Abstract

The mechanisms of bone loss around dental implants are poorly understood. The osteoclast is the most important bone-resorbing cell. Humoral factors seem able to stimulate the differentiation of osteoclasts from mononuclear phagocytes. Bacterial lipopolysaccharides seem to be directly involved in inflammatory bone loss by stimulation of the survival and fusion of preosteoclasts. Excessive load seems to be able to cause bone loss.

The aim of this paper was to evaluate the presence and number of osteoclasts in peri-implant bone in control (unloaded) and test (loaded) implants in order to determine if loading per se could be a contributing factor in peri-implant bone resorption. Forty-eight implants were inserted in the mandibles of 4 beagle dogs. After 3 months, a prosthetic superstructure was inserted on 24 implants, whereas in 24 implants only the healing screws were positioned. Twenty-four implants (12 test and 12 control) were retrieved at 6 months, and 24 implants (12 test and 12 control) were retrieved at 12 months. All implants were osseointegrated. The number of osteoclasts found in the crestal bone in the first 3 mm from the implant surface was evaluated. The mean number of osteoclasts were the following: control implants (6 months), 5.66 ± 0.81; control implants (12 months), 2.55 ± 1.05; test implants (6 months), 5.25 ± 1.55; and test implants (12 months), 2.5 ± 1.0. No statistically significant differences were observed between the control and test implants. According to our data, loading does not seem to have a relevant importance on the osteoclast activation in peri-implant bone.

RESEARCHOSTEOCLAST ACTIVITY AROUND LOADED ANDUNLOADED IMPLANTS: A HISTOLOGICAL STUDYIN THE BEAGLE DOGBartolomeo Assenza, MD, DDSAntonio Scarano, DDSGiovanna Petrone, DDS, PhDGiovanna Iezzi, DDSUlf Thams, MD, DDSFidel San Roman, MDAdriano Piattelli, MD, DDSKEY WORDSBone resorptionLoadingMicrogapOsteoclastBartolomeo Assenza, MD, DDS, is avisiting professor in the Dental School,University of Chieti, Italy, and is in privatepractice in Cologno Monzese (MI), Italy.Antonio Scarano, DDS, is a research fellowin the Dental School at the University ofChieti, Italy.Giovanna Petrone, DDS, PhD, is apostdoctoral fellow in the Dental School at theUniversity of Chieti, Italy.Giovanna Iezzi, DDS, is a research fellow inthe Dental School at the University of Chieti,Italy.Ulf Thams, MD, DDS, is in private practicein Madrid, Spain.Fidel San Roman, MD, is a professor ofsurgery and the director of the VeterinaryTeaching Hospital at Complutense University,Madrid, Spain.Adriano Piattelli, MD, DDS, is professor ofOral Medicine and Pathology in the DentalSchool in the University of Chieti, Italy.Address correspondence to Dr Piattelli at Via F.Sciucchi 63, Chieti, Italy, 66100(e-mail: apiattelli@unich.it).The mechanisms of bone loss around dental implants are poorly understood. Theosteoclast is the most important bone-resorbing cell. Humoral factors seem ableto stimulate the differentiation of osteoclasts from mononuclear phagocytes.Bacterial lipopolysaccharides seem to be directly involved in inflammatory boneloss by stimulation of the survival and fusion of preosteoclasts. Excessive loadseems to be able to cause bone loss.The aim of this paper was to evaluate the presence and number of osteoclastsin peri-implant bone in control (unloaded) and test (loaded) implants in orderto determine if loading per se could be a contributing factor in peri-implant boneresorption. Forty-eight implants were inserted in the mandibles of 4 beagle dogs.After 3 months, a prosthetic superstructure was inserted on 24 implants, whereasin 24 implants only the healing screws were positioned. Twenty-four implants(12 test and 12 control) were retrieved at 6 months, and 24 implants (12 test and12 control) were retrieved at 12 months. All implants were osseointegrated. Thenumber of osteoclasts found in the crestal bone in the first 3 mm from the implantsurface was evaluated. The mean number of osteoclasts were the following:control implants (6 months), 5.66 6 0.81; control implants (12 months), 2.55 61.05; test implants (6 months), 5.25 6 1.55; and test implants (12 months), 2.5 61.0. No statistically significant differences were observed between the control andtest implants. According to our data, loading does not seem to have a relevantimportance on the osteoclast activation in peri-implant bone.Timplants are poorly underitsrmin d an e o unt o am d un aro l e n boIhNTRODUCTIONe precossise mechanismsdentalof bone matrix.f bo2nModelingdetechangesesthestood.1 Osteoblasts and osteoclastsare involved inadaptive modeling and remodeling;these cells are able to sensetheir mechanical environment and toregulate deposition or resorption ofgeometrical form in relation to the prevailingmechanical loads and their resultingdeformations (strains).3 Remodelingrenews existing bone in a sequenceof resorption and formation;strain distributions occur during theremodeling process and show a rela-Journal of Oral Implantology 1OSTEOCLAST ACTIVITY AROUND IMPLANTS IN BEAGLE DOGStionship to the activity of osteoblastsand osteoclasts.3The osteoclast is the most importantbone-resorbing cell, and it derivesfrom the monocyte/macrophage lineage.4-6 Bipotent osteoclast precursors,which are able to form both osteoclastsand monocyte-macrophages, differentiateand become unipotent osteoblastprecursors, and these cells fuse togetherto form the multinucleated osteoclasts,which are cells activated to startbone resorption.4,7 Osteoclasts highlyexpress the alpha-beta3 integrin, whichbinds to a variety of extracellular matrixproteins including vitronectin, osteopontin,and bone sialoprotein.8These molecules tend to be located inthe sealing zone of actively resorbingosteoclasts, and they seem to have arole in linking the adhesion of osteoclaststo the bone matrix with the cytoskeletalorganization and the polarizationand activation of these cells forbone resorption.8 Different factors, localand systemic, may modulate theformation of osteoclasts, and consequentlythe extent of pathological boneresorption.7 An increase in the numberof mature bone-resorbing osteoclastsfrom macrophages is one of the cellularmechanisms that produce pathologicalbone resorption.7 Humoral factorsthat stimulate the differentiation of osteoclastsfrom mononuclear phagocytesalso are important in influencingthe extent of this bone resorption.7Bacterial infection causes signifi-cant morbidity mediated in part by theup-regulation of inflammatory cytokines,and cytokine induction isthought to stimulate osteolysis in periodontaldisease.3 Bacterial lipopolysaccharidesseem to be directly involvedin inflammatory bone loss, stimulatingthe survival and fusion of preosteoclasts.9 Moreover, potential periodontalpathogens have been found to be ableto stimulate bone resorption locallywhen placed beside a bone surface;these data may support their role inthe pathogenesis of bone loss in periodontitis.1Little is known about the impact of2 Vol. XXIX/No. One/2003loading on the peri-implant bone.10 Excessiveload seems to be able to causebone loss through the induction ofbone microdamage.10-13 Functionalloading provides a site-specific signalfor the regulation of bone mass andmorphology.14 One of the forces producedduring skeletal loading is hydrostaticpressure.15 Increased bone resorptionwas found around loaded implants,and crestal bone resorption wasrelated to overload and damage of thesupporting interfacial bone.16,17The aim of the present work was toevaluate the presence and number ofosteoclasts in peri-implant bone in control(unloaded) and test (loaded) implantsto see if loading per se could bea contributing factor in peri-implantbone resorption.MATERIALS AND METHODSSandblasted and acid-etched implants(Bone System, Milan, Italy) wereplaced in the mandible of 4 male beagledogs of at least 18 months of age.The 2 premolars and the first molarshad been extracted 3 months previously.Each dog received 12 implants inthe mandible (6 on the right side and6 on the left side). The distance betweenthe implants was at least 4 mm.All surgical procedures were performedunder general anesthesia (premedicationwith acepromazine 0.5mg/kg subcutaneously; anesthesiawith Nembutal 15 mg/kg intravenously)and antibiotic prophylaxis. The implantsites were prepared with drillsunder generously chilled saline irrigation.The implants were then insertedwith a tapping instrument. The mucosaltissues were sutured with 3-0 silksutures. In the first 2 postsurgicalweeks the oral cavities were rinsed dailywith chlorhexidine-digluconate0.12%. In addition, the dogs were feda soft diet. The sutures were removedafter 1 week. An oral hygiene regimenwas instituted, consisting of plaque removal3 times a week with a soft toothbrush and 0.2% chlorhexidine gel(SmithKline Beecham, Brentford, UK).No postoperative complications ordeaths occurred. Three months followingimplantation, second-stage surgerywas performed; on 24 implants a prostheticsuperstructure was inserted,whereas on 24 implants only the healingscrews were positioned. Two dogswere killed after 6 months, and 2 dogswere killed after 12 months. A total of24 implants were recovered.Processing of specimensThe specimens were retrieved andstored immediately in 10% bufferedformalin and processed to obtain thinground sections with the Precise 1 AutomatedSystem (Assing, Rome, Italy).18 The specimens were dehydratedin an ascending series of alcohol rinsesand embedded in a glycolmethacrylateresin (Technovit 7200 VLC, Kulzer,Wehrheim, Germany). After polymerizationthe specimens were sectionedlongitudinally along the major axis ofthe implant with a high-precision diamonddisc at about 150 mm andground down to about 30 mm. Threeslides were obtained for each implant.The slides were stained with basicfuchsin and toluidine blue. A doublestaining with von Kossa and basicfuchsin was done to evaluate the degreeof bone mineralization, and oneslide per implant, after polishing, wasimmersed in AgNO3 for 30 minutesand exposed to sunlight; the slideswere then washed under tap water,dried and immersed in basic fuchsinfor 5 minutes, and then washed andmounted.HistomorphometryHistomorphometry of the number ofosteoclasts found in crestal bone in thefirst 3 mm from the implant surfacewas carried out using a light microscope(Laborlux S, Leitz, Wetzlar, Germany)connected to a high-resolutionvideo camera (3CCD, JVC KY-F55B,JVC Professional Products, Milan, Italy)and interfaced to a monitor and PC(Intel Pentium III 1200 MMX, Intel IrelandLtd, Kildare, Ireland). This opticalsystem was associated with a digitizingpad (Matrix Vision GmbH, Oppen-TABLE 1Statistical evaluation of osteoclasts around implant at 3 mm from crestal boneMeanControl implants (unloaded)6 months (Group 1)12 months (Group 2)5.6672.5Loaded implants6 months (Group 3)12 months (Group 4)5.252.5Number of osteoclasts.*Significant at 95% (according to the analysis of variance test).weiler, Germany) and a histometrysoftware package with image capturingcapabilities (Image-Pro Plus 4.5,Media Cybernetics, Immagini ComputerSnc, Milan, Italy).Statistical evaluationThe differences in the percentages ofcrestal bone remodeling in the groupswere evaluated with the analysis ofvariance (ANOVA). The significance ofthe differences observed were evaluatedwith the Bonferroni test for multiplecomparisons (Table 1). The number ofosteoclasts were expressed as a mean6 SD and SE. Statistically significantdifferences were set at P , .05.RESULTSAll implants appeared to be osseointegratedfrom a clinical and radiographicalpoint of view. No implantswere mobile. Histologically, it was possibleto observe around all implants thepresence of compact, mature bone withwell-formed Haversian systems. Thisbone was in close contact with the metalsurface, and no gaps or fibrous tissuewere observed at the interface. Noepithelial downgrowth was observed.In some areas, marrow spaces wereabutting on the implant surface. Osteoclasts,in the process of actively resorbingbone, were present around allimplants.In Group I (unloaded implants after6 months), the mean number of osteoclastsobserved at the level of thecrestal bone in the first 3 mm from theimplant surface was 5.66 6 0.81 (Figures1 and 2).P-value SE SD0.33330.42820.81651.049.0001*0.44590.28871.5451In Group II (unloaded implants after12 months), the mean number of osteoclastswas 2.55 6 1.05 (Figures 3and 4). In Group III (loaded implantsafter 6 months), the mean number ofosteoclasts was 5.25 6 1.55 (Figures 5and 6). In Group IV (loaded implantsafter 12 months), the mean number ofosteoclasts was 2.5 6 1.0 (Figures 7and 8).Statistical evaluationThe statistical evaluation showed thatthere were statistically significant differencesin the numbers of osteoclastsin crestal bone between Group I andGroups II and IV, but no statisticallysignificant differences were observedbetween Group I and Group III (Tables1 and 2).DISCUSSIONOsteoclast-mediated bone resorptionoccurs around dental implants and hasan important role during initial andlate-healing periods, as well as in thelong-term success of an implant.19Bony craters and Howship's lacunaewere signs of bone resorption in theneck area around the dynamicallyloaded implants in comparison withthe statically loaded and control implants.11 Our results show that no differenceswere found in the number ofosteoclasts at the level of the crestalbone, in the first 3 mm from the implant,in both control (unloaded) andtest (loaded) implants. The osteoclastsdecreased only in function of time, andstatistically significant differences werefound only between the 6- and 12-Bartolomeo Assenza et alTABLE 2Significance of the differences withthe Bonferroni test for multiplecomparisons (P , .5)Significance ComparisonYesYeNos1 vs 21 vs 41 vs 33 vs 23 vs 44 vs 2YesYeNosmonth specimens. According to ourdata, loading does not seem to have arelevant importance on the osteoclastactivation in the peri-implant bonewith subsequent bone resorption.The present results agree with thehistological data reported in a study of2 plasma-sprayed, nonsubmerged implants,1 loaded for 3 months and theother left unloaded and retrieved 6months after placement, where wefound osteoclast resorption activity inboth.20 In that study, we concluded thatthe hypothesis of an osteoclast activationcorrelated to an excessive load ofthe implant must be discarded becausebone resorption was observed also inthe unloaded implant.20 This osteoclastactivation could then be related in partto the presence of bacteria found insidethe microgap between the implant andabutment.21-29 It has been reported thatsome bacteria or their products may beinvolved in periodontal bone loss.30This bone loss has been established tobe closely related to osteoclast activation.30 This activation seems to be controlledby parathyroid hormone, IL-1,IL-6, PGE2, TNF-alpha, and also byperiodontopathogens like Actinomycesactinomycetemcomitans, Porphyromonasgingivalis, and Treponema lecithinolyticum.30-32 In particular, TNF-alpha upregulatesIL-1, IL-6, and granulocytemacrophagecolony-stimulating factor.31The release of polyethylene debrisfrom hip implants induces a macrophageactivation in the joint space.33These macrophages release humoralfactors into the joint fluid, and thesefactors may stimulate the differentia-Journal of Oral Implantology 3OSTEOCLAST ACTIVITY AROUND IMPLANTS IN BEAGLE DOGSFIGURES 1-4. FIGURE 1. Unloaded implant (6 months). Resorption lacunae with osteoclasts are present (arrows; Toluidine blue and basicfuchsin, original magnification 350). FIGURE 2. Unloaded implant (6 months). At higher magnification it is possible to observe an osteoclastresorbing bone (arrow; Toluidine blue and basic fuchsin, original magnification 3400). FIGURE 3. Unloaded implant (12 months). Manyareas of remodeling are present in the crestal bone (Toluidine blue and basic fuchsin, original magnification 340). FIGURE 4. Unloadedimplant (12 months). At higher magnification it is possible to observe an osteoclast resorbing crestal bone (arrows; Toluidine blue andbasic fuchsin, original magnification 3400).4 Vol. XXIX/No. One/2003Bartolomeo Assenza et alFIGURES 5-8. FIGURE 5. Loaded implant (6 months). Many Howship's lacunae are present in the crestal bone (arrows; Toluidine blue andbasic fuchsin, original magnification 3100). FIGURE 6. Loaded implant (6 months). At higher magnification, an osteoclast is active in crestalbone resorption (Toluidine blue and basic fuchsin, original magnification 3400). FIGURE 7. Loaded implant (12 months). Vertical bone lossextending in an apical direction. In the apical portion of the pocket a few osteoblasts and osteoclasts (arrows) are present (Toluidine blueand basic fuchsin, original magnification 3100). FIGURE 8. Loaded implant (12 months). At higher magnification it is possible to observean osteoclast (arrows; Toluidine blue and basic fuchsin, original magnification 3400).Journal of Oral Implantology 5OSTEOCLAST ACTIVITY AROUND IMPLANTS IN BEAGLE DOGStion of the bone marrow cells into osteoclasts,which begin to resorb boneat the prosthesis-bone interface.34-39 Accordingto Hermann et al,26 the periimplantbone loss is determined by thecreation of a microgap between the implantand abutment. The bone will resorbwith the creation of a distancefrom the bacteria present in the microgap.27 In a retrospective study, Callanet al27 found that the peri-implant boneloss was in part related to the positionof the implant-abutment microgap.These authors reported that a bone lossequal or higher than 3 mm was observedin all implants where this microgaphad been positioned in a subgingivalposition. The penetration ofbacteria or their products through themicrogap may constitute a risk of lossof supporting bone.21 Future studiesneed to be done to try to better understandthe mechanism of peri-implantbone resorption.REFERENCES1. Zubery Y, Dunstan CR, StormBM, et al. Bone resorption caused bythree periodontal pathogens in vivo inmice is mediated in part by prostaglandin.Infect Immun. 1998;66:4158-4162.2. Kurata K, Uemura T, Nemoto A,et al. Mechanical strain effect on boneresorbing activity and messenger RNAexpressions of marker enzymes in isolatedosteoclast culture. J Bone MinerRes. 2001;16:722-730.3. Smit TH, Burger EH. Is BMUcoupling a strain-regulated phenomenon?A finite element analysis. J BoneMiner Res. 2000;15:301-307.4. Reddy SV, Roodman GD. Controlof osteoclast differentiation. CritRev Eukariot Gene Expr. 1998;8:1-17.5. Neale SD, Hatnes DR, HowleDW, Murray DW, Athanasou NA. Theeffect of particle phagocytosis and metallicwear particles on osteoclast formationand bone resorption in vitro. JArthoplasty. 2000;15:654-662.6. Lassus J, Salo J, Jiranek WA, etal. Clin Orthop. 1998;352:7-15.7. Athanasou NA, Sabokbar A.6 Vol. XXIX/No. One/2003Human osteoclast ontogeny and pathologicalbone resorption. Histol Histopathol.1999;14:635-647.8. Duong LT, Lakkakorpi P, NakamuraI, Rodan GA. Integrins and signalingin osteoclast function. MatrixBiol. 2000;19:97-105.9. Suda K, Woo JT, Takami M, SextonPM, Nagai K. Lipopolysaccharidesupports survival and fusion of preosteoclastsindependent of TNF-alpha,Il-1, and RANKL. J Cell Physiol. 2002;190:101-108.10. Gotfredsen K, Berglundh T,Lindhe J. Bone reactions adjacent to titaniumimplants subjected to staticload of different duration. A study inthe dog (III). Clin Oral Implants Res.2001;12:196-201.11. Duyck J, Ronold HJ, van OosterwyckH, Naert I, Vander Sloten J, EllingsenJE. The influence of static anddynamic loading on marginal bone reactionsaround osseointegrated implants:an animal experimental study.Clin Oral Implants Res. 2001;12:207-218.12. Barbier L, Schepers E. Adaptivebone remodeling around oral implantsunder axial and nonaxial loading conditionsin the dog mandible. Int J OralMaxillofac Implants. 1997;12:215-223.13. Wiskott HWA, Belser UC. Lackof integration of smooth titanium surfaces:a working hypothesis based onstrains generated in the surroundingbone. Clin Oral Implants Res. 1999;10:429-444.14. Rubin J, Fan X, Biskobing DM,Taylor WR, Rubin CT. Osteoclastogenesisis repressed by mechanical strainin an in vitro model. J Orthop Res. 1999;17:639-645.15. Rubin J, Biskobing D, Fan X,Rubin C, McLeod K, Taylor WR. Pressureregulates osteoclast formation andMCSF expression in marrow culture. JCell Physiol. 1997;170:81-87.16. Misch CE, Bidez MW, SharawyM. A bioengineered implant for a predeterminedbone cellular response toloading forces. A literature review. JPeriodontol. 2001;72:1276-1286.17. Hoshaw SJ, Brunski JB, CochranGVB. Mechanical loading of Branemarkimplants affects interfacialmodeling and remodeling. Int J OralMaxillofac Implants. 1994;9:345-360.18. Piattelli A, Scarano A, QuarantaM. High-precision, cost-effectivesystem for producing thin sections oforal tissues containing dental implants.Biomaterials. 1997;18:577-579.19. Minkin C, Marinho VC. Role ofthe osteoclast at the bone-implant interface.Adv Dent Res. 1999;13:49-56.20. Piattelli A, Emanuelli M, ScaranoA, Trisi P. A histologic study of nonsubmergedtitanium plasma-sprayedscrew implants retrieved from a patient:a case report. Int J Periodont Rest Dent.1996;16:139-147.21. Persson LG, Lekholm U, LeonhardtA, Dahlen G, Lindhe J. Bacterialcolonization on internal surfaces ofBranemark system implant components.Clin Oral Implants Res. 1996;7:90-95.22. Quirynen M, Bollen CM, EyssenH, van Steenberghe D. Microbialpenetration along the implant componentsof the Branemark system. An invitro study. Clin Oral Implants Res.1994;5:239-244.23. Jansen VK, Conrads G, RichterEJ. Microbial leakage and marginal fitof the implant-abutment interface. Int JOral Maxillofac Implants. 1997;12:527-540.24. Piattelli A, Scarano A, PaolantonioM, et al. Fluids and microbialpenetration in the internal part of cement-retained versus screw-retainedimplant-abutment connections. J Periodontol.2001;72:1146-1150.25. Quirynen M, van SteenbergheD. Bacterial colonization of the internalpart of two-stage implants. An in vivostudy. Clin Oral Implants Res. 1993;4:158-161.26. Hermann JS, Cochran DL,Nummikoski PV, Buser D. Crestal bonechanges around titanium implants: aradiographic evaluation of unloadednonsubmerged and submerged implantsin the canine mandible. J Periodontol.1997;68:1117-1130.27. Callan D, O'Mahony A, CobbCM. Loss of crestal bone around dentalimplants: a retrospective study. ImplantDent. 1998;7:258-266.28. Tarnow DP, Cho SC, WallaceSS. The effect of inter-implant distanceon the height of inter-implant bonecrest. J Periodontol. 2000;71:546-549.29. Cochran DL, Hermann JS,Schenk RK, Higginbottom FL, Buser D.Biologic width around titanium implants.A histometric analysis of theimplanto-gingival junction around unloadedand loaded nonsubmerged implantsin the canine mandible. J Periodontol.1997;68:186-198.30. Choi BK, Ohk SH, Lee HJ,Kang JH, Jeong GJ, Yoo YJ. Effects ofwhole cell sonicates of treponema lecithinolyticumon osteoclast differentiation.J Periodontol. 2001;72:1172-1177.31. Xu JW, Konttinen YT, Lassus J,et al. Tumor necrosis factor-alpha(TNF-alpha) in loosening of total hipreplacement (THR). Clin Exp Rheumatol.1996;14:643-648.32. Giudiceandrea F, Iacona A,Cervelli G, et al. Mechanisms of boneresorption: analysis of proinflammatorycytokines in peritoneal macrophagesfrom titanium implant. An experimentaldesign. J Craniofac Surg. 1998;9:254-259.33. Hirashima Y, Ishiguro N, KindoS, Iwata H. Osteoclast inductionfrom bone marrow cells is due to proinflammatorymediators from macrophagesexposed to polyethylene particles:a possible mechanism of osteolysisin failed THR. J Biomed Mater Res.2001;56:177-183.34. Merkel KD, Erdmann JM,McHugh KP, Abu-Amer Y, Ross FP,Teitelbaum SL. Tumor necrosis factoralphamediates orthopedic implant osteolysis.Am J Pathol. 1999;154:203-210.35. Mevoy A, Jeyam M, Ferrier G,Evans CE, Andrew JG. Synergistic effectof particles and cyclic pressure oncytokine production in human mono-Bartolomeo Assenza et alcyte/macrophages: proposed role inperiprosthetic osteolysis. Bone. 2002;30:171-177.36. Haynes DR, Crotti TN, PotterAE, et al. The osteoclastogenic proteinsRANKL and RANK are associatedwith periprosthetic osteolysis. J BoneJoint Surg Br. 2001;83:902-911.37. Bi Y, van de Motter RR, RagabAA, Goldberg VM, Anderson JM,Grenfield EM. Titanium particles stimulatebone resorption by inducing differentiationof murine osteoclasts. JBone Joint Surg Am. 2001;83-A:501-508.38. Ferrier GM, McEvoy A, EvansCE, Andrew JG. The effect of cyclicpressure on human monocyte-derivedmacrophages in vitro. J Bone Joint SurgBr. 2000;82:755-759.39. Neale SD, Fujikawa Y, SabokbarA, et al. Human bone-derived cells supportformation of human osteoclastsfrom arthroplasty-derived cells in vitro.J Bone Joint Surg Br. 2000;82:892-900.Journal of Oral Implantology 7