Prosthesis Using Antibiotic-loaded Bone Graft
H. Winkler, A. Stoiber
Osteitis Center Privatklinik Döbling, Wien, Austria
Orthopaedic Department, Weinviertel Klinikum, Schwerpunktkrankenhaus Mistelbach, Austria
Introduction
Total hip replacement (THR) has revolutionized orthopaedic surgery. Each year mil- lions of traumatic, inflammatory or degenerative hip joint lesions favourably are treated with THR worldwide. Although meanwhile a routine procedure, infections inevitably occur in a certain percentage of interventions, depending from a variety of accompanying circumstances. Today, the risk of infection for primary THR is esti- mated around 1 %, in revision cases it may rise up to 20 %. It may be expected that the real number is even higher since detection of infection not always is feasible [4, 11, 13]. The absolute number of patients with infection continuously increases as the number of patients requiring THR and revisions grows.
Treatment of infected THR most frequently includes removal of the implant and long-term antimicrobial treatment with re-implantation after all signs of infection have ceased. Such cases still are considered a disaster, both for the patient and for the treating team. Reasons for the difficulties in treatment include the specific adherence of bacteria to foreign material [8] and the poor penetration of antibiotics into infected osseous sites. After removal of necrotic bone the remaining defects must be filled. Leaving the dead space would result in early re-infection and diminished mechanical stability.
Dead space management therefore represents one critical point of septic surgery.
For filling the defects bone grafts would be most advisable, since they may restore bone stock and grant immediate mechanical support. However, usual grafts can only be applied several weeks after debridement when all signs of infection have ceased.
Otherwise the non-vital grafts immediately would become re-colonised with remain- ing bacteria.
To overcome the inferior performance of systemic application in bone infections local administration of antibiotics has been in use since long. Major concern of stud- ies investigating locally applied antibiotics has been efficacy against bacteria as well as compatibility with surrounding tissue. Among all tested antibiotics vancomycin has been suggested as most suitable, since it provides bactericidal activity against the most relevant germs and shows the least cytotoxic effect against growing osteoblasts [5, 10]. As another excellent option for local application aminoglycosides are in clini- cal use since several years and have proven efficacy and good compatibility with vital tissue [9]. For local application of the antibiotics a carrier with good storage capacity is needed.
Bone cement (Polymethylmetacrylate, PMMA) is most widely evaluated as an antibiotic carrier. It provides dead space management as well as prolonged release of antibiotics and therefore is in use in one stage procedures since more than 30 years [1]. However, several possible disadvantages must be taken into account: 1. there is no restoration of bone stock, 2. sclerotic bone, which is present in most revision cases, does not interconnect properly with PMMA, 3. in case of failure the consecutive revi- sion is rather difficult since the cement must be removed again, 4. the “empty” carrier may act as bed for new colonization with surviving selected bacteria [12]. These con- cerns certainly are some of the reasons, why this approach did not gain widespread popularity.
To overcome the disadvantages of cement uncemented implants and biological carriers have been suggested as an alternative. Cancellous bone shows a large surface after purification on which antibiotics can adhere. Witso et al. have shown, that sev- eral antibiotics may be stored and released by allograft bone [18, 19]. The same group has used netilmicin in combination with allografts for reconstruction in revision hip arthroplasty and found no adverse effects [17]. Buttaro at al. favorably used vancomy- cin impregnated grafts for reconstruction after infected THR [3]. Although concen- tration in the postoperative drainage fluid was extremely high they did not observe any adverse effects, neither systemically nor in graft incorporation [2]. However, both groups have used the grafts only in the second stage of a two stage revision after reso- lution of clinical, laboratory and radiological evidence of sepsis.
Our own in vitro studies, using a proprietary impregnation technique, revealed high initial concentrations of antibiotics in areas adjacent to antibiotic impregnated grafts with a prolonged release for several weeks [16]. The aim of our study was to investigate the performance of these compounds under the conditions of florid infec- tion together with uncemented implants.
Material and Methods
Between 1998 and 2004, 37 patients with culture proven deep infection of a THR were treated using a standardized protocol. Loose implants were removed and meticulous excision of all PMMA, granulation, necrotic and infected tissue was performed.
Cleaning was finalized using intensive pulsed lavage with saline. The cleaned sites were evaluated for bone deficiencies and consecutively filled with antibiotic impreg- nated bone graft.
The grafts originated from cadaveric donors, procured following the protocols of the European Association of Tissue Banks [6]. The retrieved cancellous parts were morsellized to granules with a diameter between 2 and 6 mm and cleaned thoroughly from marrow and adhering soft tissue leaving intact structures of bone matrix and mineral content. The bone was impregnated with high loads of antibiotic, using a spe- cific incubation technique [16].
There were two options of antibiotic impregnation: vancomycin or tobramycin, the choice being dependent on the causative pathogen isolated. Vancomycin was used in all cases, combinations with tobramycin in cases of mixed infections. The impreg- nation procedure produced an antibiotic-bone compound (ABC) with high levels of antibiotic inside the graft: For vancomycin levels were in the range of 100 mg per 1 cc
One-stage Non-cemented Revision of Septic Hip Prosthesis 195
Fig. 1. Antibiotic bone compound (ABC) ventrally and dorsally of a Zwey-mueller stem and in the acetabulum
of bone, for tobramycin in the range of 75 mg per 1 cc. On average an amount of 80 cc of ABC was used per case (range: 30 to 150 cc). After gross preparation ABC was filled into cavities using a modified technique of impaction grafting [7, 14] followed by preparation of the bone for the insertion of chosen implants. All prosthetic implants were anchored following the principles of press-fit fixation without additional use of cement.
The choice of the respective implants was dependent from the amount of accompa- nying bone defects, in uncomplicated cases a hemispheric cup was preferred in com- bination with a rectangular diameter stem. Fixation intraoperatively was qualified as stable in all cases. After insertion of the implant ABC was placed around eventually uncovered parts, in special ventrally and dorsally of the proximal parts of the stem (Fig. 1). Wounds were drained for three days and closed immediately. Perioperative antibiotic treatment consisted of second generation cephalosporines intravenously for two weeks. Postoperatively levels of vancomycin were monitored both in the drainage fluid and in serum.
Cultures taken intraoperatively revealed growth of coagulase-negative staphylo- cocci (18x), Staphylococcus aureus (11x), methicillin-resistant S. aureus (MRSA) (4x), enterococci (9x) and other Gram-positive pathogens (3x), respectively. In 6 hips Gram- negative germs were found additionally. All of them were susceptible for either vanco- mycin or tobramycin or both. Postoperative mobilisation did not differ from non-sep- tic surgery. Patients were evaluated clinically and radiographically 2 weeks, 6 weeks, 3 months, 6 months and one year after surgery and then annually. Laboratory data were collected the same time and included CRP, ESR, blood count, urea and creatinine.
Results
Postoperative serum levels of vancomycin were between 0 and 3.9 µg/ml with a median of 0.2 µg/ml. Vancomycin level in the drainage fluid were between 8 and 2243 µg/ml with a median of 345 µg/ml. Wound healing was uneventful in all cases.
No adverse side effects could be observed during the whole follow up period, in spe- cial renal function did not show any difference compared to preoperative values.
Mean hospital stay was 16 days (10 – 32 days). Rehabilitation was in the range of uncomplicated primary surgery in cases with short history of infection (up to 3 months) and prolonged in relation to duration of infection and amount of preced- ing surgery.
All patients could be followed with a minimum of 2 years and a maximum of 8 years (mean 4,1 years). 35 patients showed no sign of infection until the last follow up.
In two hips there was recurrence of infection, diagnosed 6 and 12 weeks after surgery respectively. In one of them the well fixed stem had not been exchanged, in another one a technical error during impregnation of the bone graft could be evaluated. Both could be successfully re-operated using the same technique with complete removal of implants and appropriately impregnated bone graft.
a b
Fig. 2. a Infection of a well fixed uncemented THR (Fistulography). b Postoperative radiograph showing ABC at the medial aspect of the acetabular component and the proximal part of the fem- oral component
One-stage Non-cemented Revision of Septic Hip Prosthesis 197
c
Fig. 2c. 6 years later partial resorption in unloaded areas is visible, remaining parts of the allograft are well incorporated. Lucent lines in Gruen zone 2 and 3 but no disloca- tion and no clinical sign of loosening or infection
Radiologically we could observe partial resorption of allograft bone in non-weight- bearing areas and intramedullarily (Figs. 2a – c). Incorporation of the allografts appeared normal compared to conventional grafting. There was no sign of loosening in any of the implants or dislocation of bone fragments.
Discussion
To our knowledge this is the first report on the use of antibiotic impregnated bone grafts in the treatment of infected THR in a single stage procedure. It must be empha- sized that ABC can only be considered as one tool in a complex treatment protocol consisting of 1. complete removal of all foreign material, 2. meticulous debridement of the infected site, 3. complete dead space management with allograft bone, 4. stable fix- ation of new implants and 5. adequate wound coverage. However, to our opinion only sufficient impregnation of the graft with antibiotics makes such a protocol feasible.
So far impregnated bone grafts have been used clinically only as a more or less pro- phylactic tool since it has been used in patients with high risk but without any sign of florid infection. Buttaro’s group added 500 mg of vancomycin powder to one morsel- lized femoral head, which may be estimated to represent a volume of roughly 50 cc cancellous bone. Similar techniques and concentrations were used by Witso et al.
Both groups found similar levels of antibiotic in the drainage fluid as we did in our
series. However, we could show that with our technique an amount of 5 g vancomycin may be incorporated in the same amount of bone graft, which represents about the ten-fold concentration. The reason for the comparable concentration in drainage flu- ids and serum in our opinion is, that drainage fluid can be monitored only for a few days until drainage is discontinued. During that time vancomycin is released in the same pattern, independent from the impregnation technique.
We believe that this “burst release” is finished after several days. This pattern of release seems to be sufficient in cases where no florid infection or only a low biobur- den is present. Florid infections in our opinion require a more prolonged release that should maintain antibiotic levels above the MIC for several weeks. This requirement seems to be fulfilled with our impregnation technique, which provides a double pat- tern release: the same amount of antibiotic seems to be released immediately, as in the other series, however, this amount represents only about 10 % of the total implanted dosage. An additional 9-fold amount shall be released slowly within a period between 2 and 8 weeks after surgery. These consistent levels protect both the graft and the implant against recolonisation even in highly contaminated areas during the time of release and at the same time may help in eradicating remaining bacterial colonies.
Only under these circumstances one stage revision seems to be justified.
Although very high concentrations of antibiotic were present at the operative site we did not see any adverse effects. Systemic effects seem to be avoided by the poor penetration of vancomycin between blood stream and tissue. This property always has been considered a disadvantage when vancomycin was administered intrave- nously. In our application the disadvantage turns into an advantage, because vice versa there is also poor penetration from the tissue into the vascular system which avoids quick removal of the antibiotic from the implant site, keeping serum levels low and tissue levels high. Local wound healing and remodeling of the graft seem not to be impaired due to the low cytotoxic property of vancomycin and tobramycin respec- tively.
Some authors criticize local application of antibiotics since they fear induction of resistencies. We believe this fear is not justified in our application. Moreover the opposite seems to be true: resistencies are created by sub-inhibitory concentrations of antibiotics as they are common in infected bone during systemic antibiotic ther- apy. The extremely high concentrations after local application leave no chance for bacteria to survive or even develop resistance. Development of small colony variants or similar phenotypes as created by antibiotic loaded cement [12] is unlikely since in contrast to PMMA the grafts release all the incorporated antibiotic load within sev- eral weeks while about 90 % of antibiotics inside PMMA stay there forever and are only released in subinhibitory amounts whenever cracks appear in the aging cement.
Although there are not yet significant numbers of cases and a rather short follow up period, it seems, that one stage non-cemented revision in combination with ABC provides an excellent tool in the treatment of infected THR. However, several princi- ples need to be observed. In addition to the described protocol we now recommend removing even well fixed prostheses and taking care, that at least 50 cc of well impreg- nated bone graft is implanted.
Since we have found that intraoperatively taken cultures sometimes reveal growth of bacteria that have not been found in the preoperative aspirate, we now always use vancomycin and tobramycin in combination, such covering also potentially unde-
22 One-stage Non-cemented Revision of Septic Hip Prosthesis 199
tected Gram-negative germs and taking advantage of the synergistic effect of the two antibiotics [15].
Following the described principles eradication of pathogens, grafting of bony defects and re-insertion of an uncemented implant may be accomplished in a one stage procedure. Since the graft gradually is replaced by healthy own bone, improved conditions may be expected for the long term performance and especially in the case of another revision.
References
1. Buchholz HW, Elson RA, Engelbrecht E et al (1981) Management of deep infection of total hip replacement. J Bone Joint Surg (Br) 63(3):342 – 353
2. Buttaro MA, Gimenez MI, Greco G et al (2005) High active local levels of vancomycin without nephrotoxicity released from impacted bone allografts in 20 revision hip arthroplasties. Acta Orthop 76(3):336 – 340
3. Buttaro MA, Pusso R, Piccaluga F (2005) Vancomycin-supplemented impacted bone allo- grafts in infected hip arthroplasty. Two-stage revision results. J Bone Joint Surg (Br) 87(3):314 – 319
4. Costerton W (2005) Biofilm theory can guide the treatment of device-related orthopaedic infections. Clin Orthop Relat Res (437):7 – 11. Review
5. Edin ML, Miclau T, Lester GE et al (1996) Effect of cefazolin and vancomycin on osteoblasts in vitro. Clin Orthop Relat Res (333):245 – 251
6. European Association of Tissue Banks, European Association of Muskulosceletal Transplan- tation (1999) Common Standards for Musculo Skeletal Tissue Banking. Vienna: Oebig Trans- plant
7. Gie GA, Linder L, Ling RS et al (1993) Contained morselized allograft in revision total hip arthroplasty. Surgical technique. Orthop Clin North Am 24(4):717 – 725
8. Gristina AG (1987) Biomaterial-centered infection: microbial adhesion versus tissue integra- tion. Science 237(4822):1588 – 1595
9. Miclau T, Edin ML, Lester GE et al (1995) Bone toxicity of locally applied aminoglycosides. J Orthop Trauma 9(5):401 – 406
10. Miclau T, Edin ML, Lester GE et al (1998) Effect of ciprofloxacin on the proliferation of osteo- blast-like MG-63 human osteosarcoma cells in vitro. J Orthop Res 16(4):509 – 512
11. Nelson CL, McLaren AC, McLaren SG et al (2005) Is aseptic loosening truly aseptic? Clin Orthop Relat Res (437):25 – 30
12. Neut D, van de Belt H, Stokroos I et al (2001) Biomaterial-associated infection of gentamicin- loaded PMMA beads in orthopaedic revision surgery. J Antimicrob Chemother 47(6):
885 – 891
13. Neut D, van Horn JR, van Kooten TG et al (2003) Detection of biomaterial-associated infec- tions in orthopaedic joint implants. Clin Orthop Relat Res (413):261 – 268
14. Schimmel J, Buma P, Versleyen D et al (1998) Acetabular reconstruction with impacted mor- selized cancellous allografts in cemented hip arthroplasty: a histological and biomechanical study on the goat. J Arthroplasty 13(4):438 – 448
15. Watanakunakorn C, Tisone JC (1982) Synergism between vancomycin and gentamicin or tobramycin for methicillin-susceptible and methicillin-resistant Staphylococcus aureus strains. Antimicrob Agents Chemother 22(5):903 – 905
16. Winkler H, Janata O, Berger C et al (2000) In vitro release of vancomycin and tobramycin from impregnated human and bovine bone grafts. J Antimicrob Chemother 46(3):423 – 428 17. Witso E, Persen L, Benum P et al (2004) High local concentrations without systemic adverse
effects after impaction of netilmicin-impregnated bone. Acta Orthop Scand 75(3):339 – 346 18. Witso E, Persen L, Loseth K et al (1999) Adsorption and release of antibiotics from morseli-
zed cancellous bone. In vitro studies of 8 antibiotics. Acta Orthop Scand 70(3):298 – 304 19. Witso E, Persen L, Loseth K et al (2000) Cancellous bone as an antibiotic carrier. Acta Orthop
Scand 71(1):80 – 84
