Biomaterials and Biomechanics of Oral and Maxillofacial … · 2006-04-17 · bone-implant contact...

COPYRIGHT © 2000 BY QUINTESSENCE PUBLISHING CO, INC. PRINTING

OF THIS DOCUMENT IS RESTRICTED TO PERSONAL USE ONLY. NO PART OF

THIS ARTICLE MAY BE REPRODUCED OR TRANSMITTED IN ANY FORM WITH-OUT WRITTEN PERMISSION FROM THE PUBLISHER.

The International Journal of Oral & Maxillofacial Implants 15

Biomaterials and Biomechanics of Oral and Maxillofacial Implants: Current Status and

Future DevelopmentsJohn B. Brunski, MS, PhD1/David A. Puleo, PhD2/Antonio Nanci, MSc, PhD3

Major advances have occurred over the last 3decades in the clinical use of oral and maxillo-

facial implants. Statistics on the use of dentalimplants bear this out; about 100,000 to 300,000dental implants are placed per year,1 which approxi-mates the numbers of artificial hip and knee jointsplaced per year.2 Implants are currently used toreplace missing teeth, rebuild the craniofacial skele-ton, provide anchorage during orthodontic treat-ments, and even to help form new bone in theprocess of distraction osteogenesis.

Despite the impressive clinical accomplishmentswith oral and maxillofacial implants—and theundisputed fact that implants have improved thelives of millions of patients—it is nevertheless dis-quieting that key information is still missing aboutfundamental principles underlying their design andclinical use. With some important exceptions, thedesign and use of oral and maxillofacial implants hasoften been driven by an aggressive, “copycat” mar-keting environment, rather than by basic advancesin biomaterials, biomechanics, or bone biology.

A wide variety of implants now exists for use inmany clinical indications, with over 50 companieslisted by the United States Food and Drug Adminis-tration (FDA) as being involved in the manufacture,marketing, and distribution of dental implants.While this situation is not necessarily a problem, inmany instances new companies have entered the

dental implant market by simply copying or makingminor, incremental changes to the sizes, shapes,materials, and surfaces of competitors’ products,while exaggerating the new product’s effectiveness.In addition, busy clinicians, not always equipped todiscern the difference between marketing hype andscientific advance, yet wanting to help their patientssooner rather than later, have often been too eagerto use new implants in new clinical situations beforethese new indications have been fully researchedfrom the clinical or basic science viewpoint. For bet-ter or for worse, the current state of the oral implantfield is such that a myriad of different types ofimplants are now being used in a very wide varietyof clinical indications, under largely undocumentedloading conditions in different quantities and quali-ties of bone that has healed to varying extents. It is afertile but complicated state of affairs.

Given this situation and the many variables thatcan affect the performance of oral implants, it issometimes difficult to separate fact from fiction andmake reliable predictions for the future. However, ahelpful starting point is to appreciate that the use oforal implants—and the key role of biomaterials andbiomechanics—is an excellent example of a multi-faceted design problem.

TREATMENT PLANNING WITH ORALIMPLANTS AS A DESIGN PROBLEM: AN OFTEN-IGNORED PERSPECTIVE

A guiding perspective is that the clinical use ofimplants is a design problem in the true sense of theword. Two key characteristics distinguish designproblems.3 First, design problems are open-ended,which means that they typically have more than onepossible solution: “The quality of uniqueness, soimportant in many mathematics and analysis prob-lems, simply does not apply.”3 Second, design prob-lems are ill-structured, which means that “their

1Professor, Department of Biomedical Engineering, RensselaerPolytechnic Institute, Troy, New York.

2Associate Professor, Center for Biomedical Engineering, Univer-sity of Kentucky, Lexington, Kentucky.

3Professor and Director, Laboratory for the Study of Calcified Tis-sues and Biomaterials, Faculty of Dentistry, Université de Mon-tréal, Montréal, Quebec, Canada.

Reprint requests: Dr J. B. Brunski, Department of BiomedicalEngineering, Room 7040, Rensselaer Polytechnic Institute, Troy,NY 12180-3590. Fax: 518-276-3035. E-mail: [email protected]

REVIEW ARTICLE




16 Volume 15, Number 1, 2000

BRUNSKI ET AL

solutions cannot normally be found by routinelyapplying a mathematical formula in a structuredway.”3 Moreover, design is a process typically char-acterized by a series of steps leading toward a solu-tion of the problem4:

• Identification of a need• Definition of the problem• Setting design objectives• Searching for background information and data• Developing a design rationale• Devising alternative solutions• Evaluating alternative solutions• Decision-making and communication of solutions

To help clarify the above characteristics ofdesign as applied to oral implants,5 consider the fol-lowing questions, which are often asked by clini-cians: What is the best implant system? What is thebest implant biomaterial? What is the best surfacefor an implant? What implant surface gives the bestbone-implant contact? Which grade of commer-cially pure titanium is the best? What is the bestimplant shape?

While each of these questions seems meaningful,they all miss the point in the context of design.First, without defining the word “best,” it is impos-sible to answer each question properly. This isbecause each question essentially begs anotherquestion, namely, “What do we mean by ‘best’?”Second, each question seems to have the implicitassumption that the answer to that question alonewill define the full merit of an implant, and byextension, the full merit of the implant in all clinicalsituations in which it is used.

An example illustrates the short-sightedness ofthe above thinking. To the untrained eye—and asoften implied in advertisements—it might seem thata dental implant with the largest percentage ofbone-implant contact must be the “best” implant.But some thought suggests a reason why this is notnecessarily true. One could have a short, cylindricimplant with a relatively small diameter that devel-ops 100% bone-implant contact. While 100% con-tact may be desirable, it alone does not guaranteethat the implant will work at the clinical level. Forexample, it could turn out that, because of thatimplant’s small diameter and length, it would beinadequate in total surface area to support the invivo loads, even though it has 100% bone contact.On the other hand, one could use the same lengthimplant as above but select a larger diameter plus ascrew-shaped or macro-rough geometry for 2 rea-sons: (1) it might be known ahead of time that, forwhatever reason, this implant will achieve 70%

bone-implant contact in the same bony environ-ment; and (2) this 70% contact is known to be suffi-cient for the expected in vivo loads (because of thelarger net surface area of the implant via the largerdiameter and screw shape).

An analogous example could be developedinvolving the intrinsic strength properties of tita-nium, such as yield or ultimate tensile strength.Here it might be known that there is a differencebetween the intrinsic strength properties of differ-ent commercial grades of titanium and its alloys,such as titanium-aluminum-vanadium (Ti-6Al-4V).However, a designer would appreciate that this dif-ference in strength properties could be importantbut is not necessarily paramount in a particularapplication; instead, what is critical is the overallstrength of an implant structure made of one sort oftitanium versus another. In general, a good designermight realize that it might be possible, with properengineering, to design a successful implant from arange of implant materials, given due considerationto the expected loadings and the intrinsic propertiesof the material in the design of the restoration.

The above examples make the point that theproblem with the original questions taken one at atime is that they focus on the implant per se ratherthan the entire clinical problem to be solved—whichis in general a multifaceted design problem. Anotherlimitation with these questions taken in isolation isthat they do not seem to admit the possibility thatmore than one implant might “work” equally well intreating a patient, just as different automobiles servedifferent people equally well, provided the totaldesign satisfies the design objectives.

Therefore, the relevance of each of the questionslisted above depends on the design objectives forthe clinical situation. Oral and maxillofacialimplants are chosen, placed, and restored in apatient to achieve certain design objectives. Thedesign objective is not necessarily to have the mostbone around a dental implant, although this mayhelp. Instead, the goal is to treat the patient. Whileat one level the objectives for this might be stated ina generalized manner—for example, “to improvethe patient’s quality of life”—eventually the designobjectives should be defined more specifically. Forexample, eventually one has to decide on the type,location, and number of implants; the type of pros-thesis; the nature of the expected loading, ie,whether the restoration will undergo immediate ordelayed loading; costs of the treatment; and theamount of bone that may be expected to formaround the implant. A key point involving mostdental implants is that there are usually several waysto accomplish a treatment in the same patient.





BRUNSKI ET AL

Indeed, this is why such problems are in fact designproblems. The existence of alternative solutionsmakes treatment planning quite similar to design inconventional engineering, as in the design of ahouse, a car, a laptop computer, or a computer oper-ating system, each of which can be designed in avariety of ways, depending on the objectives.Depending on the available technology, one has toidentify the different possible solutions and reviewthem in light of the design objectives, which couldinclude such factors as cost, simplicity, and speed.However, the bottom line is that the determinationof which design solution is “best” depends on howone prioritizes different objectives.

A final example may emphasize the aforemen-tioned message. Suppose that the most importantclinical design objective is to provide a patient withimplants and a final prosthesis on the same day assurgery. For example, this goal is implicit in the newBrånemark Novum System (Nobel Biocare, Göte-borg, Sweden).6 But if this is the goal, then one hasto face the problem of designing all of the aspects ofthe situation—including the number, type, and bio-material of the implant; the surgical procedures; andthe prosthesis—with this goal in mind. Moreover, itwould be useful to abandon, or at least re-examine,any preconceived notions that an implant designedfor delayed loading is automatically suitable for usein immediate loading. And indeed, in the NovumSystem, differences exist between the new implantsand those used in the more conventional 2-stageBrånemark System. The point is that implant size,shape, material, abutment connection, etc, are but afew of the many factors that must be identified andconsidered in light of the design objectives. Thistheme resurfaces repeatedly when considering therole of biomaterials and biomechanics in oralimplant design.

BIOMATERIALS

A goal of biomaterials research has been, and con-tinues to be, the development of implant materialsthat induce predictable, controlled, guided, andrapid healing of the interfacial tissues, both hardand soft. The premise is that such biomaterials willadd to the arsenal of available tools enabling thedesign of improved implant systems. In addition toan ability to positively affect normal wound healingphenomena, it would be ideal if endosseousimplants could also fulfill a design objective offorming a characteristic interfacial layer and bonematrix with adequate biomechanical properties overthe long term.

To achieve biomechanical and biologic objec-tives, however, a better understanding of events atthe interface is needed, and of the effects that bio-materials have on bone and bone cells. Such knowl-edge is essential for developing strategies to opti-mally control the phenomena that have beencollectively used as a description of the often-usedbut still somewhat undefined term osseointegration.These outcomes would allow not only faster recu-peration for the patient, but also stable fixationbetween bone and implant that would perhaps per-mit clinically reliable immediate or early loading ofthe implant. This latter type of treatment has greatpotential impact in terms of decreased patient mor-bidity, improved patient psychology, and decreasedhealth care costs. To achieve these goals, however, abetter understanding is needed of fundamentalevents surrounding tissue healing.

Events leading to integration of an implant intobone, and hence to the clinical performance of therestoration under loading, take place largely at thetissue-implant interface. Development of this inter-face is complex and involves numerous factors.These include not only implant-related factors,such as material, shape, topography, and surfacechemistry, but also mechanical loading, surgicaltechnique, and patient variables such as bone quan-tity and quality. This section of the paper reviewscurrent knowledge of the bone-biomaterial inter-face and new methods being investigated for con-trolling this interface. Because of their predominantuse as load-bearing implants, emphasis is placed onmetallic biomaterials, although most of the notedwork also applies to ceramic implant materials.

Events at the Bone-Implant InterfaceThe performance of biomaterials can be classified interms of: (1) the response of the host to the implant,and (2) the behavior of the material in the host.7Although this section emphasizes the host response,a brief review of material response is also given.

Material Response. The event that occurs almostimmediately upon implantation of metals, as withother biomaterials, is adsorption of proteins.8,9

These proteins come first from blood and tissue flu-ids at the wound site and later from cellular activityin the interfacial region. Once on the surface, pro-teins can desorb (undenatured or denatured, intactor fragmented) or remain to mediate tissue-implantinteractions. In fact, the nature of this “condition-ing film” deposited on biomaterials, along with thebiomechanical conditions surrounding the implan-tation, can be a major determinant of the hostresponse, as discussed further in the section of thispaper on biomechanics.





BRUNSKI ET AL

In addition to protein adsorption on the implant’ssurface, significant changes also occur in the mater-ial’s surface. There is ample literature that describesoxidation of metallic implants both in vivo and invitro.10,11 Although metallic implant biomaterialswere originally selected because of their stable oxidefilms, it is appreciated that the oxide surfaces stillundergo electrochemical changes in the physiologicenvironment. For example, depending on themethod of sterilization, commercially pure titanium(cp Ti) implants have an oxide thickness of 2 to 6nm before implantation.12 However, films onimplants retrieved from human tissues are 2 to 3times thicker.10,12,13 Furthermore, surface analyticstudies show that the chemical composition of theoxide film has also changed by incorporating cal-cium, phosphorus, and sulfur.12,13 Continued oxidegrowth reflects ongoing electrochemical events atthe tissue-implant interface. Another consequenceof these events is the release of metallic species intotissues.14 These corrosion by-products accumulatelocally but may also be spread systemically. Signifi-cantly elevated metal contents have been measuredboth in periprosthetic tissues15,16 and in the serumand urine of patients with orthopedic implants.17–19

For example, metal levels of up to 21 ppm titanium,10.5 ppm aluminum, and 1 ppm vanadium aroundTi-6Al-4V and up to 2 ppm cobalt, 12.5 ppmchromium, and 1.5 ppm molybdenum aroundcobalt-chromium-molybdenum (CoCrMo) havebeen measured in the fibrous membrane encapsulat-ing hip implants.15,16 The tissue may include someparticulate metal, but the ratios do not reflect thebulk composition of the alloys. Trace metals areessential for health, but they can also be toxic20 orcause hypersensitivity reactions.21 In vitro studieshave revealed that metal ions, even at sublethaldoses, interfere with differentiation of osteoblastsand osteoclasts.22–24 It remains to be determinedwhether these effects on bone cells also occur invivo. It is also unclear whether such effects are seenwith oral and maxillofacial implants, which in gen-eral have much less surface area in tissues thanorthopedic implants.

Host Response. The host response to implantsplaced in bone involves a series of cell and matrixevents, ideally culminating in tissue healing that isas normal as possible and that ultimately leads tointimate apposition of bone to the biomaterial, ie,an operative definition of osseointegration. For thisintimate contact to occur, gaps that initially existbetween bone and implant at surgery must be filledinitially by a blood clot, and bone damaged duringpreparation of the implant site must be repaired.During this time, unfavorable conditions, eg,

micromotion (a biomechanical factor discussedlater), will disrupt the newly forming tissue, leadingto formation of a fibrous capsule.25–27

Morphologic studies have revealed the hetero-geneity of the typical bone-implant interface. Onefeature often reported is the presence of an afibrillarinterfacial zone, comparable to cement lines andlaminae limitans.28–31 Although its thickness andappearance vary, this zone forms regardless of thetype of biomaterial implanted, including cp Ti,stainless steel, and hydroxyapatite.32 Early reportsindicated that the interface was rich in glycosamino-glycans.28 However, more recent high-resolutionimmunocytochemical studies demonstrated that theelectron-dense interfacial layer contains noncol-lagenous bone matrix proteins, such as osteopontin(OPN) and bone sialoprotein (BSP).32,33 Theabsence or relative paucity of serum proteins, suchas albumin, indicates a selective accumulation/depo-sition of molecules at the interface.32 Because theycontain arginine-glycine-aspartic acid (Arg-Gly-Asp) and polyacidic sequences, OPN and BSP arebelieved to play roles in cell adhesion and bindingof mineral.34–36 This interfacial zone might be thesource of whatever mechanism of “bonding” existsbetween natural hard tissue and biomaterial, as dis-cussed further in the biomechanics section of thisarticle. However, the inherent weakness of cementlines argues against this level of bonding beingstronger than a few MPa (while, for example, theultimate tensile strength of fully mineralized com-pact bone is on the order of 100 to 150 MPa).37

Osteoblasts, osteoid, and mineralized matrix havebeen observed adjacent to the lamina limitans,28–31

suggesting that bone can be deposited directly onthe surface of the implant, extending outward fromthe biomaterial. Thus, bone formation in the peri-prosthetic region occurs in 2 directions: not onlydoes the healing bone approach the biomaterial, butbone also extends from the implant toward the heal-ing bone.

Understandably, because of the complexities ofthe in vivo environment, the bone-implant interfacehas not yet been fully characterized. The hetero-geneity and patchy immunolabeling observed inmorphologic studies suggest that, even though sev-eral biomolecules have been identified at the inter-face, they are likely not the only ones present.These other biomolecules may have essential rolesin directing bone response to the implant, and fur-ther work is needed to identify them and determinetheir functions at the interface. It must be notedthat cement lines are still present on OPN-knock-out mice, suggesting that there are likely other yet-unidentified constituents in cement lines.38





BRUNSKI ET AL

In Vitro Studies. Bone cell culture models areemployed increasingly often to study bone-biomate-rial interactions. Most of the cultures have utilizedosteoblastic cells (reviewed by Cooper et al39), withonly a few using osteoclastic cells.24,40 Primary andpassaged cells from several species and anatomiclocations have been used, as well as several osteosar-coma, clonal, and immortalized cell lines. Substrate-dependent differences have been reported, but thevariety of models used makes it difficult to drawconsensus conclusions.

Whether the different bone cell models would beexpected to give comparable results is still a subjectof debate. There are few side-by-side comparisonsof different bone cell culture models on the samebiomaterial substrates. It is known that bone fromdifferent sites, developmental ages, and types showsvariabilities.41,42 Taking into account the differentmicroenvironments from which the cells come,there is no a priori reason to expect that all “osteo-blastic” cells, eg, those derived from bone marrowor calvaria, would behave the same, at least duringthe initial phases of culture. In fact, unpublishedobservations from Puleo and Nanci indicate differ-ences in adhesiveness of bone marrow– and cal-varia-derived osteogenic cells on certain substrates.Nonetheless, in vitro models have the potential tohelp elucidate events at the bone-implant interface(as reviewed by Davies43) by providing morpho-logic, biochemical, and molecular informationregarding osteoblastic development and synthesis ofmatrix at the interface with various biomaterials. Animportant consideration, however, is that the infor-mation obtained can indeed reflect in vivo events.For example, in vitro and in vivo models haveshown formation of a cement line–like layer(described earlier) and appropriate organization ofmineralized matrix during culture on various sub-strates.36,44,45

Controlling the Bone-Implant Interface by Biomaterials Selection and ModificationDifferent approaches are being used in an effort toobtain desired outcomes at the bone-implant inter-face. Many would accept the premise that an idealimplant biomaterial should present a surface thatwill not disrupt, and that may even enhance, thegeneral processes of bone healing, regardless ofimplantation site, bone quantity, bone quality, etc.As Kasemo and Lausmaa,46 among others, havedescribed, biologic tissues interact mainly with theoutermost atomic layers of an implant. Althoughsecondary and other by-product reactions willoccur, the “primary interaction zone” is generallyonly about 0.1 to 1 nm thick. Consequently, much

effort has been devoted to methods of modifyingsurfaces of existing biomaterials to achieve desiredbiologic responses. As described by Ito et al47 withrespect to polymers, the approaches can be classi-fied as physicochemical, morphologic, or bio-chemical.

Physicochemical Methods. Surface energy, sur-face charge, and surface composition are among thephysicochemical characteristics that have beenaltered with the aim of improving the bone-implantinterface. Glow discharge has been used to increasesurface free energy so as to improve tissue adhe-sion.48,49 Considering the role of electrostatic inter-actions in many biologic events, charged surfaceshave been proposed as being conducive to tissueintegration.50,51 Calcium phosphate coatings havebeen extensively investigated because of their chemi-cal similarity to bone mineral.52,53 Each approach,however, has drawbacks. Increased surface energydoes not selectively increase the adhesion of particu-lar cells or tissues, and it has not been shown toincrease bone-implant interfacial strength.54 Contra-dictory results with charged materials in bone havebeen reported; indeed, both positively50 and nega-tively51 charged surfaces were observed to allowbone formation. Although short-term clinical resultshave been encouraging,53,55 dissolution of coatingsas well as cracking and separation from metallic sub-strates remain a concern with hydroxyapatite coat-ings.56,57

Morphologic Methods. Alterations in biomaterialsurface morphology and roughness have been usedto influence cell and tissue responses to implants.Porous coatings were originally developed with therationale that, because of mechanical interlocking,bone ingrowth would increase fixation and stabilityof the implant. Many animal studies support therather obvious idea that bone ingrowth into macro-rough surfaces enhances the interfacial tensile andshear strengths (as determined from certain tests)compared to smooth surfaces (as discussed furtherin the section on biomechanics). However, datafrom retrieval studies of porous orthopedic implantsindicate that only a relatively small portion of theavailable pore volume is filled with bone.58–60 Inaddition to providing mechanical interlocking, sur-faces with specially contoured grooves can induce“contact guidance,” whereby the direction of cellmovement is affected by the morphology of the sub-strate.61 This phenomenon has applications in pre-venting epithelial downgrowth on dental implantsand directing bone formation along particularregions of an implant. Mineral deposits in bone cellcultures can also be altered by surfaces with pits andgrooves.62

Concerning surface roughness and its effects,there is a large but inconclusive literature on thebiologic and clinical effects. Using in vitro cell cul-ture systems, not all authors have come to the sameconclusions about a role for surface roughness.Cochran et al63 reported that human fibroblast andepithelial cell attachment and proliferation in vitrowere affected by surface characteristics of titanium.Martin et al64 used osteoblast-like cells (MG63) andnoted that surface roughness of titanium alteredosteoblast proliferation, differentiation, and matrixproduction. Using the same cell line, Boyan et al65

found that titanium surface roughness affected theresponsiveness of cells to hormones such as 1�,25-dihydroxy-vitamin D3. On the other hand, Castel-lani et al,66 who worked with titanium surfaces ofdiffering roughness exposed to rat bone marrowcells, “could not clearly confirm the effect of surfaceroughness on the proliferation, differentiation, andcalcification of rat bone marrow cells.”66p369 Like-wise, Sauberlich et al67 studied human gingivalfibroblasts on surfaces of titanium subjected to dif-ferent surface treatments and concluded that “amarked correlation between the cellular compatibil-ity of the modified titanium and the surface modifi-cation made did not become apparent.”67p379

In vivo studies also create an inconclusive pictureof the role of surface texture. For example, Buser etal68 placed titanium implants with 6 different sur-faces into the metaphyses of the tibiae and femoraof miniature pigs for 3 and 6 weeks. Surface treat-ments of the Ti included electropolishing (E), sand-blasting with medium grit (0.12 to 0.25 µm) andacid pickling (hydrofluoric acid/nitric acid) (SMP),sandblasting with large grit (0.25 to 0.50 µm) (SL),sandblasting with large grit and acid attack withhydrochloric acid/sulfuric acid (SLA), titaniumflame-spraying (TPS), and hydroxyapatite (HA)flame-spraying. These surfaces had not only differ-ent roughnesses, but also different surface composi-tions; the SL surface had some of the grit-blastedparticles embedded in the “highly distorted” metalsurface. A key finding of the study was that all theimplants revealed “direct bone-implant contact,”but with differing percentages of bone contact inthe cancellous bone. The highest bone-implantcontact was in the SLA and HA cases, with contactpercentages of 50 to 60% (SLA) and 60 to 70%(HA). The authors reported that “the extent of thebone-implant interface is positively correlated withan increasing roughness of the implant surface.”

Chehroudi et al62 worked with titanium-coatedepoxy replicas of 19 different micromachinedgrooved or pitted surfaces in the parietal bones ofrats and reported that “surface topography influ-

enced the frequency and amount of bone depositedadjacent to the implants.” Wong et al69 examinedbone reaction to press-fit cylindric implants madeof 3 materials and given different surface treatmentsin trabecular bone sites in knees of mature minia-ture pigs. After 12 weeks of implantation time, HA-coated implants had the highest push-out loads andthe largest surface coverage by bone, ie, 79.9% ver-sus a mean of 38.5% for the all-metal groups (cp Ti,Ti-6Al-4V, and titaninum-aluminum-niobium [Ti-6Al-7Nb]). Notably (see the section on biomechan-ics), there was an “excellent correlation . . . betweenthe average roughness of the implant surface andpushout failure load.” Ericsson et al70 made a histo-morphometric comparison in dog maxillae ofscrew-shaped dental implants with surfaces charac-terized as “machine-prepared” versus roughened byblasting with titanium oxide. At 2 months, bothtypes of implants had a mean bone-implant contactpercentage of about 40%. However, at 4 months,the roughened implants had a mean contact per-centage of 65.1% (± 17.3%), which was greater (P <.05) than the contact for the standard machinedimplant surfaces, 42.9% (± 31.2%). Wennerberg etal71 studied screw-shaped implants with 3 differentsurfaces in rabbit bone: blasted with 25-µm titaniumoxide (TiO2), blasted with 75-µm aluminum oxide,and a “turned” surface (as-machined). After 12weeks, there was a higher percentage of bone-metalcontact for implants with the 25-µm TiO2-blasting.However, there was a greater surface area of bone inthreads for the turned implants compared with theTiO2-blasted implants. Based on this short-termstudy, there was better fixation using implants withgreater roughness.

In contrast with the foregoing studies, otherwork has not revealed a major effect of roughness invivo. Jansen et al72 tested cylindric plugs of 3 differ-ent titanium alloys (cp Ti with 0.2 wt% palladium,Ti-6Al-4V, and titanium-aluminum-iron [Ti-5Al-2.5Fe]) and HA-coated Ti-6Al-4V alloy in rabbittibiae for 6 and 16 weeks. The materials did nothave identical roughness. After measuring theamount of bone apposition and other aspects of thebone reaction to the implants, the authors notedthat “the results demonstrated no marked differ-ences in bony reaction to the different implantmaterials” and that “the HA coatings showed a lossof thickness.” Caulier et al73 tested the response of“low-density” bone to threaded, uncoated, commer-cially pure titanium implants as well as the samemetal plasma-sprayed with 3 different types of cal-cium/phosphorus–containing materials (fluoroap-atite, HA, and heat-treated HA). After implantationtimes of 3 to 6 months in the maxillae of goats, no


BRUNSKI ET AL








BRUNSKI ET AL

differences were found in the histomorphometricmeasurements that were made. The authors foundno significant differences “in the bone reactionamong the various implant materials,” although allcoatings showed some decrease in thickness.

Moving to clinical studies with implants of differ-ent surface textures or roughnesses, few long-termcomparisons are available of various implant sur-faces in prospective trials. Helsingen and Lyberg74

compared the surface composition and microstruc-ture of 4 titanium implants that were identical inexternal shape (ie, screw-shaped “clones”) but madeby 4 different manufacturers: Nobel Biocare; Core-Vent (Los Angeles, CA); 3i (West Palm Beach, FL);and Osseodent (Palo Alto, CA). The study found no“substantial qualitative differences as regards chemi-cal composition” and noted that only the Core-Ventimplant’s surface was different (“more irregular”). Intheir clinical pilot study involving 22 patients whoreceived Brånemark implants on one side of the jawand 1 of the 3 other implant types contralaterally,there were no significant differences in the successrate and marginal bone level 1 to 2 years afterimplant loading. In a recent study in humans,Iamoni et al75 placed special screw-shaped implantsthat were half-coated (longitudinally) with plasma-sprayed HA. Each of 4 subjects received 2 implantsin the retromolar area. Bone cores were excised at 1,3, 6, and 12 months and analyzed histologically. Thestudy reported a “tendency toward a higher percent-age of bone contact at each healing period” for theHA-coated implants, although the number of speci-mens did not allow definitive conclusions.

Overall, these in vitro, animal, and clinical studiesas yet do not yield compelling conclusions about therole of surface composition and texture with respectto bone response at the interface. Research contin-ues in this area, for example, to formulate hypothe-ses and experiments about the role of specific geo-metric and size-related aspects of the surface andtheir role in the strength of the bone-implant inter-face.76 Davies,77 for example, hypothesizes thatrough surfaces can capture the fibrin clot more read-ily than smooth surfaces and thereby positivelyaffect the initial stages of integration. It is likely thatfuture work will continue to formulate and test cel-lular-level hypotheses involving the role of surfacetexture and composition on the basic biology of celland tissue interactions.

Biochemical Methods. Biochemical methods ofsurface modification offer an alternative or adjunctto physicochemical and morphologic methods. Bio-chemical surface modification endeavors to utilizecurrent understanding of the biology and biochem-istry of cellular function and differentiation. Much

has been learned about the mechanisms by whichcells adhere to substrate,78 and major advances havebeen made in understanding the role of biomole-cules in regulating differentiation and remodelingof cells and tissues, respectively.79 The goal of bio-chemical surface modification is to immobilize pro-teins, enzymes, or peptides on biomaterials for thepurpose of inducing specific cell and tissueresponses, or, in other words, to control the tissue-implant interface with molecules delivered directlyto the interface.

Although there are several reports of biochemicalsurface modification for modulating tissue responsesto cardiovascular materials,80–83 this approach hasreceived comparatively little, but increasing, consid-eration for orthopedic and dental applications.84–88

This methodology has great potential for control-ling initial bone-implant interactions. In contrast tocalcium phosphate coatings, biochemical surfacemodification utilizes critical organic components ofbone to affect tissue response.

One approach for controlling cell-biomaterialinteractions utilizes cell adhesion molecules. Sinceidentification of the Arg-Gly-Asp (RGD) sequenceas a mediator of attachment of cells to severalplasma and extracellular matrix proteins, includingfibronectin, vitronectin, Type I collagen, osteopon-tin, and bone sialoprotein,89 researchers have beendepositing RGD-containing peptides on biomateri-als to promote cell attachment. Cell surface recep-tors in the integrin superfamily recognize the RGDsequence and mediate attachment.79 Because ofredundancy in the affinity of integrins for adhesiveproteins and because a variety of cells possess thesame integrins, nonspecific attachment of cells toRGD-modified surfaces is a concern. Some groupsare attempting to circumvent this problem by usinglonger peptides with a particular conformation,rather than using short tetra-, penta-, or hexapep-tides.90 Others are examining non-RGD peptidesthat may be more specific for bone cells.91,92 Fur-thermore, a combination of immobilized peptideand soluble growth factor(s) might be needed toelicit specific responses.93 Presently, more studiesare needed to develop surfaces, modified with bio-active molecules, that are selective for only osteo-blastic cells.

A second approach to biochemical surface modifi-cation uses biomolecules with demonstratedosteotropic effects. A wealth of information has beenobtained about the biomolecules involved in bonedevelopment and fracture healing. Many growth fac-tors have been cloned and are recombinantlyexpressed. They have effects ranging from mito-genicity (eg, interleukin growth factor-1, FGF-2,





BRUNSKI ET AL

and platelet-derived growth factor-BB) to increasingactivity of bone cells (eg, transforming growth fac-tor-�1 [TGF-�1] enhances collagen synthesis) toosteoinduction (eg, bone morphogenetic proteins[BMPs]).94,95 By delivering one or more of thesemolecules—which normally play essential roles inosteogenesis—directly to the tissue-implant inter-face, it is possible that bone formation may be pro-moted in implant applications.

Two considerations about delivering biomole-cules to the tissue-implant interface are: (1) localcell populations must interact with the biomoleculesfor a period of time to initiate cellular events, and(2) concentrations of biomolecules must exceed cer-tain threshold levels for cellular activity.96 However,data regarding the duration of exposure or concen-tration needed for optimal activity of osteotropicbiomolecules are lacking.

To control exposure and concentration, it is pos-sible to alter retention and/or release of biomole-cules from implant surfaces by using different meth-ods, such as adsorption, covalent immobilization,and release from coatings. The simplest way todeliver biomolecules to the tissue-implant interfaceis by dipping the implant in a solution of proteinbefore placing it. In orthopedic model systems,studies using simple adsorption indicate that deliv-ery of TGF-� to the tissue-implant interface canimprove bone formation in the periprostheticgap97,98 and can enhance bone ingrowth into porouscoatings.86 Using a similar approach, alkaline phos-phatase adsorbed on titanium implants enhancedperiprosthetic bone formation.99

Besides the fact that it is difficult to control theamounts adsorbed during dipping, another draw-back to the adsorption method is that it provides lit-tle control over the delivery of molecules, includingrelease/retention and orientation. Proteins are ini-tially retained on the surface by weak physisorptionforces. Thereafter, depending on the implantmicro-environment, which varies between anatomicsites and between patients, the proteins desorb fromthe surface in an uncontrolled manner to initiatedesired responses. Considering the necessity of spe-cific receptor-ligand interactions for activity ofmany relevant biomolecules, appropriate presenta-tion of protein may also be needed. Although posi-tive responses have been observed using this simpleapproach, there is no indication that they are opti-mal for clinical applications. There is also a poten-tial problem with diffusion of osteogenic factors andformation of mineral at undesired sites.

Bonding biomolecules to implants is an alternateway of delivering them to the tissue-implant inter-face, although in this case the protein will not be

released. This approach is more complicated thanadsorption because of the chemistry involved, butthe activity of molecules immobilized on plastics hasbeen shown to equal or exceed that of soluble pro-tein.100–102 For orthopedic and dental applications,metal surfaces possess a relative paucity of the func-tional groups needed for immobilizing molecules.The passivating oxide film on these materials does,however, have surface hydroxyl groups that providelocations for bonding using silane chemistry. Thisapproach has been used to immobilize peptides,enzymes, and adhesive proteins on different bioma-terials, including Co-Cr-Mo, Ti-6Al-4V, Ti, andnickel-titanium (NiTi).88,89,103,104 Depending on theparticular silane, experimental conditions, and sub-strate used, biologically active molecules can beretained on biomaterial surfaces for several daysunder simulated physiologic conditions.89,105 Meth-ods to circumvent the problem of lack of functionalgroup diversity on metals include use of plasmatreatments, etching, and deposition of self-assem-bled monolayers (SAMs). Plasma treatment can beused not only to increase the number of hydroxylgroups, but also to deposit reactive amino and car-boxyl groups on the surface; this offers greater ver-satility for binding biomolecules using differentimmobilization chemistries. Preliminary resultsdemonstrate ultrathin plasma-deposited films thatare stable and tightly bind bioactive molecules.105,106

Stable SAMs can be formed by depositing silaneswith different terminal functional groups on modelsubstrates, eg, thiol on gold.107,108 However, forma-tion of SAMs on the imperfect surfaces of real poly-crystalline metals has not been demonstrated. Evendeposition of gold on metallic biomaterials willlikely reflect many of the defects of these surfaces.Another potential benefit of the immobilizationapproach is that it could be used to control presen-tation/orientation of biomolecules to cells. Althoughthis has not been adequately explored, it is conceiv-able that, using specific binding sites on proteins(eg, protecting all but the terminal amino groups),biomolecules could be immobilized in a particularorientation on the surface.

Coatings incorporating biomolecules are alsobeing explored for delivering biomolecules to the tis-sue-implant interface. In light of the dependence ofcell and tissue responses on the duration of exposureand concentration of biomolecules, this approach isattractive because it can be used to control theirrelease. Ethylene vinyl acetate (EVAc),109 poly(lac-tide-co-glycolide) (PLGA),110 and collagen111 areamong the coating materials being pursued. Thecontinued presence of a non-bioerodible hydropho-bic polymer, eg, EVAc, at the tissue-implant

interface after exhaustion of biomolecule, however, iscause for concern. On the other hand, bioerodiblepolymeric coatings, such as PLGA, can be used torelease protein for long periods, although there is noevidence that sustained release is required for opti-mal implant integration. Because they mimic the waymany biomolecules are normally retained in bonematrix, collagen coatings incorporating proteins havealso been investigated. Furthermore, cooperativeinteractions between BMPs and collagen have beenreported.112–114 In addition to an initial release over 1to 4 days that can initiate cell and tissue responses,biomolecules are retained in the collagen matrixcoatings and would be available for later release tosustain responses.111 The amounts of protein re-leased and retained can be controlled by the amountsof collagen and/or biomolecule. Additionally, colla-gen coatings will be turned over in vivo and replacedwith new tissue during the healing response.

Recently developed injectable, absorbable calciumphosphate cements may also be useful for biochemi-cally modifying the tissue-implant interface.115,116

These materials solidify in situ to temporarily stabi-lize the implant and allow early loading, while pro-viding an osteoconductive environment as thecement is replaced with bone. A further develop-ment of the cements would be to use them as amedium for the delivery of bioactive agents to thebone-implant interface.

BIOMECHANICS

All oral and maxillofacial implants are meant to sup-port forces in vivo, so it is obvious that biomechanicsplays a major role in implant design. The followingbiomechanical issues are among the most important:

• What are the in vivo loadings that dental implantsmust be designed to resist?

• What factors are most important in controllinghow the in vivo loads are transmitted to interfa-cial tissues? What are the stress and strain statesin bone around the implant, and how can they becontrolled?

• What are safe versus dangerous levels of stressand strain in interfacial bone? What biomechani-cal factors contribute most to implant success orfailure?

Bite Forces and In Vivo Loading of ImplantsBite Forces in the Normal Dentition. In the normaldentition without implants, mean maximal vertical(axial) bite force magnitudes in humans can be 469± 85 N at the region of the canines, 583 ± 99 N at

the second premolar region, and 723 ± 138 N at thesecond molar.117 (A vertical force is defined as theforce component acting perpendicular to theocclusal plane.) Raadsheer et al118 measured bitingforces with a triaxial piezoelectric transducer andreported average values of the “maximal voluntarybite forces” as 545.7 N in men (n = 58) and 383.6 Nin women (n = 61). The maximum force measuredin each group was 888 N in men and 576 N inwomen. The angle between the resultant bite forceand the z-axis (perpendicular to the occlusal plane)was typically 3.9 degrees. The x and y componentsof the force were much smaller than the z compo-nent (1 to 49 N).

It is interesting to compare the above data withthe vertical bite forces in other mammalian species,if only because many different animal models havebeen used for dental implant research without muchdata on the actual loadings in those species. Forexample, maximal bite forces were 550 N at poste-rior locations in a Labrador dog119 and 1,712 N atposterior locations in orangutans.120 Unfortunately,for parafunctional activities in humans, such asbruxism, few data exist on bite forces in human sub-jects,121 even though it is believed that such habitscontribute to some of the problems that candevelop with implant loading.

For lateral bite forces in the normal human den-tition, the data are less definitive. Some estimatesput the values on the order of 20 N,122 which isprobably a conservative estimate, although Raad-sheer et al’s data118 are comparable. In the incisalregion, the direction of maximum incisal bite forceis about 12 degrees to the frontal plane,123 whichsuggests that the lateral components of force on ananterior implant could be appreciable.

Bite Forces After Implant Treatment. It has beensuggested that the general features of mastication inpatients with normal and implant-restored denti-tions are approximately the same.124 However, Carrand Laney125 reported a significant improvement inboth maximum and mean biting forces in a group of14 patients who started with conventional completedentures but then received mandibular tissue-inte-grated prostheses opposing a complete maxillarydenture. For example, the mean maximal force was59.6 N in the case of conventional dentures and112.9 N for tissue-integrated prostheses. Carr andLaney also summarized other workers’ data on thesame subject and noted forces on dentures rangingfrom 57 to 120 N and forces on tissue-integratedprostheses from 137 to 190 N. Mericske-Stern andZarb126 explored maximum occlusal force and oraltactile sensibility in: (1) a group of partially edentu-lous patients restored with ITI implants supporting


BRUNSKI ET AL




fixed prostheses or single crowns, and (2) a controlgroup of fully dentate subjects with healthy naturalteeth. They found the highest maximal occlusalforce in fully dentate subjects on the second premo-lars (mean 450 N). For fixed prostheses supportedby implants, the average value of maximal occlusalforce was lower—about 200 N for first premolarsand molars and 300 N for second molars. In theirdata, the range of forces measured in the varioussubjects included values as large as 1,100 N in themolar region.

In Vivo Measurements of Bite Forces andMoments on Implants. For implants used as single-tooth implants, in vivo forces ought to replicate theforces exerted on natural teeth. This is expectedbecause in both cases, the biting would be deliveredto single, stand-alone crowns. However, factorssuch as the width of the crown occlusal table, theheight of the abutment above the bone level, andthe angulation of the implant with respect to theocclusal plane will affect the value of the momenton the implant. Also, it is self-evident that theforces and moments on an implant depend onexactly how the crown is built into an occlusalscheme. Moreover, it is also likely, but not necessar-ily clinically significant, that the nature of theocclusal surface material—eg, porcelain or acrylic—influences the dynamic character of the forcestransmitted to the implant.127

However, it is necessary to go beyond the abovefactors for free-standing implants when trying tounderstand the loadings on several implants linkedtogether by the prosthesis. Under these conditions,leverage effects exist because of geometric factorsrelating to restorations linking the implants, such asthe existence of distal cantilevers in a full-archrestoration. Such factors cause the implants to besubjected to increased bending moments as well asaxial forces that can be tensile and compressive.These facts are demonstrated in a number of repre-sentative experimental studies (and in theoreticalwork discussed shortly).

For example, Glantz et al128 used strain-gaugedabutments in vivo to measure the loading ofimplants supporting fixed prostheses retained byosseointegrated implants in 1 patient. Typical invivo loading of the implants during chewing of cel-ery, apples, and bread included axial force compo-nents from –20 N to + 25 N, with – meaning com-pression and + meaning tension. At the same time,there were bending moments in vivo up to about 20N�cm. (These data are for bending moments aboutmesiodistal and buccolingual axes. No data wereprovided concerning the moments about the centralaxis of the implant.) Moreover, when biting

occurred at the cantilever location of a prosthesissupported by 5 implants, axial forces on some of theimplants were as much as twice the biting force onthe prosthesis, a result that can be predicted fromtheoretical models discussed shortly.

Rangert et al129 used the same methods for invivo measurements of the vertical load distributionand bending moments on a 3-unit prosthesis sup-ported by a natural tooth and a single Brånemarkimplant. In 5 patients, they demonstrated that thevertical loads were distributed between the naturaltooth and the implant. For relatively large forces onthe prosthesis (greater than 100 N), the bendingmoments on the implant were 10 to 15 N�cm,which was below the acceptable limit for themechanical components of the system (50 to 60N�cm for the screw joints).

Mericske-Stern et al130 used piezoelectric trans-ducers to measure forces on implants that supportedoverdentures. The overdentures were attached tothe implants by 3 different attachment systems: U-shaped bar, round clip bar, and single telescopes.They found that the vertical component of the forcedepended on the nature of the attachments, and thatthe transverse (lateral) components were 10 to 50%of the vertical force components. These workersalso noted that “rigid bars contribute to load sharingand stress distribution onto the implants.”

Fontijn-Tekamp et al131 used bite forks andtransducers of special design to record the bitingforces during closure on mandibular overdenturessupported in different ways by implants. Theyreported that the differences in support of themandibular overdenture were not reflected in biteforce capabilities of the patients. Typical data forpatients with implants showed maximal unilateralbite forces in men and women ranging from about50 to 400 N in the molar regions and 25 to 170 Nin the incisal regions.

For fixed prostheses, Mericske-Stern and Zarb126

measured maximal occlusal force and oral tactilesensibility in 21 patients wearing complete maxillarydentures and mandibular fixed prostheses supportedby Brånemark implants. The maximal occlusal forceranged from 35 to 330 N, with the largest values atthe second premolar region. The detection thresh-old of minimal pressure was 330 g in the horizontaland 388 g in the vertical direction.

Gunne et al132 measured axial forces and bendingmoments in vivo on freestanding and connectedimplants supporting 3-unit mandibular prosthesesthat opposed complete dentures. Each of 5 patientshad 2 prostheses, one supported by 2 implants andthe other supported by 1 implant and 1 naturaltooth. The study found “no major difference in


BRUNSKI ET AL




functional load magnitudes related to the supporttype.” Also, the study reported that “the distribu-tion of load between the abutments was influencedmore by the prosthesis geometry and implant place-ment than by the difference in load characteristicsof tooth and implant,”132p335 although this conclu-sion was “limited to [the case of] one implant con-nected to a tooth.” Bending moments on implantsranged from 0 to 25 N�cm. An especially notewor-thy finding in the study was the fact that implantscould be connected with teeth in these patientswithout causing any detrimental clinical sequelae—an idea that has often been debated over the years.Also, axial forces on an implant could be tensile orcompressive, depending on their location when sup-porting a 3-unit prosthesis, and that the axial load-ing on an implant could sometimes be twice as largeas the bite force on the prosthesis because of can-tilever extensions.

Richter133 used a strain-gauged abutment to mea-sure in vivo horizontal bending moments on IMZimplants in the molar area in humans. The authorreported maximum bending moments and the cor-responding transverse force when the patientschewed various types of food, including sticky con-fections (jelly beans), sausage, carrots, and crackers.The magnitudes of the bending moments rangedfrom less than 10 N�cm to slightly more than 20N�cm. The corresponding transverse forces in buc-cal and oral directions were always less than 30 N.

Biomechanical Models for Predicting ImplantLoading. Many investigators have tried to gaininsight into implant loading by performing testsusing experimental, analytic, and computer-basedsimulations of systems of implants supporting vari-ous types of prostheses. The biomechanical model-ing of implant systems is a large literature in itsown right, and only a few highlights are reviewedbelow.

Popular experimental methods have included theuse of strain-gauged abutments (designed like thoseused in vivo by Glantz et al128 and Gunne et al132)and photoelastic models.134 Theoretical approacheshave ranged from the simple to the relatively com-plex (Table 1). Monteith135 presented a computer-ized version of the 1983 Skalak model so that clini-cians might use it more readily in case planning.More complex computer models have ranged from2D finite element (FE) idealizations to full 3D FEstress analysis with anisotropic material propertiesfor the bone, and bonded versus nonbonded bound-ary conditions for the bone-implant interface. Themain assumptions underlying the analytic and com-puter studies are summarized and discussed recentlyin Brunski and Skalak.136

From the above literature, the first key result isthat when more than 1 implant is used to support aprosthesis, the forces on the supporting implantscan sometimes exceed the forces applied to theprosthesis. This seemingly paradoxical result isexplained by the mechanics of multi-unit prosthesessupported by 2 or more abutments in various geo-metric arrangements. A second key finding from theliterature is that both forces and moments can beproduced on dental implants, which was illustratedby some of the in vivo studies reviewed previously.Third, even though bite forces are typically viewedas acting downward, toward the apex of naturalteeth, thereby tending to compress the teeth intotheir alveolar sockets, it is a fact that implants canbe exposed not only to compressive forces but alsotensile forces (in addition to moments), dependingon how the implants are arranged relative to wherethe bite force is applied to the prosthesis. Again,this fact emerges when more than 1 implant sup-ports a loaded prosthesis.

The above 3 results are demonstrated by typicalresults of experimental studies and theoretical mod-els noted in the following. Table 1 outlines similari-ties and differences among the analytic models.

One problem with the models in Table 1 is thatnone of them has been thoroughly checked againstresults from in vivo measurements. Qualitatively, thevarious studies of in vivo loads on implants suggestthat typical predictions from the models are correct,at least in broad terms. For instance, it is clear that:(1) both tensile and compressive axial loads canoccur on implants, (2) long cantilevers exacerbatethe axial loading and bending moments, and (3) sig-nificant bending moments can exist on implants invivo. However, detailed validation of the analyticmodels in view of in vivo data would be a usefultopic for future work, to allow evaluation and greaterconfidence in the predictive power of the models.

Another specific issue that needs to be investi-gated in all of these analytic models (except for oneof the models discussed by Morgan142) is theassumption of an infinitely rigid bridge. Actual pros-theses are probably not infinitely rigid in the senseused in these models. For example, recent tests ofactual prostheses143 indicate that the rigid-bridgeassumption of the 1983 Skalak model138 leads tosomewhat inaccurate results when trying to predictthe forces and moments on abutments supportingmetal-backed versus all–acrylic resin prostheses usedby Balshi and Wolfinger144 in immediate loading viaconversion prostheses. This study involved in vitroexperiments with metal-backed and acrylic resinprostheses supported by strain-gauged abutmentsand implants imbedded in plaster. In contrast to


BRUNSKI ET AL




what would be predicted by Skalak’s 1983 theory,the data showed that more of the loading was con-centrated on the implants nearest to the point wherebite force was applied to the prosthesis. Notably,this same trend was seen in the experiments onabutment loading described in recent text-books.145,146 Overall, it remains to be determinedwhether the prostheses used in actual in vivo casesare well described by the assumption of a rigidbridge. Interestingly, the prosthesis design forimmediate loading in the Novum System by Bråne-mark et al6 is shaped more like an I-beam; perhapsbecause of this shape, this prosthesis will behavemore like a rigid structure, but experiments will beneeded to confirm this.

A second validation topic for the analytic modelsis the extent to which it is realistic to assume that allof the abutments/implants have the same stiffness inbone. It is understood from the analytic models thatthe stiffness of an implant—which is further definedand discussed in the references cited initially in thissection—influences the load distribution among theabutments. For example, the Skalak-Brunski-Mendelson (SBM) theory,139 which was the first toallow for different axial and lateral stiffness valuesamong the abutments, demonstrated that the loaddistribution deviates appreciably from that pre-dicted by a model having identical stiffnesses foreach abutment (eg, the 1983 Skalak model138). Andto a large extent, the SBM theory has been success-fully validated in benchtop testing with a collectionof abutments having known but different stiffnesses.However, when it comes to actual in vivo situations,there have been no definitive validation studies ofthis model.

A third factor that needs to be validated beforeconfidently applying analytic models to clinical realityis the nature of the connections between prosthesesand implants in actual patients. All of the models inTable 1 except the Morgan-James (MJ) and Brunski-Hurley (BH) models assume ball-and-socket connec-tions between each abutment and the rigid bridge. Inreality, it is of course known that prostheses are com-monly attached to abutments by screw joints orcemented joints, neither of which is correctly ideal-ized as a ball-and-socket joint. Hence, the MJ andBH models are probably the most realistic, especiallythe BH model, which also allows for different stiff-nesses of the abutments. However, it remains tocheck these models against actual results from clinicalsituations, and to see whether the simpler modelspredict the results well enough for clinical purposes.

It should be noted that it is also possible to use theBH and MJ analytic models to analyze partially eden-tulous patients141 as well as overdenture subjects.147

For instance, a 2-implant patient was analyzed byBrunski and Skalak,136 and a 3-implant subject hasbeen analyzed.148 The latter study addresses the spe-cific problem of whether an in-line versus staggeredarrangement of 3 implants is preferred when sup-porting a 3-unit partial prosthesis. The main resultfrom that work is that a staggered arrangement isgenerally preferred because the implants tend todevelop lower bending moments when arranged in astaggered fashion. Daellenbach et al149 conducted anFE study of straight versus staggered implants withessentially the same conclusions. Renouard andRangert148 have collected a number of these ideasand clinical guidelines about implant loading and riskfactors in a recent textbook.


BRUNSKI ET AL




Table 1 Summary of Analytic Models for Predicting Implant Loading

Nature of the Axial and MomentsIdealization connection lateral force supported by

of prosthesis: between Stiffness components the prosthesis-rigid or prosthesis and of the on the abutment

Model flexible abutments abutments abutments? connection?

“See-saw” Rigid prosthesis Ball-and-socket Same No, only axial No(Rangert et al137) (supported by stiffness

no more than for both2 abutments) abutments

Skalak138 Rigid Ball-and-socket Same for all Yes NoSkalak et al139 Rigid Ball-and-socket Different Yes No

stiffnessesallowed foreach abutment

Morgan and Rigid Built-in-joint Same for all Yes YesJames140

Brunski and Rigid Built-in-joint Different for Yes YesHurley141 all abutments

Several computer models for predicting loadingon implants have appeared over the last 2 decades.While many FE models have more commonly beenused to analyze the stress and strain states in bonearound implants, only a few have been used to pre-dict the loadings on the implants.

What appears to be the first 3D FE model topredict forces and moments on implants supportingfull-arch cases was based on Mailath et al.151 Thismodel was formulated to predict 3 force compo-nents and 3 moment components at the junctionbetween each implant and the loaded prosthesis. Asimilar model by Mailath-Pokorny and Solar146

compared axial loading of 4 or 6 implants support-ing a prosthesis. This work showed data consistentwith the trend noted earlier about the role of bridgerigidity—namely, that if the bridge is not perfectlyrigid, there tends to be a concentration of moreload (relative to the predictions of the originalSkalak model) on those abutments nearest the load-ing point on the bridge.

An FE model by Benzing et al152 concluded thata more “spread out” arrangement of the implantsacross the arch gave a “more favorable” distributionof bone stresses around the implants. This modelalso pointed out that the prosthesis material anddesign (shape) affected the stress distribution aswell. Lewinstein et al153 developed a unique “IL”system for supporting the distal extension of a can-tilevered prosthesis; the system used a short implantand a special ball-type attachment that provided adistal stop when the cantilever displaced downwardunder loading. The results of the study did notfocus on the implant loadings per se, but did discussthe stresses in the bone around the implants, whichwere lower when the IL system was employed. Sert-göz154 used a 3D FE model to study the effect ofthe superstructure material and occlusal surfacematerial on the stress distribution in an implant-supported fixed prosthesis. The work reported thatusing a superstructure material with a lower modu-lus did not lead to substantial differences in thestresses in any of the parts of the model (eg, pros-thesis, screws, implants, surrounding bone),although the lower-modulus material did tend toconcentrate stresses in the retaining screws.

Overall, the main advantage of FE models versusthe analytic models is that they can more accuratelysimulate the many geometric and material complex-ities that exist in real patients. Such complexitiesinclude the 3-dimensional nature of the problem;the differing material properties of the prostheses,implants, and bone; the various boundary condi-tions inherent in how the mandible and maxilla aresupported; and the nature of the boundary condi-

tions between the bone and the implant, eg,whether the implant and bone are bonded togetheror not. However, the main disadvantage of FE mod-els is the need for expertise in relatively complexsoftware and hardware to formulate the problem,together with the need for expertise in the mechani-cal principles involved in the problem. This meansthat FE modeling is not a tool for the novice.Therefore, even though FE models would no doubtbe valuable as a clinical treatment-planning tool, itis unlikely that such models will make their way intothe clinical world anytime soon.

Misfitting Prostheses. The topic of misfittingprostheses and the biomechanics thereof is dis-cussed in depth in a companion paper by Taylor etal in this journal.155 Briefly, the misfit problemstems from the fact that dimensional errors exist inthe fabrication of the prosthesis relative to the spac-ing and angulation of the abutments. Then, whenthe final prosthesis is attached to the abutments inspite of a misfit, this “non-passive” fit induces forcesand moments on the supporting implants, evenbefore they are loaded during mastication. Theunderlying mechanical principles that come intoplay in analyzing this problem encompass a numberof topics, including the mechanics of screw jointsunder both ideal156,157 and nonideal158 conditions,as well as the properties of the structural membersheld together by the joint.159

It is interesting that even though research hasdemonstrated misfit in typical prostheses,160 andeven though there have been measurements show-ing that there can be significant loading on implantsas a result,161,162 this has not translated into sig-nificant problems at the level of the bone-implantinterface.163,164

Orthodontic and Craniofacial Biomechanics.Orthodontic and craniofacial applications of implantsinvolve many of the same issues discussed above. Forfurther analysis, see other research.165,166

Noninvasive Assessment of the Interface UsingBiomechanical Means. When loading is to begin ona given dental implant, eg, after second-stage surgery,it would be advantageous if the healing status of thebone could be measured noninvasively by some tech-nique. Radiography can give some information,along with tests of implant “mobility,” but neitherapproach provides fundamental information aboutthe structure and properties of interfacial bone at thelevel of detail needed for biomechanical predictions.The following data are needed: (1) the axial and lat-eral stiffness of the implant-bone complex; (2) thelevel of mineralization and mechanical properties ofinterfacial bone (eg, woven bone, trabecular bone, orcortical bone); (3) the spatial extent of the bone


BRUNSKI ET AL




around the implant (eg, percent bone-implant con-tact and where there is bone contact); and (4) mono-cortical or bicortical anchorage. It would be possibleto use this sort of data as input to various predictivebiomechanical models that could then help withtreatment planning—for example, the previouslynoted analytic or FE models for predicting forcesand moments on implants.

Unfortunately, there is no current technique thatprovides such information. Many workers have triedto use the Periotest device as a means to assess thestatus of the bone-implant interface, but this onlyprovides a number that has an unclear relationshipto specific properties of the surrounding bone. Forexample, Caulier et al167 found that the Periotestwas “neither able to discriminate between the firstthread nor between the total number of threads incontact with bone.” The modal analysis techniqueof Elias et al,168 which involves a small hammerwith a built-in load cell and accelerometer, has theadvantage of being able to discern trabecular versuscortical bone at an interface, or soft versus hard tis-sue, at least in vitro. The method also has a theoret-ical basis that helps with interpretation of theresulting data.169 Potentially, it also could yield datathat relate to the axial and lateral stiffness needed inmodels such as the Skalak-Brunski-Mendelson ana-lytic model. However, although it can be furtherminiaturized, it has the disadvantage of being cum-bersome to use in the mouth, and in any case hasnot been reduced to a clinically convenient device.

The so-called resonance frequency technique ofMeredith and coworkers170,171 has the significantadvantage of being convenient, involving only asmall vibratory element that can temporarily bescrewed into the implant for in vivo tests. The sig-nal analysis in this method allows it to distinguish,for example, changes in the mechanical propertiesof the interface during the polymerization of resinin which an implant is embedded. But owing to anunclear theoretical interpretation of exactly what isbeing tested by the device—eg, a combination ofbone and implant properties, and if so, exactly whatproperties?—it does not produce data of direct usein biomechanical models. Overall, the field stillneeds a clinically reliable, sensitive, noninvasive testof the biomechanical status of the interface172 thatalso provides input data for predictive models.

Biomechanics of the Bone-Implant InterfaceLoaded oral and maxillofacial implants have to besupported by the interfacial bone in which they areplaced; the chain of action-reaction from the pros-thesis to implant to interface keeps the system inmechanical equilibrium. The interfacial bone is

loaded, which produces stress and strain fields thatextend for an appreciable distance away from theimplant. Stress and strain are related by the consti-tutive equation for a material, the nature of whichhas to be determined for a given material, eg, wovenbone as opposed to mature lamellar bone. Stress andstrain are both important quantities, but whetherone focuses on stress or strain depends to an extenton the goals of the analysis, which as noted earlier,depend on the design perspective.

Probably the most important reason for discussingstress and strain is that bone, like any other material,will be damaged and possibly fail if the stresses andstrains become too high at a particular point in thematerial. Ideally, the design goal is to create animplant and its interface so that anticipated loadingdoes not damage the implant or surrounding tissueswhen the implant is loaded. Second, it is important toknow what stresses and strain develop at the exactregion where the implant surface comes into contactwith surrounding bone or other tissue. Here a rele-vant question becomes: Is there any sort of “bond”between the implant surface and tissue, and if so, whatis its strength? This specific bone-implant contactregion is only part of the interfacial zone betweenimplant and bone, but it is a part that could be dam-aged if the stress and strain states are excessive. Athird reason for studying stress and strain in interfa-cial bone is more biologic: the cells of blood and bonenear a loaded implant will also be loaded and could beaffected by the local stress-strain fields. Here thequestion is: What sorts of stress and strain states are“good” versus “bad” for the cells and biology of bone?

All 3 aspects of stress transfer above have beenexplored in a wealth of papers over the years. Thefollowing sections summarize that work by startingwith biomechanics of the early healing stages ofbone, and then ending with biomechanical issuesrelated to loading at later times.

Biomechanical Aspects of Interfacial Bone Heal-ing. Three major facts are clear: (1) drilling and cut-ting involved in oral implant surgery damages bone,(2) a cascade of wound healing events is triggered bythe surgery, and (3) not all bone sites have the samequality and quantity of bone. It is also clear that dif-ferences inevitably exist among various implants,surgical techniques, animal models, postoperativeloading protocols, graft materials, etc. Therefore, itshould not be assumed a priori that immediatelyloaded implants, implants in grafted bone, implantsin craniofacial bone, implants used with guidedbone regeneration (GBR) and membranes, etc, allhave identical bone healing sequences, rates, andinterfaces. Also, it would be presumptuous to con-clude that the load-bearing capacity or adaptability


BRUNSKI ET AL




of bone to loading will be identical in all these dif-ferent situations. Moreover, the bone in various ani-mal models is certainly not all the same and is notnecessarily easily related to human bone.

In view of the foregoing, and especially with aneye toward interest in immediately loaded implants,it is relevant to review what is known about theevents and timecourse of bone healing in: (1) gapsthat inevitably exist between implant and bone atsurgery, and (2) pre-existing bone that is damagedby surgical procedures.

Healing in Gaps, Pores, or Other Spaces that InitiallyExist Between Bone and Implant. In general, perfect3-dimensional congruity will not exist between asurgically prepared bone site and the surface of adental implant, especially in view of the multitudeof different implants that are now used. Micro- andmacro-gaps will typically exist at the interface, evenwhen the implant is supposed to have a reasonablyclose fit to the prepared bony site, as already alludedto by Brånemark.173 A common feature of any gap,pore, or other defect between bone and implant isthat it should (under the right conditions) fill with ablood clot soon after surgery. Then, as long as theimplant is stable in the site, bone will develop in thegap.77,174,175

Looking at 2 extremes of healing in bone defects,Schenk and Hunziker175 reported on intramembra-nous bone healing in small holes (0.6-mm diameter)drilled in rabbit cortical bone, while Schenk et al176

studied such healing in large (approximately 10 �10 � 7 mm) 3-wall defects in dog mandibles cov-ered by e-PTFE membranes. These and other stud-ies in various animals show that intramembranousbone formation proceeds through a well-definedsequence of steps, including blood clot formation,angiogenesis, osteoprogenitor cell migration, wovenbone formation, compaction of woven bone bydeposition of parallel-fibered and lamellar bone,and eventually secondary remodeling of the wovenbone. For small (0.6-mm diameter) drill holes inrabbit tibial cortex, woven bone (compacted withparallel-fibered and lamellar bone) filled the hole by6 weeks after surgery. Six months later, there wasappreciable secondary remodeling and creation ofnew osteons, which were replacing the woven pri-mary bone with secondary lamellar bone. Rahn177

quotes reports showing that if the gap is very small,eg, a hole up to 0.2 mm in diameter, there may notbe any woven bone, but rather healing by develop-ment of lamellar bone directly on the walls of thedrill hole, at a rate of about 1 to 2 µm per day. (Thismeans that a 0.1-mm hole would be filled in about 3to 4 weeks.) In the case of Schenk et al’s176 bonehealing in the large 3-wall, membrane-covered

defects in the dog mandible, healing followed asequence of steps that was similar to that in thelarger-diameter holes, ie, compacted woven boneexisted in the defect at 1 to 2 months, with sec-ondary remodeling of woven bone starting at about3 to 4 months after surgery. Similar events werereported in healing of a 1-mm-wide sham defect cutin mandibular bone.178

While the above studies did not have oralimplants present during bone healing, the signifi-cance for implant biomechanics is that actual stud-ies with implants confirm that the above steps dooccur in various-size gaps or spaces that inevitablyexist around implants. The significance of this isthat, for example, in the first weeks to months aftersurgery, the interface will typically be comprised ofnew woven bone (that will eventually remodel), aswell as damaged and remodeling lamellar bone;indeed, this has actually been shown in many exam-ples in the literature (eg, Plenk and Zitter179). Cer-tainly the size and location of interfacial gaps as wellas the extent of remodeling in the nearby damagedbone will vary among different implant designs,bone sites, and surgical procedures.

Another significant point about the foregoing isthat the biomechanical properties of the resultingcomposite interfacial structure are as yet unknown,yet would be needed in any detailed biomechanicalanalysis of interfacial biomechanics. One of the fewstudies to try to quantify the properties is a recentmicrohardness study of bone around implants indog femoral mid-diaphyses.180 This work docu-mented a gradient in hardness with distance awayfrom the implant surface. At 12 weeks after implan-tation, the Knoop microhardness number was about31 at 0.2 mm from the implant. Hardness valuesincreased to 45 at 1 mm from the implant, whichessentially matched the hardness of normal corticalbone far away from the implant. Exactly how thehardness data correlate with elastic modulus is notclear, but recent work with nano-indentation ofbone could provide a new technique with which toanswer this question.181

Healing of Damaged Pre-existing Bone. Implantsurgery clearly damages pre-existing bone at theimplant site, which triggers innate healing responsesin that bone. Hoshaw et al182 documented thenature and extent of microdamage to bone duringdrilling, tapping, and placement of 3.75 � 7 mm TiBrånemark implants in tibial diaphyses of NewZealand white rabbits. In a baseline group of 5 rab-bits, they measured a mean area fraction of micro-damage (eg, microcracks revealed by basic fuchsinstaining) of about 12.5% ± 2% in a 1.2-mm-wideregion of the interface adjacent to the screw. In a


BRUNSKI ET AL




healing group of 10 rabbits, implants in both tibiaewere allowed 4 weeks of healing; this group showedan area fraction of microdamage of about 9.5% ±2% (P < .03). Hoshaw et al demonstrated a correla-tion between microdamage at the interface and boneremodeling (A-R-F) at 4 weeks after implantation.(A-R-F stands for the cellular stages of the remodel-ing process: activation of osteoclastic cutter cones,resorption, and formation by osteoblasts.182)

Hoshaw et al’s observation of increased remodel-ing in relation to damage from surgical placement isconsistent with other work suggesting that micro-damage in bone stimulates bone remodeling as areparative reaction.183 Also, Roberts184 reportedthat “about 1 millimeter of compacta adjacent tothe osseous wound undergoes necrosis postopera-tively despite optimal surgical technique,” andnoted that bone remodeling would be needed torepair this bone.

The above facts are biomechanically significantbecause they indicate that damaged pre-existingbone (as well as whatever newly formed woven bonemight exist in gaps) around an implant would haveto undergo at least 1 remodeling cycle (1 “sigma”)to repair the damage. One sigma is about 1.5months in rabbits, 3 months in dogs, and 4.25months in humans.185 However, it is not clearwhether just 1 sigma is sufficient time to allowrepair of 100% of the damage caused by surgery. Itseems reasonable to suggest that the time neededfor complete repair depends on the extent of thedamage, the amount of remodeling that is triggered,and the size of the discrete packets of bone turnedover by the A-R-F process. It seems possible that100% repair of damaged bone by A-R-F turnovermight take longer than 1 sigma in dental implantpatients and experimental animals. Moreover, newbone can be deposited on old bone without priorremodeling.

This begs the question of exactly when, aftersurgery, the bone around an implant should beassumed to have settled back into a steady state ofremodeling akin to that which existed before surgery.However, there are those who question whetherinterfacial bone remodeling around an implant everreturns to presurgical remodeling rates.186,187

Garetto et al186 have reported a highly elevated (withrespect to normal) remodeling rate—on the order of500% per year—around the threads of 2-stageimplants. However, their data analysis assumes that atime of a few sigma is by definition long enoughafter surgery or loading for bone to have reached anew “steady state” of remodeling. But one couldquestion the validity of this assumption; is it certainthat just a few sigma are sufficient for all transients

to die away after surgery and/or loading? This areaof research deserves further experimentation, espe-cially in view of the data below.

The literature indicates that, in addition toremodeling of bone that is frankly damaged by thesurgery, there is heightened remodeling of corticalbone for some millimeters beyond this region ofobvious, identifiable damage. For example, in a sum-mary of bone remodeling in dog radii subjected toosteotomies, Schenk and Hunziker175 reported thata large region of bone near the osteotomy site wasundergoing remodeling via the A-R-F process; at 8weeks after the osteotomy, 62.5% of all osteons inthe entire bone’s cross section were replaced or inthe formative phase of renewal. This percentage ofactively remodeling osteons greatly exceeded thepercentage involved in remodeling in the controllimb, which was 2.5%. Similarly, for 3.75 � 7 mmBrånemark screws in dog tibiae, Hoshaw et al188

reported an increased uptake of fluorochrome bonelabel 5 weeks after surgery for 3 to 4 mm proximaland distal to the edges of the drill hole for theimplant; 3 to 4 mm is beyond the region withnoticeable microdamage from surgery. (Theincreased uptake of label was in comparison touptake at a control site defined as the opposite cor-tex of the same tibia.) Other publications,180 as wellas schematic diagrams of bone remodeling aroundimplants189 and the concept of Frost’s regional accel-eratory phenomenon (RAP),190 all suggest height-ened remodeling activity (compared to controls) ofinterfacial bone over an appreciable distance of theinterface. Related to this question of determiningwhen bone has “quieted down” after being per-turbed by surgery or loading, Frost has noted191:

Sigma designates the earliest time after chal-lenging the bone remodeling system when a newsteady state characteristic of the treatment canexist. In practice one should allow such a system 2or more sigmas to settle down, to avoid mistakingtransients for steady-state phenomena; such unwit-ting mistakes arose so often in past work . . . thatthey represent the rule rather than the exception.

Note that implicit in all of this is the idea that bonehas some sort of control system, which is a key ideadiscussed again in the last section of this paper.

From a biomechanical view, the facts above aresignificant because they suggest the considerabledifficulty in knowing (1) when, and indeed if, boneresumes some sort of steady state of remodelingafter surgery (or load-related insult); and (2) accu-rate mechanical properties of the interfacial bone,especially early after surgery. Currently there is a


BRUNSKI ET AL




lack of data about the mechanical properties ofwoven bone, although new work is shedding lighton differences in noncollagenous proteins in themakeup of woven versus lamellar bone.41,42 Also, theabove facts lead to a second key problem in studyingin vivo bone reactions during implant loadingshortly after surgery. Namely, if loading startswithin days, weeks, or months after surgery, it couldprove difficult to distinguish biologic responses thatmight be triggered by load-related variables per sefrom biologic responses triggered by innate healingreactions per se. (This issue is further discussed byBrunski.37) At this point all that can be said is thatfor immediately loaded oral implants, the structureand properties of interfacial bone will not be identi-cal to those that exist under conditions of delayedloading. Unfortunately, for either interface, detailedmechanical properties have not been determined.

Inter facial Micromotion and In Vivo TissueResponse. Micromotion (relative motion)—a rela-tive displacement between an implant and surround-ing tissue—has been fully discussed in recent reviewarticles,25–27 from which the following conclusionsemerge.

First, it is not the absence of loading per se thatis critical for osseointegration around implants, butrather the absence of excessive micromotion at theinterface. In this statement, the term osseointegrationmight best be understood to mean simply “undis-turbed bone healing around the implant.”

Second, with micromotion, it is important tospecify when the excessive micromotion occurs rela-tive to the time of implantation. To date, most liter-ature on micromotion has dealt with micromotionoccurring immediately after implantation, whenhealing events have been triggered and implant sta-bility depends on the implant’s shape and retentivedesign features (or lack thereof). Under these condi-tions, the mechanism of the oft-reported fibrous tis-sue sequela of micromotion seems to be as follows.Micromotion, if excessive, is thought to damage thetissue and vascular structures that are part of theearly stages of bone healing. Micromotion probablyinterferes with development of an adequate earlyscaffold from the fibrin clot. Also, micromotionprobably disrupts angiogenesis and the establish-ment of a new vasculature for the healing tissue,which in turn interferes with the arrival of regenera-tive cells. Eventually, excessive micromotion willpromote, in ways not fully understood, a re-routingof the healing process into repair by collagenousscar tissue instead of regeneration of bone.192

Davies77 has also suggested that excessive micromo-tion may prevent the fibrin clot from adhering tothe implant surface during early healing; it has been

theorized that surface roughness might help dimin-ish the negative effects of micromotion relative towhat would happen with a perfectly smooth surface.

Third, it is especially relevant that these findingsabout micromotion are true not only for metallicbiomaterials (with either smooth or rough surfaces),but also for ceramic biomaterials such as HA.193

Lastly, from a design viewpoint, this research onmicromotion begs a definition of “excessive” micro-motion. That is: How much micromotion can betolerated before a fibrous tissue interface willdevelop instead of an osseointegrated interface? Thereview articles noted earlier put the threshold atabout 100 µm, assuming micromotion is started andmaintained at this magnitude soon after surgery.However, maintaining the same amount of micro-motion throughout an animal experiment can beproblematic and depends on the setup. Prendergastet al194 conducted a computer simulation of work bySøballe et al193,195,196 in which fibrous interfaceschanged cellular content over time from fibroblasticto osteoblastic, nominally under the same conditionsof micromotion. However, a computer simulationsuggested that the same amount of micromotionmay not have occurred throughout Søballe et al’sexperiments; the data indicated that conditions inthe experiment probably changed from motion-con-trol to force-control because of changing tissueproperties at the implant site. Thus, it might be thatbone formation could occur once the micromotiondecreased below the threshold level.

The biomechanical relevance of the foregoing isthat in cases of immediate or early loading of oralimplants, excessive micromotion at the bone-implant interface must be avoided. Obviously, it isnot easy to demonstrate that this design goal isachieved at the start of a given case. A useful clinicalguideline is that much will depend on the inherentstability of the implant when first placed in its siteand loaded. Factors that will most likely affect sta-bility include the shape of the implant relative to itsbone site, the surface texture of the implant, theproperties of the bone, the nature of the loading onthe implant, and the splinting design for theimplants (if used in a full-arch situation), amongother factors. For example, the axial and lateralmobility of a new implant design, the Mark IVimplant (Nobel Biocare), has been measured invitro using a special system designed to measureaxial and lateral stability in trabecular bone.197

Overall, it has been difficult to gain moredetailed information about micromotion and relatedissues from the results of clinical studies of immedi-ate loading, because none of these studies have actu-ally targeted those factors in a quantitative manner.


BRUNSKI ET AL




Attachment Strength Between the Implant Sur-face and Bone. Given that it is detrimental to haveexcessive micromotion at an interface, it makessense to direct design efforts toward developing asstrong a “bond” as possible between an implant andtissue. For example, in the extreme case of an infi-nite-strength interfacial bond, it would be impossi-ble to have any interfacial micromotion because thebone and implant would remain bonded together atthe interface. However, as discussed below, the lit-erature indicates that an infinite-strength bond doesnot occur in reality. A number of studies have cen-tered on the exact nature and strength of the attach-ment that does exist, but some papers have ques-tioned whether in fact this attachment has anyappreciable physicochemical strength. That is, inthe absence of mechanical interdigitation betweensurface asperities and bone constituents, what is thestrength of attachment?

Cement Line at the Bone-Implant Interface? Recentresearch by a number of workers29,31–33,43,177,198 hasdocumented the ultrastructure of the junctionbetween the implant surface and bone in vitro andin vivo. This work shows that biochemical andstructural similarities exist between bone-biomater-ial interfaces and natural interfaces in bone itself,for example the cement line between a secondaryosteon and pre-existing bone. Davies43,77 drawsattention to this sort of cement line, but more gen-erally for mineralized tissue, cement lines (reversal,resting lines) also exist and are actually about 10times thinner than cement lines around osteons andcorrespond to what is seen around implants.

Cement Lines in Normal Bone. Even for normalbone there is uncertainty about the exact structureand properties of cement lines.199 While theosteonal cement line is known to be collagen-defi-cient and probably hypomineralized,200 the exactpercentage of collagen, mucopolysaccharides, gly-coproteins, and mineral is still under study. Afterreviewing the data, Martin and Burr suggestedthat199:

. . . the cement line represents a residuum of theground substance in osteoid, which is produced asa part of bone remodeling. Because the cementline . . . represents the remnant of the reversalphase of bone remodeling . . . the similarities incomposition [with osteoid] should not be surpris-ing.199p51–52

Significantly, cement lines in normal bone aregenerally thought of as weak points in the overallcomposite structure. For instance, yield and fracturetesting show failures at cement lines.201,202 Martin

and Burr199 attributed crack-stopping potential andenergy dissipation to the compliant nature of cementlines during fatigue of compact bone. Burr et al203

discussed a possible stiffness difference betweenbone and cement lines. Moreover, in fatigue studiesof small beams of trabecular bone versus smallbeams cut from dense cortical bone, Choi and Gold-stein204 explained the poorer fatigue behavior of tra-becular bone in terms of its “mosaic” microstruc-ture, consisting of packets of remodeled boneseparated from pre-existing bone by cement lines.However, except for these inferences that naturalcement lines are weak, little data are available on thedirect mechanical properties of cement lines.

Intrinsic Strength of the Bone-Biomaterial Interface.If a structure similar to a cement line exists at thebone-implant interface, it follows that this structuredetermines the intrinsic strength of a bone-implantinterface. Here, intrinsic strength of an interfacecan be defined as the strength in the limit of a per-fectly smooth biomaterial surface. What follows isthe thinking behind this idea as well as some resultsrelating to the shear and tensile strength of such aninterface.

Steinemann et al205 discussed interfacial bondingand distinguished between contributions to interfa-cial strength from surface roughness and bone-implant interlocking versus intrinsic biomaterial-biologic attachment. They pointed out that in theabsence of roughness to provide interdigitationbetween an implant and bone, the intrinsic strengthof an interface in shear or tension must be deter-mined by the intrinsic biomaterial-biologic attach-ment. Now, based on more modern test data, if thisattachment indeed consists of a cement line, then itfollows that the intrinsic strength of the bone-implant interface will be determined by the strengthof a cement line. However, in view of the structureof natural cement lines and their weakness in nor-mal bone, as discussed below, it follows that theintrinsic strength of bone-implant interfaces willthen be rather small when compared to the strengthof fully mineralized bone.

In attempts to quantify interfacial strength, manyinvestigators have devised a variety of methods.Brånemark206 reviewed 38 so-called pull-out orpush-out “shear” tests of implants placed intranscortical or intramedullary sites, 24 removaltorque tests, and 18 miscellaneous tests, includingtensile, crack propagation, and energy absorption.Key findings from that work were: (1) as roughnessof the implant surface decreases, interfacial shearstrength decreases; and (2) for tests of smooth-sur-faced implants, interfacial shear strengths do notreach the full shear strength of bone (which is about


BRUNSKI ET AL




68 MPa207), and interfacial tensile strengths do notreach the full tensile strength of bone (which isabout 100 to 150 MPa207).

The role of surface roughness in shear strengthtests is well demonstrated in the work of Wong etal,208 who measured shear strengths as a function ofthe roughness of different implant biomaterials inthe same animal model. For implant surfaces withdifferent roughness but the same implant shape,site, and implantation time, the results show anearly linear relationship between the “pushoutload to failure” and surface roughness in microns.Notably, test surfaces included commercially pureTi, Ti-6Al-4V, Ti-6Al-7Nb, and HA, the latter hav-ing the largest roughness. Moreover, as roughnessdecreased to zero, the push-out force alsoapproached zero (as would the shear strength com-puted in stress units, MPa, accounting for the inter-facial area). The highest shear strength in the studywas about 7 MPa, for HA with a surface roughnessof about 6 to 7 µm.

Tensile data from a number of studies are not ascomplete as the shear studies of Wong et al, owing toa wide range of animal models and implantshape/size among the studies. Steinemann et al’s205

“rough” and “plasma-coated” Ti disks (roughness ofabout 20 µm) gave bone-implant interfacial tensilestrengths of 1 to 4 MPa after 100 days of implanta-tion in the ulnae of Macaca speciosa. Using disk-shaped implants of surface roughness Ra = 0.77 µm inperiosteal sites in dog femora for 301 days, Taylor etal209 reported interfacial tensile strengths of 0.27 ±0.47 MPa for Ti-6Al-4V and 2.7 ± 0.82 MPa forHA-coated Ti-6Al-4V. Aspenberg and Skripitz210

recently reported mean tensile strengths of 0.01 MPafor untreated cp Ti and 0.29 MPa for alkali-treatedTi in rat tibiae when the surface roughness was Ra =0.48 µm. For dense HA disks with Ra = 0.32 µm,Edwards et al211 reported tensile strengths of 0.15 ±0.11 MPa at 55 days and 0.85 ± 0.55 MPa at 88 daysin a rabbit tibial model. Other investigations212,213

with dense HA polished to 400 grit (about 11 µm)reported tensile strengths of 1.32 to 1.5 MPa.

Note that in all of these tensile studies, the mea-sured bone-implant tensile strengths were all lessthan about 4 MPa, which is much less than the tensilestrength of fully mineralized bone (which, as notedbefore, is on the order of 100 to 150 MPa207), evenfor so-called bone-bonding ceramics such as HA.

While more conclusive data are not yet availableon the tensile and shear strengths of a naturalcement line in bone, it could be that the above-noted data are as close as can be obtained to suchdata for the biomaterial interface. And in any case,the biochemical makeup and structure of cement

lines would seem to preclude a very large tensile orshear strength. That is, given that a cement line—naturally occurring in bone or at a bone-implantinterface—consists of a thin layer of collagen-defi-cient, incompletely mineralized tissue betweenimplant and bone, it is unlikely that this could con-fer a high tensile or shear strength at an interface,especially with the limit of a smooth implant surface.Even if somehow the cement line’s attachment tothe implant surface per se was strong, the strengthof the material within the cement line could not bevery high, given its composition and structure.While more data need to be gathered about cementlines, it is probable that its intrinsic strength wouldset an upper boundary for the strength to beexpected at a perfectly smooth bone-implant inter-face. To date, there are no data suggesting that thisupper boundary exceeds a few MPa at best.

The Meaning of Interfacial Strength Data in Biome-chanical Models of Implants in Bone. The biomechani-cal relevance of data about intrinsic interfacestrength can be illustrated in an FE study214 thatallowed for interfacial “bond” failure according toinput data from the literature about the shear andtensile strengths of bone-implant interfaces. Themodel used a typical value of 1 MPa for both theshear and tensile strengths of the bone-titaniuminterface, which was assumed to consist of a cementline of essentially zero thickness in the computersimulation. The interfacial strength values wereused in a failure algorithm for the interface. As theimplant was loaded incrementally from 0 to 300 N,stresses at the interface were computed and com-pared with the defined interface failure functioninvolving the 1 MPa limit. The model revealedwhen and where the interface started to fail bydebonding between implant and bone. For the testcase considered, cracks started to open early in theloading, at just 30 N, forming at the thread cuspsnear the apical portion of the implant. As loadingcontinued up to 300 N, the cracks widened and alsostarted at new places along the interface, untilfinally at 300 N, the interface consisted of someregions remaining in (compressive) contact becauseof the interlocking geometry of the threads, andother regions where gaps had opened between theimplant and bone after the interfacial tensile andshear strength had been exceeded.

Another relevant point from the above type ofmodel is that the stresses and strains in bone adjacentto the interface depend strongly upon whether ornot an interfacial “bond” does or does not exist. Forexample, compared to a bonded interface, strainswere about 3 times larger in the same finite elementmodel without bonding at the interface.215–217


BRUNSKI ET AL




Therefore, it should be recognized that the presenceor absence of interfacial bonding influences interfa-cial stress and strain conditions—and potentially anybone cell reactions that might be linked somehow tothe stresses and strains.

Significance of Stress and Strain in InterfacialBone: Excessively High Strains. The motivation forstudying stress and strain in interfacial tissue comesfrom an effort to define safe versus dangerous load-ing conditions in the bone. A number of stud-ies218–221 have reported that overload of the bone-implant interface can be a factor in marginal boneloss around implants. Typical clinical symptoms ofoverload include repeated prosthetic screw loosen-ing or fracture and loss of crestal bone. A key bio-mechanical research question is the mechanism ofoverload failure of a bone-implant interface and aclarification of safe versus dangerous stress-strainstates in bone.

The likelihood of a single-cycle overload failure ofa bone-implant interface is not further consideredhere as a source of the clinically observed overload,although it is one potential type of biomechanicaloverload, especially for implants in poor quality can-cellous bone. However, no single-cycle failures haveclearly been documented in vivo as yet. Hence, it ismore relevant to discuss overload failures that havebeen linked to cyclic loading conditions over a longerperiod of time. Given the literature reviewed below,there is evidence supporting the hypothesis thatfatigue microdamage can occur in interfacial bonearound a heavily loaded dental implant, and that thismicrodamage triggers bone remodeling (and possiblyalso modeling) that may not be able to keep pacewith accumulating damage as loading continues—asituation that predisposes the bone to additionalfatigue damage and, eventually, a net loss of bone andimplant failure. The following explores the researchrelating to this hypothesis and other possibilities.

Damage to Bone. Monotonic compressive and ten-sile tests of both cortical and trabecular bone in thelaboratory (using regular specimens) have revealedthat bone can sustain various forms of mechanicaldamage when strains approach the yield point. Forcortical bone, macroscopic evidence of yieldingoccurs at a strain of about 0.75%, although there isother evidence, eg, by acoustic emission, of yieldingat lower strains such as 0.5%.202,222 For trabecularbone, the yield strain is more difficult to pinpointbecause the nominal strain of an entire specimencan differ from the strain fields that develop in indi-vidual trabeculae, as illustrated in computer modelsof trabecular bone.223 For samples of bovine trabec-ular bone, Wachtel and Keaveny224 saw transversecracks, shear bands, parallel cracks, and complete

trabecular fractures within samples tested to 2.5%strain in compression. This was 3 times more dam-age than was observed in the “yield group,” whichwas tested up to 1% strain and then unloaded.

In tensile and compressive load-controlledfatigue tests of human cortical bone, Pattin et al225

reported secant modulus degradation and increasesin cyclic energy dissipation in specimens that werecycled at “critical damage strain thresholds” of0.25% in tension and 0.4% in compression. Like-wise, trabecular bone from bovine tibiae failed infatigue when cyclically compressed for about 40,000cycles at 0.4% strain.226 Other reviews of the natureof damage that occurs in bone during the processesof yielding and fatigue appear elsewhere.201,227–229

For bone, it is known that cracks, delaminations,shear bands, and other phenomena yet to be clari-fied comprise the nature of the microdamage seenin the microscope.

A promising new avenue of research into strainin bone comes from the work of Nicolella etal,228,229 who used digital image correlation to mea-sure strains at the microstructural level in labora-tory specimens of bone. Significantly, while a straingauge measured a nominal strain of about 0.15% ina bone specimen, the microstructural level strainvalues were as large as 3.5% at various locations inthe microstructure. Note that 3.5% strain is over 20times the nominal value measured by the straingauge. This work emphasizes that values of nominalstrain in bone—as might be calculated from sampledimensions or strain gauges—may severely underes-timate actual strains occurring at the level of thebone microstructure. In other words, it is possiblethat damage might be occurring in bone eventhough routine strain analyses had estimated thatthe nominal strain values were “safe.”

Damage, Bone Remodeling, and the Possible Relation-ship to “Overload.” Mori and Burr183 established thatmicrodamage to bone stimulates repair by boneremodeling. Recent research in osteoporosis230,231

indicates that: (1) microdamage can contribute toincreased bone fragility and fracture risk; and (2) fractures can develop as the result of a viciouscycle (positive feedback mechanism) involving dam-age, remodeling-induced porosity, weakening ofbone, further damage, and so on. That is, it ishypothesized that positive feedback occurs whenbone remodeling (A-R-F) tries to repair a damagedsite in bone, but in so doing, causes increased poros-ity and a vicious cycle of worsened strain state, moredamage, more remodeling, more porosity, and soon, until failure. Conceptually, this idea could alsobe relevant to the bone-implant interface as follows.If porosity develops because of A-R- at a damaged


BRUNSKI ET AL




region, and loading continues on this already dam-aged but now remodeling region, more damage mayoccur in nearby bone as the result of the remodel-ing-induced porosity. This would then be followedby more remodeling, more porosity from the A-R-step, and so on. Although this hypothesis has notbeen conclusively tested and established in the con-text of dental implant overload, a number of clinicaland animal studies support it.

An initial study232 tried to develop an animalmodel in which controlled loads could be applied toosseointegrated implants as a means to study over-loading and related interfacial tissue responses. Thiswork used conventional 3.75-mm-diameter � 7-mm-long cp Ti Brånemark implants in dog bones.Implants in dog radii and mandibles were allowed toheal for 4 months (radii) and 7 months (mandibles)before loading by cyclic axial compression in themandibular sites (square wave, amplitude 100 N, 0.5cycles/sec, 500 cycles/day for 5 days), and cyclicaxial tension of 50 to 100 N on implants in theradial sites. The results revealed no statistically sig-nificant differences between loaded and controlinterfaces. An explanation for the lack of a differ-ence was probably the relatively low level of loading:subsequent finite element analyses suggested thatmaximum axial loads of 100 N may not have createdstrains larger than the yield strain over appreciableregions of interfacial bone for 7-mm Brånemarkimplants in these bone sites. Alternatively, it may bethat the time between loading and histologic analy-ses (20 days) was too short relative to the timeneeded for bone to develop a measurable responsevia A-R-F; recall that sigma for A-R-F in dog boneis approximately 3 months. Third, the anatomy ofmandibular and radial bone varied appreciablyamong the dogs, contributing to large standarddeviations in histologic data.

Hoshaw et al188 improved on the above study byusing more uniform, mid-diaphyseal bone of dogtibiae; a larger axial load of 300 N (tensile); a longerhealing time before loading (1 year); more detailedfinite element and strain gauge analyses of the bone;and a longer time between loading and histologicanalysis (ie, 3 months, which is about 1 sigma in thedog). The main finding was significantly more cre-stal bone loss in the loaded versus unloaded controlgroup. The results were consistent with the follow-ing proposed mechanism: (1) principal strains inexcess of bone’s yield point (approximately 0.5%)occurred over substantial regions of interfacial cre-stal bone during each of the 2,500 cycles of loadingat 300 N; (2) these large strains damaged the inter-facial bone and stimulated a cycle of remodeling inthe cortex and resorptive modeling on the peri-

osteal surface; and (3) modeling and remodeling(and more cycles of loading) eventually createdcrestal bone loss that may have been exacerbated bydowngrowth of periosteal soft tissues. The studycould not experimentally confirm that microdamageoccurred in interfacial bone because of implantloading, although this was quite likely in view of thestrain values predicted by FE models (often largerthan 4%) and the fact that 300 N was about 25% ofthe ultimate pull-out load for implants in dogtibiae.233

Isidor’s studies234,235 in monkeys were consistentwith the above hypothesis of overload. Using 5 cpTi screw-shaped implants per mandible in each of 4monkeys, Isidor created supra-occlusal contact ofthe prostheses and “excessive occlusal load” (nototherwise quantified) in the lateral rather than theaxial direction, on 2 implants per animal, starting atabout 8 months after implantation and continuingfor 18 months. Five of the 8 implants with excessiveocclusal load “lost osseointegration” by 4.5 to 15.5months after the loading commenced, while theother implants, with plaque accumulation but noloading, remained osseointegrated. Loss of osseoin-tegration was attributed to “fatigue microfracturesin the bone exceeding the repair potential.”

In an analysis of a clinical case in which overloadwas suspected, Prabhu and Brunski216,217 used FEanalyses to evaluate a case in which 2 Brånemarkimplants supported a prosthesis in the mandibularmolar region of a human with a positive history ofbruxing. The implants had healed for 6 monthsbefore loading. After about 2 to 3 months of func-tion, crestal bone loss was noted radiographicallyaround the mesial implant, which eventually frac-tured. Three-dimensional FE analysis predictedhigh strains—in excess of 1%—at the crestal regionof the mesial implant’s interface, especially for thecase of medium-size bite forces (eg, 250 N) on themesial cantilever of the prosthesis. Again, theseresults support, but do not prove, an etiology ofoverload failure by the positive feedback mechanismoutlined previously.

Computer Simulation of Overload at a Bone-ImplantInterface. To make a more detailed microstructuralexamination of mechanisms of interface failure sug-gested by the above work, Brunski and Yang236 didan FE simulation that allowed for bone microdam-age and A-R-F. Therefore, a key feature of themodel was that it allowed changes to the microstruc-ture related to the actual process of remodeling byA-R-F. The model was set up in such a way to simu-late the previously noted experiments of Hoshaw etal.187 The model included an array of 25 osteons inhealed interfacial bone. Upon implant loading, the


BRUNSKI ET AL




model computed principal strains in bone, and ifstrains exceeded a damage limit—set at 0.5% princi-pal strain (for justification, see previously noted liter-ature on bone damage)—then A-R-F was started atthose sites. Since osteoclast (OC) recruitment andactivation is the first step in A-R-F, it was assumedthat OCs must come from a blood supply, which inthis model was the Haversian canal of each osteon.As it turned out, the osteons were also frequentlysites of high strain in the FE model, because of thestress-concentrating effect of the Haversian canal.Onset of A-R- was simulated in the FE model byremoving bone matrix inside those osteons wherestrains exceeded 0.5%. This resorption created“holes” in the bone, which in turn changed the strainfields in subsequent load steps, which in turn some-times created more regions of strain above 0.5%.This caused more damage, more A-R, more poros-ity, and so on. In some cases, bone eventually weak-ened so much as a result of the ensuing porosity thatit failed completely in the next load cycle. Therefore,this FE simulation demonstrated a conceivablemechanism of overload by a positive feedback mech-anism at an osseointegrated interface. A similar FEmodel has been developed for analysis of immediateloading.237 It remains to be fully established whetherthese suggested mechanisms are in fact the in vivomechanisms.

Significance of Stress and Strain in InterfacialBone: “Wolff’s Law”? In addition to assessment ofthe effects of excessively large stress and strain inbone, there is the overarching question of the bio-logic relevance of stress and strain. “Wolff’s Law”often emerges when discussing this topic. This“law” has been translated as238: “Every change in theform and function of . . . bone(s) or of their functionalone is followed by certain definite changes in theirinternal architecture, and equally definite secondaryalterations in their external conformation, in accor-dance with mathematical laws.”238p225

This law has been recast in various ways usingmodern engineering terms. For example, some pre-sume that bone has a control system (eg, Frost’s“mechanostat,”239 analogous to a thermostat) thatacts via modeling and remodeling to maintain con-stancy of the mechanical environment of cells whenexternal loading conditions change. Or according toMattheck,240 bone is viewed as a biomechanicallyoptimized structure, with the design being governedby “probably only one single design rule . . . theaxiom of uniform stress,” which “states that onaverage, over time, stress acts uniformly over thesurface of components.”

These ideas have become so intuitively reason-able, attractive, and ingrained in the literature that

many researchers have tended to accept them a pri-ori as unquestionable biologic fact to explain resultsof loading experiments in bone with and withoutimplants present. However, at times, there has beena distressing disregard for confounding experimen-tal variables or for alternative explanations forresults in experiments; Bertram and Swartz’s critiqueis excellent on this point.241 In addition, there hasalso been a tendency to forget Chomsky’s adage,242

“It is a merit of a theory to be proven false,” ie, the-ories should be crafted so that they can actually betested. But as Currey has remarked202: “For manyworkers, it seems only necessary to show that boneis adapting, invoke Wolff’s Law, and depart, con-scious of a day’s work well done.” A full history andcritique of Wolff’s Law goes far beyond the scope ofthis paper; for additional insight see Huiskes,243

Currey,244 and Martin et al.238 The more limitedgoal of this section is to emphasize that many work-ers with dental implants have tacitly accepted aWolff ’s Law concept for bone around dentalimplants without directly addressing some concep-tual difficulties inherent in this approach.

First, the key scientific challenge is to establishwhether, in fact, bone has a control system to main-tain mechanical homeostasis or optimal design, and ifit does, to establish how it works both in normalbone and in bone around implants, specifically oraland maxillofacial implants. Among other tasks, it isnecessary to identify the following: What quantity isbeing controlled by the control system? What are thesensors of bone’s mechanical environment? If thecells are the sensors (since they are the only livingthings in bone), exactly what do they sense? And afterhaving sensed something, how do the cells controlthe quantity that is supposed to be controlled? Howquickly does this system work? A difficult problem!

An initial conceptual problem with the wholeapproach is that, when discussing the questionsabove in the context of oral (or orthopedic) implants,researchers often ignore the possibility that theanswers to the questions will depend strongly on thebiologic state of the bone. For example, while thereseems to be little question that bone in the process ofhealing can be influenced by loading—evidently, anexcellent example is distraction osteogenesis245—itdoes not necessarily follow that all types of bone—embryonic bone; bone in the developing skeleton;bone in the undisturbed, mature skeleton; bonearound immediately loaded versus delayed loadedimplants; or bone having different microstructure(eg, woven bone, plexiform bone, Haversian bone)—possess the same ability (and the same control sys-tem) to adapt to mechanical conditions. As Bertramand Swartz have aptly noted in their review241:


BRUNSKI ET AL




We suggest that bone biology would benefit froman awareness that observations . . . gatheredunder the explanatory umbrella of Wolff’s Lawmay reflect a plurality of effects which, in somecases, are related to the phenomenon suggestedby Wolff only in their general morphologicalindications.241p246

Another problem surfaces related to stress shield-ing (stress protection), an idea that has arisen primar-ily in connection with proximal bone loss aroundfemoral prostheses or bone plates in orthope-dics.246,247 It is hypothesized that bone loss occurs incertain regions near orthopedic implants because thelocal strain fields in bone become significantly lowerthan (or at least very different from) the values thatused to exist in normal bone before the implant wasplaced. The idea has now been extended to dentalimplants.248–253 This work suggests that crestal boneloss is possible around dental implants because ofabnormally low strains and stress shielding. However,a key point in the dog studies with porous-coatedimplants249 is that the bone sites in those studies wereloaded immediately or very soon after implantation;some were loaded at 4 to 8 weeks after implantation.In those situations the authors themselves noted thatwoven bone existed at the interfaces.

The problem is that whatever bone responsesmight have occurred in these animal experiments,they must have occurred within an environment ofactive bone healing in both woven bone and dam-aged lamellar bone. Therefore, the bone-implantinterface in this type of experiment is a highly com-plicated milieu, in which it would be difficult toseparate out the events related to intrinsic healingfrom events ascribed to “stress shielding.” In fact,the work by Perren et al254 strongly disputes theentire notion that stress shielding is the reason forporosis beneath bone plates; these authors explainthat impaired vascularity resulting from the surgeryand plate placement is a more likely explanation.Certainly the issue of cause and effect is key.

Another potential confusion can arise in studiesof stress shielding. Sometimes researchers seem toequate the idea of stress shielding, which is reputedto be something that happens in stress-shieldedbone around implants, with the idea of disuse atro-phy, which refers to events that happen in bone dur-ing prolonged bed rest or weightlessness duringspace flight.255 It is sometimes implied that theunderlying stimulus for bony change is the same inboth phenomena, ie, a diminished stress-strain statein bone as compared with normal conditions. How-ever, this implication can be questioned, becausedisuse atrophy evidently happens after bone starts

off in a presumed steady state and is then perturbednoninvasively by removal or diminution of load, asin the examples of prolonged bed rest or spaceflight. On the other hand, things are quite differentin bone around a recently placed implant, where thebone is assuredly not in a biologic steady state whenthe implant is placed and possibly also loaded.Indeed, bone around a freshly placed implant istraumatized by implant surgery and will be in anactive state of healing for many months aftersurgery. In view of this fact, it is not obvious thatbone loss from “stress shielding” and disuse atrophyhave similar root causes, namely understressing ofbone compared to “normal” levels.

Yet another problematic aspect of Wolff’s Law asapplied to bone around dental implants is the ideathat bone cells can somehow “know” what strainstate (or other mechanical quantity that is supposedto be controlled) is appropriate for a particular loca-tion in bone, and that if the strain state is inappropri-ate, the cells can change their bony environmentuntil the “appropriate” strain (or other signal) is re-established by the putative control system. As notedbefore, this same control-system concept is implicitin nearly all of the mathematical theories and com-puter simulations of bone adaptation to mechanicalloading.256 While the idea is intuitively appealing(and also underlies explanations of stress shieldingand disuse atrophy), it is enlightening to recall whatactually happens when implants are placed into bone.

Implantation surgery damages and sometimeskills bone, even when the gentlest procedures areused. Wound healing eventually produces new cellsand matrix, which are indeed new to the site, havingcome from marrow or other nearby tissue. Thewoven bone that forms in gaps around an implant,as well as the damaged bone that remodels nearby, ismade up of new cells and matrix that never existedexactly at that bony site before. This fact makes itdifficult to envision how the new cells can somehow“know” what used to exist at that bony site, althoughcommunication with remaining cells is a possibility.Moreover, it becomes an even greater leap of faithto claim that the new bone cells in woven bone, aswell as the new osteoclasts and osteoblasts in remod-eling lamellar bone around an implant, somehow“know” the appropriate strains (or related quanti-ties) for exactly that site—that is, they “know,” forinstance, that they are in a region that used to havenatural teeth loaded in a certain way, or that they arein the anterior or posterior region of the mandible.This caution applies with even greater force to cellsin bone surrounding an implant used in guided boneregeneration or in a bone-grafting procedure, inwhich bone from a fibula or iliac crest is grafted to


BRUNSKI ET AL




the mandible, for example. How can bone cells inthe graft (if they survive) “know” that they are nowin the mandible or maxilla?

For another example of difficulty in applying boneadaptation theories to in vivo tissue responses arounddental implants, consider the work of Rubin andMcLeod,257 who have conducted numerous experi-ments on bone adaptation using the turkey ulnamodel and other systems. Their work focuses on reg-ulating factors and morphologic goals of bone’s adap-tive process. These workers suggest that as a goal ofskeletal remodeling, “minimization of strain per senot only is not achieved, but may not be desired.”While they acknowledge that peak strain magnitudesare similar across animal species and bone types (eg,0.2 to 0.35%), and that strains above the yield strainwould be unsafe, Rubin and McLeod emphasize that:

. . . it is difficult to imagine a biologic processthat can quantify its proximity to deleteriousstrain levels, and subsequently adjust the tissue’smass to avoid them. Physiologically, it seems onlyreasonable that the resident cell populationresponsible for skeletal adaptation responds onlyto functionally induced strains, not the potentialstrain it might see should an aberrant loadingevent occur.257p101–102

Therefore, while not discounting that high strainlevels can damage bone, Rubin and McLeod empha-size that bone adaptation to lower levels of strainmight have to be mediated by signals other than justthe strain magnitude per se. After all, strain at apoint in a continuum (which raises the question ofwhether a continuum model of bone is even appro-priate) is not a scalar quantity but rather a tensorquantity, represented by a 3 � 3 matrix of values.For example, Rubin and McLeod257 have recentlyexplored an alternative factor that might be moreimportant in regulating bone remodeling, namely,the frequency spectrum of the dynamic strain signal.While it remains to be seen whether the frequencyspectrum of the dynamic strain signal will provemore useful than factors such as strain magnitude,strain rate, strain distribution, strain gradient, etc,for understanding bone reactions around loaded andunloaded implants, it is sobering to note that thesearch continues for more information about theputative control system in bone.

As a final thought about implants, bone, andWolff’s Law, consider the vast array of oral and max-illofacial implants of widely different sizes, shapes,materials, surfaces, and loading histories that havebeen placed in various bone sites in humans overseveral decades. Probably the total number of

implants in humans easily exceeds 1 million. What isperhaps most remarkable about the bone aroundthese 1 million implants is what is not observed.That is, in long-term studies, there have not beenwidespread clinical reports of any conclusive trendsin net bone formation or resorption around themany different implants in humans. Granted therehave been reports about certain load-related bonyreactions during the early healing period, or duringthe later periods, when late “overload” is observed.But apart from these reports of deleterious eventssuch as micromotion during early bone healing, orlate failure by overload (which, as noted, seems to beexplainable in terms of overt damage to bone), clini-cians have not consistently reported anything hap-pening in bone relative to loading. All the moreremarkable about this is the fact that the lack of con-clusive findings occurs in spite of the near-certaintythat 1 million differently shaped implants, under dif-ferent loading conditions in a million differentpatients, must have produced widely different stress-strain states in interfacial bone—certainly much dif-ferent than would have existed in that bone if naturalteeth had remained. Of course, this is not meant tostate that nothing happens in that bone in responseto loading. But the reports do seem to be saying thatwhatever happened in the bone in response to load-ing, the results were not readily noticeable andreported in the general clinical experience. There-fore, for bone around oral and maxillofacialimplants, it may be time to shift from a blind accep-tance of Wolff’s Law and move toward new, moretestable, specific hypotheses. For example, maybebone around oral implants is not sensitive to stressand strain except when it is healing.

SUMMARY

Research in biomaterials and biomechanics hasfueled a large part of the significant revolution asso-ciated with osseointegrated implants. Additional keyareas that may become even more important—suchas guided tissue regeneration, growth factors, andtissue engineering—could not be included in thisreview because of space limitations. All of this workwill no doubt continue unabated; indeed, it is prob-ably even accelerating as more clinical applicationsare found for implant technology and related thera-pies. An excellent overall summary of oral biologyand dental implants recently appeared in a dedi-cated issue of Advances in Dental Research.258

Many advances have been made in the under-standing of events at the interface between bone andimplants and in developing methods for controlling


BRUNSKI ET AL




these events. However, several important questionsstill remain. What is the relationship between tissuestructure, matrix composition, and biomechanicalproperties of the interface? Do surface modifica-tions alter the interfacial tissue structure and com-position and the rate at which it forms? If surfacemodifications change the initial interface structureand composition, are these changes retained? Dosurface modifications enhance biomechanical prop-erties of the interface? As current understanding ofthe bone-implant interface progresses, so will devel-opment of proactive implants that can help promotedesired outcomes.

However, in the midst of the excitement bornout of this activity, it is necessary to remember thatthe needs of the patient must remain paramount. Itis also worth noting another as-yet unsatisfied need.With all of the new developments, continuing edu-cation of clinicians in the expert use of all of theseresearch advances is needed. For example, in thearea of biomechanical treatment planning, there arestill no well-accepted biomaterials/biomechanics“building codes” that can be passed on to clinicians.Also, there are no readily available treatment-plan-ning tools that clinicians can use to explore “what-if” scenarios and other design calculations of thesort done in modern engineering. No doubt suchapproaches could be developed based on materialsalready in the literature, but unfortunately much ofwhat is done now by clinicians remains empirical. Aworthwhile task for the future is to find ways tomore effectively deliver products of research intothe hands of clinicians.

ACKNOWLEDGMENTS

The authors would like to acknowledge contributions fromtheir recent other articles in Biomaterials and Advances in Den-tal Research.

REFERENCES

1. Dunlap J. Implants: Implications for general dentists. DentEcon 1988;78:101–112.

2. Graves E. Vital and Health Statistics, Detailed Diagnosesand Procedures, National Hospital Discharge Survey, 1993.Hyattsville, MD: National Center for Health Statistics,1995.

3. Dym CL. Engineering Design: A Synthesis of Views. Cam-bridge: Cambridge University Press, 1994.

4. Eide AR, Jenison RD, Mashaw LH, Northup LL. Engineer-ing Fundamentals and Problem Solving. New York:McGraw-Hill, 1986.

5. Brunski JB. Biomechanical factors affecting the bone-dentalimplant interface. Clin Mater 1992;10:153–201.

6. Brånemark P-I, Engstrand P, Ohrnell L-O, Gröndahl K,Nilsson P, Hagberg K, et al. Brånemark Novum: A newtreatment concept for rehabilitation of the edentulousmandible. Preliminary results from a prospective clinical fol-low-up study. Clin Implant Dent Rel Res 1999;1:2–16.

7. Black J. Biological Performance of Materials. New York:Marcel Dekker, 1992.

8. Andrade JD. Principles of protein adsorption. In: AndradeJD (ed). Surface and Interfacial Aspects of Biomedical Poly-mers. New York: Plenum Press, 1985:1–80.

9. Horbett TA, Brash JL. Proteins at interfaces: Current issuesand future prospects. In: Brash JL, Horbett TA (eds). Pro-teins at Interfaces: Physiochemical and Biochemical Studies.Washington, D.C.: American Chemical Society, 1987:1–33.

10. Sundgren J-E, Bodo P, Lundstrom I. Auger electron spec-troscopic studies of the interface between human tissue andimplants of titantium and stainless steel. J Colloid InterfaceSci 1986;110:9–20.

11. Ask M, Lausmaa J, Kasemo B. Preparation and surface spec-troscopic characterization of oxide films on Ti6Al4V. ApplSurf Sci 1989;35:283–301.

12. Lausmaa J, Kasemo B, Rolander U, Bjursten LM, EricsonLE, Rosander L, Thomsen P. Preparation, surface spectro-scopic and electron microscopic characterization of titaniumimplant materials. In: Ratner BD (ed). Surface Characteriza-tion of Biomaterials. Amsterdam: Elsevier, 1988:161–174.

13. Sundgren JE, Bodo P, Lundstrom I, Berggren A, Hellem S.Auger electron spectroscopic studies of stainless-steelimplants. J Biomed Mater Res 1985;19:663–671.

14. Williams DF. Tissue reaction to metallic corrosion productsand wear particles in clinical orthopaedics. In: Williams DF(ed). Biocompatibility of Orthopaedic Implants. Boca Raton,FL: CRC Press, 1982:231–248.

15. Hennig FF, Raithel HJ, Schaller KH, Dohler JR. Nickel-,chrom- and cobalt-concentrations in human tissue and bodyfluids of hip prosthesis patients. J Trace Elem ElectrolytesHealth Disord 1992;6:239–243.

16. Dorr LD, Bloebaum R, Emmanual J, Meldrum R. Histo-logic, biochemical, and ion analysis of tissue and fluidsretrieved during total hip arthroplasty. Clin Orthop 1990;261:82–95.

17. Bartolozzi A, Black J. Chromium concentrations in serum,blood clot and urine from patients following total hiparthroplasty. Biomaterials 1985;6:2–8.

18. Michel R, Nolte M, Reich M, Loer F. Systemic effects ofimplanted prostheses made of cobalt-chromium alloys. ArchOrthop Trauma Surg 1991;110:61–74.

19. Jacobs JJ, Skipor AK, Patterson LM, Hallab NJ, PaproskyWG, Black J, Galante JO. Metal release in patients whohave had a primary total hip arthroplasty. A prospective,controlled, longitudinal study. J Bone Joint Surg [Am] 1998;80:1447–1458.

20. Friberg L, Nordberg GF, Vouk VB. Handbook on the Toxi-cology of Metals. Amsterdam: Elsevier/North-Holland,1979.

21. Merritt K, Brown SA. Distribution of cobalt chromium wearand corrosion products and biologic reactions. Clin Orthop1996;S233–S243.

22. Thompson GJ, Puleo DA. Effects of sublethal metal ionconcentrations on osteogenic cells derived from bone mar-row stromal cells. J Appl Biomater 1995;6:249–258.

23. Thompson GJ, Puleo DA. Ti-6Al-4V ion solution inhibitionof osteogenic cell phenotype as a function of differentiationtimecourse in vitro. Biomaterials 1996;17:1949–1954.


BRUNSKI ET AL




24. Nichols KG, Puleo DA. Effect of metal ions on the forma-tion and function of osteoclastic cells in vitro. J BiomedMater Res 1997;35:265–271.

25. Brunski JB. Influence of biomechanical factors at the bone-biomaterial interface. In: Davies JE (ed). The Bone-Bioma-terial Interface. Toronto: University of Toronto Press, 1991:391–405.

26. Pilliar RM. Quantitative evaluation of the effect of move-ment at a porous coated implant-bone interface. In: DaviesJE (ed). The Bone-Biomaterial Interface. Toronto: Univer-sity of Toronto Press, 1991:380–387.

27. Szmukler-Moncler S, Salama H, Reingewirtz Y, DubruilleJH. Timing of loading and effect of micromotion on bone-dental implant interface: Review of experimental literature. JBiomed Mater Res 1998;43:192–203.

28. Linder L. High-resolution microscopy of the implant-tissueinterface. Acta Orthop Scand 1985;56:269–272.

29. Nanci A, McCarthy GF, Zalzal S, Clokie CML, WarshawskyH, McKee MD. Tissue response to titanium implants in therat tibia: Ultrastructural, immunocytochemical and lectin-cytochemical characterization of the bone-titanium inter-face. Cells Mater 1994;4:1–30.

30. Davies JE, Lowenberg B, Shiga A. The bone-titanium inter-face in vitro. J Biomed Mater Res 1990;24:1289–1306.

31. Murai K, Takeshita F, Ayukawa Y, Kiyoshima T, Suetsugu T,Tanaka T. Light and electron microscopic studies of bone-titanium interface in the tibiae of young and mature rats. JBiomed Mater Res 1996;30:523–533.

32. Nanci A, McKee MD, Zalzal S, Sakkal S. Ultrastructuraland immunocytochemical analysis of the tissue response tometal implants in the rat tibia. In: Davidovitch Z, Mah J(eds). Biological Mechanisms of Tooth Eruption, Resorptionand Replacement by Implants. Boston: Harvard Society forthe Advancement of Orthodontics, 1998:487–500.

33. Ayukawa Y, Takeshita F, Inoue T, Yoshinari M, Shimono M,Suetsugu T, Tanaka T. An immunoelectron microscopiclocalization of noncollagenous bone proteins (osteocalcinand osteopontin) at the bone-titanium interface of rat tibiae.J Biomed Mater Res 1998;41:111–119.

34. Sodek J, Chen J, Kasugai S, Nagata T, Zhang Q, McKeeMD, Nanci A. Elucidating the functions of bone sialopro-tein and osteopontin in bone formation. In: Slavkin H, PriceP (eds). Chemistry and Biology of Mineralized Tissues.Amsterdam: Elsevier, 1992:297–306.

35. Butler WT, Ritchie H. The nature and functional signifi-cance of dentin extracellular matrix proteins. Int J Dev Biol1995; 39:169–179.

36. McKee MD, Nanci A. Osteopontin at mineralized tissueinterfaces in bone, teeth, and osseointegrated implants:Ultrastructural distribution and implications for mineralizedtissue formation, turnover, and repair. Microscop Res Tech1996;33:141–164.

37. Brunski J. In vivo bone response to biomechanical loading atthe bone-dental implant interface. Adv Dent Res 1999;13:99–119.

38. Rittling SR, Matsumoto HN, McKee MD, Nanci A, An X,Novick KE, et al. Mice lacking osteopontin show normaldevelopment and bone structure but display altered osteo-clast formation in vitro. J Bone Miner Res 1998;13:1101–1111.

39. Cooper LF, Masuda T, Yliheikkila PK, Felton DA. General-izations regarding the process and phenomenon of osseoin-tegration. Part II. In vitro studies. Int J Oral MaxillofacImplants 1998;13:163–174.

40. Gomi K, Lowenberg B, Shapiro G, Davies JE. Resorption ofsintered synthetic hydroxyapatite by osteoclasts in vitro. Bio-materials 1993;14:91–96.

41. Gorski JP. Is all bone the same? Distinctive distributions andproperties of non-collagenous matrix proteins in lamellar vs.woven bone imply the existence of different underlyingosteogenic mechanisms. Crit Rev Oral Biol Med 1998;9:201–223.

42. Nanci A. Content and distribution of noncollagenous matrixproteins in bone and cementum: Relationship to speed offormation and collagen packing density. J Struct Biol 1999;126:256–269.

43. Davies JE. In vitro modeling of the bone/implant interface.Anat Rec 1996;245:426–445.

44. Nanci A, Zalzal S, Gotoh Y, McKee MD. Ultrastructuralcharacterization and immunolocalization of osteopontin inrat calvarial osteoblast primary cultures. Microscop Res Tech1996;33:214–231.

45. Irie K, Zalzal S, Ozawa H, McKee MD, Nanci A. Morpho-logical and immunocytochemical characterization of primaryosteogenic cell cultures derived from fetal rat cranial tissue.Anat Rec 1998;252:554–567.

46. Kasemo B, Lausmaa J. Surface science aspects on inorganicbiomaterials. CRC Crit Rev Biocomp 1986;2:335–380.

47. Ito Y, Kajihara M, Imanishi Y. Materials for enhancing celladhesion by immobilization of cell-adhesive peptide. J Bio-med Mater Res 1991;25:1325–1337.

48. Baier RE. Surface properties influencing biological adhe-sion. In: Manly RS (ed). Adhesion in Biological Systems.New York: Academic Press, 1970:14–48.

49. Baier RE, Meyer AE. Implant surface preparation. Int J OralMaxillofac Implants 1988;3:9–20.

50. Hamamoto N, Hamamoto Y, Nakajima T, Ozawa H. Histo-logical, histocytochemical and ultrastructural study on theeffects of surface charge on bone formation in the rabbitmandible. Arch Oral Biol 1995;40:97–106.

51. Krukowski M, Shively RA, Osdoby P, Eppley BL. Stimula-tion of craniofacial and intramedullary bone formation bynegatively charged beads. J Oral Maxillofac Surg 1990;48:468–475.

52. Jarcho M. Calcium phosphate ceramics as hard tissue pros-thetics. Clin Orthop 1981; 157:259–278.

53. Dhert WJ. Retrieval studies on calcium phosphate-coatedimplants. Med Prog Technol 1994;20:143–154.

54. Wennerberg A, Bolind P, Albrektsson T. Glow-dischargepretreated implants combined with temporary bone tissueischemia. Swed Dent J 1991;15:95–101.

55. Geesink RG, Hoefnagels NH. Six-year results of hydroxyap-atite-coated total hip replacement. J Bone Joint Surg [Br]1995;77:534–547.

56. Bauer TW, Taylor SK, Jiang M, Medendorp SV. An indirectcomparison of third-body wear in retrieved hydroxyapatite-coated, porous, and cemented femoral components. ClinOrthop 1994;298:11–18.

57. Bloebaum RD, Beeks D, Dorr LD, Savory CG, DuPont JA,Hofmann AA. Complications with hydroxyapatite particu-late separation in total hip arthroplasty. Clin Orthop 1994;298:19–26.

58. Galante JO, Jacobs J. Clinical performances of ingrowth sur-faces. Clin Orthop 1992;276:41–49.

59. Cook SD, Thomas KA, Haddad RJ Jr. Histologic analysis ofretrieved human porous-coated total joint components. ClinOrthop 1988;234:90–101.


BRUNSKI ET AL




60. Engh CA, Bobyn JD, Glassman AH. Porous-coated hipreplacement. The factors governing bone ingrowth, stressshielding, and clinical results. J Bone Joint Surg [Br] 1987;69:45–55.

61. Brunette DM. The effects of implant surface topography onthe behavior of cells. Int J Oral Maxillofac Implants 1988;3:231–246.

62. Chehroudi B, McDonnell D, Brunette DM. The effects ofmicromachined surfaces on formation of bonelike tissue onsubcutaneous implants as assessed by radiography and com-puter image processing. J Biomed Mater Res 1997;34:279–290.

63. Cochran DL, Simpson J, Weber HP, Buser D. Attachmentand growth of periodontal cells on smooth and rough tita-nium. Int J Oral Maxillofac Implants 1994;9:289–297.

64. Martin JY, Schwartz Z, Hummert TW, Schraub DM, Simp-son J, Lankford J, et al. Effect of titanium surface roughnesson proliferation differentiation, and protein synthesis ofhuman osteoblasts-like cells (MG63). J Biomed Mater Res1995;29:389–401.

65. Boyan BD, Batzer R, Kieswetter K, Liu Y, Cochran DL,Szmuckler-Moncler S, et al. Titanium surface roughnessalters responsiveness of MG63 osteoblast-like cells to 1�,25-(OH) 2D3. J Biomed Mater Res 1998;39:77–85.

66. Castellani R, de Ruijter JE, Renggli H, Jansen JA. Responseof rat bone marrow cells to differently roughened titaniumdiscs. Clin Oral Implants Res 1999;10:369–378.

67. Sauberlich S, Klee D, Richter E-J, Hocker H, SpiekermannH. Cell culture tests for assessing the tolerance of soft tissueto variously modified titanium surfaces. Clin Oral ImplantsRes 1999;10:379–393.

68. Buser D, Schenk R, Steinemann S, Fiorellini JP, Fox CH,Stich H. Influence of surface characteristics on bone integra-tion of titanium implants. A histomorphometric study inminiature pigs. J Biomed Mater Res 1991;25:889–902.

69. Wong M, Eulenberger J, Schenk R, Hunziker E. Effect ofsurface topology on the integration of implant materials intrabecular bone. J Biomed Mater Res 1995;29:1567–1575.

70. Ericsson I, Johansson CB, Bystedt H, Norton MR. A histo-morphometric evaluation of bone-to-implant contact onmachine-prepared and roughened titanium dental implants.Clin Oral Implants Res 1994;5:202–206.

71. Wennerberg A, Ektessabi A, Albrektsson T, Johansson C,Andersson B. A 1-year follow-up of implants of differentsurface roughness placed in rabbit bone. Int J Oral Maxillo-fac Implants 1997;12:486–494.

72. Jansen JA, van der Waerden JPCM, Wolke JGC. Histologi-cal and histomorphometrical evaluation of the bone reactionto three different titanium alloy and hydroxyapatite coatedimplants. J Appl Biomater 1993;4:213–219.

73. Caulier H, van der Waerden JPCM, Wolke JGC, Kolle W,Naert I, Jansen JA. A histological and histomorphometricalevaluation of the application of screw-designed calciumphosphate (Ca-P) coated implants in the cancellous maxil-lary bone of the goat. J Biomed Mater Res 1997;35:19–30.

74. Helsingen AL, Lyberg T. Comparative surface analysis andclinical performance studies of Brånemark implants andrelated clones. Int J Oral Maxillofac Implants 1994;9:422–430.

75. Iamoni F, Rasperini G, Trisi P, Simion M. Histomorphomet-ric analysis of a half hydroxyapatite-coated implant inhumans: A pilot study. Int J Oral Maxillofac Implants 1999;14:729–735.

76. Hanson S, Norton M. The relationship between surfaceroughness and interfacial shear strength for bone-anchoredimplants. A mathematical model. J Biomech 1999;32:829–836.

77. Davies JE. Mechanism of endosseous integration. Int JProsthodont 1998;11:391–401.

78. Albelda SM, Buck CA. Integrins and other cell adhesionmolecules. FASEB J 1990;4:2868–2880.

79. Pimentel E. Handbook of Growth Factors. Boca Raton, FL:CRC Press, 1994.

80. Ebert CD, Lee ES, Kim SW. The antiplatelet activity ofimmobilized prostacyclin. J Biomed Mater Res 1982;16:629–638.

81. Massia SP, Hubbell JA. Human endothelial cell interactionswith surface-coupled adhesion peptides on a nonadhesiveglass substrate and two polymeric biomaterials. J BiomedMater Res 1991;25:223–242.

82. Lin HB, Zhao ZC, Garchia-Echeverria C, Rich DH,Cooper SL. Synthesis of a novel polyurethane copolymercontaining covalently attached RGD peptide. J Biomater SciPolym Ed 1992;3:217–227.

83. Liu SQ, Ito Y, Imanishi Y. Cell growth on immobilized cellgrowth factor. I. Acceleration of the growth of fibroblastcells on insulin-immobilized polymer matrix in culturemedium without serum. Biomaterials 1992;13:50–58.

84. Lynch SE, Buser D, Hernandez RA, Weber HP, Stich H,Fox CH, Williams RC. Effects of the platelet-derivedgrowth factor/insulin-like growth factor-I combination onbone regeneration around titanium dental implants. Resultsof a pilot study in beagle dogs. J Periodontol 1991;62:710–716.

85. Sumner DR, Turner TM, Purchio AF, Gombotz WR,Urban RM, Galante JO. Enhancement of bone ingrowth bytransforming growth factor-�. J. Bone Joint Surg [Am]1995;77:1135–1147.

86. Puleo DA. Biochemical surface modification of Co-Cr-Mo.Biomaterials 1996;17:217–222.

87. Endo K. Chemical modification of metallic implant surfaceswith biofunctional proteins (Part 1). Molecular structure andbiological activity of a modified NiTi alloy surface. DentMater J 1995;14:185–198.

88. Nanci A, Wuest JD, Peru L, Brunet P, Sharma V, Zalzal S,McKee MD. Chemical modification of titanium surfaces forcovalent attachment of biological molecules. J BiomedMater Res 1998;40:324–335.

89. Pierschbacher MD, Ruoslahti E. Cell attachment activity offibronectin can be duplicated by small synthetic fragments ofthe molecule. Nature 1984;309:30–33.

90. Rezania A, Thomas CH, Branger AB, Waters CM, HealyKE. The detachment strength and morphology of bone cellscontacting materials modified with a peptide sequence foundwithin bone sialoprotein. J Biomed Mater Res 1997;37:9–19.

91. Dee KC, Andersen TT, Bizios R. Design and function ofnovel osteoblast-adhesive peptides for chemical modifica-tion of biomaterials. J Biomed Mater Res 1998;40:371–377.

92. Rezania A, Healy KE. Biomimetic peptide surfaces thatregulate adhesion, spreading, cytoskeletal organization, andmineralization of the matrix deposited by osteoblast-likecells. Biotechnol Prog 1999;15:19–32.

93. Dee KC, Rueger DC, Andersen TT, Bizios R. Conditionswhich promote mineralization at the bone-implant inter-face: A model in vitro study. Biomaterials 1996;17:209–215.

94. Mohan S, Baylink DJ. Bone growth factors. Clin Orthop1991;263:30–48.


BRUNSKI ET AL




95. Lind M. Growth factor stimulation of bone healing. Effectson osteoblasts, osteomies, and implants fixation. ActaOrthop Scand Suppl 1998;283:2–37.

96. Lucas PA, Syftestad GT, Goldberg VM, Caplan AI. Ectopicinduction of cartilage and bone by water-soluble proteinsfrom bovine bone using a collagenous delivery vehicle. JBiomed Mater Res 1989;23(A1):23–29.

97. Lind M, Overgaard S, Nguyen T, Ongpipattanakul B,Bunger C, Søballe K. Transforming growth factor-� stimu-lates bone ongrowth. Hydroxyapatite-coated implants stud-ied in dogs. Acta Orthop Scand 1996;67:611–616.

98. Lind M, Overgaard S, Ongpipattanakul B, Nguyen T,Bunger C, Søballe K. Transforming growth factor-�1 stim-ulates bone ongrowth to weight-loaded tricalcium phos-phate coated implants: An experimental study in dogs. JBone Joint Surg [Br] 1996;78:377–382.

99. Piattelli A, Scarano A, Corigliano M, Piattelli M. Effects ofalkaline phosphatase on bone healing around plasma-sprayed titanium implants: A pilot study in rabbits. Bioma-terials 1996;17:1443–1449.

100. Liu SQ, Ito Y, Imanishi Y. Cell growth on immobilizedcell-growth factor. 4: Interaction of fibroblast cells withinsulin immobilized on poly(methyl methacrylate) mem-brane. J Biochem Biophys Methods 1992;25:139–148.

101. Liu SQ, Ito Y, Imanishi Y. Cell growth on immobilized cellgrowth factor: 5. Interaction of immobilized transferrinwith fibroblast cells. Int J Biol Macromol 1993;15:221–226.

102. Kuhl PR, Griffith-Cima LG. Tethered epidermal growthfactor as a paradigm for growth factor-induced stimulationfrom the solid phase. Nat Med 1996;2:1202–1207.

103. Puleo DA. Activity of enzyme immobilized on silanizedCo-Cr-Mo. J Biomed Mater Res 1995;29:951–957.

104. Puleo DA. Retention of enzymatic activity immobilized onsilanized Co-Cr-Mo and Ti-6Al-4V. J Biomed Mater Res1997;37:222–228.

105. Morra M, Cassinelli C. Organic surface chemistry on tita-nium surfaces via thin film deposition. J Biomed Mater Res1997;37:198–206.

106. Kissling RA, Puleo DA. Immobilization of protein on Ti-6Al-4V surfaces possessing amino groups [abstract]. TransSoc Biomater 1999;27:338.

107. Mrksich M, Chen CS, Xia Y, Dike LE, Ingber DE, White-sides GM. Controlling cell attachment on contoured sur-faces with self-assembled monolayers of alkanethiolates ongold. Proc Natl Acad Sci USA 1996;93:10775–10778.

108. Tanahashi M, Matsuda T. Surface functional group depen-dence on apatite formation on self-assembled monolayers ina simulated body fluid. J Biomed Mater Res 1997;34:305–315.

109. Walsh WR, Kim HD, Jong YS, Valentini RF. Controlledrelease of platelet-derived growth factor using ethylenevinyl acetate copolymer (EVAc) coated on stainless-steelwires. Biomaterials 1995;16:1319–1325.

110. Agrawal CM, Pennick A, Wang X, Schenck RC. Porous-coated titanium implant impregnated with a biodegradableprotein delivery system. J Biomed Mater Res 1997;36:516–521.

111. Puleo DA. Release and retention of biomolecules in colla-gen deposited on orthopedic biomaterials. Artif Cells BloodSubstit Immobil Biotechnol 1999;27:65–75.

112. Katz RW, Reddi AH. Dissociative extraction and partialpurification of osteogenin, a bone inductive protein, fromrat tooth matrix by heparin affinity chromatography.Biochem Biophys Res Commun 1988;157:1253–1257.

113. Muthukumaran N, Ma S, Reddi AH. Dose-dependence ofand threshold for optimal bone induction by collagenousbone matrix and osteogenin-enriched fraction. CollagenRel Res 1988;8:433–441.

114. Suzawa M, Takeuchi Y, Fukumoto S, Kato S, Ueno N,Miyazono K, et al. Extracellular matrix-associated bonemorphogenetic proteins are essential for differentiation ofmurine osteoblastic cells in vitro. Endocrinology 1999;140:2125–2133.

115. Constantz BR, Ison IC, Fulmer MT, Poser RD, Smith ST,VanWagoner M, et al. Skeletal repair by in situ formationof the mineral phase of bone. Science 1995;267:1796–1799.

116. Frankenburg EP, Goldstein SA, Bauer TW, Harris SA,Poser RD. Biomechanical and histological evaluation of acalcium phosphate cement. J Bone Joint Surg [Am]1998;80:1112–1124.

117. Van Eijden TMGJ. Three-dimensional analyses of humanbite-force magnitude and moment. Arch Oral Biol 1991;36:535–539.

118. Raadsheer MC, van Eijden TMGJ, van Ginkel FC, Prahl-Andersen B. Contribution of jaw muscle size and craniofa-cial morphology to human bite force magnitude. J DentRes 1999;78:31–42.

119. Strom D, Holm S. Bite-force development, metabolic andcirculatory response to electrical stimulation in the canineand porcine masseter muscles. Arch Oral Biol 1992;37:997–1006.

120. Lucas PW, Peters CR, Arrandale SR. Seed-breaking forcesexerted by orangutans with their teeth in captivity and anew technique for estimating forces produced in the wild.Am J Phys Anthropol 1994;94:365–378.

121. Graf H. Bruxism. Dent Clin North Am 1969;16:659–665.122. Graf H. Occlusal forces during function. In: Rowe NH

(ed). Occlusion: Research on Form and Function. AnnArbor: University of Michigan, 1975;90–111.

123. Osborn JW, Mao J. A thin bite force transducer with three-dimensional capabilities reveals a consistent change in biteforce direction during jaw muscle endurance tests. ArchOral Biol 1993;38:139–144.

124. Carlsson GE, Haraldson T. Functional response. In: Bråne-mark P-I, Zarb GA, Albrektsson T (eds). Tissue-IntegratedProstheses. Chicago: Quintessence, 1985:155–163.

125. Carr AB, Laney WR. Maximum occlusal force levels inpatients with osseointegrated oral implant prostheses andpatients with complete dentures. Int J Oral MaxillofacImplants 1987;2:101–108.

126. Mericske-Stern R, Zarb GA. In vivo measurements of somefunctional aspects with mandibular fixed prostheses sup-ported by implants. Clin Oral Implants Res1996;7:153–161.

127. Gracis S, Nicholls JI, Chalupnik JD, Yuodelis RA. Shock-absorbing behavior of five restorative materials used onimplants. Int J Prosthodont 1991;4:282–291.

128. Glantz P-O, Rangert B, Svensson A, Stafford GD, Arnvi-darson B, Randow K, et al. On clinical loading of osseointe-grated implants: A methodological and clinical study. ClinOral Implants Res 1993;4:99–105.

129. Rangert B, Gunne J, Glantz P-O, Svensson A. Vertical loaddistribution on a three-unit prosthesis supported by a nat-ural tooth and a single Brånemark implant. Clin OralImplants Res 1995;6:40–46.

130. Mericske-Stern R, Piotti M, Sirtes G. 3-D in vivo forcemeasurements on mandibular implants supporting overden-tures. A comparative study. Clin Oral Implants Res 1996;7:387–396.


BRUNSKI ET AL




131. Fontijn-Tekamp FA, Slagter AP, van’t Hof MA, GeertmanME, Kalk W. Bite force with mandibular implant-retainedoverdentures. J Dent Res 1998;77:1832–1839.

132. Gunne J, Rangert B, Glantz P-O, Svensson A. Functionalloads on freestanding and connected implants in three-unitmandibular prostheses opposing complete dentures: An invivo study. Int J Oral Maxillofac Implants 1997;12:335–341.

133. Richter E-J. In vivo horizontal bending moments onimplants. Int J Oral Maxillofac Implants 1998;13:232–243.

134. Hellden LB, Derand T. Description and evaluation of asimplified method to achieve passive fit between cast tita-nium frameworks and implants. Int J Oral MaxillofacImplants 1998;13:190–196.

135. Monteith BD. Minimizing biomechanical overload inimplant prostheses: A computerized aid to design. J Pros-thet Dent 1993;69:495–502.

136. Brunski JB, Skalak R. Biomechanical considerations forcraniofacial implants. In: Brånemark P-I,Tolman DE (eds).Osseointegration in Craniofacial Reconstruction. Chicago:Quintessence, 1998:15–35.

137. Rangert B, Jemt T, Jörneus L. Forces and moments onBrånemark implants. Int J Oral Maxillofac Implants 1989;4:241–247.

138. Skalak R. Biomechanical considerations in osseointegratedprostheses. J Prosthet Dent 1983;49:843–848.

139. Skalak R, Brunski JB, Mendelson M. A method for calculat-ing the distribution of vertical forces among variable-stiff-ness abutments supporting a dental prosthesis. In: LangranaNA, Friedman MH, Grood ES (eds). 1993 BioengineeringConference, vol 24. New York: American Society ofMechanical Engineers, 1993:347–350.

140. Morgan J, James DF. Force and bending moment distribu-tion among osseointegrated dental implants. J Biomech1995;28:1103–1109.

141. Brunski JB, Hurley E. Implant-supported prostheses: Bio-mechanical analyses of failed cases. In: Hochmuth RM,Langrana NA, Hefzy MS (eds). 1995 Bioengineering Con-ference, vol 29. New York: American Society of MechanicalEngineers, 1995:447–448.

142. Morgan MJ. Structural Analysis of an OsseointegratedDental Implant System [thesis]. Toronto: University ofToronto, 1997.

143. Smerek J, Brunski JB, Wolfinger G, Winkelman R, BalshiT. Implant loading with all-acrylic vs. metal-supported-acrylic prostheses [abstract]. J Dent Res 1997;76(specialissue):263.

144. Balshi T, Wolfinger GJ. Immediate loading of Brånemarkimplants in edentulous mandibles: A preliminary report.Implant Dent 1997;6:83–88.

145. White G. Osseointegrated Dental Technology. London:Quintessence, 1993.

146. Mailath-Pokorny G, Solar P. Biomechanics of endosseousimplants. In: Watzek G (ed). Endosseous Implants: Scien-tific and Clinical Aspects. Chicago: Quintessence,1996:291–318.

147. Brunski JB, Evaul G, Smerek J, Gigalonis-Vacario J,Wolfinger G. Biomechanics of overdentures attached toframeworks by different methods [abstract]. J Dent Res1996;75:183.

148. Daellenbach K, Hurley E, Brunski JB, Rangert B. Biome-chanics of in-line vs. offset implants supporting a partialprosthesis [abstract]. J Dent Res 1996;75:183.

149. Daellenbach K, Hurley E, Rangert B, Brunski JB. Forceand moment distribution among three abutments: Stag-gered vs. in-line arrangement. Presented at the 11th AnnualMeeting of the Academy of Osseointegration, New York,NY, Feb 29–Mar 2, 1996.

150. Renouard F, Rangert B. Risk Factors in Implant Dentistry.Chicago: Quintessence, 1999.

151. Mailath G, Schmid M, Lill W, Miller J. 3D-Finite-Ele-mente-Analyse der Biomechanik von rein implantatgetrage-nen Extensionbrucken. Z Zahnnärztl Implantol 1991;7:205–211.

152. Benzing UR, Gall H, Weber H. Biomechanical aspects oftwo different implant-prosthetic concepts for edentulousmaxillae. Int J Oral Maxillofac Implants 1995;10:188–198.

153. Lewinstein I, Banks-Sills L, Eliasi R. Finite element analy-sis of a new system (IL) for supporting an implant-retainedcantilever prosthesis. Int J Oral Maxillofac Implants1995;10:355–366.

154. Sertgöz A. Finite element analysis study of the effect ofsuperstructure material on stress distribution in an implant-supported fixed prosthesis. Int J Prosthodont1997;10:19–27.

155. Taylor TD, Agar JR, Vogiatzi T. Implant prosthodontics:Current perspective and future directions. Int J Oral Max-illofac Implants 2000;15:66–75.

156. McGlumphy E, Mendel DA, Holloway JA. Implant screwmechanics. Dent Clin North Am 1998;42:71–89.

157. Jörneus L, Jemt T, Carlsson L. Loads and designs of screwjoints for single crowns supported by osseointegratedimplants. Int J Oral Maxillofac Implants 1992;7:353–359.

158. Carr AB, Brunski JB, Hurley E. Effects of fabrication, fin-ishing, and polishing procedures on preload in prosthesesusing conventional “gold” and plastic cylinders. Int J OralMaxillofac Implants 1996;11:589–598.

159. Sakaguchi RL, Borgersen SE. Nonlinear contact analysis ofpreload in dental implant screws. Int J Oral MaxillofacImplants 1995;10:295–302.

160. Tan KB, Rubenstein JE, Nicholls JE, Yuodelis RA. Three-dimensional analysis of the casting accuracy of one-piece,osseointegrated implant-retained prostheses. Int J Prostho-dont 1993;6:533–540.

161. Isa ZM, Hobkirk JA. The effects of superstructure fit andloading on individual implant units: Part I. The effects oftightening gold screws and placement of a superstructurewith varying degrees of fit. Eur J Prosthodont RestorativeDent 1995;3:247–253.

162. Jemt T, Lekholm U. Measurements of bone and frameworkdeformations induced by misfit of implant superstructures.A pilot study in rabbits. Clin Oral Implants Res 1998;9:272–280.

163. Jemt T, Book K. Prosthesis misfit and marginal bone loss inedentulous implant patients. Int J Oral Maxillofac Implants1996;11:620–625.

164. Carr AB, Gerard DA, Larsen PE. The response of bone inprimates around unloaded dental implants supporting pros-theses with different levels of fit. J Prosthet Dent 1996;76:500–509.

165. Faulkner G, Wolfaardt J, del Valle V. Console abutmentloading in craniofacial osseointegration. Int J Oral Maxillo-fac Implants 1998;13:245–252.

166. Brunski JB, Slack J. Orthodontic loading of implants. In:Higuchi KW (ed). Orthodontic Applications of Osseointe-grated Implants. Chicago: Quintessence (in press).


BRUNSKI ET AL




167. Caulier H, Naert I, Kalk W, Jansen JA. The relationship ofsome histologic parameters, radiographic evaluations, andPeriotest measurements of oral implants: An experimentalanimal study. Int J Oral Maxillofac Implants 1997;12:380–386.

168. Elias JJ, Brunski JB, Scarton HA. A dynamic modal testingtechnique for noninvasive assessment of bone-dentalimplant interfaces. Int J Oral Maxillofac Implants1996;11:728–734.

169. Elias JJ, Carollo JS, Brunski JB, Scarton HA. Noninvasivemethod for measuring the integrity of implant-tissue inter-faces. In: Langrana NA, Friedman MH, Grood ES (eds).1993 Bioengineering Conference, June 25–29, 1993, Breck-enridge, CO. New York: American Society of MechanicalEngineers, 1993;24:327–330.

170. Meredith N, Alleyne D, Cawley P. Quantitative determina-tion of the stability of the implant-tissue interface using res-onance frequency analysis. Clin Oral Implants Res 1996;7:261–267.

171. Meredith N, Shagadi F, Alleyne D, Sennerby L, Cawley P.The application of resonance frequency measurements tostudy the stability of titanium implants during healing inthe rabbit tibia. Clin Oral Implants Res 1997;8:234–243.

172. Akimoto K. Is my implant osseointegrated? The need fornew non-invasive tests. Academy News 1999;10:1.

173. Brånemark P-I. Introduction to osseointegration. In:Brånemark P-I, Zarb GA, Albrektsson T (eds). Tissue-Integrated Prostheses. Chicago: Quintessence, 1985:11–76.

174. Schenk R. Biology of fracture repair. In: Browner BD,Jupiter JB, Levine AM, Trafton PG (eds). Skeletal Trauma,vol 1. Philadelphia: WB Saunders Co, 1992:31–75.

175. Schenk R, Hunziker EB. Histologic and ultrastructural fea-tures of fracture healing. In: Brighton CT, Friedlander G,Lane JM (eds). Bone Formation and Repair. Rosemont, IL:American Academy of Orthopaedic Surgeons, 1994:117–146.

176. Schenk RK, Buser D, Hardwick WR, Dahlin C. Healingpattern of bone regeneration in membrane-protecteddefects: A histological study in the canine mandible. Int JOral Maxillofac Implants 1994;9:13–29.

177. Rahn BA. Morphology of fracture healing and its relation-ship to biomechanics. In: Kruger E, Schilli W (eds). Oraland Maxillofacial Traumatology, vol 1. Chicago: Quintes-sence, 1982:134–145.

178. Brunski JB. Influence of biomechanical factors at the bone-biomaterial interface. In: Davies JE (ed). The Bone-Bioma-terial Interface. Toronto: Univ of Toronto Press, 1991:391–405.

179. Plenk H Jr, Zitter H. Material considerations. In: Watzek G(ed). Endosseous Implants: Scientific and Clinical Aspects.Chicago: Quintessence, 1996:63–99.

180. Huja SS, Katona T, Moore BK, Roberts WE. Microhard-ness and anisotropy of the vital osseous interface andendosseous implant supporting bone. J Orthop Res 1998;16:54-60.

181. Turner CH, Rho J, Takano Y, Tsui TY, Pharr GM. Theelastic properties of trabecular and cortical bone tissues aresimilar: Results from two microscopic measurement tech-niques. J Biomech 1999;32:437–441.

182. Hoshaw SJ, Fyhrie DP, Schaffler MB. The effect of implantinsertion and design on bone microdamage. In: DavidovitchZ (ed). The Biological Mechanisms of Tooth Eruption,Resorption and Replacement by Implants. Boston: HarvardSociety for the Advancement of Orthodontics, 1994:735–741.

183. Mori S, Burr M. Increased cortical remodeling followingfatigue microdamage. Bone 1993;14:103–109.

184. Roberts WE. Bone tissue interface. J Dent Educ 1988;52:804–809.

185. Roberts WE, Turley PK, Brezniak N, Fielder PJ. Bonephysiology and metabolism. Calif Dent Assoc J 1987;15:54–61.

186. Garetto LP, Chen J, Parr JA, Roberts WE. Remodelingdynamics of bone supporting rigidly fixed titaniumimplants: A histomorphometric comparison in four speciesincluding humans. Implant Dent 1995;4:235–242.

187. Chen J, Esterle M, Roberts WE. Mechanical response tofunctional loading around the threads of retromolarendosseous implants utilized for orthodontic anchorage:Coordinated histomorphometric and finite element analy-sis. Int J Oral Maxillofac Implants 1999;14:282–289.

188. Hoshaw SJ, Brunski JB, Cochran GVB. Mechanical loadingof Brånemark fixtures affects interfacial bone modeling andremodeling. Int J Oral Maxillofac Implants 1994;9:345–360.

189. Roberts WE, Simmons KE, Garetto L, DeCastro RA. Bonephysiology and metabolism in dental implantology: Riskfactors for osteoporosis and other metabolic diseases.Implant Dent 1992;1:11–21.

190. Frost HM. The regional acceleratory phenomenon: Areview. Henry Ford Hosp Med J 1983;31:3–9.

191. Frost HM. The origin and nature of transients in humanbone remodeling dynamics. In: Frame B, Parfitt AM, Dun-can H (eds). Clinical Aspects of Metabolic Bone Disease.Amsterdam: Excerpta Medica, 1973:124–137.

192. Spector M, Lalor P. In vivo assessment of tissue compatibil-ity. In: Ratner BD, Hoffman AS, Schoen FJ, Lemons JE(eds). Biomaterials Science: An Introduction to Materials inMedicine. New York: Academic Press, 1996:220–228.

193.Søballe K, Hansen ES, Brockstedt-Rasmussen H, JørgensenPH, Bünger C. Tissue ingrowth into titanium and hydrox-yapatite coated implants during stable and unstablemechanical conditions. Acta Orthop Scand 1992;64(suppl225).

194. Prendergast PJ, Huiskes R, Søballe K. Biophysical stimulion cells during tissue differentiation at implant surfaces. JBiomech 1997;30:539–548.

195. Søballe K, Brockstedt-Rasmussen H, Hansen ES, BüngerC. Hydroxyapatite coating modifies implant membrane for-mation. Controlled micromotion studied in dogs. ActaOrthop Scand 1992;63:128–140.

196. Søballe K, Hansen ES, Brockstedt-Rasmussen H, BüngerC. Hydroxyapatite coating converts fibrous tissue to bonearound loaded implants. J Bone Joint Surg [Br] 1993;75:270–278.

197. Liu C, Brunski JB. Axial and lateral mobility of standard vs.experimental Brånemark fixtures [abstract]. J Dent Res1999;78:246.

198. Kawaguchi H, McKee MD, Okamoto H, Nanci A.Immunocytochemical and lectin-gold characterization ofthe interface between alveolar bone and implanted hydrox-yapatite in the rat. Cells Mater 1993;3:337–350.

199. Martin RB, Burr DB. Structure, Function, and Adaptationof Compact Bone. New York: Raven Press, 1989.

200. Zioupos P, Currey JD. The extent of microcracking and themorphology of microcracks in damaged bone. J Mater Sci1994;29:978–986.

201. Carter DR, Hayes WC. Compact bone fatigue damage: Amicroscopic examination. Clin Orthop Rel Res 1977;127:265–274.


BRUNSKI ET AL




202. Currey JD. The Mechanical Adaptations of Bones. Prince-ton: Princeton University Press, 1984.

203. Burr DB, Schaffler MB, Frederickson RG. Composition ofthe cement line and its possible mechanical role as a localinterface in human compact bone. J Biomech 1992;21:939–945.

204. Choi K, Goldstein SA. A comparison of the fatigue behav-ior of human trabecular and cortical bone tissue. J Biomech1992;25:1371–1381.

205. Steinemann SG, Eulenberger J, Maeusli P-A, Schroeder A.Adhesion of bone to titanium. In: Christel P, Meunier A,Lee AJC (eds). Biological and Biomechanical Performanceof Biomaterials. Amsterdam: Elsevier, 1986:409–414.

206. Brånemark R. A Biomechanical Study of Osseointegration[thesis]. Göteborg, Sweden: Göteborg University, 1996.

207. Cowin SC, Van Buskirk WC, Ashman RB. Properties ofbone. In: Skalak R, Chien S (eds). Handbook of Bioengi-neering. New York: McGraw-Hill, 1987:2.1–2.27.

208. Wong M, Eulenberger J, Schenk R, Hunziker E. Effect ofsurface topology on the osseointegration of implant materi-als in trabecular bone. J Biomed Mater Res 1995:29:1567–1576.

209. Taylor JL, Brunski JB, Hoshaw SJ, Cochran GVB, HiguchiKW. Interfacial bond strengths of Ti-6Al-4V and hydroxy-apatite-coated Ti-6Al-4V implants in cortical bone. In:Laney WR, Tolman DE (eds). Tissue Integration in Oral,Orthopedic and Maxillofacial Reconstruction. Chicago:Quintessence, 1992:125–132.

210. Aspenberg P, Skripitz R. Tensile bonding between bone andtitanium. Trans 44th Orthop Res Soc, New Orleans,Louisiana, March 16–19, 1998. Chicago: OrthopedicResearch Society, 1998:220.

211. Edwards JT, Brunski JB, Higuchi KW. Mechanical andmorphologic investigation of the tensile strength of a bone-hydroxyapatite interface. J Biomed Mater Res 1997;36:454–468.

212. Hong L, Hengchang X, de Groot K. Tensile strength of theinterface between hydroxyapatite and bone. J BiomedMater Res 1992;26:7–18.

213. Schmikal T. Vergleichende Untersuchungen über dieKnochenhaftung verschiedener Implantatwerkstoffe mitHilfe enossaler Implantate [thesis]. Cologne: Univ ofCologne, 1984.

214. Oyola A, Brunski JB. A finite element simulation ofdebonding at an osseointegrated bone-dental implant inter-face. In: Chandran KB, Vanderby R Jr, Hefzy MS (eds).1997 Bioengineering Conference, vol 35. New York: Amer-ican Society of Mechanical Engineers, 1997:577–578.

215. Hipp JA, Brunski JB, Shephard MS, Cochran GVB. Finiteelement models of implants in bone: Interfacial assump-tions. In: Perren SM, Schneider E (eds). Biomechanics:Current Interdisciplinary Research. Hingham, MA: KluwerAcademic Publishers, 1986:447–452.

216. Prabhu A, Brunski JB. Finite element analysis of a clinicalcase involving overload of an oral implant interface. In:Chandran KB, Vanderby R Jr, Hefzy MS (eds). 1997 Bio-engineering Conference, vol 35. New York: American Soci-ety of Mechanical Engineers, 1997:575–576.

217. Prabhu AA, Brunski JB. An overload failure of a dentalprosthesis: A 3D finite element nonlinear contact analysis.In: Simon B (ed). 1997 Advances in Bioengineering, vol 36.New York: American Society of Mechanical Engineers,1997:141–142.

218. Quirynen M, Naert I, van Steenberghe D. Fixture designand overload influence on marginal bone loss and fixturesuccess in the Brånemark implant system. Clin OralImplants Res 1992;3:104–111.

219. Van Steenberghe D, Tricio J, Van den Eynde, Naert I,Quirynen M. Soft and hard tissue reactions towards implantdesign and surface characteristics and the influence ofplaque and/or occlusal loads. In: Davidovitch Z (ed). TheBiological Mechanisms of Tooth Eruption, Resorption andReplacement by Implants. Boston: Harvard Society for theAdvancement of Orthodontics, 1994:687–697.

220. Rangert B, Krogh PHJ, Langer B, van Roekel N. Bendingoverload and implant fracture: A retrospective clinicalanalysis. Int J Oral Maxillofac Implants 1995;10:326–334.

221. Esposito M, Hirsch J-M, Lekholm U, Thomsen P. Biologi-cal factors contributing to failures of osseointegrated oralimplants. II. Etiopathogenesis. Eur J Oral Sci 1998;106:721–764.

222. Carter DR, Wright TM. Yield characteristics of corticalbone. In: Ducheyne P, Hastings G (eds). Functional Behav-ior of Orthopaedic Biomaterials. New York: CRC Press,1984:9–36.

223. Fyhrie DP, Hamid MS, Kuo RF, Lang SM. The probabilitydistribution of trabecular level strains for vertebral cancel-lous bone. Trans 39th Orthop Res Soc, San Francisco, CA,Feb 15–18, 1993. Chicago: Orthopedic Research Society,1993:175.

224. Wachtel EF, Keaveny TM. Dependence of trabecular dam-age on mechanical strain. J Orthop Res 1997;15:781–787.

225. Pattin CA, Caler WE, Carter DR. Cyclic mechanical prop-erty degradation during fatigue loading of cortical bone. JBiomech 1996;29:69–79.

226. Guo XE, Gibson LJ, McMahon TA. Fatigue of trabecularbone: Avoiding end-crushing effects. Trans 39th OrthopRes Soc, San Francisco, CA, Feb 15–18, 1993. Chicago:Orthopedic Research Society, 1993:584.

227. Fazzalari NL, Forwood M, Manthey BA, Smith K, KolesikP. Three-dimensional images of microdamage in cancellousbone. In: Chandran KB, Vanderby R Jr, Hefzy MS (eds).1997 Bioengineering Conference, vol 35. New York: Amer-ican Society of Mechanical Engineers, 1997:305–306.

228. Nicolella DP, Lankford J, Jepsen K, Davy DT. Correlationof physical damage development with microstructure andstrain localization in bone. In: Chandran KB, Vanderby RJr, Hefzy MS (eds). 1997 Bioengineering Conference, vol35. New York: American Society of Mechanical Engineers,1997:311–312.

229. Nicolella DP, Nicholls AE, Lankford J. Micromechanics ofcreep in cortical bone. Trans 44th Orthop Res Soc, NewOrleans, Louisiana, March 16–19, 1998. Chicago: Orthope-dic Research Society, 1998:137.

230. Martin B. Mathematical model for repair of fatigue damageand stress fracture in osteonal bone. J Orthop Res 1995;13:309–316.

231. Burr DB, Forwood MR, Fyhrie DP, Martin RB, SchafflerMB, Turner CH. Bone microdamage and skeletal fragilityin osteoporotic and stress fractures. J Bone Miner Res1997;12:6–15.

232. Brunski JB, Hipp JA, Cochran GVB. The influence of bio-mechanical factors at the tissue-biomaterial interface. In:Hanker JS, Giammara BL (eds). Biomedical Materials andDevices. Pittsburgh: Materials Research Society, 1989:505–515.


BRUNSKI ET AL




233. Hoshaw SJ, Brunski JB, Cochran GVB. Pullout and fatiguefailure of bone-dental implant interfaces. In: Torzilli PA,Friedman MH (eds). 1989 Biomechanics Symposium. NewYork: American Society of Mechanical Engineers, 1989:205–208.

234. Isidor F. Loss of osseointegration caused by occlusal load oforal implants. A clinical and radiographical study in mon-keys. Clin Oral Implants Res 1996;7:143–152.

235. Isidor F. Histological evaluation of peri-implant bone atimplants subjected to occlusal overload or plaque accumula-tion. Clin Oral Implants Res 1997;8:1–9.

236. Brunski JB, Yang C-J. Finite element simulation of damage-induced bone remodeling at a bone-implant interface.Trans 44th Orthop Res Soc, New Orleans, Louisiana,March 16–19, 1998. Chicago: Orthopedic Research Society,1998:341.

237. Brunski JB, Kim DG. Early loading of a bone-implantinterface: Finite element simulation of damage-inducedbone remodeling. Trans 46th Orthop Res Soc (in press).

238. Martin RB, Burr DB, Sharkey NA. Skeletal TissueMechanics. New York: Springer-Verlag, 1998:225.

239. Frost HM. The mechanostat: A proposed pathogenicmechanism of osteoporosis and the bone mass effects ofmechanical and nonmechanical agents. Bone Miner 1987;2:73–85.

240. Mattheck C. Design in Nature. New York: Springer-Verlag,1998.

241. Bertram JEA, Swartz SM. The ‘law of bone transforma-tion’: A case of crying “Wolff”? Biol Rev 1991;66:245–273.

242. Chomsky N, as quoted in Jaworski ZFG. Does the mechan-ical usage (MU) inhibit bone “remodeling”? Calcif TissueInt 1987;41:239–248.

243. Huiskes R. The law of adaptive bone remodeling: A case forcrying Newton? In: Odgaard A, Weinans H (eds). BoneStructure and Remodeling. Singapore: World Scientific,1995:15–24.

244. Currey JD. The validation of algorithms used to explainadaptive bone remodeling in bone. In: Odgaard A, WeinansH (eds). Bone Structure and Remodeling. Singapore:World Scientific, 1995:9–13.

245. Goldstein SA, Wanders N, Guldberg R, Steen H, SenunasL, Goulet JA, Bonadio J. Stress morphology relationshipsduring distraction osteogenesis: Linkages between mechan-ical and architectural factors in molecular regulation. In:Brighton CT, Friedlander G, Lane JM (eds). Bone Forma-tion and Repair. Rosemont, IL: American Academy ofOrthopaedic Surgeons, 1994:405–420.

246. Van Rietbergen B, Huiskes R, Weinans H, Sumner DR,Turner TM, Galante JO. The mechanism of bone remodel-ing and resorption around press-fitted THA stems. J Bio-mech 1993;26:369–382.

247. Thakur AJ. The Elements of Fracture Fixation. New York:Churchill Livingstone, 1997.

248. Pilliar RM, Deporter DA, Watson PA, Valiquett N. Dentalimplant design—Effect on bone remodeling. J BiomedMater Res 1991;25:889–902.

249. Al-Sayyed A, Deporter DA, Pilliar RM, Watson PA,Pharoah M, Berhane K, Carter S. Predictable crestal boneremodeling around two porous-coated titanium alloy dentalimplant designs. Clin Oral Implants Res 1994;5:131–141.

250. Vaillancourt H, Pilliar RM, McCammond D. Finite ele-ment analysis of bone remodeling around porous coateddental implants. J Appl Biomater 1995;6:267–282.

251. Vaillancourt H, Pilliar RM, McCammond D. Factorsaffecting crestal bone loss with dental implants partiallycovered with a porous coating: A finite element analysis. IntJ Oral Maxillofac Implants 1996;11:351–359.

252. Skalak R. Biomechanics of osseointegration. In: BrånemarkP-I, Rydevik BL, Skalak R (eds). Osseointegration in Skele-tal Reconstruction and Joint Replacement. Chicago: Quin-tessence, 1997:45–56.

253. Van Oosterwyck H, Duyck J, Vandersloten J, Van der PerreG, De Cooman M, Lievens S, et al. The influence of bonemechanical properties and implant fixation upon bone load-ing around oral implants. Clin Oral Implants Res 1998;9:407–418.

254. Perren SM, Cordey J, Rahn BA, Gautier E, Schneider E.Early temporary porosis of bone induced by internal fixa-tion implants. Clin Orthop Rel Res 1988;232:139–151.

255. Weinreb M, Rodan GA, Thompson DD. Osteopenia in theimmobilized rat hind limb is associated with increased boneresorption and decreased bone formation. Bone 1989;10:187–194.

256. Cowin SC, Luo G. Modeling of ingrowth into implant cav-ities along the bone-implant interface. In: Brånemark P-I,Rydevik BL, Skalak R (eds). Osseointegration in SkeletalReconstruction and Joint Replacement. Chicago: Quintes-sence, 1997: 57–72.

257. Rubin CT, McLeod KJ. Biologic modulation of mechanicalinfluences in bone remodeling. In: Mow VC, Ratcliffe A,Woo S L-Y (eds). Biomechanics of Diarthrodial Joints, vol2. New York: Springer-Verlag, 1990:97–118.

258. Keller JC (ed). Oral Biology and Dental Implants. AdvDent Res 1999;13:1–190.


BRUNSKI ET AL




Biomaterials and Biomechanics of Oral and Maxillofacial … · 2006-04-17 · bone-implant contact...

Documents

Transcript of Biomaterials and Biomechanics of Oral and Maxillofacial … · 2006-04-17 · bone-implant contact...