Total joint replacement is an extremely successful surgical procedure. Nevertheless, aseptic loosening is a major problem in clinical orthopaedics unlikely to be completely solved by advances in low-wear alternative bearing surfaces or implant design.19,49 It is thought that aseptic loosening is associated with wear particle generation that initiates a proinflammatory cascade, leading to increased osteoclast differentiation and local osteolysis.20 By definition, aseptic loosening occurs in the absence of clinical or microbiologic evidence of infection. However, this does not rule out the possibility of subclinical levels of infection, especially in patients who are immunocompromised.19,41 Numerous cell culture and animal model studies support the concept that bacterial endotoxins may contribute to the proinflammatory process and the resultant implant loosening even in the absence of any evidence of infection.2,6-10,13,19,22,26,44 Therefore, adherence of lipopolysaccharide, the classic endotoxin derived from gram-negative bacteria, to metallic or polymeric wear particles substantially increases the biologic effects of the particles.2,6-10,13,19,22,26,44 Although it is unknown whether endotoxins contribute to osteolysis in patients, the prevalence of aseptic loosening can be reduced by including antibiotics in the polymethylmethacrylate (PMMA) bone cement used to provide initial fixation.16
There are at least three potential sources of bacterial endotoxins associated with aseptic loosening in patients. First, the manufacturing process likely results in substantial amounts of adherent lipopolysaccharide on at least some orthopaedic implants.36 Second, systemic lipopolysaccharide12,23 derived from gut flora, minor infections, or dental procedures may adhere to wear particles after they are generated in patients. This possibility is supported by findings that lipopolysaccharide is extremely adherent to titanium and polyethylene particles.13,36 Third, some studies have shown that bacterial biofilms exist on many implants retrieved from patients with aseptic loosening.19,47 Most of these bacteria are gram-positive and therefore would not produce lipopolysaccharide. However, gram-positive bacteria produce molecules, such as lipoteichoic acid and peptidoglycan, with biologic effects similar to those induced by lipopolysaccharide.32 Peptidoglycan likely exists in periprosthetic tissue surrounding aseptically loosened implants because it is found in synovial lining of patients with osteoarthritis or rheumatoid arthritis.42 Macrophages expressing proinflammatory cytokines were the predominant cell type found to be associated with peptidoglycan.42
The rate of clinically detectable infections of orthopaedic implants is greater in patients with inflammatory arthritis than in patients with osteoarthritis.33,39 We hypothesized that the prevalence of lipopolysaccharide in peri-prosthetic tissue surrounding aseptically loose implants is greater in patients with inflammatory arthritis than in patients with osteoarthritis.
MATERIALS AND METHODS
Periprosthetic tissues were obtained at the time of revision surgery from 10 patients with aseptic loosening and from two patients with periprosthetic gram-negative (Pseudomonas aeruginosa and Escherichia coli) infections as positive control subjects (Table 1). Tissue was obtained from areas adjacent to implants that clearly showed gross interface failure. The group of patients with aseptic loosening was subdivided by the underlying diseases: six with osteoarthritis and four with inflammatory arthritis. The inflammatory arthritis group included patients with rheumatoid arthritis, juvenile rheumatoid arthritis, and lupus. The age range of the patients at the time of revision surgery was 38 to 78 years. The time between implantation and revision surgery varied from 1.5 to 21 years. Six of the patients were women and six were men. Ten of the patients with aseptic loosening had total hip arthroplasties and two had total knee arthroplasties. Two samples were obtained from Patient 5, one from the acetabulum (5A) and one from the femur (5B). Cemented implants were used more commonly in the patients with inflammatory arthritis (Table 2).
Patients with aseptic loosening had no clinical or microbio-logic signs of infection as determined by the following measures: preoperative erythrocyte sedimentation rates, C-reactive protein levels, Gram's staining, quantitative assessment of neutrophils in frozen sections, intraoperative and postoperative cultures. Pre-operative antibiotics were avoided in all of these patients to prevent false-negative culture results (Table 2). Although postoperative culture results were not available for patients 4 and 10, sufficient information was available for all of the patients with aseptic loosening to exclude infection. Frozen sections were considered negative if there were less than five neutrophils per high-powered field.1 All procedures were approved by our Institutional Review Board.
Tissue samples were homogenized in five volumes of phosphate buffered saline (CellGro, Herndon, VA) with penicillin-streptomycin (CellGro) in Duall Tissue Grinders (Fisher, Pittsburgh, PA) that had been made endotoxin free by heating at 260°C for 18 to 20 hours.45 Tissue homogenates were heated at 100°C for 10 minutes to inactivate nonspecific amidases.25 All tissues and homogenates were processed using sterile technique and stored at −80°C before assay.
Tissue homogenates were assayed on multiple occasions, using various dilutions because it was necessary to empirically determine the appropriate dilution for each homogenate individually. Therefore, homogenates that had been overdiluted or underdiluted were reassayed at a different dilution. We considered the homogenates overdiluted if the absorbencies were less than the lowest standard and an assay of the same homogenate at a lower dilution had absorbencies greater than the lowest standard. Homogenates were considered underdiluted if the spike recovery was less than 40%, reflecting assay inhibition, and the lipopolysaccharide level without spike correction was not substantially greater than the homogenization buffer control. As a negative control, we added saline to specimen cups in the operating room and processed them identically to the tissue samples.
Conventional endotoxin assays that use Limulus amebocyte lysates are susceptible to false positives from β-glucan from fungal cell walls and from uncharacterized β-glucan-like molecules found in human tissues.11,29,38 We therefore used a modified Limulus amebocyte lysate assay known as endospecy (Associates of Cape Cod, Woods Hole, MA) that is unaffected by the β-glucan-like molecules in human tissues.29,31 Moreover, this assay responds to all tested forms of lipopolysaccharide, whereas many conventional Limulus assays are unresponsive to some forms of lipopolysaccharide.38 Briefly, various dilutions of heat-treated homogenates were assayed in triplicate by incubation with the combined lysate/substrate solution for 30 minutes at 37°C. We terminated reactions and developed assay color by diazocoupling to reduce false positives because of sample color.46 Particulates were removed by centrifugation (600 × g for 5 minutes) and absorbencies were determined at 545 nm. Lipopolysaccharide concentrations per mL of homogenate were determined by comparison with a standard curve (0.015-0.24 endotoxin units/mL) and corrected for dilution of the homogenates, sample color, and spike recovery as described.29 Assay results were considered to be positive or negative for lipopolysaccharide using the previously described criteria.29
We performed multiple endospecy lipopolysaccharide assays on most of the tissue homogenates. Each assay result was classified as positive or negative because the assay-to-assay variability prevents calculation of a precise lipopolysaccharide level.29 This variability is likely caused by the fibrous nature of the tissue samples, which makes complete homogenization difficult. More aggressive disaggregation strategies could not be used because of the necessity to avoid contaminating the samples with exogenous endotoxin.29 Using a limited number of tissue samples, Nalepka and Greenfield previously reported that results with the endospecy assay are not affected by false positives because of sample color or nonspecific amidases or false positives/false negatives because of poor spike recovery.29 Similar results also were obtained with the larger number of samples assayed in the current study (data not shown).
We used Fisher's exact test for the statistical analysis because the data are binary, unpaired, and have small expected frequencies.14
In patients with aseptic loosening, lipopolysaccharide was detected in periprosthetic tissue from all inflammatory arthritis patients but less (p < 0.05) frequently in osteoarthritis patients (Table 2; Fig 1). We found similar results when the results were reported as a percentage of the total assays performed for each group that were positive (Fig 2) rather than the percentage of patients with positive assays (Fig 1). Lipopolysaccharide also was detected in periprosthetic tissue from both infected patients included as positive control subjects (Table 2). Lipopolysaccharide was undetectable in the negative control samples.
Lipopolysaccharide was found in periprosthetic tissue from all four patients with inflammatory arthritis who had aseptically loosened implants even in the absence of any clinical or microbiologic signs of infection. Periprosthetic tissues from patients with inflammatory arthritis also likely contain gram-positive endotoxin-like molecules because peptidoglycan was found in synovial lining obtained at the time of primary arthroplasties in patients with rheumatoid arthritis.42 It is interesting to speculate that lipopolysaccharide and peptidoglycan may contribute to the elevated rate of aseptic loosening in rheumatoid arthritis reported in some studies.37,40,43 Our findings support the possibility that endotoxins derived from gram-negative bacteria may contribute to aseptic loosening in patients with inflammatory arthritis. Alternatively, loose prostheses may be more prone to subclinical infections related to chronic tissue irritation from intermittent loading-induced movement. This alternative may be especially likely in patients who are immunocompromised. Another possibility is the patients with positive lipopolysaccharide assays had been infected at an earlier time and, although the bacteria subsequently were eradicated, the endotoxin remained in the periprosthetic tissue. We consider this last alternative to be unlikely because there was no evidence of prior infection in the patients' histories. Most importantly, in any of the above-described alternatives, endotoxins from gram-negative bacteria may contribute to aseptic loosening. This is possible whether the endotoxins precede the initial stages of loosening or vice versa and is supported by the extensive cell culture and animal studies showing endotoxin increases the biologic activity of orthopaedic wear particles.2,6-10,13,19,22,26,44
Our study has several limitations, including the small number of patients studied, the spectrum of diseases in the inflammatory arthritis group, and the levels of lipopolysaccharide in the periprosthetic tissue. It is unknown whether these levels are high enough to be biologically important. It also is unknown whether the lipopolysaccha-ride is adherent to the wear particles or is in the surrounding tissue.29 However, this limitation is of relatively minor importance because, in either case, the lipopolysaccharide would be expected to act together with the wear particles to increase the inflammatory response. A final limitation is the possibility that the lipopolysaccharide might be attributable to contamination of the periprosthetic tissues during preparation for the assays. However, because the samples from patients with osteoarthritis rarely contained detectable lipopolysaccharide, it is unlikely contamination would preferentially occur in samples from patients with inflammatory arthritis. Moreover, we were unable to detect endotoxin in saline that was added to specimen cups in the operating room and then processed identically to the tissue samples.
The higher levels of peptidoglycan42 and lipopolysaccharide in patients with inflammatory arthritis compared with levels in patients with osteoarthritis may be related to subclinical infections because these patients have higher rates of clinically detectable implant infections.33,39 Immune system dysregulation in rheumatoid arthritis,3 juvenile rheumatoid arthritis,27 and lupus4 may contribute to higher rates of subclinical infections, other transient bacteremia, and increased translocation of bacteria or lipopolysaccharide from the intestine.5,15,48 Moreover, bacteria have been implicated as a contributing factor in rheumatoid arthritis (reviewed in42), and reduced clearance of bacterial and other antigens because of complement system deficiencies is associated with rheumatoid arthritis and with lupus.17,28 An alternative possibility is the frequent use of cemented implants in patients with inflammatory arthritis patients creates a microenvironment favorable for establishing subclinical infections. Cemented implants might increase subclinical infections because PMMA increased infection in a canine model, especially when the PMMA was polymerized in vivo.34 This is likely because of the observation that PMMA polymerization causes local tissue necrosis.18 Taken together, these concepts suggest subclinical infections and/or dysregulation of the immune system may explain the presence of lipopolysaccharide in patients with rheumatoid arthritis, juvenile rheumatoid arthritis, and lupus.
The bacterial biofilm found on many aseptically loose implants is unlikely to be the source of the lipopolysaccharide because the biofilm usually consists of gram-positive bacteria,19,47 and therefore would not produce lipopolysaccharide. An alternative source is lipopolysaccha-ride adherent to the implants before surgery.36 Unless clearance of this lipopolysaccharide is profoundly reduced in inflammatory arthritis, it probably would be cleared after implantation because lipopolysaccharide is rapidly cleared in animal models.44 Therefore, lipopolysaccharide initially adherent to the implant surfaces is more likely to inhibit early osseointegration rather than to increase osteolysis.
Lipopolysaccharide was detected in only one of the six patients with osteoarthritis who had aseptic loosening. Nevertheless, bacterially derived endotoxins may contribute to aseptic loosening in patients with osteoarthritis for at least three reasons. First, we cannot exclude the possibility that periprosthetic tissue from patients with osteoarthritis contains lipopolysaccharide at levels below the sensitivity of our methods. The second reason bacterially derived endotoxins may contribute to aseptic loosening in patients with osteoarthritis despite our findings comes from a study of greater than 40,000 total hip replacements.16 That study found including antibiotics in bone cement reduces the frequency of aseptic loosening in patients with osteoarthritis, suggesting a role for subclinical levels of bacteria.16 In addition, the number of doses of antibiotics administered systemically during the initial 3 days after hip replacement surgery dose dependently reduced aseptic loosening.16 The long-term effect of early antibiotic administration, in the PMMA cement or administered systemically, is likely because of the race for the surface of the implant between the patient's osteoblasts and biofilm-forming bacteria.21 The winner of the race at each portion of the implant is thought to establish a colony that inhibits adherence of the other cell type and therefore has long-term effects on implant stability.21 An additional mechanism that may account for the long-term effects of antibiotics in the PMMA cement is that antibiotic elution from the cement occurs for a much longer time in patients than in vitro. Therefore, substantial antibiotic concentrations are found in cement recovered 5 years after implantation.30 The third reason bacterially derived endotoxins may contribute to aseptic loosening in patients with osteoarthritis despite our findings is because periprosthetic tissue from these patients likely contains endotoxin-like molecules such as lipoteichoic acid and peptidoglycan derived from the gram-positive bacterial biofilm found on many aseptically loose implants.19,47 In support of this possibility, gram-positive bacteria secrete a highly active form of lipoteichoic acid.24 Also, patients with infected implants produce antibodies specific for this secreted form of lipoteichoic acid,35 indicating it activates the patients' immune system. In addition, because peptidoglycan is detectable in the synovial lining of patients with osteoarthritis,42 it is likely that peptidoglycan also exists in periprosthetic tissue surrounding aseptically loosened implants in these patients.
We showed that lipopolysaccharide exists in periprosthetic tissues from patients with inflammatory arthritis who have clinical and laboratory findings of aseptically loose implants. In contrast, periprosthetic tissue from most patients with osteoarthritis who have aseptically loose implants does not contain detectable levels of lipopolysaccharide. Therefore, lipopolysaccharide is more likely to contribute to aseptic loosening in patients with inflammatory arthritis. Although the relative importance of lipopolysaccharide and other bacterial endotoxins in aseptic loosening compared with wear particles per se or other factors is undetermined, this topic merits additional investigation, especially in patients with inflammatory arthritis and compromised immunity.
We thank Dr. Donald Goodfellow, Dr. William Petersilge, and Dr. Roger Wilber for providing additional tissue samples; and Dr. Andrew Islam, Renata Kadlcek, Dr. James Ninomiya, and Dr. Matthew Smith for helpful discussions during preparation of the manuscript. We also thank Dr. Ronald Berzofsky (BioWhit-taker), Dr. Marilyn Gould (Associates of Cape Cod), and Dr. Thomas Novitsky (Associates of Cape Cod) for advice on endotoxin assays and Associates of Cape Cod for donation of Endospecy assay kits; and Barbara Calabro, Patricia Conroy-Smith, Christine Mullins, Cathy Ostrander, and Rebecca Thomas for assistance in retrieving demographic information.
1. Abdul-Karim FW, McGinnis MG, Kraay M, Emacipator SN, Goldberg V. Frozen section biopsy assessment for the presence of polymorphonuclear leukocytes in patients undergoing revision of arthroplasties. Mod Pathol
2. Akisue T, Bauer TW, Farver CF, Mochida Y. The effect of particle wear debris on NFkappaB activation and pro-inflammatory cytokine release in differentiated THP-1 cells. J Biomed Mater Res
3. Albani S, Carson DA, Roudier J. Genetic and environmental factors in the immune pathogenesis of rheumatoid arthritis. Rheum Dis Clin North Am
4. Arnett FC, Reveille JD. Genetics of systemic lupus erythematosus. Rheum Dis Clin North Am
5. Berg RD. Bacterial translocation from the gastrointestinal tract. Trends Microbiol
6. Bi Y, Collier TO, Goldberg VM, Anderson JM, Greenfield EM. Adherent endotoxin mediates biological responses of titanium particles without stimulating their phagocytosis. J Orthop Res
7. Bi Y, Seabold JM, Kaar SG, Ragab AA, Goldberg VM, Anderson JM, Greenfield EM. Adherent endotoxin on orthopedic wear particles stimulates cytokine production and osteoclast differentiation. J Bone Miner Res
8. Brooks RA, Wimhurst JA, Rushton N. Endotoxin contamination of particles produces misleading inflammatory cytokine responses from macrophages in vitro. J Bone Joint Surg Br. 2002;84:295-299
9. Charlebois SJ, Daniels AU, Smith RA. Metabolic heat production as a measure of macrophage response to particles from orthopedic implant materials. J Biomed Mater Res
10. Cho DR, Shanbhag AS, Hong CY, Baran GR, Goldring SR. The role of adsorbed endotoxin in particle-induced stimulation of cytokine release. J Orthop Res
11. Cooper JF, Weary ME, Jordan FT. The impact of non-endotoxin LAL-reactive materials on Limulus amebocyte lysate analyses. PDA J Pharm Sci Technol
12. Cooperstock MS, Tucker RP, Baublis JV. Possible pathogenic role of endotoxin in Reye's Syndrome. Lancet
13. Daniels AU, Barnes FH, Charlebois SJ, Smith RA. Macrophage cytokine response to particles and lipopolysaccharide in vitro. J Biomed Mater Res
14. Dawson B, Trapp RG. Basic and Clinical Biostatistics
. New York, NY: Lange Medical Books/McGraw-Hill; 2001:399.
15. Deitch EA. Role of bacterial translocation in necrotizing enterocolitis. Acta Paediatr Suppl
16. Engesaeter LB, Lie SA, Espehaug B, Furnes O, Vollset SE, Havelin LI. Antibiotic prophylaxis in total hip arthroplasty. Acta Orthop Scand
17. Garred P, Madsen H, Halberg P, Petersen J, Kronberg G, Svejgaard A, Andersen V. Mannose-binding lectin polymorphisms and susceptibility to infection in systemic lupus erythematosus. Arthritis Rheum
18. Goodman SB, Schatzker J, Sumner-Smith G, Fornasier VL, Goften N, Hunt C. The effect of polymethylmethacrylate on bone: an experimental study. Arch Orthop Trauma Surg
19. Greenfield EM, Bi Y,Ragab AA. Goldberg VM, Nalepka JL, Seabold JM. Does endotoxin contribute to aseptic loosening of orthopedic implants? J Biomed Mater Res B Appl Biomater
. 2005;72: 179-185.
20. Greenfield EM, Bi Y, Ragab AA, Goldberg VM, Van De Motter RR. The role of osteoclast differentiation in aseptic loosening. J Orthop Res
21. Gristina AG. Implant failure and the immuno-incompetent fibro-inflammatory zone. Clin Orthop Relat Res
22. Hitchins VM, Merritt K. Decontaminating particles exposed to bacterial endotoxin (LPS). J Biomed Mater Res
23. Inada K, Endo S, Takahashi K, Suzuki M, Narita T, Yoshida T, Suda H, Komuro T, Yoshida M. Establishment of a new perchloric acid treatment method to allow determination of the total endotoxin content in human plasma by the limulus test and clinical application. Microbiol Immunol
24. Jones KJ, Perris AD, Vernallis AB, Worthington T, Lambert PA, Elliott TSJ. Induction of inflammatory cytokines and nitric oxide in J774.2 cells and murine macrophages by lipoteichoic acid and related cell wall antigens from Staphylococcus epidermidis. J Med Microbiol
25. Kawamura S, Hirayama K, Mishima M, Akiyama T, Mitsuoka T. Endotoxic activity in faeces of mice from different microbiological environments. Res Microbiol
26. Lee DH, Park JC, Suh H. Synergistic induction of cyclooxygenase-II by bacterial lipopolysaccharide in combination with particles of medical device materials in a murine macrophage cell line J774A.1. J Biomed Mater Res
27. Lovell DJ. Pediatric rheumatic diseases B. Juvenile Rheumatoid Arthritis and Juvenile Spondyloartropathies. In: Klippel JH, ed. Primer on the Rheumatic Diseases
. Atlanta, GA: Arthritis Foundation; 2001:534-542.
28. Moulds JM, Krych M, Holers VM, Liszewski MK, Atkinson JP. Genetics of the complement system and rheumatic diseases. Rheum Dis Clin North Am
29. Nalepka JL, Greenfield EM. Detection of bacterial endotoxin in human tissues. Biotechniques
30. Neut D, van de Belt H, van Horn JR, van der Mei HC, Busscher HJ. Residual gentamicin-release from antibiotic-loaded polymethylmethacrylate beads after 5 years of implantation. Biomaterials
31. Obayashi T, Tamura H, Tanaka S, Ohki M, Takahashi S, Arai M, Masuda M, Kawai T. A new chromogenic endotoxin-specific assay using recombined limulus coagulation enzymes and its clinical applications. Clin Chim Acta
32. O'Neill LA. Toll-like receptor signal transduction and the tailoring of innate immunity: a role for Mal? Trends Immunol
33. Peersman G, Laskin R, Davis J, Peterson M. Infection in total knee replacement: a retrospective review of 6489 total knee replacements. Clin Orthop Relat Res
34. Petty W, Spanier S, Shuster J, Silverthorne C. The influence of skeletal implants on incidence of infection: experiments in a canine model. J Bone Joint Surg Am
35. Rafiq M, Worthington T, Tebbs SE, Treacy RB, Dias R, Lambert PA, Elliott TS. Serological detection of Gram-positive bacterial infection around prostheses. J Bone Joint Surg Br. 2000;82: 1156-1161
36. Ragab AA, Van De Motter R, Lavish SA, Goldberg VM, Ninomiya JT, Carlin CR, Greenfield EM. Measurement and removal of adherent endotoxin from titanium particles and implant surfaces. J Orthop Res
37. Ranawat CS, Deshmukh RG, Peters LE, Umlas ME. Prediction of the long-term durability of all-polyethylene cemented sockets. Clin Orthop Relat Res
38. Roslansky PF, Novitsky TJ. Sensitivity of Limulus amebocyte ly-sate (LAL) to LAL-reactive glucans. J Clin Microbiol
. 1991;29: 2477-2483.
39. Salvati EA, Valle AGD, Masri BA, Duncan CP. The infected total hip arthroplasty. Instr Course Lect
40. Sarmiento A, Ebramzadeh E, Gogan W, McKellop H. Total hip arthroplasty with cement: a long-term radiographic analysis in patients who are older than fifty and younger than fifty years. J Bone Joint Surg Am
41. Savarino L, Baldini N, Tarabusi C, Pellacani A, Giunti A. Diagnosis of infection after total hip replacement. J Biomed Mater Res B Appl Biomater
42. Schrijver IA, Melief MJ, Tak PP, Hazenberg MP, Laman JD. Antigen-presenting cells containing bacterial peptidoglycan in synovial tissues of rheumatoid arthritis patients coexpress costimulatory molecules and cytokines. Arthritis Rheum
43. Severt R, Wood R, Cracchiolo A, Amstutz HC. Long-term follow-up of cemented total hip arthroplasty in rheumatoid arthritis. Clin Orthop Relat Res
44. Skoglund B, Larsson L, Aspenberg PA. Bone-resorptive effects of endotoxin-contaminated high-density polyethylene particles spontaneously eliminated in vivo. J Bone Joint Surg Br. 2002;84: 767-773
45. Sweet BH, Huxsoll JF. Depyrogenation by Dry Heat Depyrogenation
. Philadelphia, PA: Parenteral Drug Association; 1985:101-108.
46. Tamura H, Tanaka S, Obayashi T, Yoshida M, Kawai T. A new sensitive method for determining endotoxin in whole blood. Clin Chim Acta
47. Tunney MM, Patrick S, Curran MD, Ramage G, Hanna D, Nixon JR, Gorman SP, Davis RI, Anderson N. Detection of prosthetic hip infection at revision arthroplasty by immunofluorescence microscopy and PCR amplification of the bacterial 16S rRNA gene. J Clin Microbiol
48. Wells CL, Maddaus MA, Simmons RL. Proposed mechanisms for the translocation of intestinal bacteria. Rev Infect Dis
. 1988;10: 958-979.
49. Wright TM, Goodman SB. Implant Wear in Total Joint Replacement
. Rosemont, IL: American Academy of Orthopaedic Surgeons; 2001:224.