Background: The aim of this study was to determine the optimal formulation of antibiotic-loaded bone cement for knee periprosthetic joint infection. We used both in vitro and in vivo models incorporating various broad-spectrum antibiotics and tested their efficacy against gram-positive and gram-negative bacteria.
Methods: Bone cement specimens loaded with 4 g of either vancomycin or teicoplanin and 4 g of ceftazidime, imipenem, or aztreonam were studied to measure their in vitro antibiotic release characteristics and antibacterial capacities against methicillin-susceptible Staphylococcus aureus, methicillin-resistant Staphylococcus aureus, Staphylococcus epidermidis, Pseudomonas aeruginosa, and Escherichia coli. Bone cement spacers loaded with the antibiotics with the superior in vitro antibacterial capacity were then implanted into 8 patients (4 women and 4 men between 51 and 79 years of age) diagnosed with chronic knee periprosthetic joint infection. The antibiotic concentrations and antibacterial activities in the joint fluid at the site of the infection were measured following spacer implantation.
Results: Cement samples loaded with vancomycin and ceftazidime exhibited in vitro antibacterial activity against the test microorganisms that lasted for as long as or longer than that of cement loaded with the other antibiotic combinations. Joint fluid samples exhibited activity against bacteria including American Type Culture Collection (ATCC) strains and clinically isolated strains.
Conclusions: Bone cement loaded with vancomycin and ceftazidime provided broad-spectrum antibacterial capacity both in vitro and in vivo and was shown to be a potentially effective therapeutic measure in the treatment of knee periprosthetic joint infections.
Clinical Relevance: This study confirmed the potential effectiveness of drug delivery from bone cement spacers impregnated with vancomycin and ceftazidime.
Total joint arthroplasty is currently one of the most frequently performed and most successful surgical procedures in orthopaedics1,2. Infection, however, remains the most common cause of failure of total knee arthroplasty, with 25.2% of revisions due to infection3. In much of the world, 2-stage exchange arthroplasty is considered the gold-standard treatment of chronic periprosthetic joint infection, and it has become common to apply an antibiotic-loaded bone cement spacer during the first stage of the procedure4,5.
Gram-positive pathogens including Staphylococcus aureus (S. aureus) and Staphylococcus epidermidis (S. epidermidis) are the most common organisms implicated in periprosthetic joint infection6,7. However, gram-negative bacteria cause 6% to 23% of all periprosthetic joint infections7,8. Pseudomonas aeruginosa (P. aeruginosa) and Escherichia coli (E. coli) are the most common gram-negative bacteria in periprosthetic joint infections7,8. When using antibiotic-loaded bone cement to eradicate both gram-positive and gram-negative pathogens in periprosthetic joint infections, antibiotics with a broad spectrum of antibacterial activity are the most appropriate choice9.
In our study, we used an in vitro model to measure the antibiotic release characteristics of bone cement loaded with vancomycin and ceftazidime, vancomycin and imipenem, vancomycin and aztreonam, teicoplanin and ceftazidime, teicoplanin and imipenem, or teicoplanin and aztreonam as well as to compare their antibacterial activities against methicillin-susceptible S. aureus (MSSA), methicillin-resistant S. aureus (MRSA), S. epidermidis, P. aeruginosa, and E. coli. We then implanted bone-cement spacers, loaded with the antibiotic combination found to have the superior in vitro antibacterial capacity, into 8 patients with a knee periprosthetic joint infection. The antibiotic concentrations and antibacterial activities of joint fluid at the site of the infection were measured during the initial period following spacer implantation to determine the antibiotic combination with the best antibacterial capacity against periprosthetic joint infections.
Materials and Methods
In Vitro Study
Antibiotic-Loaded Bone Cement Specimens
Surgical Simplex bone cement (Stryker Orthopaedics) was tested after the addition of 4 g of vancomycin or teicoplanin and 4 g of ceftazidime, imipenem/cilastatin, or aztreonam. Hsieh et al. showed that 8 g of antibiotic powder with 40 g of cement polymer is the highest mixture ratio allowing bone cement to be introduced into the mold and formed into a spacer without difficulty4,5. The cement-antibiotic mixture was hand-mixed and then manually pressed into a plastic mold to form uniform test cylindrical specimens 20 mm in height and 15 mm in diameter.
Antibiotic Elution Assay
Every day, for a 60-day period, each cement cylinder was washed with 0.9% saline solution, immersed in a test tube filled with 5 mL of phosphate-buffered saline solution (1× PBS), and shaken in a rotator at 37°C. Daily antibiotic release was measured at 14 time points (days 0, 1, 2, 3, 4, 5, 6, 7, 10, 14, 21, 28, 40, and 60), and the cumulative release over the 60 days was calculated as well. The elution testing was conducted in triplicate.
Bioassay of Antibiotic Activity of in Vitro Samples
The biologic activity of the released antibiotics was studied using aliquots of the samples and a modified microtube dilution bioassay. The following strains were selected as test organisms: MSSA American Type Culture Collection (ATCC) 25923, MRSA ATCC 43300, S. epidermidis ATCC 14990, P. aeruginosa ATCC 27853, and E. coli ATCC 25922. In brief, the in vitro samples were inoculated with 105 colony forming units (CFUs) of bacteria per milliliter in 96-well culture dishes and incubated at 37°C for 24 hours. The growth of bacteria associated with the different concentrations of antibiotics was compared visibly with each other and against the positive control (without antibiotic)10. The minimum inhibitory concentration of each antibiotic against each bacterium was determined by microtube dilution bioassay (see Appendix).
In Vivo Study
The study protocol was approved by our institutional review board, and written informed consent was obtained from all patients prior to their participation in the study.
Between September 2013 and March 2014, patients diagnosed with a chronic knee periprosthetic joint infection were asked to participate in this study. All patients had evidence of clinical infection that met the new definition of periprosthetic joint infection according to the Musculoskeletal Infection Society11. Any patient who had a malignant tumor, had received immunosuppressive agents, or had a history of allergy to vancomycin or ceftazidime was excluded from the study. Eight patients (4 women and 4 men between 51 and 79 years of age) were enrolled. Their characteristics are summarized in Table I.
A bone cement spacer loaded with vancomycin and ceftazidime was implanted in the 8 patients. That combination of antibiotics was chosen for the in vivo study because its antibacterial activity against the test microorganisms had lasted for as long as or longer than that of the other antibiotic combinations in the in vitro study.
All bone cement spacers were made during surgery by thoroughly hand-mixing 4 g of vancomycin powder with 4 g of ceftazidime powder per 40-g pack of polymethylmethacrylate (PMMA) polymer (Fig. 1).
After implantation of the PMMA spacer, 10-mL aliquots of drainage were collected under sterile conditions every 24 hours until the daily drainage amount was <50 mL; 10-mL specimens of peripheral venous blood were also collected along with the first 24-hour drainage sample. The second surgery was carried out when the C-reactive protein level had returned to normal and the soft tissue lacked local heat, erythema, swelling, and any infection-related symptoms and signs.
Bioassay of Antibiotic Activity of Joint Fluid Samples
The antibiotic concentrations in the joint fluid were measured daily for 7 days using high-performance liquid chromatography. The bioactivity of joint fluid was determined using an agar disk diffusion bioassay. Discs (PDM Diagnostic Disks; AB Biodisk) containing 35 μL of joint fluid and standardized concentrations of vancomycin or ceftazidime were placed on agar seeded with MSSA ATCC 25923, MRSA ATCC 43300, S. epidermidis ATCC 14990, P. aeruginosa ATCC 27853, or E. coli ATCC 25922. The discs were incubated overnight at 37°C12, after which the inhibitory activity of the disks was determined.
The demographic and experimental data are represented using descriptive analysis. Values for continuous variables were expressed as the mean and standard deviation (SD).
In Vitro Antibiotic Release Characteristics of Antibiotic-Loaded Bone Cement
High-performance liquid chromatography showed that vancomycin, teicoplanin, and ceftazidime were released from the antibiotic-loaded bone cement for a longer duration (60 days) than either aztreonam (21 days) or imipenem (6 days) (Table II).
All of the antibiotics retained their antibacterial effects after incorporation into the PMMA. All cement samples exhibited antibacterial effects against the tested gram-positive bacteria (MSSA, MRSA, and S. epidermidis) for 60 days, regardless of the combination of antibiotics loaded into the cement (Fig. 2). The antibacterial effects against the gram-negative organism P. aeruginosa lasted for 4 days (vancomycin and imipenem or teicoplanin and imipenem), 10 days (vancomycin and aztreonam), 14 days (teicoplanin and aztreonam), or 30 days (vancomycin and ceftazidime or teicoplanin and ceftazidime). The antibacterial effects against the other gram-negative organism tested (E. coli) lasted for 60 days (vancomycin and ceftazidime or teicoplanin and ceftazidime), 14 days (vancomycin and aztreonam or teicoplanin and aztreonam), or 6 days (vancomycin and imipenem or teicoplanin and imipenem).
Antibiotic Concentrations in the Joint Fluid of Patients with Knee Periprosthetic Joint Infection
Vancomycin and ceftazidime were detected in the joint fluid of each patient. Very high levels of the antibiotics were detected in the joint fluid during the first week following implantation of the PMMA spacer (Table III). The mean vancomycin concentration (± SD) was 352 ± 118 μg/mL on the first day and decreased to 76 ± 89 μg/mL by the seventh day. The mean ceftazidime concentration was 432 ± 119 μg/ mL on the first day and decreased to 41 ± 21 μg/mL by the seventh day (Table III).
The joint fluid samples of all patients exhibited antibacterial activity against the test microorganisms (Fig. 3). In the disk diffusion assay, the disk loaded with joint fluid exhibited a substantially larger inhibitory zone against all of the test microorganisms than the disk loaded with the serum of peripheral venous blood (Fig. 3). We also evaluated the antibacterial activity of the joint fluid samples against 4 different strains of microorganisms (MSSA, MRSA, Enterococcus faecalis, and Serratia marcescens) that had been clinically isolated from the enrolled patients (Table I). The joint fluid samples of all of the enrolled patients exhibited antibacterial activity against the clinically isolated microorganisms (Fig. 4).
No patient in this series exhibited any allergy, renal insufficiency, hepatic dysfunction, or other adverse systemic effect related to the use of the antibiotics. For example, the average serum creatinine level was 0.6 ± 0.2 mg/dL before the first-stage operation and 0.8 ± 0.2 mg/dL before the second-stage operation (Table I). The average serum alanine transaminase level was 31.8 ± 8.9 U/L before the first-stage operation and 29.4 ± 9.9 U/L before the second-stage operation (Table I). During the second-stage surgery, there was no visible evidence of infection and the bacterial cultures of tissue samples taken in the operating room were all negative. No evidence of recurrent infection was observed in any patient.
Our in vitro results demonstrated that bone cement loaded with high doses of vancomycin and ceftazidime provided antibacterial activity against MSSA, MRSA, S. epidermidis, P. aeruginosa, and E. coli for as long as or longer than the bone cement samples loaded with the other antibiotic combinations. Furthermore, after knee joint implantation, the bone cement loaded with vancomycin and ceftazidime provided a prompt, abundant release of these antibiotics. The antibiotic levels in the joint fluid far exceeded therapeutic requirements between the stages of the exchange arthroplasty. In addition, despite the high concentrations of antibiotics (>500 μg/mL in some cases) achieved at the site of infection, no toxic systemic effect was noted in any patient.
Renal toxicity of vancomycin is notorious, and acute renal injury has been attributed to the use of antibiotic-loaded bone cement in the treatment of periprosthetic joint infection13,14. However, Hsieh et al. noted no systemic side effects after using high doses of vancomycin and aztreonam in bone cement in 46 patients with a hip periprosthetic joint infection15. Also, Springer et al. reported no systemic adverse effects from the use of high doses of vancomycin and gentamicin in cement spacers in a series of 36 knees with periprosthetic joint infection16.
There have been several reports concerning the in vivo elution characteristics of antibiotics from antibiotic-loaded bone cement15,17,18. However, bioassays of the released antibiotics against common microorganisms involved in knee periprosthetic joint infections were not performed in these studies. In our study, we found therapeutic levels of antibiotics in the joint fluid against common microorganisms such as MSSA, MRSA, S. epidermidis, P. aeruginosa, and E. coli as well as against clinically isolated bacteria obtained through aspiration of the infected joints of our enrolled patients. We also found that our antibiotic-loaded bone cement spacer released high concentrations of antibiotics during the initial period after implantation (Table III).
When impregnated with antibiotics, bone cement spacers not only serve as a temporary prosthesis to retain soft-tissue tension, maintain joint function, and facilitate subsequent reimplantation, but also act as local drug delivery systems to provide substantially higher antibiotic levels as compared with systemic administration at the infection site19. However, the optimal combination of antibiotics in antibiotic-loaded bone cement is still unclear. Furthermore, the antibiotic release efficacy, antibacterial duration, and antibacterial spectrum of bone cement spacers loaded with various combinations of antibiotics have not been well investigated. In order to provide a wide antibacterial spectrum and long antibacterial duration, antibiotic-loaded bone cement spacers used as therapeutic measures commonly incorporate high doses of 2 different antibiotics, most commonly aminoglycosides and vancomycin20. Vancomycin-loaded and teicoplanin-loaded bone cement have both been widely used for prophylaxis against and treatment of gram-positive periprosthetic joint infections12,21. Gentamicin is one of most common aminoglycosides used in bone cement. However, it is available in only liquid form in some countries, and liquid antibiotics have been shown to significantly compromise the mechanical properties of cement and to increase the risk of fracture of cement spacers22.
Ceftazidime (a third-generation cephalosporin)23, imipenem (a carbapenem antibiotic)24, and aztreonam (a monobactam antibiotic) all have strong activity against susceptible gram-negative bacteria. Ceftazidime, imipenem, and aztreonam all retain their antibacterial capacities after being loaded in bone cement. The efficiency of antibiotic release from bone cement is a critical factor that determines the antibacterial activity of the cement. Bone cement specimens loaded with ceftazidime, imipenem, or aztreonam exhibited different antibacterial durations even though the same antibiotic dose had been used. The kinetics of antibiotic release from antibiotic-loaded bone cement depends on the penetration of dissolution fluids into the polymer matrix and subsequent diffusion of the dissolved drug from the cement. In this study, the differences in the release efficacies among the bone cement samples loaded with different antibiotics might have been due to differences in the penetration by dissolution fluid10.
Our study had several limitations, including a small sample size. Also, because of the short follow-up period and relatively preliminary in vivo results, we could not determine recurrent infection rates over the mid or long term. In addition, the bioassay against microorganisms was performed in an in vitro or ex vivo laboratory environment, which does not necessarily reflect actual clinical circumstances. The quantity and flow of body fluid, limb mobility, host response, and antibiotic stability in vivo were not taken into consideration. Finally, only 1 type of bone cement and dose of antibiotic was used, and only MSSA, MRSA, S. epidermidis, P. aeruginosa, E. coli, and the clinically isolated bacteria from the enrolled patients were selected as test organisms in the experiment. We do not know the bioactivity against other biofilm-forming organisms, such as Propionibacterium acnes.
The bone cement samples loaded with vancomycin and ceftazidime and those loaded with teicoplanin and ceftazidime provided equal antibacterial duration and spectra in our in vitro model. Teicoplanin, although widely distributed in the majority of European and Asian countries, is still not available in North America, including the U.S. and Canada. Therefore, we did not test the teicoplanin and ceftazidime combination in our in vivo model.
In conclusion, in our in vitro assay, vancomycin and ceftazidime-loaded bone cement provided equal or better antibacterial activity against MSSA, MRSA, S. epidermidis, P. aeruginosa, and E. coli strains compared with cement specimens loaded with vancomycin and imipenem, vancomycin and aztreonam, teicoplanin and ceftazidime, teicoplanin and imipenem, or teicoplanin and aztreonam. Monitoring antibiotic elutions from knee cement spacers loaded with vancomycin and ceftazidime and implanted in vivo showed high local antibiotic levels and biologic activity over the short term.
A figure showing the inhibitory effects against bacteria exhibited by the different concentrations of antibiotics is available with the online version of this article as a data supplement at jbjs.org.
Investigation performed at the Chang Gung Memorial Hospital, Linko, Taiwan
Disclosure: This work was supported by Chang Gung Memorial Hospital (Grants CMRPG391581, CMRPG391582, and CMRPG391583). The authors report no conflicts of interest. The Disclosure of Potential Conflicts of Interest forms are provided with the online version of the article.
- Copyright © 2017 by The Journal of Bone and Joint Surgery, Incorporated