- Email: [email protected]
Cell surface characteristics of microbiological isolates from human percutaneous titanium implants in the head and neck
Biomaterials 20 (1999) 1319}1326
Cell surface characteristics of microbiological isolates from human percutaneous titanium implants in the head and neck K.M. Holgers *, As . Ljungh Department of Anatomy and Cell Biology, University of GoK teborg, GoK teborg, Sweden Department of Audiology, Sahlgren+s University Hospital, University of GoK teborg, GoK teborg, Sweden Department of Infectious Diseases and Medical Microbiology, Lund University, Lund, Sweden Received 11 February 1998; accepted 5 February 1999
Abstract Percutaneous implants are commonly associated with several problems, and di!erent failure modes have been described. Infections constitute one serious complication which may lead to the removal of the implant. In contrast to infections around polymer implants, infections around skin-penetrating titanium implants anchored in the temporal bone are often cured by local treatment. Coagulase-negative staphylococci are the most common etiological agents in infections related to polymers whereas Staphylococcus aureus is considered as the main pathogen in infections around metallic implants. Microbial adhesion is a prerequisite for an infection. In the present study, the cell surface of microbes isolated from the skin around skin-penetrating titanium implants, with and without signs of infection, was characterized with respect to expression of cell surface hydrophobicity and to binding of immobilized "bronectin, vitronectin and collagen type 1 which could mediate adhesion. Expression of protein binding was similar in strains isolated from the two groups. No strain expressed a hydrophobic cell surface as determined by two-phase separation, and we conclude that the microenvironment around a titanium implant promotes expression of a hydrophilic rather than a hydrophobic cell surface which in turn makes many infections around a titanium implant curable by local treatment. 1999 Elsevier Science Ltd. All rights reserved Keywords: Titanium; Percutaneous; Implants; Infection; Biomaterials; Clinical
1. Introduction Foreign materials are introduced into the human body for temporary or clinical use with an increasing frequency. Examples are continuous ambulatory peritoneal dialysis catheters, intravenous catheters, intraocular lenses and attachment of prostheses for amputated limbs or craniofacial defects. After implantation, an in#ammatory reaction is induced in the tissue which also a!ects the biomaterial [1, 2]. When the biomaterial also penetrates the skin additional load is applied on the system since the skin barrier is broken and exogenous agents more easily colonize the material and the surrounding
* Corresponding address: Department of Audiology, Sahlgren's University Hospital, University of GoK teborg, GoK teborg, Sweden. Fax: 0046-31-829811.
tissue [3]. Skin-penetrating implants can be considered as a foreign body penetrating the skin through a surgically created defect. Percutaneous implants are often associated with di!erent failure modes such as marzupialization, permigration, avulsion and infections [4]. Established infections around implants usually lead to removal of the implant. In contrast to infections associated with polymers, infections around skin-penetrating titanium implants are often cured with local treatment [5, 6]. Within fractions of a second after implantation host proteins adsorb to the biomaterial surface. The protein adsorption depends on physicochemical properties of the material, like hydrophobicity and charge, and the composition of the host proteinaceous #uid. Microbial adhesion preferentially occur to these surface-adsorbed structures. Fibronectin (Fn) has been proposed to mediate adhesion of staphylococci to materials in the blood
0142-9612/99/$ - see front matter 1999 Elsevier Science Ltd. All rights reserved. PII: S 0 1 4 2 - 9 6 1 2 ( 9 9 ) 0 0 0 3 3 - 2
1320
K.M. Holgers, As. Ljungh / Biomaterials 20 (1999) 1319}1326
stream [7]. We have reported that vitronectin (Vn) could mediate adhesion of coagulase-negative staphylococci (CNS) to polymers in cerebrospinal #uid [8]. CNS, like Staphylococcus epidermidis are the most common etiological agents in infections related to polymers [9, 10] whereas S. aureus dominates in infections around metal implants [11]. S. aureus was shown preferentially to colonize metal alloys (Ti6A14V) whereas S. epidermidis prefers polymer surfaces [12]. In 14 patients with infections associated with external pin "xation of fractures, S. aureus was isolated from 8 patients and gram-negative rods from 6. The isolation of CNS was not reported, nor were analyses of patients without clinical infection done. In patients with skin-penetrating titanium implants we reported on a polymicrobial #ora. S. aureus was the most commonly isolated species in patients with and without clinical signs of infection [3]. S. aureus can speci"cally bind Fn, Vn, collagen (Cn) and other extracellular matrix (ECM) proteins. Binding of several ECM proteins has been described by CNS strains, preferentially to proteins immobilized on a surface, by Pseudomonas aeruginosa and several gram-negative species, and by anaerobic species [13]. Hydrophobic surface proteins have also been proposed to mediate adhesion [14 ]. The aim of the present study was to characterize microbes isolated around percutaneous titanium implants with respect to expression of binding of immobilized ECM proteins and of cell surface hydrophobicity. We further wanted to correlate these "ndings to signs and degree of infection around the implants. Elucidation of adhesive processes between microbes and biomaterials is vital for the creation of surfaces which are less prone to become infected.
2. Materials and methods Surgical procedures. The installation of the titanium implant is made in two stages. In the "rst, a threaded implant of commercially pure titanium is inserted into the temporal bone behind the ear-canal. The hole is threaded and the screw-shaped "cture is installed. After suturing the skin the implant is left unloaded to heal for 3}4 months. The skin-penetrating part is adapted in a second operation. To get a thin skin-penetration area it is generally necessary to surgically reduce the thickness of the subcutaneous tissue and to place a thin hairless skin graft at the site of the implant. Three or four weeks later the hearing aid is "tted. Patients. Twenty-six patients (14 women and 12 men) with permanent skin-penetrating titanium implants for bone anchored hearing aids (BAHA) were studied. The duration of the skin penetration was 24}207 months (mean 94 months, SD 54). The skin condition was classi"ed as grade 0*no signs of infection around the implant, grade 1*slightly reddish skin around the implant, and
Table 1 The relation between the clinical "ndings and isolation of di!erent microbiological strains around the percutaneous titanium implants
CNS S. aureus Propionibacterium acnes Bacteroides urealyticus Proteus Klebsiella E. coli Peptostreptococcus
Grade 0
Grade 1
Grade 2
19 2 2 1 2 1 1 1
15 2 9
2 2
grade 2*reddish and moist skin around the implant. The classi"cation of our patients in grade 0}2 is presented in Table 1. Microbiology. Specimens were obtained with a sterile cotton swab (1.0 mm in diameter) from the skin immediately adjacent to the implant anchoring the BAHA. Touching the titanium implant during sampling was unavoidable. The specimens were inoculated on blood agar (5% v/v horse erythrocytes, Columbia agar base, ACU, Bethesda, MD, USA) for 48 h at 373C in aerobic and anaerobic atmospheres. Isolated colonies were typed to the species level according to the ASM Manual of Clinical Microbiology [15]. CNS were typed to the species level with an in-house panel of biochemical tests [16]. The isolates were lyophilized. For further testing the strains were dissolved in distilled water and cultured on blood agar as above. Bacterial strains were washed twice in 0.02 M potassium phosphate bu!er, pH 6.8, and resuspended in the same bu!er to approximately 10 colony forming units/ml (CFU/ml). Cell surface hydrophobicity. Polyethylene glycol (7.13% w/v, PEG 6000; KEBO, Stockholm, Sweden) and Dextran (8.75% w/v in 0.015 M NaCl, pH 6.8; molecular weight 48 000, Sigma Chemical Co, St Louis, MO, USA) were prepared in an aqueous two-phase system as earlier described [16]. Partitioning was performed by adding 100 ll of bacterial suspension (approx. 5;10 CFU/ml) to 0.9 ml of the phase system, mixing by gentle shaking, and allowing phase separation at 203C for 1 h. The concentration of bacterial cells was measured turbidimetrically in the PEG-rich top phase and in the dextran-rich bottom phase at 540 nm, and was expressed as percentage of the original concentration of added cells. Negative values denote hydrophobicity and positive values hydrophilicity. Particle agglutination assay (PAA). Vn, Fn and Cn were separately immobilized on latex particles (0.8 lm diameter; Difco Laboratories, Detroit, MI, USA) as earlier described [17]. After adsorption of 100 lg protein to 1 ml latex beads (303C, 1 h) on a horizontal shaker the mixtures were centrifuged (9200 g, 5 min, 43C) the pellets
K.M. Holgers, As. Ljungh / Biomaterials 20 (1999) 1319}1326
were resuspended in 2 ml glycine bu!er (0.17 M glycine NaOH, pH 8.2) with ovalbumin (0.01%; Sigma Chem Co.) and merthiolate (0.01%) and kept at 43C. The assay was performed by gently mixing equal volumes (20 ll) of latex reagent and bacterial suspension on glass slides. The reactions were scored after 2 min and assigned a numerical value: 0"no agglutination, negative, 1"moderate agglutination, positive, and 2" strong agglutination. Chemicals. Vn was puri"ed from human urea-treated plasma on heparin}sepharose [18] and Fn from human plasma on gelatin}sepharose [19]. Vitrogen (containing 95% collagen type 1 and 5% collagen type III) was purchased from Collagen Corporation, Palo Alto, CA, USA. Statistical methods. The Student's t-test was used when appropriate.
3. Results Thirty bacterial strains were isolated from 13 patients with clinical signs of infection, and 28 strains from 13 patients without signs of infection. Aerobic and anaerobic bacteria were isolated from patients with and without signs of infection (Table 1). CNS were the most commonly isolated species in this group of patients with percutaneous implants. S. aureus was isolated from 2 patients each with grade 2, grade 1 and grade 0. There were more anaerobic isolates in the group with skin irritation.
1321
No fungi were isolated. Two strains (Peptostreptococcus sp and CNS) could not be recovered from the lyophilized state, and were hence not further studied. The relation between the clinical "ndings and the distribution of bacterial isolates is presented in Table 1. In patients without sign of infection (grade 0) the bacterial #ora contained a higher variety of species than in the groups with signs of infection. The relation between clinical "ndings and expression of binding of Cn, Fn and Vn is presented in Figs. 1}6. Of 24 strains isolated from patients without clinical signs of infection 11 expressed binding of Cn type I and Fn, and 13 of Vn. In the groups with clinical signs of infection binding of Cn type I, Fn and Vn was detected in 12, 10 and 12 strains, respectively, out of 29 strains. The expression of binding of Cn type I, Fn and Vn was hence the same in strains isolated from patients with or without signs of infection. Two strains of P. acnes and 1 S. aureus from patients without clinical signs of infection agglutinated spontaneously in the PAA, and 1 P. acnes from one patient with signs of infection. The expression of cell surface hydrophobicity/hydro philicity by strains isolated from skin-penetrating implants is presented in Fig. 7. No strain expressed a hydrophobic cell surface. However, strains isolated from the group of patients with clinical signs of infection were less hydrophilic compared with isolates from patients without signs of infection. The mean value in the group with in#ammation was 4.29 (SE 0.73) and in the group
Fig. 1. Binding of collagen type I in clinically non-irritated skin. The expression of binding is described as: no agglutination (0), positive (1) or strongly positive (2). The number of strains are presented.
1322
K.M. Holgers, As. Ljungh / Biomaterials 20 (1999) 1319}1326
Fig. 2. Binding of collagen type I in clinically irritated skin. The expression of binding is described as: no agglutination (0), positive (1) or strongly positive (2). The number of strains are presented.
Fig. 3. Binding of "bronectin in clinically non-irritated skin. The expression of binding is described as: no agglutination (0), positive (1) or strongly positive (2). The number of strains are presented.
without clinical signs of infection 4.75 (SE 0.92). The di!erence in expressing hydrophilicity/hydrophobicity between isolates from grade 0 and isolates from grade 1 and 2 is not statistically signi"cant.
4. Discussion We have earlier described a polymicrobial #ora around percutaneous titanium implants with S. aureus as
K.M. Holgers, As. Ljungh / Biomaterials 20 (1999) 1319}1326
1323
Fig. 4. Binding of "bronectin in clinically irritated skin. The expression of binding is described as: no agglutination (0), positive (1) or strongly positive (2). The number of strains are presented.
Fig. 5. Binding of vitronectin in clinically non-irritated skin. The expression of binding is described as: no agglutination (0), positive (1) or strongly positive (2). The number of strains are presented.
the most commonly isolated bacterial species in samples from patients with and without clinical signs of infection [20]. In the present study we also report on a polymicrobial #ora around the implants but here, the pre-
dominating species is CNS. The skin conditions in the group with infection was less severe in this study than in one earlier study [20]. Anaerobic bacteria were isolated from patients with and without infection. However, the
1324
K.M. Holgers, As. Ljungh / Biomaterials 20 (1999) 1319}1326
Fig. 6. Binding of vitronectin in clinically irritated skin. The expression of binding is described as: no agglutination (0), positive (1) or strongly positive (2). The number of strains are presented.
number of anaerobic isolates was higher in the group with clinical signs of infection. CNS, S. aureus and many anaerobic species form part of the normal microbial #ora on skin and mucous membranes. Hence, species from the normal micro#ora commonly colonize skin-penetrating devices. ECM proteins, especially Fn, have been proposed to mediate adhesion to surfaces [7, 8]. Expression of binding of ECM proteins has been described by S. aureus, CNS, P. acnes [21] and several other microbial species [22]. Expression of binding is in#uenced by the microbial growth conditions. It has earlier been shown that bacteria commonly express binding after growth in nutrient-poor conditions, like the skin [23]. In the former study we analyzed expression of binding of Fn, Vn and Cn type 1 in soluble form [20]. Binding was expressed by 86% of the strains but we did not "nd any di!erences between strains isolated from patients with or without clinical infection. In the present study, binding of ECM proteins immobilized on a polymer surface (latex) was analyzed since this seems more relevant to biomaterialrelated infections. No di!erence in expression of binding was found between strains isolated from the groups with or without infection. Possibly, a di!erence in binding may exist in strains of P. acnes, anaerobic bacteria or gram-negative rods but the number of isolates is too small to draw conclusions. The percentage of strains expressing binding of these immobilized proteins was much lower compared to the expression of soluble pro-
teins earlier reported [20]. One explanation might be that proteins immobilized on the polymer surface expose other domains than soluble proteins, and possibly not microbial binding domains. Furthermore, proteins immobilized on polymer surfaces may expose other domains than when adsorbed on to a titanium surface. The latter is hydrophilic [24] whereas polymer commonly express a hydrophobic surface. It is not known if microbes colonize by adhering to ECM proteins in the surrounding tissue or to proteins adsorbed on the titanium implant. The ECM proteins all have the Arg}Gly}Asp central domain mediating binding of eukaryotic cells. If microbes obscure this domain the tissue integration will be severely impaired. No bacteria have been identi"ed to bind to the Arg}Gly}Asp domain but we showed that P. acnes bind to a central domain of Fn, close to the heparin binding domain and Arg}Gly}Asp domain [21]. In the present study, no isolates expressed a hydrophobic cell surface. However, the isolates from clinically infected implants were less hydrophilic than those from non-in#amed tissue. We also earlier isolated hydrophobic strains from patients with more severe infections around percutaneous titanium implants than from noninfected tissue [20]. The di!erence in hydrophobicity and hydrophilicity of di!erent implants are of substantial importance. Metals have hydrophilic surfaces whereas polymers usually have hydrophobic surfaces. It is more energitically favorable for ECM proteins to adsorb to
K.M. Holgers, As. Ljungh / Biomaterials 20 (1999) 1319}1326
1325
wettability. We still encounter problems with percutaneous titanium implants despite the clinically acceptable results e.g. with anchorage of bone conductive hearing aids and craniofacial epistheses. The failure mode described by von Recum in 1984 is to a certain extent applicable for titanium implants [4]. There is epithelial downgrowth around titanium implants [27] but since these are anchored in bone tissue failure due to marsupialization has not been described. Infections occur but are often cured by local treatment. 20}30% of patients with percutaneous titanium implants required locally administered antibiotic treatment (terracortril with polymyxin) [5, 6, 28], and 1.5}4% of the implants were removed due to infection [5, 6]. In one patient with a severe skin infection during several months around a percutaneous implant the bone tissue appeared una!ected [29]. One might speculate that there is some kind of structural barrier at the periostium level. In#ammation induced by mechanical forces causing disruption of the interface area (avulsion) probably also occurs around bone anchored titanium implants. With other indications for percutaneous implants, e.g. cochlear implants the demand for a safer percutaneous coupling is evident. The pathogenesis of infection in association with titanium implants need to be elucidated in order to modify surfaces to become less prone to microbial colonization but still allow adhesion of eukaryotic cells, and hence tissue integration. Acknowledgements Fig. 7. The expression of cell surface hydrophobicity/hydrophilicity of microbial isolates, present around percutaneous titanium implants and their relation to the clinical "nding (grade 0, grade 1 and grade 2) are presented. Positive value denotes hydrophilicity. Expression of hydrophobicity (negative values) was not found.
a hydrophobic than to a hydrophilic surface [25]. This probably also in#uences the degree of bio"lm formation which generally is much lower on metal implants. When a skin-penetrating titanium implant becomes infected this is commonly treated successfully with soap and water [5, 6]. This is likely to be due to the hydrophilicity of the implant surface and the minor bio"lm formation, and to the cell surface hydrophilicity expressed by colonizing microbes. Proteins adhere with higher a$nity to hydrophobic polymer surfaces than to hydrophilic, and one can but speculate that hydrophobic microbes colonize polymer surfaces with higher a$nity than hydrophilic surfaces such as metal biomaterials. Also, the more severe histopathological reaction around percutaneous polymer implants for peritoneal dialysis [26] compared to our "ndings around titanium implants [27] can, to some extent, be explained by the di!erence in
This study was supported by the Board for Technical development (NUTEK) and the Swedish Medical Research Council (11229, 9495, 12286). We thank Dr. Gunilla Zachrisson for valuable help at the Clinical Bacteriology Laboratory, Sahlgrenska Hospital, GoK teborg, Sweden. References [1] Black J. Biological performances of biomaterials, 2nd ed. New York: Marcel Decker Inc, 1992. [2] Thomsen P, Ericson LEE. In#ammatory cell response to bone implant surfaces. In: Davies JE, editor. The bone}biomaterial interface. Toronto: University of Toronto Press, 1991:153}63. [3] Holgers KM. Soft tissue reactions around clinical skin-penetrating titanium implants. PhD thesis, University of GoK teborg, 1994. [4] von Recum A. Applications and failure modes of percutaneous devices: a review. J Biomed Mater Res 1984;18:323}36. [5] Holgers KM, TjellstroK m A, Bjursten LM, Erlandson BE. Soft tissue reaction around percutaneous implants: a clinical study on skin-penetrating titanium implants used for bone-anchored auricular prostheses. Int J Oral Maxillofac Implants 1987;2:35}9. [6] Holgers KM, TjellstroK m A, Bjursten LM, Erlandson BE. Soft tissue reactions around percutaneous implants: a clinical study on soft tissue conditions around skin-penetrating implants for boneanchored hearing aids. Am J Otol 1988:9:56}9.
1326
K.M. Holgers, As. Ljungh / Biomaterials 20 (1999) 1319}1326
[7] Vaudeaux P, Pittet D, Haeberlu A, Lerch PG, Morgenthaler J-J, Proctor RA, Waldvogel FA, Lew DP. Fibronectin is more active than "brin or "brinogen in promoting Staphylococcus aureus to inserted intravascular catheters. J Infect Dis 1993; 167:633}41. [8] Lundberg F, Schliamser S, Ljungh As . Vitronectin may mediate adhesion of coagulase-negative staphylococci to biomaterials exposed to cerebrospinal #uid. J Med Microbiol 1997;46: 285}96. [9] Bisno AL, Waldvogel FA. Infections associated with indwelling medical devices. Washington DC: American Society for Microbiology, 1989. [10] Christensen GD, Baddour LM, Hasty DL, Lowrance JH, Simpson WA. Microbial and foreign body factors in the pathogenesis of medical device infections. In: Bisno AL, Waldvogel FA, editors. Infections associated with indwelling medical devices. Washington DC: American Society for Microbiology, 1989:27}60. [11] Gristina AG. Biomaterial-centered infection: microbial adhesion versus tissue integration. Science 1987;237:1588}95. [12] Barth E, Myrvik QM, Wagner W, Gristina AG. In vitro and in vivo camparative colonization of Staphylococcus aureus and Staphylococcus epidermidis on orthopaedic implant materials. Biomaterials 1989;10:325}8. [13] SzoK ke I, Pascu C, Nagy E, Ljungh As , WadstroK m T. Binding of extracellular matrix proteins to the surface of Bacteroides spp. Anaerobe 1997;3:91}5. [14] Hogt AH, Dankert J, Feijen J. Adhesion of Staphylococcus epidermidis and Staphylococcus saprophyticus to a hydrophobic biomaterial. J Gen Microbiol 1985;131:2485}91. [15] Lennette EH, Balows A, Hausler WJ Jr, Shadomy HJ, editors. Manual of clinical microbiology, 4th ed. Washington DC: American Society for Microbiology, 1985:263}77. [16] Paulsson M, Petersson AC, Ljungh As . Serum and tissue protein binding and cell surface properties of Staphylococcus lugdunensis. J Med Microbiol 1993;38:96}102. [17] Naidu AS, Paulsson M, WadstroK m T. Particle agglutination assay for rapid detection of "bronectin, "brinogen and collagen receptors on Staphylococcus aureus. J Clin Microbiol 1988;26: 1549}54.
[18] Yatohgo T, Izumi M, Kashiwagi H, Hayashi M. Novel puri"cation of vitronectin from human plasma by heparin a$nity chromatography. Cell Struct Funct 1988;13:281}92. [19] Vuento M, Vaheri A. Puri"cation of "bronectin from plasma by a$nity chromatography under non-denaturing conditions. Biochem J 1979;183:331}7. [20] Holgers KM, Paulsson M, TjellstroK m A, Bjursten LM, Ljungh As . Selected microbial "ndings in association with percutaneous titanium implants. Int J Oral Maxillofac Implants 1994;9:565}70. [21] Yu J-L, Ma nsson R, Flock J-I, Ljungh A> . Fibronectin binding by Propionibacterium acnes. FEMS Immunol Microbiol 1997; 19:247}53. [22] Ljungh A> , Moran AP, WadstroK m T. Interactions of bacterial adhesins with extracellular matrix and plasma proteins: pathogenic implications and therapeutic possibilities. FEMS Immunol Med Microbiol 1996;16:117}26. [23] Liang OD, Ascencio F, Vazquez-Juarez R, WadstroK m T. Binding of collagen, "bronectin, lactoferrin, laminin, vitronectin and heparan sulfate to Staphylococcus aureus strain V8 at various growth phases and under nutrient stress conditions. ZBl Bakt Hyg 1993;179:180}90. [24] Lausmaa J, Kasemo B. Surface spectroscopic characterization of titanium implant materials. Appl Surf Sci 1990;44:135}46. [25] Lestelius M. Tailor-made monolayer asemblies for in vitro studies of blood protein}surface interaction. LinkoK ping studies in Science and technology, PhD dissertation, No 455, 1996:85}109. [26] Twardowski Z, Dobble JM, Moore HL, Nichols WK, Despain JD, Anderson PC, Khanna R, Nolph KD, Loy TS. Morphology of peritoneal dialysis catheter tunnel: macroscopy and light microscopy. Perit Dial Int 1991;11:237}51. [27] Holgers KM, Thomsen P, TjellstroK m A, Ericson L, Bjursten LM. Morphological evaluation of clinical long-term percutaneous titanium implants. Int J Oral Maxillofac implants 1994;9:689}97. [28] TjellstroK m A, GranstroK m G. One stage procedure to establish osseointegration: 0}5 years #ow up report. J Laryngol Otol 1995;109:593}8. [29] Holgers KM, Thomsen P, TjellstroK m A. Persistent soft tissue irritation around an osseointegrated titanium implant. A case report. Scand J Plast Reconstr Surg 1994;28:225}30.
Cell surface characteristics of microbiological isolates from human percutaneous titanium implants in the head and neck K.M. Holgers *, As . Ljungh Department of Anatomy and Cell Biology, University of GoK teborg, GoK teborg, Sweden Department of Audiology, Sahlgren+s University Hospital, University of GoK teborg, GoK teborg, Sweden Department of Infectious Diseases and Medical Microbiology, Lund University, Lund, Sweden Received 11 February 1998; accepted 5 February 1999
Abstract Percutaneous implants are commonly associated with several problems, and di!erent failure modes have been described. Infections constitute one serious complication which may lead to the removal of the implant. In contrast to infections around polymer implants, infections around skin-penetrating titanium implants anchored in the temporal bone are often cured by local treatment. Coagulase-negative staphylococci are the most common etiological agents in infections related to polymers whereas Staphylococcus aureus is considered as the main pathogen in infections around metallic implants. Microbial adhesion is a prerequisite for an infection. In the present study, the cell surface of microbes isolated from the skin around skin-penetrating titanium implants, with and without signs of infection, was characterized with respect to expression of cell surface hydrophobicity and to binding of immobilized "bronectin, vitronectin and collagen type 1 which could mediate adhesion. Expression of protein binding was similar in strains isolated from the two groups. No strain expressed a hydrophobic cell surface as determined by two-phase separation, and we conclude that the microenvironment around a titanium implant promotes expression of a hydrophilic rather than a hydrophobic cell surface which in turn makes many infections around a titanium implant curable by local treatment. 1999 Elsevier Science Ltd. All rights reserved Keywords: Titanium; Percutaneous; Implants; Infection; Biomaterials; Clinical
1. Introduction Foreign materials are introduced into the human body for temporary or clinical use with an increasing frequency. Examples are continuous ambulatory peritoneal dialysis catheters, intravenous catheters, intraocular lenses and attachment of prostheses for amputated limbs or craniofacial defects. After implantation, an in#ammatory reaction is induced in the tissue which also a!ects the biomaterial [1, 2]. When the biomaterial also penetrates the skin additional load is applied on the system since the skin barrier is broken and exogenous agents more easily colonize the material and the surrounding
* Corresponding address: Department of Audiology, Sahlgren's University Hospital, University of GoK teborg, GoK teborg, Sweden. Fax: 0046-31-829811.
tissue [3]. Skin-penetrating implants can be considered as a foreign body penetrating the skin through a surgically created defect. Percutaneous implants are often associated with di!erent failure modes such as marzupialization, permigration, avulsion and infections [4]. Established infections around implants usually lead to removal of the implant. In contrast to infections associated with polymers, infections around skin-penetrating titanium implants are often cured with local treatment [5, 6]. Within fractions of a second after implantation host proteins adsorb to the biomaterial surface. The protein adsorption depends on physicochemical properties of the material, like hydrophobicity and charge, and the composition of the host proteinaceous #uid. Microbial adhesion preferentially occur to these surface-adsorbed structures. Fibronectin (Fn) has been proposed to mediate adhesion of staphylococci to materials in the blood
0142-9612/99/$ - see front matter 1999 Elsevier Science Ltd. All rights reserved. PII: S 0 1 4 2 - 9 6 1 2 ( 9 9 ) 0 0 0 3 3 - 2
1320
K.M. Holgers, As. Ljungh / Biomaterials 20 (1999) 1319}1326
stream [7]. We have reported that vitronectin (Vn) could mediate adhesion of coagulase-negative staphylococci (CNS) to polymers in cerebrospinal #uid [8]. CNS, like Staphylococcus epidermidis are the most common etiological agents in infections related to polymers [9, 10] whereas S. aureus dominates in infections around metal implants [11]. S. aureus was shown preferentially to colonize metal alloys (Ti6A14V) whereas S. epidermidis prefers polymer surfaces [12]. In 14 patients with infections associated with external pin "xation of fractures, S. aureus was isolated from 8 patients and gram-negative rods from 6. The isolation of CNS was not reported, nor were analyses of patients without clinical infection done. In patients with skin-penetrating titanium implants we reported on a polymicrobial #ora. S. aureus was the most commonly isolated species in patients with and without clinical signs of infection [3]. S. aureus can speci"cally bind Fn, Vn, collagen (Cn) and other extracellular matrix (ECM) proteins. Binding of several ECM proteins has been described by CNS strains, preferentially to proteins immobilized on a surface, by Pseudomonas aeruginosa and several gram-negative species, and by anaerobic species [13]. Hydrophobic surface proteins have also been proposed to mediate adhesion [14 ]. The aim of the present study was to characterize microbes isolated around percutaneous titanium implants with respect to expression of binding of immobilized ECM proteins and of cell surface hydrophobicity. We further wanted to correlate these "ndings to signs and degree of infection around the implants. Elucidation of adhesive processes between microbes and biomaterials is vital for the creation of surfaces which are less prone to become infected.
2. Materials and methods Surgical procedures. The installation of the titanium implant is made in two stages. In the "rst, a threaded implant of commercially pure titanium is inserted into the temporal bone behind the ear-canal. The hole is threaded and the screw-shaped "cture is installed. After suturing the skin the implant is left unloaded to heal for 3}4 months. The skin-penetrating part is adapted in a second operation. To get a thin skin-penetration area it is generally necessary to surgically reduce the thickness of the subcutaneous tissue and to place a thin hairless skin graft at the site of the implant. Three or four weeks later the hearing aid is "tted. Patients. Twenty-six patients (14 women and 12 men) with permanent skin-penetrating titanium implants for bone anchored hearing aids (BAHA) were studied. The duration of the skin penetration was 24}207 months (mean 94 months, SD 54). The skin condition was classi"ed as grade 0*no signs of infection around the implant, grade 1*slightly reddish skin around the implant, and
Table 1 The relation between the clinical "ndings and isolation of di!erent microbiological strains around the percutaneous titanium implants
CNS S. aureus Propionibacterium acnes Bacteroides urealyticus Proteus Klebsiella E. coli Peptostreptococcus
Grade 0
Grade 1
Grade 2
19 2 2 1 2 1 1 1
15 2 9
2 2
grade 2*reddish and moist skin around the implant. The classi"cation of our patients in grade 0}2 is presented in Table 1. Microbiology. Specimens were obtained with a sterile cotton swab (1.0 mm in diameter) from the skin immediately adjacent to the implant anchoring the BAHA. Touching the titanium implant during sampling was unavoidable. The specimens were inoculated on blood agar (5% v/v horse erythrocytes, Columbia agar base, ACU, Bethesda, MD, USA) for 48 h at 373C in aerobic and anaerobic atmospheres. Isolated colonies were typed to the species level according to the ASM Manual of Clinical Microbiology [15]. CNS were typed to the species level with an in-house panel of biochemical tests [16]. The isolates were lyophilized. For further testing the strains were dissolved in distilled water and cultured on blood agar as above. Bacterial strains were washed twice in 0.02 M potassium phosphate bu!er, pH 6.8, and resuspended in the same bu!er to approximately 10 colony forming units/ml (CFU/ml). Cell surface hydrophobicity. Polyethylene glycol (7.13% w/v, PEG 6000; KEBO, Stockholm, Sweden) and Dextran (8.75% w/v in 0.015 M NaCl, pH 6.8; molecular weight 48 000, Sigma Chemical Co, St Louis, MO, USA) were prepared in an aqueous two-phase system as earlier described [16]. Partitioning was performed by adding 100 ll of bacterial suspension (approx. 5;10 CFU/ml) to 0.9 ml of the phase system, mixing by gentle shaking, and allowing phase separation at 203C for 1 h. The concentration of bacterial cells was measured turbidimetrically in the PEG-rich top phase and in the dextran-rich bottom phase at 540 nm, and was expressed as percentage of the original concentration of added cells. Negative values denote hydrophobicity and positive values hydrophilicity. Particle agglutination assay (PAA). Vn, Fn and Cn were separately immobilized on latex particles (0.8 lm diameter; Difco Laboratories, Detroit, MI, USA) as earlier described [17]. After adsorption of 100 lg protein to 1 ml latex beads (303C, 1 h) on a horizontal shaker the mixtures were centrifuged (9200 g, 5 min, 43C) the pellets
K.M. Holgers, As. Ljungh / Biomaterials 20 (1999) 1319}1326
were resuspended in 2 ml glycine bu!er (0.17 M glycine NaOH, pH 8.2) with ovalbumin (0.01%; Sigma Chem Co.) and merthiolate (0.01%) and kept at 43C. The assay was performed by gently mixing equal volumes (20 ll) of latex reagent and bacterial suspension on glass slides. The reactions were scored after 2 min and assigned a numerical value: 0"no agglutination, negative, 1"moderate agglutination, positive, and 2" strong agglutination. Chemicals. Vn was puri"ed from human urea-treated plasma on heparin}sepharose [18] and Fn from human plasma on gelatin}sepharose [19]. Vitrogen (containing 95% collagen type 1 and 5% collagen type III) was purchased from Collagen Corporation, Palo Alto, CA, USA. Statistical methods. The Student's t-test was used when appropriate.
3. Results Thirty bacterial strains were isolated from 13 patients with clinical signs of infection, and 28 strains from 13 patients without signs of infection. Aerobic and anaerobic bacteria were isolated from patients with and without signs of infection (Table 1). CNS were the most commonly isolated species in this group of patients with percutaneous implants. S. aureus was isolated from 2 patients each with grade 2, grade 1 and grade 0. There were more anaerobic isolates in the group with skin irritation.
1321
No fungi were isolated. Two strains (Peptostreptococcus sp and CNS) could not be recovered from the lyophilized state, and were hence not further studied. The relation between the clinical "ndings and the distribution of bacterial isolates is presented in Table 1. In patients without sign of infection (grade 0) the bacterial #ora contained a higher variety of species than in the groups with signs of infection. The relation between clinical "ndings and expression of binding of Cn, Fn and Vn is presented in Figs. 1}6. Of 24 strains isolated from patients without clinical signs of infection 11 expressed binding of Cn type I and Fn, and 13 of Vn. In the groups with clinical signs of infection binding of Cn type I, Fn and Vn was detected in 12, 10 and 12 strains, respectively, out of 29 strains. The expression of binding of Cn type I, Fn and Vn was hence the same in strains isolated from patients with or without signs of infection. Two strains of P. acnes and 1 S. aureus from patients without clinical signs of infection agglutinated spontaneously in the PAA, and 1 P. acnes from one patient with signs of infection. The expression of cell surface hydrophobicity/hydro philicity by strains isolated from skin-penetrating implants is presented in Fig. 7. No strain expressed a hydrophobic cell surface. However, strains isolated from the group of patients with clinical signs of infection were less hydrophilic compared with isolates from patients without signs of infection. The mean value in the group with in#ammation was 4.29 (SE 0.73) and in the group
Fig. 1. Binding of collagen type I in clinically non-irritated skin. The expression of binding is described as: no agglutination (0), positive (1) or strongly positive (2). The number of strains are presented.
1322
K.M. Holgers, As. Ljungh / Biomaterials 20 (1999) 1319}1326
Fig. 2. Binding of collagen type I in clinically irritated skin. The expression of binding is described as: no agglutination (0), positive (1) or strongly positive (2). The number of strains are presented.
Fig. 3. Binding of "bronectin in clinically non-irritated skin. The expression of binding is described as: no agglutination (0), positive (1) or strongly positive (2). The number of strains are presented.
without clinical signs of infection 4.75 (SE 0.92). The di!erence in expressing hydrophilicity/hydrophobicity between isolates from grade 0 and isolates from grade 1 and 2 is not statistically signi"cant.
4. Discussion We have earlier described a polymicrobial #ora around percutaneous titanium implants with S. aureus as
K.M. Holgers, As. Ljungh / Biomaterials 20 (1999) 1319}1326
1323
Fig. 4. Binding of "bronectin in clinically irritated skin. The expression of binding is described as: no agglutination (0), positive (1) or strongly positive (2). The number of strains are presented.
Fig. 5. Binding of vitronectin in clinically non-irritated skin. The expression of binding is described as: no agglutination (0), positive (1) or strongly positive (2). The number of strains are presented.
the most commonly isolated bacterial species in samples from patients with and without clinical signs of infection [20]. In the present study we also report on a polymicrobial #ora around the implants but here, the pre-
dominating species is CNS. The skin conditions in the group with infection was less severe in this study than in one earlier study [20]. Anaerobic bacteria were isolated from patients with and without infection. However, the
1324
K.M. Holgers, As. Ljungh / Biomaterials 20 (1999) 1319}1326
Fig. 6. Binding of vitronectin in clinically irritated skin. The expression of binding is described as: no agglutination (0), positive (1) or strongly positive (2). The number of strains are presented.
number of anaerobic isolates was higher in the group with clinical signs of infection. CNS, S. aureus and many anaerobic species form part of the normal microbial #ora on skin and mucous membranes. Hence, species from the normal micro#ora commonly colonize skin-penetrating devices. ECM proteins, especially Fn, have been proposed to mediate adhesion to surfaces [7, 8]. Expression of binding of ECM proteins has been described by S. aureus, CNS, P. acnes [21] and several other microbial species [22]. Expression of binding is in#uenced by the microbial growth conditions. It has earlier been shown that bacteria commonly express binding after growth in nutrient-poor conditions, like the skin [23]. In the former study we analyzed expression of binding of Fn, Vn and Cn type 1 in soluble form [20]. Binding was expressed by 86% of the strains but we did not "nd any di!erences between strains isolated from patients with or without clinical infection. In the present study, binding of ECM proteins immobilized on a polymer surface (latex) was analyzed since this seems more relevant to biomaterialrelated infections. No di!erence in expression of binding was found between strains isolated from the groups with or without infection. Possibly, a di!erence in binding may exist in strains of P. acnes, anaerobic bacteria or gram-negative rods but the number of isolates is too small to draw conclusions. The percentage of strains expressing binding of these immobilized proteins was much lower compared to the expression of soluble pro-
teins earlier reported [20]. One explanation might be that proteins immobilized on the polymer surface expose other domains than soluble proteins, and possibly not microbial binding domains. Furthermore, proteins immobilized on polymer surfaces may expose other domains than when adsorbed on to a titanium surface. The latter is hydrophilic [24] whereas polymer commonly express a hydrophobic surface. It is not known if microbes colonize by adhering to ECM proteins in the surrounding tissue or to proteins adsorbed on the titanium implant. The ECM proteins all have the Arg}Gly}Asp central domain mediating binding of eukaryotic cells. If microbes obscure this domain the tissue integration will be severely impaired. No bacteria have been identi"ed to bind to the Arg}Gly}Asp domain but we showed that P. acnes bind to a central domain of Fn, close to the heparin binding domain and Arg}Gly}Asp domain [21]. In the present study, no isolates expressed a hydrophobic cell surface. However, the isolates from clinically infected implants were less hydrophilic than those from non-in#amed tissue. We also earlier isolated hydrophobic strains from patients with more severe infections around percutaneous titanium implants than from noninfected tissue [20]. The di!erence in hydrophobicity and hydrophilicity of di!erent implants are of substantial importance. Metals have hydrophilic surfaces whereas polymers usually have hydrophobic surfaces. It is more energitically favorable for ECM proteins to adsorb to
K.M. Holgers, As. Ljungh / Biomaterials 20 (1999) 1319}1326
1325
wettability. We still encounter problems with percutaneous titanium implants despite the clinically acceptable results e.g. with anchorage of bone conductive hearing aids and craniofacial epistheses. The failure mode described by von Recum in 1984 is to a certain extent applicable for titanium implants [4]. There is epithelial downgrowth around titanium implants [27] but since these are anchored in bone tissue failure due to marsupialization has not been described. Infections occur but are often cured by local treatment. 20}30% of patients with percutaneous titanium implants required locally administered antibiotic treatment (terracortril with polymyxin) [5, 6, 28], and 1.5}4% of the implants were removed due to infection [5, 6]. In one patient with a severe skin infection during several months around a percutaneous implant the bone tissue appeared una!ected [29]. One might speculate that there is some kind of structural barrier at the periostium level. In#ammation induced by mechanical forces causing disruption of the interface area (avulsion) probably also occurs around bone anchored titanium implants. With other indications for percutaneous implants, e.g. cochlear implants the demand for a safer percutaneous coupling is evident. The pathogenesis of infection in association with titanium implants need to be elucidated in order to modify surfaces to become less prone to microbial colonization but still allow adhesion of eukaryotic cells, and hence tissue integration. Acknowledgements Fig. 7. The expression of cell surface hydrophobicity/hydrophilicity of microbial isolates, present around percutaneous titanium implants and their relation to the clinical "nding (grade 0, grade 1 and grade 2) are presented. Positive value denotes hydrophilicity. Expression of hydrophobicity (negative values) was not found.
a hydrophobic than to a hydrophilic surface [25]. This probably also in#uences the degree of bio"lm formation which generally is much lower on metal implants. When a skin-penetrating titanium implant becomes infected this is commonly treated successfully with soap and water [5, 6]. This is likely to be due to the hydrophilicity of the implant surface and the minor bio"lm formation, and to the cell surface hydrophilicity expressed by colonizing microbes. Proteins adhere with higher a$nity to hydrophobic polymer surfaces than to hydrophilic, and one can but speculate that hydrophobic microbes colonize polymer surfaces with higher a$nity than hydrophilic surfaces such as metal biomaterials. Also, the more severe histopathological reaction around percutaneous polymer implants for peritoneal dialysis [26] compared to our "ndings around titanium implants [27] can, to some extent, be explained by the di!erence in
This study was supported by the Board for Technical development (NUTEK) and the Swedish Medical Research Council (11229, 9495, 12286). We thank Dr. Gunilla Zachrisson for valuable help at the Clinical Bacteriology Laboratory, Sahlgrenska Hospital, GoK teborg, Sweden. References [1] Black J. Biological performances of biomaterials, 2nd ed. New York: Marcel Decker Inc, 1992. [2] Thomsen P, Ericson LEE. In#ammatory cell response to bone implant surfaces. In: Davies JE, editor. The bone}biomaterial interface. Toronto: University of Toronto Press, 1991:153}63. [3] Holgers KM. Soft tissue reactions around clinical skin-penetrating titanium implants. PhD thesis, University of GoK teborg, 1994. [4] von Recum A. Applications and failure modes of percutaneous devices: a review. J Biomed Mater Res 1984;18:323}36. [5] Holgers KM, TjellstroK m A, Bjursten LM, Erlandson BE. Soft tissue reaction around percutaneous implants: a clinical study on skin-penetrating titanium implants used for bone-anchored auricular prostheses. Int J Oral Maxillofac Implants 1987;2:35}9. [6] Holgers KM, TjellstroK m A, Bjursten LM, Erlandson BE. Soft tissue reactions around percutaneous implants: a clinical study on soft tissue conditions around skin-penetrating implants for boneanchored hearing aids. Am J Otol 1988:9:56}9.
1326
K.M. Holgers, As. Ljungh / Biomaterials 20 (1999) 1319}1326
[7] Vaudeaux P, Pittet D, Haeberlu A, Lerch PG, Morgenthaler J-J, Proctor RA, Waldvogel FA, Lew DP. Fibronectin is more active than "brin or "brinogen in promoting Staphylococcus aureus to inserted intravascular catheters. J Infect Dis 1993; 167:633}41. [8] Lundberg F, Schliamser S, Ljungh As . Vitronectin may mediate adhesion of coagulase-negative staphylococci to biomaterials exposed to cerebrospinal #uid. J Med Microbiol 1997;46: 285}96. [9] Bisno AL, Waldvogel FA. Infections associated with indwelling medical devices. Washington DC: American Society for Microbiology, 1989. [10] Christensen GD, Baddour LM, Hasty DL, Lowrance JH, Simpson WA. Microbial and foreign body factors in the pathogenesis of medical device infections. In: Bisno AL, Waldvogel FA, editors. Infections associated with indwelling medical devices. Washington DC: American Society for Microbiology, 1989:27}60. [11] Gristina AG. Biomaterial-centered infection: microbial adhesion versus tissue integration. Science 1987;237:1588}95. [12] Barth E, Myrvik QM, Wagner W, Gristina AG. In vitro and in vivo camparative colonization of Staphylococcus aureus and Staphylococcus epidermidis on orthopaedic implant materials. Biomaterials 1989;10:325}8. [13] SzoK ke I, Pascu C, Nagy E, Ljungh As , WadstroK m T. Binding of extracellular matrix proteins to the surface of Bacteroides spp. Anaerobe 1997;3:91}5. [14] Hogt AH, Dankert J, Feijen J. Adhesion of Staphylococcus epidermidis and Staphylococcus saprophyticus to a hydrophobic biomaterial. J Gen Microbiol 1985;131:2485}91. [15] Lennette EH, Balows A, Hausler WJ Jr, Shadomy HJ, editors. Manual of clinical microbiology, 4th ed. Washington DC: American Society for Microbiology, 1985:263}77. [16] Paulsson M, Petersson AC, Ljungh As . Serum and tissue protein binding and cell surface properties of Staphylococcus lugdunensis. J Med Microbiol 1993;38:96}102. [17] Naidu AS, Paulsson M, WadstroK m T. Particle agglutination assay for rapid detection of "bronectin, "brinogen and collagen receptors on Staphylococcus aureus. J Clin Microbiol 1988;26: 1549}54.
[18] Yatohgo T, Izumi M, Kashiwagi H, Hayashi M. Novel puri"cation of vitronectin from human plasma by heparin a$nity chromatography. Cell Struct Funct 1988;13:281}92. [19] Vuento M, Vaheri A. Puri"cation of "bronectin from plasma by a$nity chromatography under non-denaturing conditions. Biochem J 1979;183:331}7. [20] Holgers KM, Paulsson M, TjellstroK m A, Bjursten LM, Ljungh As . Selected microbial "ndings in association with percutaneous titanium implants. Int J Oral Maxillofac Implants 1994;9:565}70. [21] Yu J-L, Ma nsson R, Flock J-I, Ljungh A> . Fibronectin binding by Propionibacterium acnes. FEMS Immunol Microbiol 1997; 19:247}53. [22] Ljungh A> , Moran AP, WadstroK m T. Interactions of bacterial adhesins with extracellular matrix and plasma proteins: pathogenic implications and therapeutic possibilities. FEMS Immunol Med Microbiol 1996;16:117}26. [23] Liang OD, Ascencio F, Vazquez-Juarez R, WadstroK m T. Binding of collagen, "bronectin, lactoferrin, laminin, vitronectin and heparan sulfate to Staphylococcus aureus strain V8 at various growth phases and under nutrient stress conditions. ZBl Bakt Hyg 1993;179:180}90. [24] Lausmaa J, Kasemo B. Surface spectroscopic characterization of titanium implant materials. Appl Surf Sci 1990;44:135}46. [25] Lestelius M. Tailor-made monolayer asemblies for in vitro studies of blood protein}surface interaction. LinkoK ping studies in Science and technology, PhD dissertation, No 455, 1996:85}109. [26] Twardowski Z, Dobble JM, Moore HL, Nichols WK, Despain JD, Anderson PC, Khanna R, Nolph KD, Loy TS. Morphology of peritoneal dialysis catheter tunnel: macroscopy and light microscopy. Perit Dial Int 1991;11:237}51. [27] Holgers KM, Thomsen P, TjellstroK m A, Ericson L, Bjursten LM. Morphological evaluation of clinical long-term percutaneous titanium implants. Int J Oral Maxillofac implants 1994;9:689}97. [28] TjellstroK m A, GranstroK m G. One stage procedure to establish osseointegration: 0}5 years #ow up report. J Laryngol Otol 1995;109:593}8. [29] Holgers KM, Thomsen P, TjellstroK m A. Persistent soft tissue irritation around an osseointegrated titanium implant. A case report. Scand J Plast Reconstr Surg 1994;28:225}30.