cement - journal of clinical pathology · contrast media in bone cement table i clinical details...

_4 Clin Pathol 1992;45:984-989

Histopathological and microanalytical study ofzirconium dioxide and barium sulphate in bonecement

C E Keen, G Philip, K Brady, J D Spencer, D A Levison

AbstractAims: To report the appearances of zirco-nium dioxide and barium sulphate ininterface membranes, synovium, andother tissues around joint prostheses.Methods: Histological sections from 23specimens were reviewed by light micro-scopy and polarisation. Scanning electronmicroscopy and x ray microanalysis wereperformed on routinely processed paraf-fin wax sections.Results: Polyethylene, metals, and poly-methylmethacrylate cement debris wereeasily recognisable. Almost all the cementremnants contained either zirconiumdioxide or barium sulphate, confirmed bymicroanalysis. The contrast media hadcharacteristic light microscopic appear-ances. Zirconium was identified in macro-phages away from cement remnants.Conclusion: The presence ofradiographiccontrast media in tissues around pros-thetic joints is common but not widelyrecognised.

( Clin Pathol 1992;45:984-989)

Department ofHistopathologyC E KeenG PhilipD A LevisonDepartment ofOrthopaedic SurgeryJ D SpencerElectron MicroscopyUnitK BradyUnited Medical andDental Schools, Guy'sCampus, LondonCorrespondence to:Dr C E Keen, Departmentof Histopathology,Lewisham Hospital, LondonSE13 6LH.

Accepted for publication28 April 1992

Contrast media are added to polymethyl-methacrylate cement used in joint replacementoperations to confer radiopacity. The commonmedia used are zirconium dioxide and bariumsulphate. Although the acrylic cement is largelylost during routine processing, these contrastmedia can be identified in the resultant spacesin histological sections of synovium and otherperiarticular tissues excised during biopsy orrevision of joint replacement. We could find nodetailed accounts of the histological appear-ances of these media in such tissues. In a studyof the reaction of bone to methacrylate Willertet al detected the presence of contrast media byelectron microprobe studies and described thezirconium dioxide particles as "mulberry-shaped". 1 Revell described the presence ofcontrast media as a useful indicator that tissuespaces were the site ofbone cement.2 Meachimet al, in experimental studies of synovialresponse to acrylic cement, used the presenceof barium sulphate as a marker of the sites ofacrylic cement dissolved in processing.3 AGerman publication documents histiocyticreaction and intracytoplasmic localisation ofzirconium oxide and polymethylmethacrylatewear particles in interface membranes bytransmission electron microscopy and lasermicroprobe mass analysis.4 The same groupidentified and graded wear debris, including

zirconium in draining lymph nodes, in nec-ropsy material.5 Another group discussed theeffect of the contrast media on the tissuereaction in relation to joint protheses.6 Experi-ments have been reported showing thatscratching damage to stainless steel femoralhead components is predominantly caused bythe abrasive effect of x ray contrast media inacrylic cement." The scratched femoral headcomponent then causes increased wear in thepolyethylene acetabular cup.We undertook a critical light, polarisation,

and scanning electron microscopic review of23 cases of cement or other wear particledeposition in relation to prosthetic joints, tosearch for these media and to characterise theirappearances. The specificity of our findingswas supported by microanalysis ofparaffin waxsections by energy dispersive x ray spectro-scopy.

MethodsThe 23 cases that form the basis of this studywere collected from the referral files of one ofthe authors (DAL) and the surgical files ofGuy's and Lewisham hospitals. The clinicaldetails are shown in table 1. The study was setup following the unexpected analytical findingof zirconium in four consecutive synovialbiopsy specimens from patients with jointprostheses. We included all our samples(accrued over two years) of periarticular tissuecontaining foreign material from patients witha prosthetic joint or other implant.

Representative haematoxylin and eosinstained sections were examined by lightmicroscopy and with polarised light.

Scanning electron microscopy (SEM) inconjunction with x ray energy dispersive spec-troscopic microanalysis (EDS) was performedin all cases. In brief, dewaxed sections (stainedor serial unstained) were transferred from glassslides to a Perspex substrate using an organicbased glue (Beecham UHU). The sectionswere carbon coated and inserted into thespecimen chamber of an Hitachi S520 micro-scope equipped with an S4548 high efficiencyscintillator type backscatter detector and aKevex 4460 delta class analyser. Secondaryelectron images and backscattered electronimages (providing atomic number related aswell as topographic images) of the contrastmedia and wear particles were photographed.Numerous particles of various sizes were ana-lysed in each section. Analysis was generallyperformed at an accelerating potential of 30KV and for 100 seconds. To detect oxygen, a

984

on March 27, 2020 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.45.11.984 on 1 Novem

ber 1992. Dow

nloaded from

http://jcp.bmj.com/

Contrast media in bone cement

Table I Clinical details of cases studied

Case Interval to revi- Elements on micro-No Age Initial operation (cement used) sion Site of biopsy analysis

1 41 Titanium ring prosthesis Hip joint ZrTi2 64 Capsule of hip Zr Ti3 88 THR Zimmer D 4 years Darkly stained periarticular fibrous Zr Ti

tissue. Revision of loose hipprosthesis

4 69 TKR Uncemented PCA 3 years Mass in knee (stiff & painful) ZrFragment of cement from joint

5 61 TKR Stanmore 9 years Synovium of knee at revision Zr Co/Cr6 52 Hemiarthroplasty (Palacos) 2 years Synovium at Girdlestone excision Zr Ti Co/Cr7 63 THR Charnley. Revised after 1 13 years Interface membrane (loose prosthesis) Zr Cr/Fe/Ni Co/Cr

year at Girdlestone excision arthroplasty8 69 THR Zimmer D 9 years Hip Zr Co/Cr9 69 THR twice 16 years Loose prosthesis, Femoral shaft Zr Ba/S

interface membrane10 44 THR Zimmer (CMW) 5 years Biopsy of inflamed synovium and Ti Ba/S

capsule11 76 TKR twice (Palacos R in 2nd 2 years Granulation tissue at arthrodesis Zr Co/Cr Fe/Ni

operation)12 76 THR Membrane from acetabulum at Zr

revision13 38 Irradiated lymphoma. Articular capsule at revision to total Cr/Fe/Ni

Pathological fracture neck of hip replacementfemur Dynamic hip screw

14 57 THR Ultralock (Palacos R) 18 months Tissue from hip capsule Zr Ti15 87 TKR Synovium and intramedullary lining Co/Cr Cr/Fe/Ni16 Attenborough TKR (Simplex) 10 years Tissue from around painful prosthesis Co/Cr17 76 THR Tissue from joint space Zr Ti18 64 THR 18 months Inflamed granulation tissue around Ba/S

painful ?infected joint19 73 THR Ultralock (Palacos R) 4 years Synovium of hip ZrTi20 32 THR 7 years Synovium of hip at revision Zr21 55 THR Zimmer D then 7 years Synovium and capsule. Revision of ZrTi Ba/S

Stanmore; (Palacos) loose prosthesis22 64 TKR PCA (Simplex) 2 years Synovium Co/Cr Ba/S23 82 TKR Attenborough Synovium of knee at revision to Co/Cr

cemented TriconTHR

TKR = Total knee replacement. THR = Total hip replacement.

much lower potential was used, to obtainoptimal overvoltage, the ratio between theaccelerating potential in the column and theenergy of the emitted x rays. Oxygen is near thelower atomic number limit for detection in ourEDS system, and minimal column contamina-tion affects our ability to detect it.

Samples of three different named brands ofcement were also examined. These are shownin table 2. The cement powder was sprinkledon a slide in each case and mounted in DPX.Double sided tape was used to attach samplesof the powder to a Perspex substrate for SEMand EDS.Bulk samples of used cement removed dur-

ing prosthesis revision as well as fragments cutfrom a polyethylene acetabular cup were exam-ined by SEM and EDS.

Correlation was made between the appear-ances of the human tissue samples and theunused cement samples by light, polarisationand electron microscopy, and EDS findings.An attempt was made to detect putative

contaminating radioactivity in the zirconiumcontaining cement. A full packet (40 g) ofPalacos bone cement containing 6-0 g ofzirconium dioxide was counted for 50000seconds (14 hours) in a large volume counter,using a sodium iodide detector housed in ashielded container and connected to a multi-channel analyser.

Table 2 Different brands of cement examined

Elements onTrade name Manufacturer microanalysisPalacos LV-40 Kulzer & Co Zr/OCMW 3 Bone Cement CMW Ba/SSurgical Simplex P Howmedica Ba/S

ResultsThe foreign material identified in the sectionsincluded polyethylene, polymethylmethacry-late cement, radiographic contrast media andvarious metal or metal alloy particles. Individ-ual appearances are described below and ana-lytical findings are summarised in table 1.SEM of paraffin wax sections showed that itwas easy to identify correlating morphologicalfeatures in any particular section and hence toconfirm by EDS the elemental composition ofmuch of the foreign material identified by lightand polarising microscopy. However, littleadditional information was obtained by thistechnique on the elemental composition of thecement or the polyethylene as these are com-posed exclusively ofcommon organic elementsmost of which are too light for detection in anEDS system.

Colourless, transparent, strongly birefrin-gent, organic material in rough elongate shardselicited a florid foreign body-type giant cellreaction. Seen in 10 cases, this material,resistant to routine processing, and giving noEDS reading above background, was assumedto be polyethylene or other plastic wear debrisfrom the prostheses. This interpretation was

supported by analysis of fragments cut from a

prosthetic acetabular cup; these were birefrin-gent and gave no clear EDS signal abovebackground, consistent with polyethylenewhich is composed only of carbon andhydrogen.

Polymethylmethacrylate cement was largelydissolved out by routine processing, but thespaces left were identifiable by the admixtureof zirconium dioxide particles or barium sul-phate. These spaces were usually ovoid, oftenseptate, and of the order of 0-2 mm-2 mm in

985



j.com/

J Clin P


ber 1992. Dow

nloaded from

http://jcp.bmj.com/

Keen, Philip, Brady, Spencer, Levison

Figure 1 Zirconiumdioxide particles in cementspace. (A) Lightmicroscopy (haematoxylinand eosin); (B) polarisedlight (haematoxylin andeosin); (C) low powerscanning electronmicrograph ofparaffin waxsection (backscatteredelectron image); (D) highpower secondary electronimage.

*e

.-'

maximum diameter. All but three of the casesincluded had recognisable cement spaces. Thespaces were surrounded by attenuated giantcells, and in all but one case (case 23),contained particles of contrast material. Whereit was not completely dissolved away thecement appeared as transparent refractileshards without birefringence. Again analysis ofthese shards showed no qualitative differencefrom the perspex substrate. Analysis of thepolymethylmethacrylate portion of the bulkcement fragments gave a significant peak foroxygen, again consistent with its elementalcomposition-namely carbon, oxygen, andhydrogen.The zirconium dioxide particles (16 cases)

consisted of globular aggregates 5-50 ,um indiameter (fig 1). At low power these appeareddark, but at high power there was sometransparency in the centre of the globules. Onpolarisation the particles exhibited a character-istic "flat" whitish or pale lilac birefringence.Occasionally there was a more discrete partic-

ulate birefringence within the aggregates. Inone case the zirconium had a more finelydivided granular appearance that was moredifficult to distinguish from barium sulphate.However, in this case the granular material wasdarker than barium and was birefringent likezirconium dioxide.

Microanalysis of all these particles showedthat the overwhelming signal was of zirconiumbut trace amounts of hafnium were also pres-ent, (fig 2A). Under optimal circumstances,oxygen, consistent with zirconium dioxide,could also be detected (fig 2B).

In most cases zirconium could be detectedin the cytoplasm of histiocytes away from thecement spaces, (fig 3). This was confirmed bylight microscopy and polarisation in sevencases; small particles typical of zirconiumdioxide were present in mononuclear phago-cytes, often along with metal shards.Barium sulphate (five cases) (fig 4) was

found in similar polymethylmethacrylatecement spaces with surrounding attenuated

986



j.com/

J Clin P


ber 1992. Dow

nloaded from

http://jcp.bmj.com/


14-nev-1" loasmis7CDS 91

Vert. 7176 cavta $lIp' IZr

Pretat. 56 aeElapead- t16 es

~~~~~~~~~~Zr1

4- CM ftse 10.2 W> MM1rl la~~~~Stegr1 * * msl6-i

Figure 2 Spectra obtained on microanalysis. (A)Zirconium dioxide analysed at 20 KV Note smaUl peaksfor hafnium. (B) Oxygen is shown at 10 KV in thezirconium compound. (C) Barium sulphate.

Figure 3 Zirconium dioxide in a macrophage(backscattered electron image). Both the larger particleand the finer granules give a clear zirconium signal onmicroanalysis.

giant cells, but was often very inconspicuous.Sparse collections of particles each about 5 ,umin diameter were found, and in some cases theyhad only a very weak hint of the golden-brownhue typical of the barium sulphate usually seen

in the intestine. There was a little birefringenceon polarisation, but the best method of high-lighting the presence of barium was to lowerthe microscope condenser lens, revealingstrong refractility. Clear peaks for barium andsulphur were obtained by EDS of this material(fig 2C).

In the three cases that contained bothzirconium dioxide and barium sulphate thesecontrast media were found in separate cementspaces, never mixed. Zirconium dioxide andbarium sulphate were confined to the cases

showing cement spaces, confirming their asso-

ciation with cement. In only one case were

cement spaces identified without finding zirco-nium dioxide or barium sulphate, confirmingthe very widespread use of cement with addedcontrast media. This case was one of thosefrom which bulk fragments of cement wereobtained, and it was confirmed by analysis ofthese that the cement in this case contained nocontrast medium. In all cases where it wasknown what cement was originally used, thefindings on microanalysis were consistent.Three sorts of metal deposits were noted by

EDS and the appearances of these depositswere reviewed by light microscopy. Titanium(nine cases) was finely divided in fairly uniformsmall black shards in the cytoplasm of sheets ofmononuclear, and sometimes multinucleate,vacuolated macrophages. Titanium was oftenassociated with minor aluminium and vana-

dium signals, expected in titanium alloys.Cobalt/chromium alloy (eight cases) appearedsimilar, with indistinguishable tissue reaction,but also a tendency to an admixture of largershards up to 50 pm in diameter, often extra-cellular. Four cases contained particles consist-ing of chromium/iron/nickel (stainless steel)and in at least one case (case 13) these were

deposited without macrophage reaction. All ofthese metals or alloys gave a similar pattern on

polarisation, with a weak pinkish linear glint atthe edge of many particles. Four cases showedthe presence of two types of metal. Both thecobalt/chrome alloy and the stainless steel gavesmall EDS peaks for molybdenum, in keepingwith the usual composition of these alloys. Inpractice, we found it impossible to predictreliably the type of metal or alloy from thehistological appearance.The appearances of the contrast media in

the unused cement samples (table 2) corre-lated precisely with those in the human tissues.Light microscopical examination showed thatzirconium dioxide was present in globularaggregates identical with those in the cementspaces in the tissues. The barium sulphate was

also easily recognisable as such and was much

more abundant in the unused cement samplesthan in the tissue sections. The refractile andpolarising properties of the contrast mediawere also similar to those in the tissues. Inall of the samples the contrast media were

admixed with transparent, refractile but non-

(A)

(B)

(C)

IZ 14 5 5 l 1Z IZ114 1184-6.,6 Rng.. 28.466 Set 23.228 6.

_9soseesateral234

12-N.-191 17:61:26CDI 233 31 Prcc. 5U acesVert- 0671 eoat* Slago I Elapsed- 1a owes

Zr

987



j.com/

J Clin P


ber 1992. Dow

nloaded from

http://jcp.bmj.com/

Keen, Philip, Brady, Spencer, Levison

Figure 4 Bariumsulphate particles in cementspace. (A) Polarised light(haematoxylin and eosin);(B) condenser racked down(haematoxylin and eosin);(C) scanning electronmicrograph ofparaffin waxsection (backscatteredelectron image).

W..

..

e..

birefringent spheres of polymethylmethacry-late. Scanning electron microscopy showedthat there was similar morphological correla-tion (fig 5) and in the backscatter mode theatomic number contrast highlighted the zirco-nium dioxide and barium sulphate with respect

Figure S Scanningelectron micrograph ofunused cement sample(Palacos) containingzirconium dioxide particlesand showing part of apolymethylmethacrylatecement sphere below.

to the organic polymethylmethacrylate spheresand the Perspex substrate. The elements foundby microanalysis (table 2) corresponded tothose expected from the product information.Again traces of hafnium were found associatedwith the zirconium, but neither radium noractinium were detected. Radioactivity was notdetected in the zirconium containing cementby the sensitive method used.

DiscussionThe appearances in histological sections ofjoint prosthesis wear particles of polyethylene,metal, and polymethylmethacrylate cementhave been well documented.29 We have addeddetailed descriptions of the appearances bylight and polarisation microscopy of the com-mon radiopaque contrast media mixed withpolymethylmethacrylate cements. These mediaare found in the spaces left by removal of theacrylic cement in processing.Zirconium dioxide is found as small globular

aggregates exhibiting a characteristic "flat"whitish birefringence. The shape and situation(mainly in cement spaces) of zirconium diox-ide particles in periprosthetic tissues are char-acteristic and should permit confidentrecognition by the histopathologist reportingsuch specimens. The polarisation properties,while distinctive and providing useful con-firmation, are not specific. A very similar "flat"whitish birefringence, sometimes with a pink-ish hue in stained sections, is seen in tissuespecimens containing titanium dioxide.'0 1IOther metallic oxides probably show similarpolarisation characteristics by light micro-scopy.Barium sulphate granules are often sparse

and inconspicuous, and show little birefrin-gence, but are strongly refractile. They, too, arefound in cement spaces. The specificity ofthese observations was confirmed by EDS andgood morphological correlation was obtainedbetween the light and scanning electron micro-

988



j.com/

J Clin P


ber 1992. Dow

nloaded from

http://jcp.bmj.com/


scopic appearances.The occurrence of both media in two cases

(9 and 21) in separate cement spaces suggestedthe use of different cements either at one ormore operations, rather than one cementcontaining a mixture of contrast media.Indeed, each of these patients was found onclinical review to have had three hip replace-ment operations, although the details of thecement used on every occasion were notavailable.We did not find specific tissue reaction

patterns for the different contrast media. Thetissue reaction to the cement whether it con-tained barium sulphate, zirconium dioxide, orno contrast medium was the same. We wereable to show the presence of zirconium inmany cases in the cytoplasm of histiocytes asdescribed by Bos et al.4 We are, however,unwilling to speculate on any possible patho-genetic role that these contrast media mayhave.On first detecting zirconium in tissues, it

took detailed enquiry to discover its source.The fact that barium sulphate or zirconiumdioxide is added to bone cement seems not tobe widely known by general histopathologistsor orthopaedic surgeons. We feel it is worth-while drawing attention to this fact so that theappearances reported may be recognised moreeasily. Other sources of zirconium in tissuesaround prostheses may include second genera-tion zirconia ceramics." It is also possible thatzirconium or its compounds may be found inother types of specimens. For example, it hasbeen used in some dental media, as well as (inthe past) for radiographic contrast of thealimentary tract (Zirconotrast which containedzirconium dioxide'3). Hence zirconium mightbe expected to be found in oral biopsy speci-mens and in the gastrointestinal tract. Zirco-nium dioxide and other zirconium salts, suchas sodium zirconium lactate and zirconiumoxychloride, have been used in deodorantpreparations and are said to have causedallergic granulomatous reactions,'4 but wehave no information about the appearances ofsuch compounds in the skin. Despite this weare confident that such particles could beshown to contain zirconium by EDS in asystem such as ours.We were unable to detect radioactivity in the

zirconium containing cement sample that wetested. Such radioactivity, due to contamina-tion of natural zirconium by radium andactinium, was reported by Hopf et al.'5 16Theyemphasised that only barium sulphate orhighly purified zirconium dioxide free of thesecontaminating radioactive elements should beused in bone cements for human implantation.The levels of radioactivity they detected inzirconium containing cements were very low,but they could not exclude a risk of cumulativedamage. It seems unlikely that the rare cases ofsarcoma associated with plastic or metalimplants'7"-" could have been related to

endogenous radioactivity of the implant or

contrast medium in bone cement. It is alsointeresting that we were able to detect haf-nium, a common non-radioactive contaminantof natural zirconium ores. The manufacturersof Palacos are currently seeking alternativesources of zirconium dioxide with sufficientlylow residual radiation and without adverseeffect on the mechanical properties (Sheard J;Schering-Plough Ltd, personal communica-tion).We conclude that the diagnostic pathologist,

armed with our descriptions, will easily be ableto recognise zirconium dioxide and bariumsulphate in tissue samples such as ours. As thismaterial becomes more widely recognised,further meaningful pathological studies intothe implications of its use may be feasible.

We thank Sarah Allen, senior physicist in the Department ofNuclear Medicine, Guy's Hospital, for performing the radio-activity studies on the samples of cement.

1 Willert H-G, Ludwig J, Semlitsch M. Reaction of bone tomethacrylate after hip arthroplasty. A lonig-term gross,light microscopic, and scanning electron microscopicstudy. J Bone joint Surg 1974;56-A:1368-82.

2 Revell PA. Tissue reactions to joint prostheses and and theproducts of wear and corrosion. In: Berry CL, ed. Boneand joint disease. Current topics in pathology 17. Berlin:Springer, 1982:73-101

3 Meachim G, Brooke G. The synovial response to intra-articular acrylic cement particles in guinea pigs. Bio-materials 1984;5:69-74.

4 Bos I, Lindner B, Seydel U, et al. Untersuchungen uber dieLockerungsursache bei zementierten Huftgelenkendo-prothesen. Licht- und elektronenmikroskopische Unter-suchung und Laser-Mikrosonden-Massenanalyse. ZOrthop 1990;128:73-82.

5 Bos I, Johannisson R, L6hrs U, Lindner B, Seydel U.Comparative investigations of regional lymph nodes andpseudocapsules after implantation of joint endopros-theses. Path Res Pract 1990;186:707-16.

6 Rudigier J, Reck R, Walde H-J, Degreif J. Der Einflui3 vonRontgenkontrastmitteln in Knochenzementen auf Bind-egewebe und Knochenstruktur. Aktuel Probl Chir Orthop1987;31:181-3.

7 Isaac GH, Atkinson JR, Dowson D, Kennedy PD, SmithMR. The causes of femoral head roughening in explantedCharnley hip prostheses. Engineering in Medicine1987;16: 167-73.

8 Caravia L, Dowson D, Fisher J, Jobbins B. The influence ofbone and bone cement debris on counterface roughnessin sliding wear tests of ultra-high molecular weightpolyethylene on stainless steel. Proceedings of the Institute ofMechanical Engineers 1990;204:65-70.

9 Revell PA. The synovial biopsy. In: Anthony PP, MacSweenRNM, eds. Recent advances in histopathology Vol 13.Edinburgh: Churchill Uvingstone, 1987;79-93.

10 Morgan CA, Mullick FG, Ishak KG, Johnson FB, HummnerWB. Identification of titanium in human tissues: Probablerole in pathologic processes. Hum Pathol 1991 ;22:450-4.

11 Coelho Filho JC, Moreira RA, Crocker PR, Levison DA,Corrin B. Identification of titanium pigment in drugaddicts' tissues. Histopathology 1991 ;19: 190-2.

12 Christel PS. Zirconia: The second generation of ceramicsfor total hip replacement. Bulletin of the Hospital for JointDiseases Orthopaedic Institute 1989;49: 170-7.

13 Dalheimer AR, Kaul A, Said MD. Analysis of tissue samplescontaining colloidal thorium dioxide (Thorotrast) orzirconium dioxide (Zirconotrast): Radiochemical prep-aration and alpha-spectrometry. Sci Total Environ1988;70:335-53.

14 Reynolds JEF, ed. Martindale. The extra pharmacopoeia. 29thedn. London: The Pharmaceutical Press, 1989:937.

15 HopfW, Hopf CG, Globel B. About radioactivity in somePMMA bone cements. Acta Orthop Bedg 1990;56:443-4.

16 Hopf T, Scherr 0, Globel B, Hopf C. Vergleichendetierexperimentelle Untersuching zur Gewebsvertraglich-keit und Messungen der Radioaktivitat verschiedenerRontgenkontrastmittel. Z Orthop 1989;127:620-4.

17 Enzinger FM, Weiss SW. Soft tissue tumours. 2nd ed. St.Louis: CV Mosby, 1988:2.

18 Calandruccio RA. Arthroplasty of hip. In: Crenshaw AH,ed. Campbell's operative orthopaedics. 7th edn. St Louis:CV Mosby, 1987:1236.

19 Wapner KL. Implications of metallic corrosion in total kneearthroplasty. Clin Orthop Rel Res 1991;271:12-20.

989



j.com/

J Clin P


ber 1992. Dow

nloaded from

http://jcp.bmj.com/

cement - journal of clinical pathology · contrast media in bone cement table i clinical details...

Documents