We investigated the activity of LY333328 alone and combined with gentamicin, both in vitro and in a rabbit model of experimental endocarditis, against the susceptible strain Enterococcus faecalis JH2-2 and its two glycopeptide-resistant transconjugants, BM4316 (VanA) and BM4275 (VanB). MICs of LY333328 and gentamicin were 2 and 16 microgram/ml, respectively, for the three strains. In vitro, LY333328 alone was bactericidal at 24 h against JH2-2 at a concentration of 2 microgram/ml and against BM4316 and BM4275 at a concentration of 30 microgram/ml. The combination of LY333328 and gentamicin (4 microgram/ml) was synergistic and bactericidal after 24 h of incubation against the three strains at LY333328 concentrations of 2 microgram/ml for JH2-2 and 8 microgram/ml for BM4275 and BM4316. The combination of LY333328 and gentamicin was the only regimen demonstrating in vitro bactericidal activity against BM4316. In vivo, intravenous treatment with LY333328 alone, providing peak and trough serum levels of 83.3 +/- 1.3 and 3.8 +/- 0.2 microgram/ml, respectively, was inactive against BM4316 and BM4275 and selected mutants resistant to LY333328 in half of the rabbits infected with the VanA-type strain (MICs, 8 to 20 microgram/ml). However, the LY333328-gentamicin combination was active against the three strains and prevented the emergence of mutants resistant to both components of the combination. We conclude that the LY333328-gentamicin combination might be of interest for the treatment of enterococcal infections, particularly against VanA-type strains.

Activity and diffusion of LY333328 in experimental endocarditis due to vancomycin-resistant Enterococcus faecalis

The activity of LY333328 against Enterococcus faecalis JH2-2, which is susceptible to glycopeptides, and against its transconjugants E. faecalis BM4281 and BM4316, with VanB and VanA phenotypes, respectively, was investigated. LY333328 was active in vitro against the three strains, for which MICs were 2 microg/ml on agar and 0.25 microg/ml in broth. LY333328 was bactericidal in broth against E. faecalis JH2-2 and BM4281 at a concentration of 8 microg/ml and against BM4316 at a concentration of 30 microg/ml. The protein binding of LY333328 to rabbit serum was >99%, and the bactericidal activity of LY333328 in broth was reduced when it was tested in the presence of 90% rabbit serum. Autoradiographic studies performed in rabbits with enterococcal endocarditis showed that 14[C]LY333328 was distributed heterogeneously throughout cardiac vegetations. In rabbits with aortic endocarditis, a regimen of 20 mg of LY333328 per kg of body weight administered intramuscularly twice a day for 5 days after a loading dose of 40 mg/kg was active against the three strains in vivo (P < 0.01), whereas vancomycin was not active against the VanB-type strain and teicoplanin was not active against the VanA-type strain. We conclude that the activity of LY333328 is not significantly modified by acquired resistance to glycopeptides in E. faecalis either in vitro or in experimental endocarditis.

A new model of experimental prosthetic joint infection due to methicillin-resistant Staphylococcus aureus: a microbiologic, histopathologic, and magnetic resonance imaging characterization

Partial knee arthroplasty was done in rabbits with a silicone-elastomer implant. Immediately after closing the surgical wound, 5 x 10(6) cfu of methicillin-resistant Staphylococcus aureus was injected into the joint. Disease evolution was studied at different stages of infection up to 8 weeks. Prosthetic infection developed in all animals. Gross pathology and histopathologic changes were characteristic of joint and bone infection. Quantitative bacterial counts from infected bone confirmed disease chronicity. The mean number of colony-forming units per gram of bone +/- SD 1 week after infection was 4.84 +/- 0.24 log10 cfu/g and remained stable from week 1 to week 8. Magnetic resonance imaging showed evidence of prosthetic infection as of week 1, while only mild radiologic changes of bone were seen 2 weeks after infection. This model produces a prosthetic infection that is reproducible and close to that of human prosthetic infection.

Influence of the pre-treatment duration of infection on the efficacies of various antibiotic regimens in experimental streptococcal endocarditis

The influence of the pre-treatment duration of infection on the efficacies of three different antibiotic regimens was investigated in a rabbit model of subacute endocarditis caused by a novel, nutritionally-variant species, Streptococcus adjacens strain GaDT. Treatment was initiated either 6 or 10 days after bacterial inoculation (days 7 and 11 respectively) and comprised procaine penicillin (150,000 IU/kg bd), alone or combined with tobramycin (12 mg/kg od), teicoplanin (10 mg/kg bd), all administered by the intramuscular route for 4 days. The MICs and MBCs of penicillin, tobramycin and teicoplanin were 0.015 and 1 mg/L, 8 and 16 mg/L and 0.25 and 256 mg/L respectively. In the control rabbits, the mean (+/- S.D.) weights of the vegetations were 25 +/- 16 mg on day 7 and 45 +/- 34 mg on day 11 (P = 0.06). The mean (+/- S.D.) reductions in the number of cfu in the vegetations of the treated groups of animals after completion of therapy which had been started on days 7 and 11, compared with the mean numbers of cfu in the vegetations of the untreated controls on days 7 and 11 (delta log10 cfu/g), were 4.0 +/- 1.3 and 2.1 +/- 1.5 respectively for penicillin (P < 0.05), 3.2 +/- 1.8 and 2.4 +/- 1.8 respectively for teicoplanin and 5.4 +/- 1.2 and 5.2 +/- 1.2 respectively for the combination of penicillin and tobramycin. The increase in the size of the vegetations and changes in the metabolism of the bacteria within the vegetations between days 7 and 11, as demonstrated by electron microscopy, might explain why penicillin was more effective earlier in the course of the disease and why the influence of the duration of infection before treatment was initiated, varied according to the antibiotic regimen. These results suggest that the use of bactericidal regimens, such as the combination of penicillin and tobramycin, which are equally effective in reducing the bacterial counts in vegetations which have been infected for both long and short periods could minimize the risk of relapse in patients with endocarditis in whom there have been long delays before initiating treatment and/or who have large vegetations.

Efficacy and ocular penetration of sparfloxacin in experimental streptococcal endophthalmitis

Gram-positive cocci are the most common pathogens in severe human eye infections. Streptococcal endophthalmitis is a devastating infection, and intravitreal antibiotic therapy is limited by retinal toxicity. Because few systemic antistreptococcal antibiotics penetrate into the vitreous, sparfloxacin, a newer quinolone with improved antistreptococcal activity, might be of interest. We therefore assessed its efficacy by the intravitreal route in a rabbit model of streptococcal endophthalmitis. The vitreal bacterial count (mean +/- standard deviation log10 CFU per milliliter) was significantly reduced after an intravitreal injection of 800 micrograms of sprafloxacin (4.9 +/- 0.7) relative to the counts in untreated control (7.1 +/- 0.7) and pefloxacin-treated (7.8 +/- 1.2) eyes. After systemic administration to rabbits, the maximum concentration of sparfloxacin in serum was 5.6 micrograms.ml-1 and the half-life was 7.5 h. Sparfloxacin exhibited very good penetration ratios in the vitreous (54%), cornea (76%), and lens (36%). In the vitreous, the levels of sparfloxacin remained greater than the MICs for most gram-positive cocci for up to 18 h. Further experimental studies are warranted to determine the efficacy of systemic sparfloxacin as adjuvant therapy in the treatment of human endophthalmitis.

Daptomycin or teicoplanin in combination with gentamicin for treatment of experimental endocarditis due to a highly glycopeptide-resistant isolate of Enterococcus faecium

Using an experimental endocarditis model, we studied the activity of daptomycin used alone or in combination with gentamicin against an Enterococcus faecium strain that was highly resistant to glycopeptides and susceptible to gentamicin. In vitro, the MIC of daptomycin was 1 micrograms/ml. In vivo, daptomycin appeared to be effective only when it was used in a high-dose regimen, i.e., 12 mg/kg of body weight every 8 h (-2.5 log10 CFU/g versus controls; P < 0.05), particularly when it was combined with gentamicin (-5.0 log10 CFU/g versus controls; P < 0.01). Since the distribution of daptomycin into cardiac vegetations, as evaluated by autoradiography, appeared to be homogeneous, the poor in vivo activity of daptomycin was considered to be related to its high degree of protein binding, as suggested by killing curves studies. Since the MIC of teicoplanin for the vancomycin-resistant E. faecium strain used in the study was only 64 micrograms/ml and since an in vitro synergy between teicoplanin at high dose and gentamicin was observed, a high-dose regimen of teicoplanin, i.e., 40 mg/kg every 12 h, was also assessed in vivo. This treatment provided marginal activity only when it was combined with gentamicin (-2.3 log10 CFU/g versus controls; P < 0.05). These results suggest that the levels of daptomycin or teicoplanin in serum required to cure experimental endocarditis caused by a highly glycopeptide-resistant strain of E. faecium would not be achievable in humans.

Optimal aminoglycoside dosing regimen for penicillin-tobramycin synergism in experimental Streptococcus adjacens endocarditis

The combination of penicillin and aminoglycoside is the recommended therapy for endocarditis caused by nutritionally variant streptococci (NVS). However, the optimal aminoglycoside dosing regimen remains controversial. We compared the efficacies of four regimens of tobramycin alone or combined with procaine penicillin in the therapy of rabbits with endocarditis caused by Streptococcus adjacens, a new species of NVS. Animals were injected intramuscularly for 4 days with procaine penicillin (150,000 U/kg of body weight twice daily) or tobramycin at a low dose (3 mg/kg every 24 h) or a high dose (12 mg/kg every 24 h) either once or three times daily (t.i.d.) alone or in combination with procaine penicillin. Additional groups of animals were treated with the combination regimens for a shorter period of time (2 days) in order to demonstrate a possible difference in the rapidity of efficacy between the regimens. The MICs and MBCs were 0.015 and 1 micrograms/ml and 8 and 16 micrograms/ml for penicillin and tobramycin, respectively. The mean peak tobramycin levels in plasma were 2.4 +/- 1.3 (1 mg/kg t.i.d.), 5.4 +/- 3.7 (4 mg/kg t.i.d.), and 25 +/- 9.3 (12 mg/kg once daily). The mean penicillin levels in serum were always above the MIC. In vitro kill curves plotted at the time that peak concentrations were reached in plasma showed a concentration-dependent killing effect of tobramycin alone but not in combination with penicillin. In vivo, low-dose tobramycin was significantly less effective than the high dose. Results for the combinations of the different dosing regimens of tobramycin with procaine penicillin were not significantly different. Our results suggest that (i) against susceptible strains of streptococci, aminoglycoside alone exhibits a concentration-dependent killing effect both in vitro and in vivo; (ii) against NVS strains, combinations of penicillin and high- or low-dose tobramycin are equally effective; and (iii) aminoglycoside given once daily or at a low dose t.i.d. with penicillin could be a cost-effective alternative with reduced toxic risk for patients with NVS endocarditis when the bacteria are susceptible to the killing activities of both compounds.

Efficacy of temafloxacin in experimental Streptococcus adjacens endocarditis and autoradiographic diffusion pattern of [14C]temafloxacin in cardiac vegetations

Temafloxacin, a new fluoroquinolone, alone or in combination with tobramycin, was compared with penicillin, tobramycin, and their combination in the therapy of rabbits with endocarditis caused by Streptococcus adjacens GaDT, a new species of nutritionally variant streptococci. Animals were injected intramuscularly for 4 days with temafloxacin (50 mg/kg of body weight twice daily [b.i.d.]) alone or combined with tobramycin (12 mg/kg once daily), with procaine penicillin (150,000 U/kg b.i.d.) alone or combined with tobramycin (12 mg/kg once daily), or with tobramycin (12 mg/kg once daily) alone. Another group of animals was treated with a higher dose of temafloxacin (100 mg/kg b.i.d.). Temafloxacin, penicillin, and tobramycin MICs and MBCs were 1 and 2, 0.015 and 1, and 8 and 16 micrograms/ml, respectively. Time-kill curves showed that the addition of tobramycin to penicillin or temafloxacin increased the killing rate. In vivo, treatment with temafloxacin (50 and 100 mg/kg b.i.d.) alone reduced the bacterial counts in vegetations (3.9 +/- 0.9 and 3.1 +/- 0.8 log10 CFU/g of vegetation) compared with those in the vegetations of control animals (7.5 +/- 0.9 log10 CFU/g of vegetation). This result was similar to that obtained with penicillin alone (4.5 +/- 0.8 log10 CFU/g of vegetation). The combination of temafloxacin (50 mg/kg) and tobramycin was as effective as penicillin plus tobramycin (2.5 +/- 0.3 versus 2.3 +/- 0.4 log10 CFU/g of vegetation, respectively). The autoradiographic pattern of [14C]temafloxacin diffusion into infected cardiac vegetations was studied. Thirty minutes after the end of infusion of 250 microCi of [14C]temafloxacin, the [14C]temafloxacin was homogeneously distributed throughout the vegetations. These data support further evaluation of quinolones in experimental endocarditis.

Piperacillin, tazobactam, and gentamicin alone or combined in an endocarditis model of infection by a TEM-3-producing strain of Klebsiella pneumoniae or its susceptible variant

The efficacy of tazobactam, a beta-lactamase inhibitor, in combination with piperacillin, was studied in vitro and in rabbit experimental endocarditis due to a Klebsiella pneumoniae strain (KpR) producing an extended-spectrum beta-lactamase, TEM-3, or its nonproducing variant (KpS). In vitro, piperacillin was active against KpS (MIC = 4 micrograms/ml, MBC = 8 micrograms/ml with 10(7)-CFU/ml inoculum) but not against KpR (MIC = MBC = 256 micrograms/ml). Tazobactam (1 microgram/ml) restored the activity of piperacillin against KpR (MIC = 2 micrograms/ml, MBC = 4 micrograms/ml). Gentamicin was active against both strains (MIC = 0.25 and 0.5 micrograms/ml for KpS and KpR, respectively). The piperacillin-tazobactam-gentamicin combination was synergistic in vitro. The piperacillin/tazobactam ratio in plasma and in vegetations was always lower than the 4/1 injected dose ratio. In vivo, piperacillin (300 mg/kg of body weight four times a day [QID]) was active against KpS but not against KpR. Tazobactam (75 mg/kg QID) was able to restore the in vivo effect of piperacillin (300 mg/kg QID) against KpR (-3.0 log10 CFU/g of vegetation versus that of controls). Gentamicin (4 mg/kg twice a day [BID]) was active against both strains. Compared with controls, the combination of gentamicin plus piperacillin against KpS (-5.6 log10 CFU/g of vegetation), and the gentamicin-piperacillin-tazobactam combination against KpR (-4.4 log10 CFU/g of vegetation) achieved the greatest decrease in bacterial counts in vegetations and were the only regimens that significantly increased the proportion of sterile vegetations. It is concluded that (i) tazobactam was able to restore the effect of piperacillin against a TEM-3 extended-spectrum Beta-lactamase-producing strain of K. pneumoniae, both in vitro and in a severe experimental infection with high inoculum, when used in a 4/1 piperacillin/tazobactam dose ratio; (ii) gentamicin alone was effective because of the high peak/MBC ratio in plasma; (iii) piperacillin-tazobactam-gentamicin, probably because of the effect of gentamicin in reducing bacterial inoculum in vivo, as stressed by the results obtained by piperacillin-gentamicin against KpS, may be the most effective regimen against KpR.