13

"Technical , Notes" in BPR originated when certain Progress Reports went beyondthe usual scope to present a wealth of detail about some single aspect of the re-search, typically a matter of instrumentation or materials or procedure.

With the establishment of the category, authors have been finding the departmenta useful channel of communication. Material available has tended to grow in intel-lectual and technical stature . BPR's editorial posture has changed in response, sothat the earlier informal verification by staff and occasional external review hastended to give way to a more formal procedure which involves reviewers from thepublication's Editorial board, ad hoc reviewers, or both, as may be appropriate todeal with complex or wide-ranging subject matter . This "double-anonymous" revieweffort frequently equals that given to all regular feature articles in this publication.

Therefore, while these more ambitious Technical Notes may lack the number ofsubjects, weight of data, and discovery of or replication of new knowledge requiredfor a traditional scientific article, they have been handled with the greatest possiblerespect for the authors' devotion and the readers' requirements.

TECHNICAL NOTES

The Wear Particles of Synovial Flu d:Their Ferrographic Analysisand Pathophysiological Significance a

CHRISTOPHER H. EVANS, Ph. D., C. CHEM., M .R .S .C . bAssistant Professor of Orthopaedic Surgery

DANA C. MEARS, S.M., Ph . D., M .R .C.P.Associate Professor of Orthopaedic Surgery

Dept. of Orthopaedic SurgeryUniversity of PittsburghPittsburgh, Pennsylvania 15261

Abstract—This article reviews recent progress in ourinvestigations into the wear particles which accumulate inprosthetic and natural diarthrodial joints . Ferrographicanalysis of these particles is providing new insight into themanner in which joints undergo wear in situ . It also has thepotential to serve as a diagnostic and prognostic technique,with special application to monitoring very early changes inthe wear status of joints and distinguishing the variousarthritides . Wear particles interact with periarticular cellsand tissues, provoking the release of lytic enzymes andeliciting other biochemical changes, which exacerbate thedestruction of the joint.

Introduction

Recently a novel analytical tool, ferrography, has enabledus to undertake a non-invasive study of the very earlydeterioration that accompanies various arthritides whenapplied to an aspirated sample of synovial fluid . The methodpermits a critical separation of diverse types of biologicalwear particles generated from the bearing surfaces of anarticular joint.

'This investigation was supported by the Veterans Administrationand by NIH Grant AM20697.

'Address correspondence concerning this paper to C. H . Evans, Ph. D .,Department of Orthopaedic Surgery, University of Pittsburgh, 986Scaife Hall, Pittsburgh, PA 15261 .

Ferrography is a technique devised originally for theanalysis of wear particles in machines . Ferrous wearparticles are retrieved from a sample of the lubricating oilunder the influence of an external magnetic field.Microscopic examination of the particles thus retrievedpermits accurate diagnosis of the wear status of themachine (q .v.).

Given the success of the ferrographic technique withmachines, it was suggested that ferrography could beprofitably applied to the analysis of wear particles containedin the synovial fluid of natural and artificial joints . Forprosthetic joint replacements, this would allow a non-surgical ; sophisticated analysis of the rate and type of wearof the implant and prognostication of its future performance.With natural joints, there is the enticing prospect of vastlyimproving the differential diagnosis of arthritis. Particularlyimportant is the potential of ferrography to facilitate an earlydiagnosis, before the disease has progressed to conspicuousradiological manifestations, by which time potentiallyprophylactic treatments may not longer suffice and surgicalprocedures remain the only solution.

Wear particles in joints are of additional significance . Byinfluencing the metabolism of cells in the surroundingtissues, particles eroded from prosthetic devices couldcontribute to various of the adverse reactions which result,for example, in loosening of the implant, or infection . Innatural joints, the wear particles may be mediators of tissuedestruction during arthritis, in which case their investigationshould provide insights into the disease process itself.

In this article, we review progress made here and atFoxboro Analytical, Burlington, MA, in the application offerrography to the study of wear particles derived fromhuman and prosthetic joints, and in establishing theirpossible role in the initiation and progression of arthritis.

The Analysis of Wear by Ferrography

Ferrography is a technique of proven worth in evaluatingtribological processes by studying the wear particles theyproduce . Developed by Vernon Westcott of FoxboroAnalytical, Burlington, Massachusetts, ferrography is basedupon the magnetic retrieval and separation of wear particles

14TECHNICAL NOTE : EVANS AND MEARS

FIGURE 1The ferrograph analyser. The photograph at the top of the page showsa "dual" analyser capable of processing two samples simultaneously.The diagram illustrates the interaction of the suspension of wearparticles with the magnetic field during the ferrographic analysis.

FIGURE 2A ferrogram. This ferrogram has been deliberately overloaded withcartilaginous wear particles and stained to illustrate the technique.The substrate is 60 mm long . The entry point is at the left of thephotograph, and contains the highest concentration of particles .

from lubricating fluids (1) . It was originally designed tomonitor wear in machines, where the particles are typicallyferrous and thus intrinsically susceptible to an externalmagnetic field . Most biological materials are, of course,diamagnetic, so, as described later, special techniques havehad to be developed to enable their magnetic separation.

With machines, a small volume (usually 1—3 ml) oflubricating oil is slowly pumped along a thin glassmicroscope slide (the substrate) which lies in a high-gradientmagnetic field (Fig . 1) . The substrate is raised slightly at thepoint where it first comes into contact with the sample (theentry point), so that particles encounter a perpendicularmagnetic force which is weakest at the entry point andwhich increases progressively down the length of thesubstrate . Consequently, for particles of the same unitvolume susceptibility, the largest particles deposit nearestthe entry point, with a grading of particles according to sizedown the length of the substrate . A washing solution ispumped over the particles on the substrate, which is thenallowed to air dry. The resultant ferrogram (Fig . 2) can thenbe examined by various optical and electron microscopictechniques.

Particles are classified on the basis of their size, shapeand other morphological features, to indicate the operationof certain wear modes in the machine from which thesample was drawn . Once the "running-in" period of themachine is over, it is normal to find a moderate number ofelongated particles (Fig . 3) indicative of rubbing wear.Abnormal conditions are indicated by increased numbers ofparticles of greater size and of distinctive morphologyindicative of cutting wear, fatigue, or other disturbances(Fig . 3) . These particles appear on ferrograms beforeadverse symptoms affect the machine (2) : As differentcomponents of engines are made from different alloys,identification of the constitution of the wear particlespermits the site of excessive or abnormal wear to belocated.

The ferrographic analysis of engine oil has beenextensively reviewed (2,3).

Ferrographic Analysis of Human Joints

With the foregoing background, ferrography seemed ofpotential value in studying wear in human artificial andnatural joints . Wear particles accumulate in synovial fluid inmuch the same way as they do in engine oil . Non-invasive,objective, diagnostic and prognostic, this technique yieldsimportant information pertaining to the rates of wear, typesof wear and reasons for failure of prosthetic jointreplacements . This information may greatly aid thedevelopment of improved implants, and would permitconvenient, non-invasive, serial assessment of the progressof implanted prostheses with minimum discomfort to thepatient . Ferrography also has potential application to studiesof osteoarthritis, where mechanical wear of the jointsurfaces occurs.

Ferrographic analysis of prosthetic joint replacements—Ferrography was first applied to the examination of synovialfluids and washings obtained from replacement joints . Suchmetal-on-metal or metal-on-plastic arthroplasties constitute atype of machine confined within the body, and provide the

15Rolle in of Prosthetics Research BPR 10-36 Fall 1981

opportunity for a logical extension of the originalferrographic technique.

Synovial fluids, or saline washings, from these sourcescan be processed by a method analogous to that used tomake ferrograms from engine oil . This is so because the tinymetallic particles become embedded in the surroundingmaterials, thereby imparting a magnetic susceptibility whichis sufficiently high for normally diamagnetic wear particlesto be retrieved under the influence of the magnet of theferrograph . Under these conditions, wear particles ofcartilage, bone, plastic, polymethylmethacrylate andsynovium can be analyzed without the need for artificialmeans of magnetisation (4).

Synovial fluid is obtained by sterile needle aspiration fromjoints after arthroplasty. If synovial fluid is unobtainable, thejoint can be flushed with sterile saline . About 2 ml of thesynovial fluid are pumped over the substrate as describedbelow.

Following washing and drying, the ferrogram is firstexamined by optical microscopy. A useful instrument in thiscontext is the bichromatic microscope, providing bothreflected and transmitted illumination (1) . Using theappropriate filters, the transmitted light is green and thereflected light red . Under the bichromatic microscope,metallic wear particles appear red, due to their attenuationof the green transmitted light and reflection of the red,

direct light . Non-metallic materials appear green, yellow orpink, depending upon their density or thickness . Polarizedlight is advantageous when examining particles of bone,cartilage, rnethylmethacrylate and plastic.

Following optical inspection, particles can be examinedwith greater morphological precision using the scanningelectron microscope (SEM) and their elemental compositioncan be determined by energy dispersion X-ray analysis.

Ferrographic analysis of saline washings of failedprostheses (Fig . 4) reveals the presence of metallic, plasticand polymethylmethacrylate particles in synovial fluid(Fig . 5) . The relative proportions vary with the type ofprosthesis and the severity and type of wear which hasoccurred . Thus, larger numbers of metallic particles arefound when both articular surfaces are composed of metal.With metal rubbing on polyethylene, most of the particlesare plastic, and range from 1–10 gm in diameter up toshredded fibres several hundred micrometers long . Metallicparticles range in size from under 025«m wherovvoa, isrelatively small, to 1 mm in length where fatigue or abrasivewear has occurred . Polymethylmethacrylate particles aregenerally cuboid with large variations in size from 1 gm to1 mm or more . Sometimes, these particles adhere tometallic fragments. Particles of bone, cartilage and synovialtissue are also found . Histological examination of synoviumdemonstrates the presence of embedded wear particles,

FIGURE 3Wear particles from engine oil . (A) Normalrubbing wear particles, man . 450x . (a)Cutting wear particles, scanning electronmicroscopy, 225x . (Cl Fatigue wear vau '

-~~mag. 450x . (D) Laminar wear parti-cle, mag . l000x.

3-A

16

TECHNICAL NOTE : EVANS AND MEARS

FIGURE 4Macroscopic appearances of wear in failed prosthetic joint replace-ments . At top of page, erosion of the polyethylene surface of a failedtotal knee replacement is seen ; the depressions are filled with po-lymethylmethacrylate . Above, a failed total hip replacement . Again,there is pitting of the polyethylene surface with polymethylmethac-rylate infilling . Note that the metallic component is less severelyeroded .

FIGURE 5Wear particles from prosthetic joints, recovered by ferrography . Attop, metallic wear particles, mag . 1000x . Center, polyethylene wearparticle under polarized light, mag. 400x. Immediately above, po-lymethylmethacrylate particles under polarized light, mag. 400x .

17Bulletin of Prosthetics Research BPR 10-36 Fail 1981

while phagocytic cells containing endocytosed wearendocytosed wear debriscan be retrieved from the oynovial fluid of affected joints(Fig . 8).

X-ray elemental analysis of the metallic particles confirmsthat they originate from both the arthroplasty and thenungical instruments, most especially haemostats, scissors,and ooteotomoa(4).

This study revealed interesting and encouragingsimilarities to the ferrographic analysis of the lubricating oilof machines . In both cases, excessive wear is signalled byan increase in the numbers and sizes of particles, and thepresence of wear particles representing altered andabnormal modes of wear. Thus, ferrography appears to offerthe same advantages to studying wear in prosthetic joints asit presently does to studying wear in machines . Thepresence of wear debris embedded in the synovium (Fig . 6)indicates how rapid prosthetic wear could provoke a painful,

FIGURE 6Interaction of particleswearrounding cells and tissues . Above, right:a polarized light micrograph of synov-iurn shows embedded wear particles.Mag . 100x . At right, a transmissionelectron micrograph of a polymor-phonuclear leukocyte retrieved from thesynovial fluid . Engulfed particles occurin intracellular vesicles . Mag . 38,300x .

18

TECHNICAL NOTE : EVANS AND MEARS

proliferative synovitis. Such information has stimulatedfurther modification of implant design, and it has indicatedone other pathogenesis of a painful joint replacement.

Ferrographic analysis of wear in natural joints—Before thebiological wear particles present in the synovial fluid ofnatural joints can be analyzed effectively by ferrography, it isnecessary to make them susceptible to external magneticfields. Research conducted at Foxboro Analytical hasproduced methods of achieving this (5) . The 'magnetisation'process is based on the sorption of paramagnetic cations ofthe rare earth element erbium (Ill) to particles of bone andcartilage (6) . Because Er a+ interacts strongly with manyother substances in synovial fluid, producing troublesomeprecipitates, the particles are first collected by centrifugingand washed three times by resuspension and recentrifugingin saline (0 .9% wlv) . Even this may be insufficient toadequately remove hyaluronic acid, a sticky macromoleculewhich may pose problems by tenaciously coating theparticles . For this reason, the washed particles are treatedmildly with a highly specific hyaluronidase before analysis.(That step is usually unnecessary for particles obtainedthrough saline washings of joints .)

Bloody samples are difficult to process . If the degree ofcontamination is slight, the erythrocytes can be lysed indistilled water prior to ferrography. This leaves a residue ofcellular debris which, in modest quantities, is not a problem.Samples of synovial fluid quickly form gelatinousprecipitates . For this reason, all such specimens must becentrifuged, washed and resuspended in saline beforestorage. Sodium azide (final concentration 0 .02% w/v)should be added to stored saline suspensions of wearparticles to prevent microbial contamination.

Peculiarities of "biological" ferrography—Althoughbiological particles achieve considerable positive magneticsusceptibilities on treatment with magnetising solutionscontaining Er 3 ', the induced susceptibilities are many timesless than those of corresponding metallic particles . For thisreason, it is necessary to position the substrate flat againstthe magnet (Fig . 1) to ensure adequate recovery of thebiological material . Unfortunately, this arrangement reducesthe longitudinal gradient in magnetic force that is exerted onthe particles as they flow along the substrate, therebyinterfering with the size-grading of particles along theferrogram . This is exacerbated by the chemicalheterogeneity among the wear particles themselves whichinfluences their uptake of Er a+ and thus their individualmagnetic susceptibilities. Such heterogeneity results fromintrinsic variations in chemistry at different locations withinthe cartilage, from different mechanisms of wear, and fromalterations produced by the disease. (Eventually, theresultant change in ferrographic behaviour of such particlesmay prove of diagnostic value .)

Dust-free work area—When making and examiningferrograms, it is necessary to work in an environment whichis free from dust . Particles of dust are often optically activeand in many cases resemble cartilaginous wear particles.Extraneous particulates can be prevented by situating theferrograph analyser and the bichromatic microscope in alaminar flow hood .

Notes on Optical Analysis—Examination of theferrograms is first undertaken with the bichromaticmicroscope, as described in the previous section.Ferrograms made from synovial fluid aspirates reveal avariety of deposited materials, not all of which are yetcompletely understood.

As bone and cartilageare both optically active, polarizedlight microscopy is especially useful . Whereas mostcartilaginous and osseous wear particles are conspicuousunder polarized light, soft tissue components such assynovium or other cellular material are not . Osseousparticles usually have high optical activity and a compact,chunky appearance ; much of this is due to the scattering oflight when it encounters the polycrystalline mineral of bone.Consequently, osseous particles remain bright as thedirection of prolongation is changed while maintainingcrossed polars. On the other hand, materials exhibitingbirefringence, in which the molecular structure of theparticle is organized over a period of several micrometers,will exhibit a change in light intensity as a function of thedirection of prolongation . Cartilaginous particles varygreatly in their optical activity—some resemble bone inhaving high optical activity, while others are barely visibleunder crossed polars . This may reflect their degree ofmineralization, the extent to which the collagen fibres areoriented along a single axis, and the thickness of theparticles.

Occasionally, especially in the larger cartilaginousparticles, one can observe lacunae which, in life, contain thechondrocytes. Phase contrast microscopy facilitates suchobservations.

Histological and chemical notes—Various histologicaltechniques 'have been brought to bear on the complexproblem of particle identification . It would seem, at firstsight, that these various types of particle could beelucidated by standard histological techniques, many ofwhich have been specifically devised for differentiallystaining the tissues of the joint. This, however, has notproved as straightforward as it appears.

The reasons for this are several .° The chemistry of thewear particles probably differs from that of the intact tissue.Specifically, metachromatic staining of articular cartilage ismarkedly reduced in early osteoarthritis, reflecting a loss ofthe proteoglycans which are responsible for many of thestaining properties of cartilage. Also, once released into thesynovial fluid, further biochemical modifications of theminute articular particles may occur. However, some successhas been obtained with a modified Movat's pentachromestain (7).

X-Ray analysis of materials—Particles of bone andcartilage which have not undergone excessive degradationcan be distinguished readily by energy dispersion X-rayanalysis. As shown in Figure 7, cartilaginous particles give astrong sulphur emission from their sulphatedglycosaminoglycans . Bone provides strong calcium andphosphorous peaks (Fig . 8), while mineralized cartilage

Procedures required to avoid staining problems, when Er a+ has beenemployed to impart magnetic susceptibility to tissue particles, arediscussed in (7) .

19

Bulletin of Prosthetics Research BPR 10-36 Fall 1981

FIGURE 7—cartilaginous wear particle (arrowed), mag . 200x, with(at right) its energy dispersion X-ray analysis spectrum . The analysisshows the presence of sulphur, from the sulphated glycosaminog-lycans of cartilage, and erbium, which is used to magnetize the par-ticles . The sulphur peak is irregular due to subtraction of thebackground spectrum produced by the glass substrate.

contains both of these and sulphur. Figure 8 compares partof the elemental spectrum of bone (background) with softtissue (foreground) . The bone gives a major peak of calcium,while the soft tissue has only a small calcium peak, which isdwarfed by that of potassium . A difference in their relativeamounts of chlorine is also seen.

While this technique is rather tedious for routine use, itserves to resolve the identity of enigmatic particles and tocross-check the various other methods tested in the searchfor improved identification techniques.

Comments on an Analysis of 50 Synovial Samples

Having developed a technique for the ferrographicanalysis of synovial fluid aspirates, its applicability was

investigated by an analysis of about 50 synovial samples.This survey yielded encouraging results (7) . A variety ofdifferent particles deposited on the ferrograms, indicatingthat discrete wear particles do indeed exist in humansynovial fluid, and that they are susceptible to ferrographicanalysis . It was also encouraging to observe that wearparticles were not pieces of random detritus, but that manyof them fell into one of several different, identifiablemorphological categories.

On the basis of that preliminary work, we are beginning toappreciate the different types of wear particle in joints, thewear conditions that produce them, and their significance. Inthe work outlined below, the ferrographic analysis wascompared to radiographic, arthroscopic, and physical

8-C

FIGURE 8—Osseous particles under (A) unpolarized and (B) polarized light, mag . 200x, with (C) their energy dispersion X-ray analysis . Thebackground spectrum is that of bone showing a marked calcium peak. In the foreground is, for comparison, non-osseous tissue which hasonly a little calcium .

examination of the joint from which the sample was

in Figure 9a is the entry deposit, viewed under polarizedobtained .

light

100x magnification,

ferrogram made from theFerno8,ams made from the wear debris of joints with even

synovial fluid of p patient with d` d malacia patella.modest degrees of articular erosion often contain a sizable

Closer examination of samples such as these enables manyand varied population of wear particles, Some of these are

of the particles to be categorized.indicated in Figures 9 (a—h> and 10 (a —e)Tho array shown

FIGURE 9Types of cartilaginous wear particlesare shown in the photographs at rigIand below

(a) Entry deposit on ferrogram from chon-

(b) Thin, angular lamellae of superficial car-umma!aoiu patella, polarized light, muo

tilage, unpo!anzoulight, maoouox.1nu« .

(e) Spherical cartilaginous particles . Mag.400x, unpolarized light.

(a) Cartilaginous wear particles from a mod-

(d) Fibrous cartilaginous wear particle, mag.eratelveroded knee joint. m,wz0C ' pola,

nm'vmuri =ed light.izao!ight.

(f) Triangular fragment of cartilage and car-

(g) Cartilaginous wear particles retrieved

(h) Large cartilaginous wear particle from,i!aninv"uxphone, polarized light, mam . 400x .

from osteoarthritic joint, mao . 200x, polar-

osteoarthritic joint, mao . 100x ' u"po/anoodized light,

light .

21Bulletin of Prosthetics Research BPR 10-36 Fall 1981

Benign forms of low-grade wear in joints appear togenerate small (< 40 gm) flakes of thin, angular lamellae ofcartilage with weak optical activity. These occur on mostferrograms, even where damage is light, and they are likelyto originate from the superficial layers of the cartilage . AnSEM micrograph of one such particle is shown in Figure 9b

More severe damage—When damage to the articulatingsurfaces is more severe, the lamellae are larger (> 80 gm),more numerous, and sometimes give the impression ofbeing laminate . More severe wear also produces largerparticles of higher optical activity and a more compactappearance, as shown in Figure 9c under polarized lightPossibly, these are derived from the deeper zones of thecartilage or are produced by different wear mechanisms, orboth . Other cartilaginous wear particles are elongate rods,some of which have a coarse fibrous nature (Fig . 9d) . (Wesuspect that many of the latter are derived from themeniscus .)

Spherical particles—Under certain conditions, sphericalcartilaginous particles occur. The ones shown in Figure 9eare about 5«m in diameter and were retrieved from theknee of a patient with chondromalacia patella . The sphere,

shown by SEM in 9f is larger (—25 pm diameter) and ofstronger optical activity.

The mechanism of production of cartilaginous spheresremains to be determined . In machines, spherical wearparticles are shaped by their rolling in minute cracks in thesurface of one of the components of the machine . Sphericalparticles in synovial fluid may be formed from non-sphericalprecursor particles In an analogous manner, or they mayhave been spherical to begin with . In the latter case, anunusually precise wear process must have carved them outof the cartilage, or, more probably, pre-existing sphereswithin the cartilage might have been released by the wearprocess In several examples, there was evidence of somedegree of mineralization associated with the spheres (7).This tentative suggestion may bear some relationship to theobservations of Pautard (8) on the morphology ofcrystallized matenul within bone . In one instance, a highconcentration of mineralized spheres of diameter 25—30 gmwas seen on a ferrogram made from the aynovial fluid of apatient with bursitis and primaryooteoarth,itis (Fig . lOe).McCarty has observed slightly smaller spheres ofhydroxyapatite in the shoulder joint of certain patients (9).

FIGURE 10Photographs (a) through (e) show ex-amples of other types of deposits en-countered on ferrograms .

(a) Optically inactive memuo,~mag . 100x '(b) Amorphous debris, mag. 100^ ' vnpo~~unpolarized light .

ized light.

(c) As (b), polarized light .

(d) Crystals, mag . 400x . (e) Crystalline spheres, mag . 200x, polar-ized light. shows SEM of one sphere,mag . 1000x .)

22TECHNICAL NOTE : EVANS AND MEARS

Other evidence of crystal deposition in osteoarthritic joints isgrowing . This would provide an ideal mechanism for thegeneration of the "cutting wear particles" of cartilage wehave observed on ferrograms made from the fluid ofosteoarthritic joints . (Wear particles could conceivably serveas foci for the deposition of minerals .)

When articular erosion is severe, larger particles areproduced which do not fall into distinct morphologicalcategories such as the lamellae, spheres, and rodsmentioned earlier. The triangular particle shown in Figure 9fis one such example . Others are shown in Figures 9g and9h . Lacunae are visible in the particle shown in Figure 9g.Part of another large cartilaginous particle is shown inFigure 9h under polarized light ; again, lacunae are evident.In some cases, cartilaginous wear particles have a blue-greysheen under polarized light (7).

Ferrograms contain other types of wear particle inaddition to the cartilaginous ones described above . Thin,amorphous membranes without optical activity (Fig . 10a)probably represent fragments of synovium . In contrast,amorphous debris of the type shown in Figure 10b hasappreciable optical activity (Fig . 10c) . Preliminary energydispersion X-ray analysis suggests an appreciable level ofassociated sulphur, indicating the presence of cartilageproteoglycans . These deposits may thus constitute adegraded form of cartilage, possibly that subjected toenzymic attack . In agreement with this interpretation, thefew samples of rheumatoid synovial fluid which we haveanalysed by ferrography, and where the enzymic mode ofdegradation of cartilage would be expected to be operative,have high amounts of this material (7).

Minerals in the form of crystals (Fig . 10d) or spheres (Fig.10e) also occur on ferrograms . Those containing phosphatebecome especially magnetic (6) as Er a ' has a high affinityfor phosphate groups.

The sizes of the synovial wear particles are generallymuch larger than their metallic counterparts . Even in jointswhere the damage is light, wear particles reached 20–30 µmin size, with fibres being larger than this . As with machines,increased severity of wear produced particles of increasedsize, particles up to 2 mm in length often appearing onferrograms . Even larger fragments occur in severely arthriticjoints . (The upper size limit for detection by ferrography isset by the gauge of the hypodermic needle used to aspiratethe sample, and the bore of the turret tube used in theferrographic analysis (Fig . 1) .) Along with the increased sizewhich accompanies more advanced degeneration, there isincreased morphological variety among the particles, whichthus become difficult to classify.

Studies of Selected Patient Groups

The preliminary survey already described has confirmedthe applicability of ferrography to synovial fluid analysis : thewear particles fell into a number of unique categories whichvaried from one patient to another ; the sizes andmorphologies of the particles altered with the severity ofwear. However, no clear-cut differences could be foundbetween the various groups of diseases studied(chondromalacia patella, torn meniscus, osteoarthritis,rheumatoid arthritis) (7) . From these findings, it became

clear that realising the diagnostic and predictive potentialsof ferrography would require close scrutiny of selectedgroups of patients whose medical histories and arthriticconditions were minutely detailed.

To develop that phase of the project, in collaboration withDr. Carl Stanitski of the University of Pittsburgh HealthCenter, we are presently limiting ourselves to ferrographicanalysis of saline washings of knee joints recovered duringarthroscopic examination of the knee . This eliminates manyvariables—we are dealing with only one joint (the knee) andwith a limited number of disorders . Most patients sufferfrom a torn meniscus, chondromalacia patella, or earlyosteoarthritis . On occasion, arthroscopic examination willreveal no joint abnormalities, thus providing valuable"normal control" samples, which are otherwise difficult toobtain . Arthroscopy also has the advantage of permittingclose visual examination of the articulating surfaces andsynovium, thus providing the detailed information needed ina study of this kind . Ancillary historical data, physicalexaminations, and radiological assessments can becorrelated with the ferrographic analyses and arthroscopies.

The most interesting group of patients in this study havebeen those whose arthroscopic examinations revealedessentially no damage to the cartilaginous surfaces of thejoint. One illuminating subset contained three patients ofequivalent age (11, 12, 14 years) . The initial arthroscopicexamination classified each of these as normal . However,ferrographic analysis revealed marked differences betweenone of these patients, whose knee contained very little weardebris, and the two others, whose ferrograms containedevidence of cartilaginous damage . On arthroscopic re-evaluation, one of the latter two patients was found to havea possible slight softening of the patella, and the other abarely detectable softening of the femoral condyle . Thus,ferrography seems much more sensitive than arthroscopy tosubtle changes in the integrity of the cartilage.

Examination of five patients with torn anterior cruciateligaments reinforced that conclusion . Ferrograms made fromsaline washings of the affected knees contained anappreciable amount of wear debris . Although the number ofwear particles was elevated, their size remained small ; somewere rounded particles only 1–5 µm in diameter. In eachcase, arthroscopy failed to detect this 'micro-damage' to thecartilage . The suggestion has been made that such damagemay arise from the alteration in biomechanical forces withinthe joint as a result of the ligamentous injury. Withligamentous laxity, sliding of the articular surfaces maysupplement the normal rolling motion and greatly alter thewear pattern . In the presence of peripheral ligamentousinjury (or with abnormal joint congruity perturbed bytraumatic, congenital, or arthritic change) the local sites oflinear or point contact on the bearing surfaces might resultin excessive forces on those bearing surfaces, so thatabnormal wear modes ensue . This suggestion finds supportin the analysis of the wear particles in the knees of twopatients in which the torn cruciate ligamentous injury wassuperimposed on another defect; in one of these this wassoftening of the articular cartilage and in the other it was aslightly "ragged" meniscus . In both cases, the numbers andtypes of wear particle revealed damage in excess of thatcaused by torn ligaments or the other injuries alone .

23Bulletin of Prosthetics Research BPR 10-36 Fall 1981

Patients with articular damage of sufficient severity to bedetected by arthroscopy yielded elevated numbers of wearparticles of large size and varied morphology.

These studies illustrate the extreme sensitivity offerrography to articular damage. Its superiority overarthroscopy has two aspects . First, there is themagnification factor ; ferrograms are routinely examinedunder the optical microscope under magnifications of 100x-400x . With the electron microscope, much highermagnifications are possible . Arthroscopy, however, scans forarticular damage at magnifications of only 3x-5x . A secondaspect is the greater resolution permitted by the meansused for examining the wear particles, in contrast to thatused to view the bulk material, when searching for evidenceof damage.

It appears, then, that even grossly normal knee joints docontain some wear particles . This observation raisesquestions about normality. Preliminary analysis of a fewsamples seems to indicate that the "background" weardebris of asymptomatic joints increases with age . This tiesin with the observation that most, if not all, jointsdegenerate with age, even though only a fraction sufferfrom clinically overt arthritis . We are presently trying toanalyse ferrographically a sufficient number of samplesfrom asymptomatic knee joints to permit an accurateassessment of this matter.

The Pathological Significance of Wear Particles : Observa-tions and Hypotheses

Specialized phagocytic cells exist for the removal ofparticulate matter from the body. Most active in this respectare macrophages and polymorphonuclear leucocytes, bothof which are derived from the stem cells in the bonemarrow. Their phagocytic activities are thought to feature indefense against invading microorganisms . Several othertypes of cell are able to internalize particulate materials . Ofrelevance to arthritis is the ability of synoviocytes to do this.

Phagocytosis has a number of biochemical andphysiological consequences . One is the release of varioushydrolytic enzymes, several of which degrade cartilage . Avariety of particulate stimuli have been shown to elicit thisresponse, including inert substances such as latex beads(10), asbestos (11), minerals (12), precipitatedimmunoglobulins (13) and collagen (14) . Putting these twoobservations together synthesises the hypothesis thatphagocytosis of wear particles provokes the release of lyticenzymes which attack the articular surfaces . This would"soften"- them, thus potentiating the mechanical release ofmore particles . Such circumstances could alter the wearmodes in operation, thus producing particles of alteredmorphology. These "secondary" particles could also differchemically from bulk cartilage as a result of the enzymicattack which had facilitated their production . Such particlesmight have an altered propensity to provoke the phagocyticcells into releasing enzymes. From these considerations, it iseasy to appreciate the possibility of complex interactionsbetween mechanical wear, cellular and biochemical events,and arthritis.

In initial experiments to examine these possibilities,macrophages and synovial cells were grown in tissue

culture and exposed to wear particles retrieved fromsynovial fluid aspirates . The conditioned culture media werethen examined for various proteolytic enzymes of relevanceto the breakdown of cartilage (15) . The results aresummarized in Table 1 . Both macrophages and synovial cellsreleased proteinases in response to the wear particles.Collagenase may be important in breaking down thecollagenous component of cartilage. It is interesting to notethat synovial cells released more collagenase thanmacrophages; cultures of rheumatoid synovium are a richsource of this enzyme (16). This difference is reflected in theinability of macrophages to liberate as much hydroxyprolinefrom cartilage as synoviocytes . Both types of cell producedequivalent activities of the other proteinases tested andwere able to release quite large amounts of degradedproteoglycan from cartilage (Table 1).

In every case, mild trypsinisation of the media revealedenzyme latency (Table 1) . From these data it is not possibleto determine to what extent the newly synthesised enzymesare latent, as their autoactivation occurs duringconcentration and storage of the conditioned media.

With these initial observations, attention is being turnedtowards delineating which features of the wear particles areresponsible for producing the cellular effects . Wear particlesare very heterogeneous in size and shape. Their precisechemical make-up also varies depending on whether theyare of meniscal or articular cartilage, or osseous, and towhat extent they have been subject to enzymic attack.

The first of these variables to be investigated was size.Synovial fluid aspirates were not used as a source ofparticles for these experiments, as the yield of particles ofany one size range is small . Furthermore, even particles ofone size may be chemically heterogeneous . Thus, tominimize the number of variables, meniscal cartilage wasused as a source of particles . This was powdered and addedto cultures of cells . The response of cells exposed tocartilaginous particles of this type was compared to thatelicited by latex beads of diameter 1µm and 45 µm . Fromthe results of such experiments (Table 2), it was concludedthat internalization of particles by the cells was not anecessary condition for enzyme release. Particles which aretoo large for endocytosis adhere to the cell surfaces andpromote the release of enzymes (17) . This observation tiesin with the demonstration by Harris et al . (18) that uratecrystals need not be internalized by synovial cells tostimulate the production of collagenase . Thus it seems that,qualitatively, the observed tissue response is not limited toone particular size of wear particle, although there may wellbe qualitative differences due to variations in surface area tovolume ratios, etc.

Cartilaginous particles provoked a greater release ofneutral proteinases than did latex beads (19), which areassumed to be biochemically inert . It thus appears thatchemical, as well as physical, stimuli are involved in theeffects produced by cartilaginous wear particles on culturesof macrophages.

Chondroitin sulphate is a major component of cartilage,which appears to be one of these chemical stimuli (19).Purified chondroitin sulphate produced a marked elevationin the secretion of both lysosomal hydrolases and neutralproteinases by cultured macrophages (Table 3) . The effect is

24TECHNICAL NOTE : EVANS AND MEARS

TABLE 1.Rates of production of proteinases by cells in the presence or absence of wear particles.

Productions of Proteinases

(mUb/106 cells/day)

Cell Types ConditionsCollagenase Gelatinase Azocaseinase Pz-Peptidase

without particles 0 .73 ± 0 .06 2.37 ± 0 .18 1 .53 ± 0 .18 1 .92 ± 0 .17

without particles, 0 .97 ± 0 .10 2 .87 t 0 .21 2 .18 t 0.20 2 .41

± 0 .21

MurinePeritonealMacrophages

trypsinised

with particles 1 .52 ± 0 .21 4 .84 ± 0.25 2 .91

± 0 .23 3.53 ± 0 .28

with particles, 1 .91 ± 0 .52 6 .56 ± 0 .43 4 .28 t 0 .35 5.08 ± 0 .46

trypsinised

without particles 0 .33 2 .52 2 .05 1 .83

without particles, 0 .83 3 .15 3.10 2 .42

Human Blood

MononuclearPhagocytes

trypsinised

with particles 1 .61 4.72 3 .22 3.69

with particles, 1 .85 5 .98 4.86 5 .11

trypsinised

without particles 1 .31 ± 0 .14 3 .00 t 0 .21 1 .33 t 0 .21 2 .64 ± 0 .19

without particles, 2 .38 ± 0 .21 5.19 ± 0 .46 2 .56 ± 0.30 3 .82 ± 0.30

HumanSynovi alCells

trypsinised

with particles 5 .52 ± 0 .35 6 .52 t 0 .53 2 .84 ± 0 .38 5 .31

± 0 .42

with particles, 9 .95 ± 0 .89 11 .56

t.

1 .03 5 .20 ± 0 .51 8.18 .t 0 .78

trypsinised

aResults shown for murine peritoneal macrophages are the means t SEM of six replicate

experiments ; for synovial cells, four replicates ; and for human blood phagocytes, the

average of two replicate experiments.

b l unit of collagenase, gelatinase or Pz-peptidase degrades lug substrate per min at 37°.1 unit of azocaseinase degrades lmg azocasein per hour at 37°.

reversible ; within three days of changing the cultures ofmacrophages back to a medium which does not containchondroitin sulphate, the production of neutral proteinasesand lysosomal hydrolases returns to normal.

Certain lines of evidence suggest that these effects arenot limited to the artificial conditions of tissue culture . Wearparticles exist in considerable amounts in manyosteoarthritic joints (7) . Observations such as those shownin Figure 6 demonstrate the interaction of these particleswith periarticular cells and tissues, which may explain thesynovitis and other signs of inflammation that often occur inosteoarthritic joints . Furthermore, the activities of several ofthe enzymes measured in the work described here are alsoelevated in tissues taken from osteoarthritic joints (20).

Animal Model : Preliminary Results-We have recentlyemployed laboratory animals to determine, in a more direct

manner, whether wear particles have arthritogenicproperties (21) . Intra-articular injections of particles of rabbitarticular cartilage into the knees of recipient rabbits provokean intense inflammatory arthritis . After 3-4 months ofreceiving 3 mg of articular cartilage per week, the recipientknees become swollen and inflamed . Histologicalexamination of the synovium reveals a marked cellularinfiltrate and the presence of cartilaginous wear particles,presumably from the injected material . Organ cultures of thesynovium from knees receiving wear particles secrete muchhigher levels of both neutral proteinases and lysosomalhydrolases than do cultures of control synovium . Thecartilage of particle-injected knees is discoloured and itsmetachromatic staining properties are attenuated . Thesechanges are much more marked than those reported byChrisman et al . (22) who conducted similar experiments withdogs. In certain rabbits, the synovium appears to have

25

Bulletin of Prosthetics Research BPR 10-36 Fall 1981

TABLE 2.Production of extracellular proteinases by macrophages in response to particles of cartilage and latex beads of different sizes.

Enzyme Production

(munits/10 6cells/day)

Collagenase Gelatinase Azocaseinase Pz-Peptidase

Control Macrophages 0 .57 2 .1 1 .62 1 .85

Macrophages and Cartilaginous Particles 1 .26 5 .9 3.83 4 .52

Macrophages and Small

Latex Beads 0 .92 4 .3 3.21 3 .93

Macrophages and Large Latex Beads 0 .63 2 .9 2 .03 2 .16

One unit of collagenase, gelatinase or pz-peptidase breaks down 1 ug subsubstrate/min at 37° C . Oneunit of azocaseinase breaks down lmg azocasein/hr at 37°C . Enzyme assays on conditioned media were usually

incubated for 18 hr. Small latex beads were added to cultures at a concentration of 100 beads/cell ; thesame weight of large latex beads was added to parallel cultures . Powdered meniscus was added at a concen-tration of 5mq/75 cu cm culture vessel containing approx . 10' cells.

formed an invasive pannus, as found in human rheumatoidarthritis and in symptomatic artificial knee joints.

Experiments are underway to determine whether animmune response has been mounted to the injectedmaterial, or whether the effects can be accounted for by thecellular release of catabolic enzymes, and other factors,provoked by the particles . Preliminary results suggest that asystemic response to the injected particles has not occurred;the recipient rabbits give a negative skin test and attemptsto demonstrate circulating lgG antibodies against theparticles have failed . Furthermore, the timing and severity ofsymptoms is remarkably constant between animals, afeature which is not normally seen in experimental diseasesthat rely on an immune reaction.

Evidence that chondroitin sulphate elicits cellularresponses in vivo comes from experiments in which purifiedchondroitin sulphate was injected into the peritoneal cavitiesof mice (19) . After 4 days, peritoneal macrophages wereharvested, counted, cultured and their release of acid andneutral lytic enzymes measured . Chondroitin sulphate

elicited a concentration-dependent increase in theproduction of all enzymes measured except lysozyme ; thisfinding agrees with previous reports that the production oflysozome by macrophages is not greatly affected by theirstate of activation.

It is noteworthy that this stimulation of enzymeproduction occurred without any increase in the number ofcells per mouse. That is unusual, as most agents whichactivate peritoneal macrophages, such as thioglycollate,recruit additional cells, so that the yield of macrophages permouse is greatly enhanced . Chondroitin sulphate is notantigenic (23) . It may, however, have the important propertyof reversibly stimulating the local secretion of lytic enzymes,without triggering a full-blown inflammatory response.Workers in Sledge's laboratory (24) have found that theability of anionic polysaccharides, such as chondroitinsulphate, to provoke synovitis in rabbit knees increases withtheir molecular weight and charge density . Chondroitinsulphate was shown to produce a transient inflammation;4 weeks after the termination of the chondroitin sulphate

TABLE 3.Effect of chondroitin sulphate on the production of acid hydrolases and neutral proteinases by cultured macrophages.

Concentration of

Chondroitin Sulphate

ENZYMIC ACTIVITY OF CONDITIONED MEDIUM*Release from cartilage of:

in culture medium (mg/ml) Chondroitin Sulphate Hydroxyproline Azocaseinase 9-glucuronidase

(pH

7 .2) (pH

7 .2) (pH

5 .1) (pH

5 .1)

0 .0 4.40 1 .04 84 54 .7

0 .1 7 .47 3 .09 98 73 .6

1 .0 8.65 4 .67 126 94.9

10 .0 9 .14 4 .77 142 98.3

*Figures quoted are ug of substrate (azocasein, phenolphthalein-glucuronic acid) degraded, or jigof product (chondroitin sulphate, hydroxyproline) released per 10 6 macrophages per hour.

26

TECHNICAL NOTE : EVANS AND MEARS

injections, the synovium continued to produce slightlyelevated amounts of catabolin-like activity, but theproduction of lysosomal marker enzymes equalled that of

controls (25).

General Discussion

Despite its long history and widespread debilitatingeffects, osteoarthritis remains a problematic disease . It isdifficult to diagnose at an early stage, and it resists effectivetreatment . Diagnosis relies largely on patient symptoms andX-ray data supplemented, in certain instances, by techniquessuch as arthrotomy and arthroscopy. But the occurrence of

symptoms is peculiar. Pain, for instance, is not directlyrelated to the progression of osteoarthritis . Aroentgenographical survey of Americans over the age of 75years revealed an 85% incidence of articular degeneration(26) . Yet the incidence of arthritic symptoms is much lessthan this . Under such circumstances, osteoarthritis isfrequently diagnosed when it is too late to initiate theappropriate anti-inflammatory regime or undertake theoptimal reconstructive procedures, such as osteotomies torealign the joint or to restore joint congruity. Treatment isunsatisfactory, in that symptoms rather than the underlyingdisease processes receive attention . Thus, an arthriticpatient with advanced disease may be treated by resort to awalking aid, while the pathophysiological undertow flowsunchecked . Alternatively, an unnatural replacement joint oflimited anticipated functional period may be the onlyrealistic method of treatment . This, of course, reflects an

ignorance of the aetiology and mode of progression ofosteoarthritis . Of the existing treatments for severeosteoarthritis, prosthetic hip replacements are outstandinglysuccessful for a period of 5 to 10 years. Much effort is

presently being directed towards extending the range ofsuch prostheses to include knees, fingers and elbows and

other types of joints . An important determinant is thebiomechanical properties of the materials forming the

prosthesis . In particular, it is essential that the articulatingsurfaces have the appropriate rheological and tribologicalproperties. Accurate evaluation of these functions, especiallywhen the implants are performing in situ, is needed.

From the findings reviewed in this paper, we feel that theferrographic analysis of wear particles and studies of theirbiochemical and cellular properties can aid the resolution ofmany of these problems.

REFERENCES

1. Seifert WW and Westcott VC : A Method for the Study of WearParticles in Lubricating Oil . Wear 21 :27-42, 1972.

2. Scott D, Seifert WW and Westcott VC : The Particles of Wear.Sci Amer 230 :88-97, 1974.

3. Bowen ER, Scott D, Seifert WW and Westcott VC : Ferrography.Tribology International 9 :109-115, 1976.

4. Mears DC, Hanley EN, Rutkowski R and Westcott VC : Ferro-graphic Analysis of Wear Particles in Arthroplastic Joints . JBiomed Biomater Res 12 :867-875, 1978.

5. Evans CH, Bowen ER, Bowen J, Tew WP and Westcott VC : Syn-ovial Fluid Analysis by Ferrography. J Biochem Biophys Meth2 :11-18, 1980.

6. Evans CH and Tew WP : Isolation of Biological Materials UsingErbium (111)-Induced Magnetic Susceptibilities . Science . 213 :653-654, 1981.

7. Evans CH, Mears DC and McKnight JL : A Preliminary Ferro-graphic Survey of the Wear Particles in Human Synovial Fluid.Arthritis Rheum 24 :912-918, 1981 .

8. Pautard FGE : Data presented at the "Inorganic BiochemistryDiscussion Group" Meeting, Birkbeck College, University ofLondon, December 1980.

9. Halverson PB, Cheung HS, McCarty DJ, Garancis J, and MandelN : "Milwaukee Shoulder" : Association of Microspheroids Con-taining Hydroxyapatite Crystals, Active Collagenase and Neu-tral Protease with Rotator Cuff Defects . II . Synovial Fluid Studies.Arthritis Rheum 24:474-483, 1981.

10. Werb Z and Reynolds JJ : Stimulation by Endocytosis of theSecretion of Collagenase and Neutral Protease from RabbitSynovial Fibroblasts. J Exp Med 140 :1482-1497, 1974.

11. Miller K : The Effects of Asbestos on Macrophages . CRC CritRev Toxicol 5 :319-354, 1978.

12. Cheung HS, Halverson PB and McCarty DJ : Release of Colla-genase, Neutral Protease and Prostaglandins from CulturedMammalian Synovial Cells by Hydroxyapatite and Calcium Py-rophosphate Dihydrate Crystals . Arthritis Rheum 24 :1338-1344,1981.

13. Ragsdale CG and Arend WP : Neutral Proteinase Secretion byHuman Monocytes . Effect of Surface-Bound Immune Com-plexes . J Exp Med . 149 :954-968, 1979.

14. Biswas C, Dayer JM : Stimulation of Collagenase Production byCollagen in Mammalian Cell Cultures . Cell 18 :1035-1041, 1979.

15. Evans CH, Mears DC and Cosgrove JL : Release of Neutral Pro-tenases by Mononuclear Phagocytes and Synovial Cells in Re-sponse to Cartilaginous Wear Particles in Vitro . Biochim BiophysActa 677 :287-294, 1981.

16. Harris ED, Brinckerhoff CE and Vater A : Release of Collagenasefrom Synovial Tissues in Rheumatoid Arthritis . In (Panagi GSand Johnson PM, eds .) Immunopathogenesis of RheumatoidArthritis . Chertsey (Surrey) Reedbooks 40 :147-154, 1979.

17. Evans CH : Cellular Mechanisms of Hydrolytic Enzyme Releasein Osteoarthritis . Sem Arthritis Rheum Vol . 11, Suppl . 1 :93-95,1981.

18. Harris ED, Vater CA, Brinckerhoff CE, McMillan RM and Has-selbacher P : Patterns of Regulation of Collagen Breakdown inArticular Cartilage . Sem Arthritis Rheum Vol . 11, Suppl . 1 :69-70, 1981.

19. Evans CH and Mazzocchi RA : Chondroitin Sulphate Elicited Re-lease of Catabolic Enzymes from Macrophages . Trans OrthopRes Soc 7 . Page 212, 1982.

20. Erlich MG, Houle PA, Vigliani G and Mankin HJ : CorrelationBetween Articular Cartilage Collagenase Activity and Osteoar-thritis . Arthritis Rheum 21 :761-766, 1978.

21. Evans CH, Mazzocchi RA, Nelson DD and Rubash HE : Experi-mental Arthritis Induced by Cartilaginous Wear Particles . TransOrthop Res Soc 7 . Page 191, 1982.

22. Chrisman OD, Fessel JM and Southwick WD : Experimental Pro-duction of Synovitis and Marginal Articular Extoses in the KneeJoints of Dogs. Yale J Biol Med 37 :409-412, 1965.

23. Keiser H and Sandson J : Immunology of Cartilage Proteogly-can . Fed Proc 32 :1474-1477, 1973.

24. Aaron RK, Sledge CB and Shortkroff S : Experimental ArthritisInduced by Polysaccharide Molecules . Arthritis Rheum. In press,1981.

25. Steinberg J, Sledge CB, Noble J and Gilles G : Dextran SulphateInduced Arthritis in Rabbits. Ann Rheum Dis 39 :387-391, 1980.

26. Health Statistics ser B no . 36 US Dept Health, Education andWelfare, 1962 . (Chronic Conditions Causing Limitations ofActivities .)