Biomaterials 23 (2002) 517–526

Inflammatory responses to orthopaedic biomaterialsin the murine air pouch

Paul H. Wooleya,b,*, Robert Morrena, John Andarya, Sudha Suda, Shang-You Yanga, LoisMaytona, David Markela, Allison Sievinga, Sam Nassera,b

aDepartment of Orthopaedic Surgery, Wayne State University School of Medicine, Hutzel Hospital, One South, 4707 St. Antoine Boulevard,

Detroit, MI 48201, USAbThe Veterans Administration Medical Center, Detroit, MI 48201, USA

Received 4 December 2000; accepted 3 April 2001

Abstract

An in vivo model of the inflammatory response to orthopaedic biomaterials was used to examine cellular and cytokine responsesto polymer particles of ultra high molecular weight polyethylene (UHMWPE) and polymethylmethacrylate (PMMA), and metalparticles of cobalt–chrome (Co–Cr) and titanium alloy (Ti–6Al–4V). Responses were determined separately and in combinations, toexamine interactions between different forms of biomaterials. Murine air pouches were injected with particle suspensions, and

reactions evaluated using histological, immunological, and molecular techniques. All particulate biomaterials caused significantincreases in membrane thickness compared with control (saline) air pouches, with the highest reaction seen in response to Ti–6Al–4V particles. A synergistic increase in membrane thickness was observed when PMMA was combined with UHMWPE, suggesting

that multiple biomaterial stimuli markedly increase the inflammatory reaction. Cellular analysis indicated that all particlesincreased the absolute number and the percentage of macrophages in the membrane over the control level, with the mostpronounced increase due to individual biomaterial occurring with UHMWPE particles. Cytokine analysis revealed that biomaterials

provoked a strong IL-1 response. Ti–6Al–4V stimulated the highest IL-6 gene transcription and the lowest IL-1 gene transcription.The data suggest that synergism in the inflammatory response to biomaterials may be important in adverse responses to orthopaedicwear debris. r 2001 Elsevier Science Ltd. All rights reserved.

Keywords: Orthopaedic biomaterials; Particulates; Inflammation; Polymers; Alloys

1. Introduction

A critical aspect of adverse responses to the bioma-terials used in the construction of orthopaedic prostheticcomponents is the stimulation of cells in the peri-prosthetic tissue by particles resulting from wear of theprosthesis. The peri-prosthetic tissue serves as an inter-face between the prosthesis and the bone, and containscells of the macrophage/monocyte lineage. Thus smallparticles are readily phagocytosed by the resident cells ofthe peri-prosthetic membrane. Plastics and metal areimpervious to enzymatic destruction, and wear particlesdisturb the biological degradative function of phago-

cytes that ingest the debris [1]. Repeated phagocytosis ofthe particles results in activated cells that secrete bothproteolytic enzymes and high levels of the inflammatorycytokines IL-1 and TNFa, which may in turn contributeto the osteolytic process by providing activation signalsto osteoclasts [2]. Biomaterial debris also appears toprovoke other biological effects, including granulomaformation and inflammatory cell influx, which may alsocontribute to bone resorption, osteolysis and theeventual loss of the prosthesis support. The tissue fromthe areas of osteolysis shows a synovial-like layer at thecement surface, and the presence of macrophages andforeign-body giant cells invading the femoral cortices.This appearance shares some histological characteristicsof both rheumatoid arthritis (RA) and a foreign bodyreaction. Microscopic examination of specimens ob-tained at revision surgery of failed hip replacements has

*Corresponding author. Tel.: +1-313-745-6828; fax: +1-313-993-

0857.

E-mail address: ad8754@wayne.edu (P.H. Wooley).

0142-9612/02/$ - see front matter r 2001 Elsevier Science Ltd. All rights reserved.

PII: S 0 1 4 2 - 9 6 1 2 ( 0 1 ) 0 0 1 3 4 - X

revealed a varied cellular composition of the pseudosy-novium, with histiocytes, giant cells, lymphocytes,plasma cells and neutrophils all present, with the areasaround the loosened prostheses characterized as aggres-sive granulomatous lesions consisting of well organizedtissue containing histiocytic–monocytic and fibroblasticreactive zones [3]. Furthermore, immunohistologicalevaluation has revealed the presence of multinucleatedgiant cells and C3bi-receptor bearing monocyte–macro-phages [4]. Particles of acrylic cement and shardsof polyethylene appear incorporated into the histio-cyte/macrophage or giant cell population, resulting in‘‘foci’’ of cellular activity within the synovial-likemembrane [5].It is recognized that any type of particulate debris,

including that generated by poor surgical technique, lossof mechanical fixation of the polymethylmethacrylate(PMMA) bone cement, or wear at the ultra-highmolecular weight polyethylene (UHMWPE)–metal in-terface, may lead to the aseptic loosening process.However, the critical aspects of biomaterial debris thatprovoke the most severe cellular responses remainunclear, and quantitative analysis of cell types presentin this periprosthetic tissue revealed considerableheterogeneity between tissues from different individuals.Hence a predictable model of the inflammatory responseto biomaterials in particulate form would be useful toexamine the molecular pathogenesis of the steps leadingto the osteolytic process. We have adapted a murinemodel of inflammation that closely resembles the peri-prosthetic tissue encountered in aseptically loosenedprosthetic components to evaluate the cellular responseto biomaterials in vivo. The rodent air pouch has beenidentified as a useful model for the evaluation of theresponse to orthopaedic biomaterials [6,7], providingcellular infiltration and mediators of inflammation thatappear to closely resemble the pseudosynovium asso-ciated with aseptic loosening. In this study we haveexamined the cellular and cytokine reactions to bothmetal and polymeric biomaterials, and demonstrate thatthe model is sensitive differences in the materialcomposition of the particles under investigation.

2. Methods

2.1. Murine air pouches

Air pouches were generated according to themethods of Sedgewick et al. [8] in groups of 10–15female BALB/c mice. An area of the dorsal skin (2 cm2)was cleaned with alcohol and shaved to provide thepouch site. A subcutaneous injection of 3ml of air wasinjected at a single site with 25-gauge needle and 3mlsyringe. The air pouches were injected with 1ml of airon alternate days for 5 days to establish a definitive fluidfilled pouch. On day 6, the pouches were injected with500 ml of suspension containing 5% weight/volumebiomaterial particles. Control pouches received 500 mlof sterile PBS alone. The mice were sacrificed 48 h afterthe introduction of particles into the pouch. Thepouches were lavaged with 1ml of saline, and the lavagefluid rapidly frozen at @801C for evaluation ofcytokines. The pouches were then dissected free fromthe surrounding tissue. The pouch was divided, withone-half fixed in formalin for histological evaluation,and the remainder snap-frozen in liquid nitrogen forRNA extraction.

2.2. Biomaterial particles

Four orthopaedic biomaterials in particulate formwere evaluated for reactions in the murine air pouch,alone and in combinations (Table 1). The size anddistribution of the particles was evaluated with aCoulter particle counter equipped with interchangeable(100, 30, and 15 mm pore) attachments, and by scanningelectron microscopy (SEM). Particles for SEM analysiswere dispersed on a 0.1 mm Isopore membrane filter, anddried for 24 h. PMMA and UHMWPE samples weregold coated using a Fullam sputter coater prior to SEMimaging using an S-2400 Hitachi scanning electronmicroscope. Samples were imaged at 800� magnifica-tion to visual particle size characteristics and particleconcentration distribution, which was analyzed usingthe Image Pro software package (Media Cybernetics,

Table 1

Experimental study groups. Mice were injected with 500ml of suspension containing 5% weight/volume particles. Groups with particle combinations

were injected with 500ml comprised of 250ml from each individual biomaterial suspension

Group Injection suspension Mean diameter (mm) Size range (mm)

1 Vehicle control (sterile PBS) na na

2 UHMWPE 3.6 2.0–23

3 PMMA 4 0.1–10

4 Co–Cr 5.7 1.0–20

5 Ti–6Al–4V 2.3 0.1–68

6 UHMWPE+PMMA 3.7 0.1–23

7 UHMWPE+Co–Cr 4.2 1.0–23

8 UHMWPE+Ti–6Al–4V 3.1 0.1–68

P.H. Wooley et al. / Biomaterials 23 (2002) 517–526518

Maryland). SEM imaging (Fig. 1), image analysis, andparticle analysis revealed all materials to be predomi-nantly spherical in shape. Ultra-high molecular weightpolyethylene (UHMWPE) particles (the generous gift ofDr. John Cuckler, University of Alabama, Birmingham,AL) had a mean particle diameter of 3.6 mm, withdiameters ranging from 2.0 to 23 mm. Cobalt–chromiumalloy particles (mean particle size 5.7 mm, range 1–20 mm) were the generous gift of Dr. Jack Parr (WrightMedical Inc., Memphis, TN). Titanium alloy particles(Ti–6Al–4V) (mean particle size 2.3 mm, size range 0.1–68 mm) were the generous gift of the Zimmer Corpora-tion (Warsaw, Indiana). PMMA particles with (meandiameter 4.0 mm, range 0.1–10 mm) were obtained from acommercial source (Polyscience, Warrington, PA).The metal and PMMA particles were washed in70% ethanol solution to remove bound endotoxin [9],and the absence of endotoxin was confirmed using theLimulus assay (Endosafe; Charles Rivers, Charlestown,SC). Particles were suspended in sterile PBS forinjection.

2.3. Histological evaluation and image analysis

Tissue samples were fixed, dehydrated, and embeddedin paraffin blocks with particular care to preserve the

original shape of the pouch tissue. Sections were cutalong a longitudinal axis at approximately the pouchmidline, mounted and stained with hematoxylin andeosin (HaE) and Esterase stain. After staining, theslides were permanently bonded with coverslips. Mini-mums of three separate sections per specimen wereevaluated in a blinded fashion using the Image Proimage analysis software package. Pouch thickness wasdetermined at six points on each section, with an evendistribution of measurement on the proximal side, distalside, and transition curve of the pouch. Total number ofcells (based upon nucleus count) was determined as cellsper mm2, and cells differentiated into fibroblastic andhistiocytic type based upon nuclear morphology. Alldata was exported to an Excel spreadsheet, and meancellular tissue densities calculated.

2.4. Cytokine gene activation

Cytokine RNA signals were determined in pouchmembrane tissues using reverse transcription–polymer-ase chain reactions (RT–PCR) techniques describedpreviously [10,11]. Tissues were homogenized with aPolytron RT 2000 in 7.5m guanidium–HCl for 3min,and the sodium lauryl sarcosinate added to a finalconcentration of 0.5%. After centrifugation to remove

Fig. 1. Scanning electron microscopy (SEM) appearance of the particle preparations. 800� .

P.H. Wooley et al. / Biomaterials 23 (2002) 517–526 519

debris, and the addition of 2m potassium acetate and1m acetic acid, RNA was precipitated by the addition ofcold absolute ethanol. Quantity and purity of RNA wasdetermined by absorbance on a spectrophotometer(Beckman Instruments, Fullerton, CA.) at 260 and280 nm. Samples with ratios >1.7 were accepted foranalysis. To analyze the inflammatory cytokine geneexpression by the RT–PCR method, RNA was firstreverse transcribed into complementary DNA (cDNA).PCR amplification was performed with primers specificfor murine IL-1, IL-6 and TNFa (Clontech Inc.).Reactions for an individual sample were performedsimultaneously to ensure consistency in the analysis.One hundred nanograms of each cDNA was combinedin a reaction mixture with each primer pair, PCRbuffer, MgCl2, dNTPs, and AmpliTaq DNA polymerasewas added to a final concentration of 5U/ml.The reaction was started by heating at 941C for1min to denature the RNA–cDNA hybrid, thenannealing the primers at 551C for 1min, and extendingthe primer sequence at 721C for 1min. This cyclewas repeated 35 times using the DNA thermalcycler (Perkin Elmer, Norwalk, CT). Following ampli-fication, 10 ml of each reaction solution was mixed with1 ml of loading buffer (Perkin Elmer), loaded on a 1.8%agarose gel containing 0.5 mg/ml ethidium bromide(Sigma, St. Louis, MO) and run at 50V for 30min.All PCR assays were repeated to confirm the presence orabsence of the bands. The amplified PCR products werevisualized by ultraviolet light. The densities of all thevisualized bands of cytokine gene products weremeasured with an ISO 2000 Digital Imaging system(Alpha Innotech, San Leandro, CA). Each cytokinesignal was normalized relative to the PCR product froma housekeeping gene (GADPH) using cDNA from thesame sample.

2.5. Cytokine levels

Levels of murine IL-1 and TNFa in pouch lavage fluidwere determined by ELISA as described previously [12].Commercial kits (Genzyme, Cambridge, MA) were usedfor these experiments, in accordance with the manufac-turers instructions. The levels of each cytokine in thesample were calculated by regression analysis from astandard curve provided with the kit, assayed on thesame plate concomitantly.

2.6. Data analysis

Statistical comparisons of the experimental para-meters between groups were made using the ANOVAtest, with the Scheffe formula for post hoc multiplecomparisons, using the SPSS software package (SPSS,Chicago, IL).

3. Results

3.1. Air pouch membrane

The histological appearance of the control (saline) airpouch membrane is illustrated by Fig. 2. The controlmembrane was characterized by an outer fibrous layerpopulated mainly by fibroblastic cells and an innerinflammatory layer predominantly comprised of macro-phages. The mean thickness of control pouch mem-branes was 62.0 mm (S.D. 11.6 mm) while the number ofnucleated cells was 960/mm2 with a standard deviationof 54 cells. The differential count revealed the controlmembrane was comprised of 65% cells with fibroblasticmorphology, with the remaining cells (320 cells/mm2)classified as macrophage/monocytes based upon cellularmorphology and esterase histochemical staining. Giantcells were not observed in the saline-injected air pouchmembrane.

3.2. Tissue response to individual biomaterials

The introduction of any biomaterial particles into theair pouch caused dramatic increases in histologicalparameters of membrane thickness and cellularity,compared with control pouch membranes. Thesechanges are illustrated by Fig. 3A–D. The objectivemeasurements of changes in response to the differentbiomaterials (membrane thickness and cell count),determined by image analysis of the histologicalsections, are summarized in Figs. 4 and 5. Membranethickness in pouches stimulated by PMMA injection(Fig. 3A) was significantly increased (225.6 mm, S.D.22.7) compared to saline injected controls (po0.001),and the number of cells per mm2 were also elevated(1180 cells/mm2), which was also statistically significant

Fig. 2. The histological appearance of the control (saline) air pouch

membrane. The membrane is characterized by an outer fibrous layer

populated mainly by fibroblastic cells and an inner inflammatory layer

predominantly comprised of histiocytic cells.

P.H. Wooley et al. / Biomaterials 23 (2002) 517–526520

(po0.05). Fig. 3B illustrates the membrane changes thatoccur when the air pouch was injected with UHMWPE.Membrane thickness was increased (141 mm, S.D. 23.2)compared with control, and UHMWPE caused the

highest increase in the density of cells infiltrating thepouch membrane (1550 cells/mm2). Both of thesechanges were significant (po0.01) compared withcontrol values. Co–Cr also increased both membrane

Fig. 3. The histological appearance of particle-stimulated air pouch membranes. Five day established air pouches were injected with (A) PMMA, (B)

UHMWPE, (C) Co–Cr, or (D) Ti–6Al–4V particles.

Fig. 4. Membrane thickness of control and particle-stimulated air

pouch membranes.

Fig. 5. Membrane cellularity of control and particle-stimulated air

pouch membranes.

P.H. Wooley et al. / Biomaterials 23 (2002) 517–526 521

thickness (177 mm, S.D. 11.1) and membrane cellularity(1460 cells/mm2) to a highly significant degree oversaline control (po0.001 and o0.01, respectively), asshown in Fig. 3C. The highest increase in membranethickness was observed using Ti–6Al–4V particles(249 mm, S.D. 16.1) as illustrated in Fig. 3D. Ti–6Al–4V also provoked a significant increase (po0.05) in thecell count within the membrane (1280 cells/mm2). Whenindividual biomaterials were compared for variations inthe membrane responses, there were no substantialdifferences among the different particles with respect tothe total cell count per unit area within the membrane.This may suggest that an equivalent degree of cellstimulation occurred using the different particle sources.However, it was observed that Ti–6Al–4V particlescaused a significant increase (po0.05) membrane thick-ness compared with UHMWPE particles, suggestingthat this metallic alloy provokes a stronger membraneresponse compared to UHMWPE.

3.3. Tissue response to biomaterials in combination

The objective measurements of changes (membranethickness and cell count) in response to the combinationof different biomaterials with UHMWPE are shown inFigs. 4 and 5. While combinations of metallic particleswith UHMWPE did not provoke increased membranethickness over either particle alone, the combination ofUHMWPE with PMMA proved synergistic, with theresulting membrane thickness markedly elevated overfindings with either individual biomaterial. This in-creased thickness with the combination of UHMWPEand PMMA was significantly different from eitherPMMA or UHMWPE alone (po0.001).Cellular infiltration in membranes with combined

particles was elevated compared with individual parti-cles, but this increase was not significant compared withUHMWPE particles alone. However, the increases inmembrane cellularity were significant (po0.02) whenthe respective combinations were compared to PMMAalone or Ti–6Al–4V alone.

3.4. Differentiation cellular response to biomaterials

A differential count of the cellular infiltrate wasdetermined using histochemical staining and imageanalysis of nuclear morphology. In contrast to thepredominantly fibroblastic composition of the control(saline) membrane, the introduction of biomaterials intothe pouch resulted in a major accumulation of macro-phages (Fig. 6) and the appearance of eosinophils andoccasional giant cells (o1% of total cells). Thedifferential count revealed that the significant increasein cellularity of the membrane in response to biomater-ials was due to the macrophage component; there wereno significant changes in the fibroblast component of the

membrane when biomaterial stimulated pouches werecompared with either the saline control or one another.The level of significance of the increase in macrophagesin biomaterial-stimulated membranes was increasedcompared with the statistic for the total cellular counts,with p values ranging from po0.01 to o0.001. How-ever, there were no significant variations in membranemacrophage count when the different biomaterials werecompared with each other. This finding was also true forbiomaterial combinations (data not shown). Eosinophilswere invariably absent from the saline control mem-brane, and were present at consistent low levels inparticle stimulated membranes. The highest level ofeosinophil accumulation was seen in PMMA stimulatedmembrane (4.4% of total cells), although this level wasnot significantly elevated compared with other bioma-terials.

3.5. Cytokine gene activation in the air pouch membrane

The result of a typical RT–PCR using membraneextracted mRNA is shown in Fig. 7. Positive bandswere generated using primers specific for murine IL-1and IL-6, but weak or negative results were observedusing primers specific for TNFa. This surprisingdifference between the two predominant inflammatorycytokines did appear to be specific for particle stimula-tion; the introduction of 2 mg of lippopolysachaccarideinto the air pouch resulted in positive TNFa geneactivation within 24 h (not shown). In all instances, thestrength of the PCR signal for IL-1b appeared to behigher than the strength of the PCR signal for IL-6,suggesting that the major inflammatory cytokine re-sponse was mediated via IL-1. The addition ofbiomaterial particles to the air pouch resulted in anincrease in strength of the cytokine gene PCR signalsover the control air pouch membrane. No marked

Fig. 6. Differential count of the cellular infiltrate of control and

particle-stimulated air pouch membranes.

P.H. Wooley et al. / Biomaterials 23 (2002) 517–526522

variations in cytokine production were observed be-tween the different biomaterials, with the exception ofthe results with Ti–6Al–4V particles. Membranesstimulated with Ti–6Al–4V appeared to be skewedtowards an IL-6 response; Ti–6Al–4V provoked thelowest IL-1b band density, and the highest IL-6 banddensity. The combination of Ti–6Al–4V withUHMWPE also generated a cytokine gene activationthat was weighted towards an IL-6 response.

3.6. Cytokine levels in the air pouch fluid

The levels of IL-1b and TNFa in pouch fluiddetermined by ELISA assay are shown in Fig. 8. Notall biomaterials provoked increased IL-1b levels abovethe results seen in the saline control pouch (230 pg/ml),although stimulation with PMMA and Co–Cr, andparticles in combination did result in significantincreases in the fluid cytokine levels (po0.05–o0.01).There was reasonable agreement between the bandstrength of IL-lb gene activation in the pouch membraneand the level of cytokine protein detected in the pouchfluid, with the exception of the findings for UHMWPEalone, where protein levels of IL-Ib were unexpectedlylow. Levels of TNFa in pouch fluid were low, aspredicted by the gene activation with the membrane. Nosignificant increases in the level of TNFa in pouch fluiddue to particle stimulation were observed.

4. Discussion

In order to evaluate the mechanisms of the biologicalresponse to orthopaedic biomaterials, it is important toestablish accurate models of the disease process. Wehave utilized the murine air pouch model, initiallydeveloped to investigate the response to orthopaedicmaterials by Schumacher et al. [6,7], for the investiga-tion of the inflammatory response to particulate debris.This model exhibits cellular infiltration and mediators ofinflammation that appear to closely resemble thepseudosynovium associated with aseptic loosening[6,13,14], and has been demonstrated to be sensitive todifferences both in the material composition and thephysical form of the particles under investigation [6,15].The kinetics of our study were based on the reports ofthe histological features of this experiment model, andthe 5 day air pouch was selected to provide an organizedmechanical barrier that retained both the particles andproducts of the inflammatory response [8]. We observedmarked responses to both polymeric and metallicparticles, and synergistic responses to mixed particletypes. The appearance of the particulate stimulatedmembrane, with biomaterials deeply embedded withinthe tissue and surrounded by a predominantly macro-phage infiltrate, is strongly reminiscent of peri-prosthe-tic tissue recovered during arthroplasty revisionprocedures [16,17]. The order of the severity of thetissue reactions to the biomaterials cannot be preciselyranked in this study, since UHMWPE provokeda greater cellular infiltrate than PMMA, but resultedin a less thickened membrane; while Ti–6Al–4Vprovoked a thicker membrane than Co–Cr but a lowerintensity of macrophage accumulation. The response toTi–6Al–4V in the air pouch appears somewhat differentfrom the findings observed in subcutaneous implanta-

Fig. 7. RT–PCR using primers specific for murine IL-1b and mRNAextracted from control and particle-stimulated air pouch membranes.

Fig. 8. The levels of IL-1b and TNFa in pouch fluid from extractedfrom control and particle-stimulated air pouch membranes.

P.H. Wooley et al. / Biomaterials 23 (2002) 517–526 523

tion in rodents [18], and in vitro studies with Ti–6Al–4Vparticles suggest that they may be less provocative ofinflammatory responses [19]. However, strong macro-phage reactions to particulate Ti–6Al–4V have beenobserved in periprosthetic tissue recovered duringrevision arthroplasty [20–23].Our data suggest that the composition of the

particulate stimulus has a marked effect upon theinflammatory reaction within the pouch membrane.This is in agreement with the observations of Gelb et al.[6], who noted that the introduction of PMMA particlescomposed of different sizes and shapes resulted invariations in the production of cytokines and othermediators of inflammation. They observed that small,irregularly shaped PMMA particles (o20 mm) elicited asignificantly greater inflammatory reaction than largeparticles (>50 mm), and the findings were influenced bythe dose of particles introduced within the pouch. Theyhypothesize that the total surface area of the PMMAparticles is critical in the inflammatory response, andpropose that a threshold exists at which the inflamma-tory response increased dramatically. We did notexamine the dose response to particles in this study,however our mean particle size was fairly consistentamong the different materials studied. There was nocorrelation between the density of the biomaterial andthe level of the biological response, suggesting that thevariations between the different materials were notassociated with absolute particle number. In combina-tion experiments, the total particle load for eachmaterial was one half of that used for the individualbiomaterial experiments, so the increase due to particlecombinations cannot be attributed to increased particlenumbers. In the combination of UHMWPE andPMMA, synergistic effects were observed with themembrane reactions significantly greater than eithermaterial alone. These observations suggest that thecomposition of the biomaterial may influence the leveland profile of the inflammatory reaction, although thedose and physical properties of the particles maysignificantly alter the degree of biological reactivity,probably due to alterations in the degree ofphagocytosis [24]. However, there is no readilyaccepted method of comparing polymer and metaldebris. If mass is chosen, surface area or particulatenumber will vary. If particle number is chosen, surfacearea or mass must differ. We have chosen a constant 5%weight/volume, recognizing these limitations. Eventhough the particle numbers are not equal, the responseto combinations of materials clearly suggests synergy.This result must be extended cautiously to the patientsituation. The range of the particles under investigationin this study represents a large proportion of the sizerange of wear debris extracted from peri-prosthetictissue associated with failed joint components. However,the mean particle size studied was in excess of 1 mm,

which may be larger than the mean particle size inperiprosthetic tissue, and the macrophage response tothe particles under study clearly includes both uptake(phagocytosis) and extra-cellular adhesion. However,histological evaluation of peri-prosthetic tissue asso-ciated with failed implants does reveal a similarpathology.Many of the tissue changes associated with particle

stimulation may be associated with the effects of IL-lbproduction. Isaji [25] used the rodent air pouch to studythe local tissue response to recombinant human inter-leukin 1 alpha and beta (rIL-1 beta), and observed a 10-to 100-fold increase in the cellular accumulation.Repeated exposure to IL-1b caused the accumulationof large amounts of fluid within preformed pouches andan apparent thickening of the connective-tissue lining ofthe pouch, with deposition of large quantities of extra-cellular collagen within the pouch wall. They concludedthat the air pouch changes were consistent with IL-1bregulating the development and perpetuation of aninflammatory reaction. In a chronically stimulated airpouch, Dawson et al. [26] observed large numbers ofmacrophages accumulated after 3 days, accompanied byfibroblast proliferation and new blood vessels. Thereaction was significantly inhibited by dexamethasonebut not by indomethacin, suggesting that chronicallystimulated inflammatory reactions in this model werecorrelated with the increase of reactivity at the site ofinflammation and the exudative reaction was notmediated by cyclo-oxygenase products. Induction ofIL-1 in activated macrophages at the prosthesis interfacein response to the presence of metallic wear debris is afrequent finding in peri-prosthetic tissue recoveredduring revision arthroplastic procedures [27], and IL-1and IL-6 levels in these cells have been correlated withmarkers of the monocyte/macrophage lineage [28].Jiranek et al. [29] examined patient tissue and demon-strated IL-1b mRNA production occurred predomi-nantly in macrophages, and not in fibroblasts to anymajor extent. However, IL-1b protein was bound onboth macrophages and fibroblasts, indicating thatmacrophages are the source of the cytokine, whichsubsequently binds to both fibroblasts and macro-phages. Our model appears to resemble this proposedmechanism, with the response to particles occurringpredominantly in macrophages with a concomitantincrease in IL-1b production. The lack of a sustainedTNFa response to the particulate stimulation in ourstudy was unexpected, particularly since a TNFaresponse has been observed in response to ceramicparticle injection into the rodent air pouch [7]. However,this report also noted that synthetic monosodium uratecrystals did not elicit a TNFa response, suggested thatthe composition of the inflammatory stimulus is criticalin the generation of a specific cytokine reaction. Gelbet al. [6] suggested that the cellular mechanism for the

P.H. Wooley et al. / Biomaterials 23 (2002) 517–526524

TNF component of the inflammatory response isdifferent from the mechanism resulting in cell accumula-tion and other mediators of inflammation; a hypothesisthat our data strongly supports. Although thecellular component of the membrane is implicatedfor the production of the inflammatory cytokines, thecontribution of the fluid cells should be considered. Notall pouch fluids were analyzed for cell contentin our study, but an inflammatory exudate wasusually seen. Control pouch fluids containedPMNs (57%), lymphocytes (41%) and monocytes(2%). Fluid cells from PMMA pouches contained11% eosinophils, 56% PMNs and 33% lympho-cytes, while fluid cells from UHMWPE pouchescontained 53% lymphocytes, 44% PMNs, 3% eosino-phils and 1% monocytes. The relative contribution ofthese cells to the inflammatory cytokine levels remainsto be elucidated.The unusual pattern of cytokine elicitation using Ti–

6Al–4V, with high IL-6 production and lower IL-1bproduction is intriguing, since the presence of Ti–6Al–4V particles seems to accelerate bone loss and loosening[30]; however, Ti–6Al–4V particles have been observedto specifically up-regulate IL-6 production in vitro usinga variety of cell types [31,32], which may imply a specificcytokine profile associated with the response to thismetallic alloy. Variations in cytokine stimulation amongmetals has been demonstrated in vitro by Haynes et al.[33] indicating that while Co–Cr, stainless steel andtitanium alloy all induced TNFa and IL-1b, stainlesssteel particles were the most potent stimulators of IL-1band Ti–6Al–4V particles were the strongest stimulatorsof IL-6 and PGE2. However, these findings were notconfirmed by Liu et al. [34], who demonstrated thatexposure of leukocytes to Co2+ ion increased the releaseof TNFa, IL-6, and PGE2, but leukocyte exposure toTi3+ ions was associated with a decreased TNFa andPGE2 release and a minimal change in IL-6 release.Overall, our findings suggest that the murine air

pouch provides an accurate model for the biologicalresponse to particulate biomaterials, with a markedmacrophage accumulation and strong elicitation of theinflammatory cytokines IL-1b and IL-6. Our datasuggests that the variations in the material compositionof the particulate stimulus may result in differences inthe tissue response, and that combinations of differentbiomaterials may elicit synergy in the level of theinflammatory response. In order to accurately assess theosteolytic component of debris-induced aseptic loosen-ing, it is desirable to include bone tissue within themodel. To accomplish this, we are currently investigat-ing particle-stimulated air pouches with femoral headimplantation. These models may prove useful in thedevelopment of novel biomaterials for orthopaedicapplications, and therapeutic strategies for the treatmentof aseptic loosening.

Acknowledgements

The authors gratefully acknowledge the contributionsof W. Dwayne Lawrence, M.D. and the PathologyImage Analysis laboratory at Hutzel Hospital inestablishing the image techniques used in this study,and the grant support from the Veterans Administration(Rehabilitation Study Section) and the Arthritis Foun-dation.

References

[1] Wooley PH, Nasser S, Fitzgerald Jr RH. The immune response

to implant materials in humans. Clin Orthop Relat Res 1996;326:

63–70.

[2] Shanbhag AS, Jacobs JJ, Black J, Galante JO, Glant TT. Cellular

mediators secreted by interfacial membranes obtained at revision

total hip arthroplasty. J Arthroplasty 1995;10:498–506.

[3] Kaufman RL, Tong I, Beardmore TD. Prosthetic synovitis:

clinical and histologic characteristics. J Rheumatol 1985;12:

1066–74.

[4] Santavirta S, Konttinen YT, Bergroth V, Eskola A, Tallroth K,

Lindholm TS. Aggressive granulomatous lesions associated with

hip arthroplasty. Immunopathological studies. J Bone Jt Surg-

FAm Vol 1990;72:252–8.

[5] Eftekhar NS, Doty SB, Johnston AD, Parisien MV. Prosthetic

synovitis. Hip 1985;169–83.

[6] Gelb H, Schumacher HR, Cuckler J, Ducheyne P, Baker DG. In

vivo inflammatory response to polymethylmethacrylate particu-

late debris: effect of size, morphology, and surface area. J Orthop

Res 1994;12:83–92.

[7] Nagase M, Udagawa E, Schumacher HR, Baker DG. Prolonged

inflammatory reactions induced by ceramic powders in the rat air

pouch model. Nippon Seikeigeka Gakkai ZasshiFJ Jpn Orthop

Assoc 1990;64:602–11.

[8] Sedgwick AD, Sin YM, Edwards JC, Willoughby DA. Increased

inflammatory reactivity in newly formed lining tissue. J Pathol

1983;141:483–95.

[9] Bi Y, van de Motter RR, Ragab AA, Goldberg VM, Greenfield

EM. The role of adherent endotoxin in stimulation of osteoclast

differentiation by orthopaedic wear particles. Trans Orthop Res

Soc 1999;45:5.

[10] Chadha HS, Wooley PH, Fitzgerald RH. Cellular proliferation

and cytokine responses to polymethylmethacrylate (PMMA)

following joint replacement. Inflammat Res 1995;44:145–51.

[11] Wooley PH, Sud S, Langendorfer A, Calkins C, Christner PJ,

Peters J, Jimenez SA. T cells infiltrating the skin of Tsk2

scleroderma-like mice exhibit T cell receptor bias. Autoimmunity

1997;27:91–8.

[12] Schaefer CJ, Whalen JD, Knapp T, Wooley PH. The influence of

silicone implantation on type II collagen-induced arthritis in mice.

Arthritis Rheum 1997;40:1064–72.

[13] Nagase M, Baker DG, Schumacher HR. Prolonged inflammatory

reactions induced by artificial ceramics in the rat air pouch model.

J Rheumatol 1988;15:1334–8.

[14] Wooley PH, Morren R, Andary J, Sud S, Song Z, Mayton L,

Fitzgerald RH, Nasser S. A model of cellular responses in aseptic

loosening: UHMWPE and PMMA particles provoke different

responses in the murine air pouch. Orthop Trans 1996;21:500.

[15] Wooley PH, Mayton L, Sud S, Nasser S. Synergistic inflamma-

tory responses to metallic and polymeric particles in the murine

air pouch model of aseptic loosening. Arthritis Rheum

1996;39:175.

P.H. Wooley et al. / Biomaterials 23 (2002) 517–526 525

[16] Bos I, Fredebold D, Diebold J, Lohrs U. Tissue reactions to

cemented hip sockets. Histologic and morphometric autopsy

study of 25 acetabula. Acta Orthop Scand 1995;66:1–8.

[17] Boss JH, Shajrawi I, Soudry M, Mendes DG. Histological

features of the interface membrane of failed isoelastic cementless

prostheses. Int Orthop 1990;14:399–403.

[18] Akagawa Y, Hashimoto M, Kondo N, Yamasaki A, Tsuru H.

Tissue reaction to implanted biomaterials. J Prosthet Dent 1985;

53:681–6.

[19] Gonzales JB, Purdon MA, Horowitz SM. In vitro studies on the

role of titanium in aseptic loosening. Clin Orthop Relat Res

1996;330:244–50.

[20] Agins HJ, Alcock NW, Bansal M, Salvati EA, Wilson Jr. PD,

Pellicci PM, Bullough PG. Metallic wear in failed titanium-alloy

total hip replacements. A histological and quantitative analysis.

J Bone Jt SurgFAm Vol 1988;70:347–56.

[21] Korovessis P, Repanti M. Evolution of aggressive granulomatous

periprosthetic lesions in cemented hip arthroplasties. Clin Orthop

Relat Res 1994;300:155–61.

[22] Lombardi Jr. AV, Mallory TH, Vaughn BK, Drouillard P.

Aseptic loosening in total hip arthroplasty secondary to osteolysis

induced by wear debris from titanium-alloy modular femoral

heads. J Bone Jt SurgFAm Vol 1989;71:1337–42.

[23] Witt JD, Swann M. Metal wear and tissue response in failed

titanium alloy total hip replacements. J Bone Jt SurgFBr Vol

1991;73:559–63.

[24] Goodman SB, Fornasier VL, Lee J, Kei J. The histological effects

of the implantation of different sizes of polyethylene particles in

the rabbit tibia. J Biomed Mater Res 1990;24:517–24.

[25] Isaji M, Momose Y, Naito J. Enhancement of inflammatory

reactions in a nonimmunological air pouch model in rats. British

J Exp Pathol 1989;70:705–16.

[26] Dawson J, Sedgwick AD, Edwards JC, Lees P. Lymphocyte

kinetics in a murine model of chronic inflammation. Agents

Actions 1989;27:461–4.

[27] al Saffar N, Revell PA. Interleukin-1 production by

activated macrophages surrounding loosened orthopaedic im-

plants: a potential role in osteolysis. Br J Rheumatol 1994;33:

309–16.

[28] Perry MJ, Mortuza FY, Ponsford FM, Elson CJ, Atkins RM.

Analysis of cell types and mediator production from tissues

around loosening joint implants. Br J Rheumatol 1995;34:

1127–34.

[29] Jiranek WA, Machado M, Jasty M, Jevsevar D, Wolfe HJ,

Goldring SR, Goldberg MJ, Harris WH. Production of cytokines

around loosened cemented acetabular components. Analysis with

immunohistochemical techniques and in situ hybridization.

J Bone Jt SurgFAm Vol 1993;75:863–79.

[30] Buly RL, Huo MH, Salvati E, Brien W, Bansal M. Titanium wear

debris in failed cemented total hip arthroplasty. An analysis of

71 cases. J Arthroplasty 1992;7:315–23.

[31] Haynes DR, Rogers SD, Howie DW, Pearcy MJ, Vernon-Roberts

B. Drug inhibition of the macrophage response to metal wear

particles in vitro. Clin Orthop Relat Res 1996;323:316–26.

[32] Morris CJ, Blake DR, Hewitt SD, Lunec J. Macrophage ferritin

and iron deposition in the rat air pouch model of inflammatory

synovitis. Ann Rheum Dis 1987;46:334–8.

[33] Haynes DR, Boyle SJ, Rogers SD, Howie DW, Vernon-Roberts

B. Variation in cytokines induced by particles from different

prosthetic materials. Clin Orthop 1998;352:223–30.

[34] Liu HC, Chang WH, Lin FH, Lu KH, Tsuang YH, Sun JS.

Cytokine and prostaglandin E2 release from leukocytes in

response to metal ions derived from different prosthetic materials:

an in vitro study. Artif Organs 1999;23:1099–106.

P.H. Wooley et al. / Biomaterials 23 (2002) 517–526526