Efficacy of Teicoplanin and Autoradiographic Diffusion Pattern of [14C ...

ANTIMICROBIAL AGENTS AND CHEMOTHERAPY, Nov. 1998, p. 2830–2835 0066-4804/98/$04.0010 Copyright © 1998, American Society for Microbiology. All Rights Reserved.

Vol. 42, No. 11

Efficacy of Teicoplanin and Autoradiographic Diffusion Pattern of [14C]Teicoplanin in Experimental Staphylococcus aureus Infection of Joint Prostheses ´ MIEUX,1* RE ´ MY BLETON,1 FARID ISMAEL,1 AZZAM SALEH MGHIR,1 ANNE CLAUDE CRE 2 1 MICHEL MANTEAU, SOPHIE DAUTREY, LAURENT MASSIAS,1 LOUIS GARRY,1 ` RE,2 AND CLAUDE CARBON1 NICOLE SALES,2 BERNARD MAZIE Ho ˆpital Bichat Claude-Bernard, Institut National de la Sante´ et de la Recherche Me´dicale, Unite´ 13, Paris,1 and Service Hospitalier Fre´de´ric-Joliot, Commissariat a ` l’Energie Atomique, Orsay,2 France Received 28 April 1998/Returned for modification 11 June 1998/Accepted 24 August 1998

Prosthesis infections are difficult to cure. Infection with methicillin-resistant staphylococci is becoming more common in patients with orthopedic implants. Using a recently developed model of methicillin-resistant Staphylococcus aureus (MRSA) infection of a knee prosthesis, we compared the efficacies of teicoplanin and vancomycin. [14C]teicoplanin diffusion in this model was also studied by autoradiography. A partial knee replacement was performed with a silicone implant fitting into the intramedullary canal of the tibia, and 107 CFU of MRSA was injected into the knee. Treatment with teicoplanin or vancomycin (20 or 60 mg/kg of body weight, respectively, given intramuscularly twice daily) was started 7 days after inoculation and was continued for 7 days. The teicoplanin and vancomycin MICs for MRSA were 1 mg/ml. Mean peak and trough levels in serum were 39.1 and 23.5 mg/ml, respectively, for teicoplanin and 34.4 and 18.5 mg/ml, respectively, for vancomycin. Fifteen days after the end of therapy, the animals were killed and their tibias were removed, pulverized, and quantitatively cultured. Teicoplanin and vancomycin significantly reduced (P < 0.05) the bacterial density (2.7 6 1.3 and 3.3 6 1.6 log10 CFU/g of bone, respectively) compared to those for the controls (5.04 6 1.4 log10 CFU/g of bone). The bacterial covents of teicoplanin- and vancomycin-treated rabbits were comparable. The [14C]teicoplanin autoradiographic diffusion patterns in rabbits with prostheses, two of which were uninfected and two of which were infected, were studied 15 days after inoculation. Sixty minutes after the end of an infusion of 250 mCi of [14C]teicoplanin, autoradiography showed that in the infected animals, the highest levels of radioactivity were located around the prosthesis and in the periosteum, bone marrow, and trabecular bone. Radioactivity was less intense in epiphyseal disk cartilage, femoral cartilage, articular ligaments, and muscles and was weak in compact bone. A similar distribution pattern was seen in uninfected rabbits. Thus, teicoplanin may represent an effective alternative therapy for the treatment of these infections. ministered as outpatient i.v. therapy for chronic osteomyelitis (8) or prosthetic joint infections (5). However, the efficacy of teicoplanin compared with that of vancomycin in this setting remains unknown. Because of the wide variability of factors conditioning the response to therapy, it is difficult to compare treatment results from the different small series published in the literature. Furthermore, in this context, prospective randomized trials are difficult to design. Experimental studies comparing the efficacies of various antibiotics for the therapy of prosthetic joint infection are needed. Using the recently described model of experimental methicillin-resistant S. aureus (MRSA) infection of a knee prosthesis that closely mimics the infection in humans (1), we compared the efficacies of teicoplanin and vancomycin. We also studied the autoradiographic diffusion patterns of [14C]teicoplanin in this model in an attempt to correlate the in vivo efficacy with the bone diffusion pattern. (This work was presented in part at the 36th Interscience Conference on Antimicrobial Agents and Chemotherapy, New Orleans, La., 15 to 18 September 1996 [18].)

Prosthetic joint infection is a severe complication of total joint arthroplasty. Perioperative contamination is responsible for most infections, which are mainly due to Staphylococcus aureus or Staphylococcus epidermidis (16). Surgical debridement, removal of all foreign material, and prolonged high-dose intravenous (i.v.) and oral antimicrobial therapy are necessary for successful treatment (2). Infection with methicillin-resistant staphylococci is increasingly common among patients with implanted orthopedic devices (9). These organisms are often resistant to many of the other commonly used antibiotics. Glycopeptide antibiotics are used as the first-line therapy and, because resistant to oral antimicrobial agents, such as a fluoroquinolone or rifampin, is frequent, parenteral treatment must be administered for 4 to 6 weeks (17). To avoid prolonged hospitalization solely for the administration of parenteral antibiotics, home i.v. infusion therapy is being used more and more frequently (3). Teicoplanin offers several advantages over vancomycin for home therapy: it can be administered by i.v. bolus injection once daily and does not require a central venous catheter (5). Several open studies have shown that it is effective for the treatment of bone and joint infections and can be safely ad-

MATERIALS AND METHODS Test strain. An MRSA strain was used in this study. The strain was isolated from a patient with an infected knee prosthesis treated at Bichat Hospital, Paris, France. Its virulence was maintained by intraperitoneal injection into mice. In vitro antibiotic susceptibility testing. The MICs were determined in Mueller-Hinton broth (Diagnostics Pasteur, Marnes-la-Coquette, France) by the tube

* Corresponding author. Mailing address: Ho ˆpital Bichat ClaudeBernard, 46, rue Henri-Huchard, 75877 Paris Cedex 18, France. Phone: 33 (1) 40 25 87 00. Fax: 33 (1) 40 25 88 45. E-mail: u13bcb @magic.fr. 2830

VOL. 42, 1998

TEICOPLANIN IN EXPERIMENTAL JOINT PROSTHESIS INFECTION

macrodilution method (13). The antibiotics tested were teicoplanin (Marion Merrell S.A., Levallois-Perret, France) and vancomycin (Lilly France S.A., SaintCloud, France). Each tube contained twofold dilutions of the antibiotic and a final bacterial inoculum of 106 CFU/ml. The tubes were incubated for 18 h at 37°C. The MIC was defined as the lowest concentration of antibiotic that prevented turbidity in the test tube after incubation. The minimum bactericidal concentration (MBC) was defined as the lowest concentration of antibiotic that killed at least 99.9% of the organisms after incubation, as determined by plating 0.1 ml from each clear broth tube onto Trypticase soy agar and incubating the plates for 18 h at 37°C. Experimental prosthesis infection. Thirty-one New Zealand White rabbits, each weighing between 2.5 and 3 kg, were used. They were housed in individual cages with a natural light-dark cycle. The experimental protocol was in keeping with French legislation on animal experimentation. This model has been described in detail elsewhere (1). Briefly, the rabbit underwent a partial right knee replacement with a tibial component performed by an orthopedic surgeon. The operation was carried out under general anesthesia induced by intramuscular (i.m.) injection of ketamine (25 mg/kg of body weight) and then by continuous inhalation of 1% isoflurane. A silicone-elastomer implant, commonly used in arthroplasty of the first metatarsophalangeal joint (Silastic, Great toe implant HP; Swanson Design, Dow-Corning, provided by Ortho Technique, Cre´teil, France) was implanted as a tibial prosthetic component. The stem of the nail-shaped silicone implant (14 mm long) was inserted into the intramedullary canal of the tibia, with the implant head (15 mm in diameter and 5 mm high) replacing the tibia plateaus. The skin overlying the right leg was shaved 24 h before the operation. Prior to surgery, the skin was cleaned with an iodine solution. A longitudinal skin incision was made and the knee joint was exposed. After dislocation of the tibia, the epiphyseal plates were removed. The metaphysis was exposed, the cancellous bone of the medullary cavity of the proximal metaphysis was reamed (15 mm) and filled with the prosthesis, and then the deep fascia and the skin were closed. Immediately after surgery, animals were inoculated with of 5 3 107 CFU of MRSA in 0.5 ml, which was injected into the knee close to the prosthesis. Each rabbit was given oral acetaminophen (80 mg/day) as postoperative analgesia for 2 days. Therapeutic studies. (i) Treatment and evaluation of therapy. Seven days after inoculation, the following i.m. injections were started: teicoplanin (20 mg/kg of body weight twice daily [b.i.d.]), and vancomycin (60 mg/kg b.i.d.). Each regimen was administered for 7 days. The animals were killed by i.v. injection of pentobarbital 2 weeks after the end of therapy (day 28) in order to take into account regrowth of the bacteria after the end of therapy while avoiding the persistence of residual antibiotic in bone. Untreated control rabbits were also killed on day 28. At this time, the right hind leg was dissected and the tibia and the femur were separated from the surrounding soft tissues. A smear of the prosthesis was made on a blood-agar plate. For quantitative bacterial counts, the upper third (length, 3 cm) of the tibia including compact bone and marrow was isolated, split with a bone crusher, weighed, and then cut into little pieces, frozen in liquid nitrogen, and crushed in an autopulverizer (Spex 6700, Freezer/Mill Industries Inc., Metechen, N.J.). The pulverized bone was suspended in 10 ml of sterile saline, and serial dilutions were made and plated on Trypticase soy agar. After overnight incubation at 37°C, the number of viable organisms was determined. The results are expressed as mean 6 standard deviation (SD) log10 CFU per gram of bone. (ii) Serum antibiotic levels. Antibiotic concentrations in the sera of uninfected rabbits were determined. An i.m. injection of teicoplanin (20 mg/kg) or vancomycin (60 mg/kg) was administered to eight rabbits (four in each group). Blood was drawn 1 and 12 h later. Teicoplanin concentrations were measured by high-performance liquid chromatography with UV detection (l 5 224 nm) (10). Vancomycin concentrations were determined by a fluorescence polarization immunoassay (TDX; Abbott, Rungis, France). The lower limits of detection were 2 mg/ml for teicoplanin and 0.6 mg/ml for vancomycin. Quantitative autoradiography of [14C]teicoplanin diffusion into infected bone. Fifteen days after inoculation, 250 mCi of [14C]teicoplanin was injected i.v. over 30 min into two infected rabbits and two uninfected rabbits. The rabbits were killed 60 min (one infected rabbit and one uninfected rabbit) or 180 min (one infected rabbit and one uninfected rabbit) after the end of the infusion. After skinning, both hind limbs of each animal were frozen in liquid nitrogen and were stored at 270°C. Sections (50 mm thick) of the limb (including the prosthesis, the bones, the ligaments, and part of the surrounding muscles) were cut at 225°C on an LKB macrocryostat and were collected on adhesive tape. Autoradiographic films (Hyper film b-Max; Amersham, Les Ulis, France) that had been exposed to the leg sections for 6 weeks were developed and autoradiographic images were quantified. For each hind limb, four different sections were quantified by computed densitometry (Autorad; Imstar, Paris, France). The results are expressed as the mean 6 SD concentration of radioactivity in tissue/ concentration of radioactivity in blood. The radioactivity level in blood was determined in the heart ventricles of the animals by the same autoradiographic imaging technique. Statistics. Mean bacterial densities in bone for the various experimental groups were compared by an analysis of variance, followed by Scheffe’s test for multiple comparisons. Results are expressed as means 6 SDs. A P value of ,0.05 was considered significant.

2831

TABLE 1. Effect of antibiotic treatment on experimental MRSA infection of prosthetic knee in rabbits Treatmenta

No. of rabbits

No. of rabbits with sterile bone

log10 CFU/g of bone (mean 6 SD)

None Teicoplanin Vancomycin

8 11 12

0 4 1

5.0 6 1.4 2.7 6 1.3b 3.3 6 1.6b

a Rabbits were treated i.m. for 7 days with teicoplanin (20 mg/kg b.i.d.) or vancomycin (60 mg/kg b.i.d.). b Significantly different from untreated controls (P , 0.05).

RESULTS In vitro studies. The MICs of teicoplanin and vancomycin were 1 mg/ml. The MBCs of teicoplanin and vancomycin were 2 mg/ml. Serum antibiotic levels in uninfected animals. Injection of teicoplanin (20 mg/kg i.m.) into uninfected animals resulted in mean 6 SD peak and trough concentrations in serum of 39.07 6 3.68 and 23.45 6 3.31 mg/ml, respectively, which were obtained 1 and 12 h after injection, respectively. At the same times after an injection of vancomycin (60 mg/kg i.m.), the mean 6 SD peak and trough concentrations in serum were 34.40 6 14.25 and 18.47 6 4.39 mg/ml, respectively. Therapeutic studies. All experimental animals were infected, exhibiting positive prosthesis smear cultures with a mean 6 SD bacterial count of 5.0 6 1.4 log10 CFU/g of bone (Table 1). In the teicoplanin-treated group, 4 of 11 animals had sterile bone; their mean bone bacterial density was significantly lower than that for control animals (P , 0.05). In the group of animals treated with vancomycin, 1 of 12 animals had sterile bone and the bacterial count in bone was significantly different from that in control animals (P , 0.05). The mean bacterial counts in the teicoplanin- and vancomycin-treated animals were not significantly different. No teicoplanin- or vancomycinresistant strain emerged in the bones of treated animals. [14C]teicoplanin diffusion. The distributions of [14C]teicoplanin 60 min after the end of an infusion given 15 days after inoculation were visualized and quantified (Fig. 1A and B and 2A,, respectively). [14C]teicoplanin diffusion into the different tissues of the left leg (without a prosthesis) was similar to that in the right leg (with a prosthesis). The highest levels of radioactivity were detected around the prosthesis, in the periosteum, and in the bone marrow and trabecular bone; these levels were slightly higher or in the same range as those in blood. Radioactivity was less intense in epiphyseal disk cartilage, femoral cartilage, muscles, and ligaments. Cortical bone was practically devoid of radioactivity. Autoradiographs obtained 3 h after the end of [14C]teicoplanin injection did not differ from those obtained 60 min after the end of an infusion (data not shown). The teicoplanin distributions in uninfected animals (Fig. 1C and D and 2B) were similar to those observed in infected animals. DISCUSSION Prosthetic joint infection is an infrequent but severe complication of total joint arthroplasty. Several questions remain unsolved, such as the optimal therapeutic antibiotic regimen for cure of these infections. In fact, to date there have been no adequately designed, randomized, controlled prospective trials with sufficient numbers of patients and long-term follow-up to guide medical therapy. Because a number of factors can influence treatment outcome, such as the age of the prosthesis, the

2832

SALEH MGHIR ET AL.

ANTIMICROB. AGENTS CHEMOTHER.

FIG. 1. Autoradiography of the right knee (with prosthesis) (A and C) and the left knee (B and D) taken from [14C]teicoplanin-treated rabbits 60 min after the end of i.v. infusion 15 days after injection of MRSA (A and B) or saline (C and D). Radiotracer disintegration blackens the film, with the gradient of darkness indicating the concentration: the blacker the image, the higher the concentration of radioactivity. Magnifications, 31.5.

duration of infection, the organism responsible for the infection, the quality of the surgical debridement, and the surgical procedure, it is probable that most of the information used to guide the choice of antibiotics will continue to rely upon in vitro and experimental data. Our experimental model reproduces a prosthetic knee infection close to that observed in the human situation and is suitable for comparative evaluation of antibiotic regimens (4). MRSA is a common cause of postoperative prosthetic joint infection. These organisms represent a difficult therapeutic

challenge because they are resistant to other antibiotics commonly used in bone infections, for example, fluoroquinolones, which can be given orally for several weeks in an outpatient setting (7). MRSA strains are also often resistant to rifampin, which has been shown to be, in combination, the most effective antibiotic regimen for the treatment of experimental osteomyelitis (15) or experimental foreign body infection (19). In 1995, 96% of MRSA strains isolated in Bichat ClaudeBernard Hospital were resistant to fluoroquinolones and 68.5% were resistant to rifampin (11).

VOL. 42, 1998


2833

FIG. 1—Continued.

Glycopeptide antibiotics, like vancomycin and teicoplanin, were poorly effective as monotherapy in an experimental model of S. aureus osteomyelitis (15) or in experimental tissue cage infection (20). However, in many instances they appear to

be the only available agents that can be used, as demonstrated by the multidrug resistance pattern observed in vitro. Teicoplanin offers advantages over vancomycin, especially its easier administration. Indeed, it can be given once daily in an i.v.

2834

SALEH MGHIR ET AL.

ANTIMICROB. AGENTS CHEMOTHER.

FIG. 2. Mean ratio of concentration of radioactivity in tissue/concentration of radioactivity in blood in the right leg (prosthesis side) and the left leg (control side) of a rabbit 15 days after MRSA inoculation (A) or saline injection (B).

pulse. Although some clinical studies suggest that it is effective in the treatment of bone and joint infections (5, 8), at present, no data comparing the efficacies of vancomycin and teicoplanin in prosthetic joint infection have been reported. In our experimental model, both regimens were shown to be equally effective at significantly reducing the numbers of CFU in bone compared to those in the bones of controls. With teicoplanin, 4 of 11 animals had sterile bone at the end of therapy. The doses used were chosen to obtain mean trough concentrations of about 15 mg/ml for vancomycin and 20 mg/ml for teicoplanin. In the experimental model of osteomyelitis described by Norden et al. (14), in spite of the high doses used, neither teicoplanin nor vancomycin was able to sterilize bone. The model used in this study differs from the rabbit model of chronic osteomyelitis described by Norden et al. (14). We induced an essentially periprosthetic infection that primarily involved the joint and adjacent bone marrow (7 days postinoculation). This situation is similar to human prosthetic joint infection: although the septic process around a total joint prosthesis can involve bone, it is not usually associated with extensive bone destruction, except when reintervention is contraindicated and prolonged suppressive antimicrobial therapy is administered.

In the rabbit model of osteomyelitis, on day 14, when therapy began, the bone lesions were more progressive (advanced) and thus could be more difficult to eradicate. Thus, the results obtained with chronic osteomyelitis models can hardly be extrapolated to joint prosthesis infections (6). The different results obtained with the experimental models can also be explained by the [14C]teicoplanin diffusion into the infected tissue observed by autoradiography. The [14C]teicoplanin concentrations in the artificial joint space in contact with the prosthesis and in the adjacent bone marrow were at least equal to or higher than those in blood. The radioactivity level was also suprisingly high in the periosteum. However, as observed with other antibiotics (4), teicoplanin penetrated very poorly into compact bone. This diffusion pattern, obtained 60 min after the end of infusion, was not modified 180 min after infusion, arguing against a slow diffusion process. The high [14C]teicoplanin concentration in the artificial joint space in contact with the prosthesis was observed in infected and uninfected rabbits and thus reflects the inflammatory reaction due to the presence of the prosthesis and not to the presence of bacteria. Teicoplanin has been shown to penetrate into human neutrophils (12), and thus, its diffusion could be favored by the flow of cells into the inflammatory site.

VOL. 42, 1998


Finally, the heterogenous diffusion of teicoplanin into the different infected tissues could explain the difficulty in interpreting bone drug levels assessed in homogenates by conventional methods (15), for example, the discordance observed between the teicoplanin levels in diseased bone, which were eightfold higher than the MIC for the infecting strain of S. aureus, and the inefficacy of the antibiotic in the model of chronic osteomyelitis (14). In conclusion, the autoradiographic analysis of [14C]teicoplanin diffusion demonstrated the antibiotic’s good penetration into the pus surrounding the prosthesis and the different components of bone. Moreover, teicoplanin appears to be as effective as vancomycin in our experimental joint prosthesis model and thus could be an attractive alternative therapy, especially when considering the possibility of home i.v. therapy.

12.

ACKNOWLEDGMENTS

13.

This work was supported by a grant from Laboratoires Roussel Diamant, Puteaux, France. We thank Wright Medical Technology Inc., Arlington, Tenn., and Ortho Technique, Creteil, France, for kindly providing the silastic toe implants.

14.

7. 8. 9. 10. 11.

15. REFERENCES 1. Belmatoug, N., A. C. Cre´mieux, A. Volk, R. Bleton, A. Saleh-Mghir, M. Grossin, L. Garry, and C. Carbon. 1996. A new model of experimental prosthesis joint infection due to methicillin-resistant Staphylococcus aureus: a microbiologic, histopathologic and magnetic resonance imaging characterization. J. Infect. Dis. 174:414–417. 2. Brandt, C. M., W. W. Sistrunk, M. C. Duffy, A. D. Hanssen, J. M. Steckelberg, D. M. Hstrup, and D. R. Osmon. 1997. Staphylococcus aureus prosthetic joint infection treated with debridement and prosthesis retention. Clin. Infect. Dis. 24:914–919. 3. Brown, R. R. 1991. Selection and training of patients for out patient intravenous antibiotic therapy. Rev. Infect. Dis. 13(Suppl. 2):S147–S151. 4. Cre´mieux, A. C., A. Saleh-Mghir, R. Bleton, M. Manteau, N. Belmatoug, L. Massias, L. Garry, N. Sales, B. Mazie`re, and C. Carbon. 1996. Eficacy of sparfloxacin and autoradiographic pattern of diffusion of 14C-sparfloxacin in an experimental Staphylococcus aureus joint prothesis infection. Antimicrob. Agents Chemother. 40:2111–2116. 5. Davey, P. G., D. R. Rowley, and G. A. Phillips. 1992. Teicoplanin home therapy for prosthetic joint infections. Eur. J. Surg. Suppl. 567:23–25. 6. Fitzgerald, R. H. 1983. Experimental osteomyelitis: description of a canine

16. 17.

18.

19. 20.

2835

model and the role of depot administration of antibiotics in the prevention and treatment of sepsis. J. Bone Joint Surg. 65A:371–380. Gentry, L. O. 1991. Oral antimicrobial therapy for osteomyelitis. Ann. Intern. Med. 114:986–987. Graninger, W., C. Wenisch, E. Wiesiger, M. Menschik, J. Karimi, and E. Persterl. 1995. Experience with outpatient intravenous teicoplanin therapy for chronic osteomyelitis. Eur. J. Clin. Microbiol. Infect. Dis. 14:643–647. James, P. J., L. A. Butcher, E. R. Cardner, and D. Hamblen. 1994. Methicillin-Resistant Staphylococcus epidermidis in infection of hip arthroplastics. J. Bone Joint Surg. Br. 76:725–727. Jehl, F. 1990. High-performance liquid chromatography of antibiotics. J. Chromatogr. 531:509–548. Joly-Guillou, M. L., D. Decre´, and P. Weber. 1996. Etat des re´sistance bacterienne `a l’ho ˆpital. Existe-t il des re´lations avec les re´sistances observee´s on ville, p. 7–23. In M. Wolff, A. C. Cre´mieux, C. Carbon, F. Vachon, J. P. Couland, and J. L. Vilde (ed.), Du bon usage des antibiotiques a` l’ho ˆpital. Arnette Blackwell, Paris, France. Maderazo, E. G., S. P. Breaux, L. C. Woronick, R. Quintiliani, and C. H. Nightingale. 1988. High teicoplanin uptake by neutrophils. Chemotherapy (Basel) 34:248–255. National Committee for Clinical Laboratory Standards. 1985. Methods for dilution antimicrobial susceptibility tests for bacteria that grow aerobically. Approved standard. Publication M7-A. National Committee for Clinical Laboratory Standards, Wayne, Pa. Norden, C. W., K. Niedereiter, and E. M. Shinners. 1986. Treatment of experimental chronic osteomyelitis due to Staphylococcus aureus with teicoplanin. Infection 14:136–138. Norden, C. W. 1988. Lessons learned from animal models of osteomyelitis. J. Infect. Dis. 40:103–110. Norden, C. W., J. D. Nelson, J. T. Mader, and G. B. Calandra. 1992. Evaluation of new anti-infection drugs for the treatment of infection of prosthetic hip joint. Clin. Infect. Dis. 15:S177–S181. Osmon, D. R., J. M. Steckelberg, M. P. Wilhelm, M. R. Kealing, R. C. Walken, A. D. Hanssen, and W. R. Wibes. 1993. Medical management of total knee arthroplasty infection, p. 377–392. In J. A. Rand (ed.), Total knee arthroplasty. Raven Press, New York, N.Y. Saleh-Mghir, A., A. C. Cre´mieux, M. Manteau, R. Bleton, L. Garry, N. Sales, B. Mazie`re, and C. Carbon. 1996. Autoradiographic pattern of diffusion of 14 C-teicoplanin and efficacy of teicoplanin in an experimental Staphylococcus aureus joint prothesis infection, abstr. A38, p. 8. In Program and abstracts of the 36th Interscience Conference on Antimicrobial Agents and Chemotherapy. American Society for Microbiology, Washington, D.C. Widmer, A. F., A. Gaechter, P. E. Ochsner, and W. Zimmerli. 1992. Antimicrobial treatment of orthopedic implant-related infections with rifampin combinations. Clin. Infect. Dis. 14:1251–1253. Zimmerli, W., R. Frei, A. F. Widmer, and Z. Rajacic. 1994. Microbiological tests to predict treatment outcome in experimental device-related infections due to Staphylococcus aureus. J. Antimicrob. Chemother. 33:959–967.