THE ROLE OF IMPLANT SURFACE CHARACTERISTICS

IN THE HEALING OF BONE

K. Kieswetter
OsteoBiologics, Inc., San Antonio, Texas
Z. Schwartz
Department of Periodontics, Hebrew University, Hadassah Hospital, P.O. Box 12272, Jerusalem, Israel; University of Texas Health Science Center at San Antonio, San Antonio, Texas
D.D. Dean
B.D. Boyan*
Department of Orthopaedics, The University of Texas Health Science Center at San Antonio, 7703 Floyd Curl Drive, San Antonio, Texas 78284-7774; f to whom correspondence should be
addressed

ABSTRACT: The surface of an implant determines its ultimate ability to integrate into the surrounding tissue. The composite
effect of surface energy, composition, roughness, and topography plays a major role during the initial phases of the biological
response to the implant, such as protein adsorption and cellular adherence, as well as during the later and more chronic pha-
ses of the response.
For bone, the successful incorporation (and hence rigid fixation) of an alloplastic material within the surrounding bony bed
is called osteointegration. The exact surface characteristics necessary for optimal osteointegration, however, remain to be elu-
cidated. This review will focus on how surface characteristics, such as composition and roughness, affect cellular response to
an implant material. Data from two different culture systems suggest that these characteristics play a significant role in the
recruitment and maturation of cells along relevant differentiation pathways. In the case of osteointegration, if the implant sur-
face is inappropriate or less than optimal, cells will be unable to produce the appropriate complement of autocrine and
paracrine factors required for adequate stimulation of osteogenesis at the implant site. In contrast, if the surface is appropri-
ate, cells at the implant surface will stimulate interactions between cells at the surface and those in distal tissues. This, in turn,
will initiate a timely sequence of events which include cell proliferation, differentiation, matrix synthesis, and local factor pro-
duction, thereby resulting in the successful incorporation of the implant into the surrounding bony tissue.

Key words. Surface characteristics, wound healing, osteointegration.

Introduction (Homsy et al., 1968; Homsy, 1970; Homsy and

arly investigations into the interaction between Armediades, 1971; Hulbert et al., 1973a,b; Hench and
E implant materials and tissues were limited to exam-
inations of the inherent biocompatibility characteristics
Ethridge, 1982). Variations in the thickness of the cap-
sule may occur as a function of micromotion at the inter-
of the material. While these classic studies were general- face (Hench and Ethridge, 1982).
ly limited to morphological observations, they nonethe- Often, materials which elicit formation of thin con-
less resulted in a set of materials that are currently nective tissue capsules are mistakenly termed "bioinert"
accepted as "suitable" for implantation. These materials because a severe chronic inflammatory response is not
can be subdivided into three general classes—biotoler- evident. These materials do in fact interact with sur-
ant, bioactive, and bioresorbable—based on their level rounding tissue, but often not to the extent of being able
of interaction with the surrounding tissue and the type of to cause changes in either the cellular structure or mate-
cellular response elicited. The formation of thin connec- rial at the light microscopic level. Indeed, the presence of
tive tissue capsules (from 0.1 to 10 jjim) around implant- connective tissue at the interface demonstrates that
ed devices is generally characteristic of blotolerant mate- these materials have elicited a biologic response, caus-
rials. The capsule formed does not adhere to the materi- ing the cells in adjacent tissue to synthesize, secrete, and
al. It is continuous, composed of acellular matrix, and maintain the connective tissue interface. Some of the
separates the implant from surrounding viable tissue materials that are in current use and which come into

7(4):329~345 (1996) Crit Rev Oral Biol Med 329

Downloaded from cro.sagepub.com at University of Sydney on November 18, 2015 For personal use only. No other uses without permission.
direct contact with bone are polymers such as polytetra- The use of porous coatings on implants was investi-
fluoroethylene (PTFE) and polymethylmethacrylate gated as early as 1969 (Hirschhorn and Reynolds, 1969)
(PMMA), ceramics such as alumina and zirconia, and and was reported by several other groups shortly there-
metals such as the titanium and cobalt-chrome-molyb- after. The initial indications suggested that a mechanical
denum alloys (Bechtol et al, 1959; Homsy, 1970; Homsy interlock between the implant and the surrounding tis-
and Armediades, 1971; Hench and Ethridge, 1982). sue produced sufficient fixation in the bone to obviate
Bioactive materials comprise another general cate- the need for polymethyl methacrylate cements in total
gory of materials used in bone in vivo. The surfaces of joint replacements (Galante et al, 1971; Cameron et al,
these materials are designed to interact with the sur- 1973; Robertson et al, 1976; Bobyn et al, 1980a,b).
rounding tissue in such a way as to induce bone forma- These porous surfaces were produced by a multitude
tion in direct contact with the implant. Ideally, these of techniques. In some cases, spherical beads of various
materials are bone-bonding and form a seamless junc- sizes were sintered onto metal substrates (Bobyn et al,
tion between the biomaterial and the extracellular matrix 1980a), while in others, metal fibers were bonded togeth-
of the adjacent bone cells. In most cases, however, only er and then incorporated into the device. A variety of
a tight apposition or interdigitation occurs at this junc- studies was conducted that, at least during the initial
tion. The successful incorporation (and hence rigid fixa- post-implantation phase, showed better pull-out
tion) of an alloplastic material within its surrounding strength with porous-coated surfaces (Cook et al, 1988;
bony bed is generally referred to as osteointegration or Cook, 1991). Over the last 20 years, this has translated
osseointegration. into the clinical use of porous-coated devices in younger
Bioactive materials can be derived from natural tis- patients who have good bone stock.
sue, such as replamineform-treated hydroxylapatite Development of porous coatings brought to light the
materials, or they may be entirely man-made, such as the role of localized strains in the formation of bone
bioglasses and synthetic calcium phosphates, including (Cameron et al, 1973). Excessive motion at the bone-
hydroxylapatite (Beder and Eade, 1956; Hall, 1970; implant interface has been shown to have a detrimental
Hulbert et al, 1971, 1972, 1973b, 1974; Hench and effect on the amount of bone ingrowth into porous-coat-
Ethridge, 1982). Whether metals such as commercially ed implants (Cameron etal, 1973; Pilliaret al, 1986; Cook
pure titanium can be bioactive materials is still a matter et al, 1988; Cook, 1991). Previous studies have shown
of discussion (Takatsuka et al, 1995), but the interaction that various parameters of motion also affect the nature
between a titanium surface and cells cultured on these of this tissue growing into the pores of the implant.
surfaces, as well as possible implications for osteogene- These parameters include the amplitude and frequency
sis, are addressed in greater detail below. of the motion (Aspenberg et al, 1992; Goodman and
Biodegradable materials are designed to be ulti- Aspenberg, 1992; Goodman et al, 1993a,b).
mately replaced by autologous tissue. Many of these Some authors have shown that short periods of
materials are composites of biological agents such as micromotion can promote bone accretion, while exces-
cells, tissue extracts, or proteins, and a structural sup- sive microstrain can have the opposite effect (Rubin and
port system. Ideally, after the material is resorbed, no Lanyon, 1984; Goodship and Kenwright, 1985; Lanyon,
trace of the implant should exist. These materials, there- 1987; Kenwright and Goodship, 1989; Rubin and
fore, are generally limited to compositions that can be McLeod, 1994).
readily metabolized by the body (Nelson et al, 1977; It was apparent during the early biocompatibility
Hench and Ethridge, 1982), such as collagen-derived studies on porous-coated implants that excessive micro-
products, polylactic acid and polyglycolic acid polymers, motion at the implant site enhanced the formation of
and processed bone graft. fibrous tissue. However, variations in pore sizes of the
While the classic studies of the Fifties, Sixties, and coatings affected both the quantity and quality of the
Seventies (Beder and Eade, 1956; Bechtol et al, 1959; new bone formed at the site. In their classic study exam-
Hall et al, 1966; Homsy et al, 1968; Homsy, 1970; Homsy ining the effect of pore size on the fixation of porous-
and Armediades, 1971; Hulbert et al, 1971, 1972, 1973b; coated implants in bone, Bobyn et al (1980b) found the
Natiella et al, 1977; Nelson et al, 1977) have served to greatest shear strength (approximately 17 MPa) with
define the nature of biocompatibility and determine implants containing pores in the 50-400-jnm range.
what materials meet these criteria, they did not examine Implants with smaller pores did not have enough inter-
what characteristics made these materials effective in stitial volume to accommodate uniform bone formation.
vivo. Interest in the surface characteristics of implant Implants with larger pores (400-800 jim) had lower shear
materials and their effects in vivo emerged during the lat- strengths at eight weeks and increased shear strengths at
ter years of this period. Initially, these studies were 12 weeks, due to the fact that a greater volume of bone
macroscopic and focused on the creation of a porous had to form for creation of the mechanical lock, but once
surface which allowed for bony ingrowth. formed, the increased bone surface was more effective.

330 Crit Rev Oral Biol Ued 7(4):329~345 (1996)

Downloaded from cro.sagepub.com at University of Sydney on November 18, 2015 For personal use only. No other uses without permission.
The authors concluded that use of more particles per unit not been quantitatively determined, theories exist sta-
volume to achieve an intermediate pore size would result ting that a certain amount of strain is required for the for-
in an increased likelihood of particle-to-bone contact mation of bone as opposed to fibrous tissue. Exceeding
between the device and the surrounding tissue and these "bone strain" levels results in the production of
increase the frictional forces between the two bodies, fibrous tissue.
thereby decreasing the possibility of micromotion at the All of this, of course, follows Wolff's law. The notion
interface. that strain affects cell characteristics has been studied in
The relationship between implant surface character- vitro and, to a lesser extent, in vivo. Some proponents
istics and local strains was also examined in studies believe that it is the mechanical characteristics at a site,
looking at the effects of porous coatings on femoral as opposed to the inherent material characteristics, that
stems used in total hip replacements (THR). Clear evi- dictate implant success. The truth of the matter may be
dence was seen that the location of the porous coating somewhere in between, since surface characteristics will
had a significant effect on the clinical outcome (Brown determine the local strain environments.
and Ring, 1985). Profound effects on proximal fixation More recently, studies have focused on surface char-
were evident in THRs where the entire femoral compo- acteristics of implant materials and their ability to affect
nent had been coated with sintered beads. In these clinical outcome favorably. It is hoped that an under-
patients, fixation was noted in the lower and mid- standing of how the characteristics of the material direct
femoral regions, but not in the proximal femur. The cellular response will lead to the development of sur-
authors concluded that rigid fixation in the middle and faces and materials that facilitate successful incorpora-
lower portions of the femoral component resulted in tion (whether by bone-bonding or by osteointegration)
massive stress shielding, and hence resorption osteoly- of the device into the surrounding tissue.
sis, of the proximal region. In this same study, patients Regardless of whether a material is biotolerant,
who had received non-porous-coated devices showed no bioactive, or bioresorbable, it is almost universally
change in radiolucency indicative of osteolysis. implanted during a surgical treatment. Further, the
A common theory is that the presence of these resulting outcome of the surgery is heavily dependent
rougher surfaces allows the cells to migrate into the upon the device's ability to influence both the acute and
pores and, while in this environment, to be exposed to chronic inflammatory phases of the healing process. In
proper strain levels for bone formation, thereby promo- general, the very first steps of the acute healing phase
ting a solid mechanical interlock. While these levels have will be the same in the the presence or absence of a for-

Protein
Adsorption
Surface
Composition
Cell
\ Adhesion
\
Cell
\ Proliferation
Surface w Surface
Roughness
Energy
Cell
Differentiation
\ -(
Matrix
/ Production
Surface
Topography
Calcification

Figure 1. Surface composition, roughness, topography, and energy are interelated surface characteristics that dictate the biological
response to an implanted device. At the onset of implantation, surface energy plays a critical role in determining the nature of the pro-
i .i . • i i i \ A/i *i f _l _______ I __ T_ ' II..I ____.___.__.___. ______• _______ ___££__.__! i_» . . - . . . _ ~ I L . _~_-» __^_J_

osteointegra-
and,
ultimately, calcification remain to be examined.

7(4)-.329-345 (1996) Crit Rev Oral Biol Med 331

Downloaded from cro.sagepub.com at University of Sydney on November 18, 2015 For personal use only. No other uses without permission.
eign body (e.g., the implant). This is because the wound- protein, Coulomb interactions between the material and
healing response must be mounted in response to the the protein, and structural rearrangements in the adsor-
surgical trauma. From this point on, however, the mater- bing proteins. As the binding orientation of ions, miner-
ial characteristics of the device dictate the steps toward als, water, protein, and other molecules is altered, there
successful wound healing. In the case of a material is a corresponding change in the cell's attachment to the
implanted into bony tissues, successful healing includes surface via cell adhesion molecules, resulting in a change
incorporation or osteointegration of the implant into the in cell shape, and finally, the behavior of the cell.
surrounding bony bed. Incorporation into the surround- A variety of implant materials is used in bone with
ing tissue is also important when materials are implant- the expectation that osteogenesis will occur.
ed into soft tissue or other sites; discussion of these Immediately upon implantation, the device becomes
other tissues, however, is beyond the scope of this paper. coated with a layer of organic and inorganic components
The. body interacts with the implant's surface. from the plasma. These proteins, lipids, sugars, and ions
Characteristics such as surface composition, surface adsorb to the surface of the implant. Not only are the
topography, surface roughness, and surface energy are of types of molecules adsorbed important, but also the
importance (Schwartz and Boyan, 1994). These charac- sequence in which these molecules adsorb to the surface
teristics are highly interrelated, and it is difficult to dif- is critical in orchestrating the body's response to the
ferentiate among the effects of the individual character- implant (Vroman and Adams, 1969; Hench and Paschall,
istics. Initially, the role of surface energy, as dictated by 1973; Vroman et al, 1980; Pearson et al, 1988; Ziats et al,
the surface roughness, topography, and composition of 1988; Pankowsky etal, 1990). Fibronectin, a protein found
the implant, may play a major role in determining which in plasma at 2-3 mg/dL (Pankowsky et al, 1990), has been
proteins are adsorbed onto the surface, as well as found to bind almost instantaneously to a number of
whether or not the cells themselves adhere to the surface materials (Pearson et al, 1988). This protein has an argi-
(Fig. 1). Thus, surface energy can have an effect on the nine-glycine-aspartic acid (RGD) binding-site sequence,
latter stages of bone formation and calcification by influ- thereby allowing it to mediate mesenchymal cell attach-
encing the types of cells that initially attach to the mate- ment. Other proteins, with similar binding-site
rial and differentiate at the implant-cell interface. sequences, can also adsorb to the surface. For example,
Ultimately, it is the combined effect of the material and
attachment of bone-derived cells to stainless steel, tita-
the early-responding cell populations that dictate the
nium, alumina, or polyethelene-terephthalate, in vitro,
body's long-term response to materials (Schwartz and
has been shown to be dependent on the prior adsorption
Boyan, 1994).
ofvitronectin to the surface (Howlett et al, 1994). In addi-
In order for either osteointegration or bone bonding
tion, proteins such as albumin and IgG are also adsorbed
to occur, a series of events ranging from protein adsorp-
tion to osteoid synthesis must occur at the implant sur- onto the material at early time points (Pankowsky et al,
face. In short, these events can be summarized as protein 1990).
adsorption, cellular adherence, proliferation, differentia- It appears that the composition of the proteins at
tion, matrix production, and calcification. The effects of the surface may be altered with time (Vroman and
implant surface characteristics on these various phases Adams, 1969; Vroman et al, 1980). Meyer et al (1988)
will be discussed here. demonstrated that the absorption of lipids and proteins
onto various biomaterials, including titanium, occurs
PROTEIN ABSORPTION within five minutes in vivo, while cell attachment does not
Cells, either in vitro or in vivo, do not encounter a com- occur until after several hours have elapsed. This sug-
pletely clean surface. At all times, the material is covered gests a requirement for prior conditioning of the surface
or conditioned by various components of the fluid in before cell attachment can occur—a finding consistent
which it is immersed, whether it is serum, saliva, crevic- with increased attachment by osteoblasts to titanium
ular fluid, or cell culture media. Which proteins, or other after the metal is coated with extracellular matrix pro-
components of the bathing medium, coat the surface of teins (Schneider and Burridge, 1994). A considerable
the material is partially dependent on the surface energy amount of literature, well beyond the scope of this
of the material. However, the potential for proteins to review, exists on the adsorption of proteins to various
undergo structural rearrangement on the surface itself is materials, especially those materials used in cardiovas-
also an important factor. Therefore, the behavior of cular applications and their subsequent effects in cell
adsorbed proteins may be related not only to the inter- attachment.
action between the charge of the material and the pro-
tein, but also to the protein's potential for change once CELLULAR ADHERENCE
adsorbed onto the surface. This process is governed by While proteins adsorbed to the implant surface have an
changes in the hydration of the material surface and the effect on the subsequent binding of cells, the composi-

332 Crit Rev Oral Biol Ued 7(4):329~345(1996)

Downloaded from cro.sagepub.com at University of Sydney on November 18, 2015 For personal use only. No other uses without permission.
tion of the material itself may also affect cellular attach- slow and appeared to be mediated by RGD-containing
ment. The extent to which surface composition can affect proteins. These studies suggest the possibility that cel-
cell adherence is well-illustrated in a recent study by lular adhesion may be mediated by one of two methods:
Gabriel et al. (1994). In this study, micro-organisms, com- Adherence can occur when the cell itself directly binds to
monly associated with device-centered infections the surface, or when it binds to RGD-containing proteins
(Staphylococcus epidermidis), were incubated in the presence that adhere to the surface.
of micromachined titanium (Ti), aluminum (Al), and The surface energy of a material can readily be
vanadium (V) substrates for 15 to 60 minutes. changed by some of the processing techniques used in
Quantitative analysis indicated tbat these organisms the manufacture and preparation of devices for implan-
adhered preferentially to V surfaces. The numbers of tation. For example, several studies have shown that
adherent micro-organisms per unit area of V were approx- steam sterilization causes deposition of organic sub-
imately two and three times, respectively, the numbers of stances on the implants, rendering the surfaces less than
organisms on unit areas of Ti and Al. These investigators clean and directly affecting wetting profiles. Changes in
also found that bacteria have a higher affinity for Ti surface wetting profiles have direct effects on the biolo-
grains containing the oc + p phase than for those con- gy and adherence of materials to the implant surface
taining solely the a phase in Ti alloy (Ti6Al4V). The a + p (DePalma, 1976; DePalma and Baier, 1978; Baier et al,
phase regions of the alloy have higher concentrations of 1988). This can result in poor tissue adhesion and devel-
V than do a phase regions. This is because, in general, opment of peri-implant scar tissue. Thus, it is possible
the level of V in the Ti alloy exceeds the solubility limit of that a material may preferentially adsorb one protein
the low-temperature p phase. Vanadium, however, serves over another simply because of the treatment it receives
to stabilize the high-temperature p phase. The ability of during fabrication. This, in turn, may affect the popula-
microbes to distinguish between a and a + p grains of Ti tion of cells that colonize the device.
alloy may be due to the surface energy of the two com- Given the same composition, materials with smaller
positions. This, in turn, may affect the differential grain sizes are expected to have a greater amount of sur-
adsorption of plasma components (Pankowsky et al, face energy than those with larger grains. Greater levels
1990) and/or alter their conformation on the surface of mechanical energy input during materials processing
(Vroman and Adams, 1969; Vroman et al, 1980). result in smaller grain structures. This energy is stored in
The effect of fluoride treatment on dental enamel grain boundaries. Thus, an increase in surface energy
provides a good example of this phenomenon. It has results from an increase in grain boundary surface area
been shown that exposure of dental enamel to stannous (per unit volume). Kilpadi and Lemons (1994), however,
fluoride (SnF) results in a substantial increase in nega- found that the critical surface tension (CST) of two com-
tive surface charge density (Kambara and Norde, 1995). mercially pure titanium (cpTi) materials of different grain
The presence of fluorapatite at the surface can alter for- sizes (23 jim vs. 70 |im) did not vary. Since titanium mate-
mation of enamel pellicle, i.e., bound salivary proteins, rials rapidly form an amorphous titanium oxide layer on
and subsequent binding of oral micro-organisms their surfaces, it is, in effect, the CST of the amorphous
(Ericsson and Ericsson, 1967; Rolla et al, 1977; Kambara titanium oxide layer and not that of the underlying metal
and Norde, 1995). It is possible that similar mechanisms substrate that is being examined.
may play a role in implant materials as well. Although the surface energy of cpTi (and its amor-
Studies by Hanein et al (1993, 1994) suggest that phous titanium oxide layer) was not noticeably affected
eukaryotic cells can also discern differences at this level. by mechanical processing, as evidenced by grain size in
In these studies, dramatic differences were seen in the the example above, it was affected by surface treatment.
ability of A6 kidney cells to adhere differentially to two Materials that had been cleaned via radio-glow dis-
different faces of calcium (R,R)-tartrate tetrahydrate crys- charge, for instance, had higher CSTs than those that had
tals. These crystals have two chemically equivalent, but been passivated and dry-heat-sterilized. Both of these
structurally unique, crystallographic faces. Although the had higher CSTs than untreated materials (Kilpadi and
same atoms are present on the surface of the crystal, Lemons, 1994).
their distribution results in the {011} and {101} faces hav- When the CST of unpolished, passivated, dry-heat-
ing different distributions of lattice water molecules and, sterilized, unalloyed titanium flats with an average sur-
hence, electrochemical charge. Although the two crystal- face roughness (Ra) of 1.3 jim was compared with that of
lographic surfaces are chemically equivalent, they exhib- polished (Ra = 0.2 jim) samples that were otherwise
it remarkable differences with respect to cellular adhe- identical, the smoother surfaces were found to have
sion characteristics. Initially, cells preferentially attach to greater CSTs (Kilpadi and Lemons, 1994). Although dif-
and spread on the {011} face. This appeared to be pro- ferences were noted among the various treatment
tein-independent and was followed by cell death. groups, it should be noted that all of the materials had
Attachment to the {101} surface, however, was extremely CST values within the relatively narrow range of 30 to 50

7(4):329-345 (1996) Crit Rev Oral Biol Med 333

Downloaded from cro.sagepub.com at University of Sydney on November 18, 2015 For personal use only. No other uses without permission.
dyne/cm. This relatively narrow band of CSTs may et al, 1994; Martin et al, 1995; Kieswetter et al, 1996;
account for the excellent biocompatibility of titanium Schwartz etal, 1996).
materials that have been subjected to considerably dif- To evaluate the effect of surface composition on cel-
ferent manufacturing regimens (Baier and Meyer, 1988). lular response, investigators have cultured rat costo-
Although cpTi is considered to have excellent bio- chondral chondrocytes and MG63 cells (osteoblast-like
compatibility (Beder and Eade, 1956; Albrektsson et al, cells derived from hyman osteosarcoma) on thin films of
1981; Williams, 1981; Branemark, 1985; Donley and various implant materials sputter-coated onto standard
Gillette, 1991), studies have shown that cells can discern polystyrene culture dishes (Windeler et al, 1991;
differences in surface characteristics. In a study utilizing Hambleton et al, 1994). The target materials were titani-
human fibroblasts, Keller et al. (1990) found that sterili- um, alumina, zirconium, and calcium phosphate (Ca/P =
zation techniques significantly altered cell adherence 1.67). All surfaces were 800 A thick and had the same sur-
characteristics. While the percentage of cells attached to face morphology as the underlying polystyrene. Of the
acid-passivated cpTi was approximately the same as that various coatings, differences between two coatings pro-
on tissue culture polystyrene after 15, 30, and 60 minutes duced from the titanium target were the most intriguing.
in culture, those on surfaces sterilized by autoclaving, The films prepared in the inert environment had an
ethanol, or ethylene oxide exhibited considerably less amorphous TiO2 passivation layer at the surface, while
cell attachment. those prepared in the presence of oxygen were com-
Adherence to cpTi is altered not only by sterilization posed of crystalline TiO2 (Windeler et al, 1991). Thus,
techniques but also by surface roughness. Studies using these films allowed the responses of cells on surfaces
osteoblast-like cells from rat calvarial explants (Bowers et with identical chemistries, but different crystallographic
al, 1992) have shown that cells are sensitive to differ- structures, to be evaluated.
ences in roughness. These cells were cultured on materi- The response of MG63 to these surfaces suggested
als with Ra's ranging from 0.14 \im to 1.15 |um. Of the var- that cell proliferation alone was unable to assess the
ious surfaces, a sandblasted surface with intermediate effect of surface chemistry on cell response adequately
roughness (Ra = 0.87 jim) had consistently higher levels (Windeler et al, 1991). Phenotypic expression of the
of percent cell attachment than did the plastic controls MG63 cells, as indicated by changes in alkaline phos-
and the other titanium surfaces. These data, in conjunc- phatase specific activity, was modified by surface com-
tion with the observation that cells grown on sandblast- position, but differences in cell number and collagen
ed surfaces were morphologically different from those on production were not evident, demonstrating the impor-
the other surfaces, support the premise that cpTi surface tance of the examination of multiple parameters.
roughness has an effect on initial cell attachment and This was made even more evident when non-trans-
subsequent spreading (Kelleretal, 1990; Ben-Zeev, 1991; formed chondrocytes were examined on the sputter-
Bowers et al, 1992). Alteration in cell morphology has coated surfaces. Cells were derived from two regions of
been shown to alter the phenotypic expression of cul- rat costochondral cartilage which are at two distinct
tured cells (Folkman and Moscona, 1978; Watt et al, stages of endochondral maturation in vivo. Cells from the
1988). As will be discussed below, titanium surface resting zone are relatively immature, while those from
roughness has been shown to modulate cell proliferation the growth zone are more mature. The differential phe-
and differentiation, matrix synthesis, and local factor notype of the two cell populations is maintained through
production (Cochran et al, 1994; Martin et al, 1995; four passages in culture, thereby allowing for compari-
Kieswetter^ta/., 1996; Schwartz et al, 1996). son of cell responses to various factors as a function of
cell maturation (Boyan et al, 1988, 1992a,b, 1994;
PROLIFERATION Schwartz and Boyan, 1988; Schwartz et al, 1988a). By
The enormous number of permutations available with using this model, we have been able to demonstrate that
respect to the effects of surface energy, roughness, tex- chondrocytes are sensitive to surface composition.
ture, and composition on the efficacy of biocompatible Specifically, proliferation of both resting zone and growth
materials has led to attempts by several laboratories zone chondrocytes on alumina and calcium phosphate
(Davies et al, 1990; Keller et al, 1990; Ricci et al, 1991; materials was inhibited relative to that on plastic con-
Bowers etal, 1992; Cochran et al, 1994; Martin etal, 1995; trols, but not on either of the TiO2 surfaces described
KieswettereU!., 1996; Schwartz et al, 1996) to isolate one above (Hambleton et al, 1994). Despite the similar
of these characteristics and closely examine its effects on responses of these two cell types with respect to prolif-
cellular response in vitro. In our laboratory, we have used eration, the resting zone and growth zone cells exhibited
two separate experimental systems to investigate the differences in cellular morphology and other phenotypic
effects of surface composition and roughness on the pro- markers indicative of their cartilage maturation state
liferation, differentiation, and matrix production of cells when cultured on the same surface.
of mesenchymal origin (Windeler et al, 1991; Hambleton Chondrocytes and osteoblasts are also sensitive to

334 Crit Rev Oral Biol Ued 7(4):329-345 (1996)

Downloaded from cro.sagepub.com at University of Sydney on November 18, 2015 For personal use only. No other uses without permission.
surface roughness. We have examined the responses of ber and DNA synthesis with increasing surface rough-
rat chondrocytes and osteoblast-like MG63 cells utilizing ness. In contrast, growth zone cells grown on the three
an experimental system consisting of cpTi disks with var- roughest surfaces (FA, TPS, and CA) all had significantly
ious surface roughnesses, but essentially the same greater levels of [3H)-thymidine incorporation than seen
chemistry and surface oxide thicknesses (Martin et al, on tissue culture plastic, although cell number exhibited
1995; Kieswetterrtal., 1996; Schwartz etal, 1996). Four of a decreasing trend with increasing surface roughness.
the five surfaces varied in roughness, and the fifth had a Lian and Stein (Stein et al, 1990; Lian and Stein,
roughness that was comparable with that of one of the 1992, 1993) have hypothesized that decreased prolifera-
other test surfaces, but varied with respect to topograph- tion precedes the expression of the more differentiated
ical distribution of the rough areas. The different surface phenotype in culture. In light of this observation, it
roughnesses were produced by alterations in the manu- appears that, at least with respect to changes in prolifer-
facturing techniques. All disks were pre-treated with ation, cells grown on rougher surfaces may be at a more
hydrofluoric acid-nitric acid and washed (PT). PT disks advanced stage of development than their counterparts
were then: electropolished (EP), sandblasted with a fine on smoother surfaces. The differences between the
alumina grit, etched with HCl and H2SO4 and washed responses of the less mature cells (MG63s and resting
(FA), sandblasted with a coarse alumina grit, etched with zone cells) and those of the more mature growth zone
HCl and H2SO4 and washed (CA), or titanium-plasma- cells suggest that cell maturation state may be more
sprayed (TPS). The surfaces, from smoothest to roughest, important to cellular response than either cell type or
ranked as follows: EP, PT, FA, CA, and TPS. species. This is a rather intriguing concept, given that
Initial studies examined the effect of surface rough- when a material is implanted into the body, it is more
ness on osteoblast-like MG63 cells. All assays were con- likely to interact with relatively undifferentiated mes-
ducted 24 hours after cells reached confluence on stan- enchymal cells participating in the wound-healing
dard tissue culture plastic surfaces. As has been noted by response than with fully differentiated osteoblasts.
others, cell morphology was distinctly altered as a func-
tion of roughness. Cells grown on the rough TPS surfaces DIFFERENTIATION
were rounded in appearance and had cytoplasmic exten- Once the inflammatory response is resolved and bone
sions which spanned the distance between surface wound healing is well under way, it is critical that the
peaks. In contrast, those grown on the smooth EP sur- progenitor cells which have migrated into the site differ-
faces had a well-spread, flattened morphology similar to entiate into osteoblasts. Differentiation of mesenchymal
that of cells grown on plastic. cells is dependent on a highly variable and complex
Cell number also varied with surface roughness. The interplay among surrounding cells and various cytokines
number of osteoblasts on the roughest surface (TPS) was and growth factors present in the milieu. One marker of
significantly lower than that on tissue culture plastic, differentiation for endochondral chondrocytes and
while the number on the smoothest surface (EP) was osteoblasts is increased alkaline phosphatase specific
significantly greater. The three intermediate surfaces, PT, activity (Stein et al, 1990; Lian and Stein, 1992, 1993). In
FA, and CA, did not differ from the plastic controls with our lab, we have used the costochondral chondrocyte
respect to cell number (Martin et al, 1995). The DNA syn- and MG63 osteoblast-like cell models to examine the
thesis of cells grown on all titanium surfaces was signifi- effects of surface composition and roughness on differ-
cantly lower than that of those grown on plastic. In gen- entiation (Windeler et al, 1991; Hambleton et al, 1994;
eral, the extent of [3H]-thymidine incorporation was Martin et al, 1995; Schwartz et al, 1996).
inversely related to surface roughness. While the proliferative responses of MG63
Cellular response to titanium surface characteristics osteoblast-like cells grown on sputter-coated surfaces of
has also been shown to vary with cell type. The prolifer- different compositions were essentially equivalent, the
ative responses of fibroblasts and epithelial cells are sig- extent of differentiation, as assessed by alkaline phos-
nificantly affected by titanium surface characteristics phatase specific activity, varied with surface composi-
(Cochran et al, 1994). Examination of resting zone and tion. Of particular interest were the responses of the cells
growth zone chondrocyte response to surface roughness when cultured on the two surfaces produced from the
suggests that the observed effects may be cell-matura- titanium target. Alkaline phosphatase specific activity in
tion-dependent as well (Schwartz et al., 1996). With the cell layer (cells plus matrix vesicles, organelles asso-
respect to proliferation, the characteristics of the less ciated with initial calcification) of cultures grown on
mature resting zone chondrocytes (Schwartz et al, 1996) amorphous TiO2 was three times that of similarly pre-
are very similar to those of MG63 cells (Martin et al, pared cultures grown in plastic. In addition, the enzyme
1995), which have been hypothesized to be an immature specific activity of isolated cells (no matrix vesicles)
osteoblast-like cell line (Franceschi et al, 1985). In both which had been cultured on the amorphous TiO2 was
cases, the general trend was one of decreasing cell num- approximately six times that of cells grown on poly-

7(4):329-345 (1996) Crit Rev Oral Biol Med 335

Downloaded from cro.sagepub.com at University of Sydney on November 18, 2015 For personal use only. No other uses without permission.
TABLE 1 TABLE 2
Cellular Responses of Resting Zone Cellular Responses of Growth Zone
Chondrocytes (RCs) to Surfaces of Chondrocytes (RCs) to Surfaces of
Various Roughnesses Various Roughnesses
EP PT FA CA TPS EP PI FA CA IPS
Proliferation Proliferation
Cell number — — ii i u Cell number — —
[3H]-thymidine incorporation — — i i i [3H]-thymidine incorporation — — T TT
Differentiation Differentiation
Cell layer ALPase — — i i i Cell layer ALPase — — — — 14
Matrix production Matrix production
Collagen synthesis — — — T T Collagen synthesis — — 4 4 4
Comparisons of the PT, FA, CA, and TPS are relative to the EP sur- Comparisons of the PT, FA, CA, and TPS are relative to the EP sur-
face. RC cells were cultured on one of the following five titanium face. GC cells were cultured on one of the following five titanium
surfaces, ranked from smoothest to roughest: electropolished (EP), surfaces, ranked from smoothest to roughest: electropolished (EP),
pre-treated surface (PT), fine-grit-blasted (FA), coarse-grit-blasted pre-treated surface (PT), fine-grit-blasted (FA), coarse-grit-blasted
(CA), and titanium-plasma-sprayed (TPS). ALPase: alkaline phos- (CA), and titanium-plasma-sprayed (TPS). ALPase: alkaline phos-
phatase specific activity. phatase specific activity.

styrene. In contrast, MG63 cells grown on the crystalline phatase specific activity in the intact cell layer was ana-
TiO2 surface had reduced levels of both cellular and cell lyzed. For growth zone chondrocytes cultured on the
layer alkaline phosphatase specific activity relative to rough CA surface, there was a substantial difference in
controls (Windeler et al, 1991). Cellular and cell layer the enzyme activity in the intact cell layer compared with
alkaline phosphatase activity in cultures of both resting that of isolated cells. This indicated that alkaline phos-
zone and growth zone chondrocytes grown on these sur- phatase activity in the matrix must have been very high.
faces was similar to that of the osteoblasts (Hambleton As discussed previously, the high level of cell-layer alka-
etal, 1994). line phosphatase activity can be attributed to the pres-
Since the presence of extracellular matrix vesicles ence of matrix vesicles. The higher levels of matrix vesi-
contributes significantly to cell layer alkaline phos- cle enzyme specific activity suggest that the rougher sur-
phatase, but not to that of the isolated cells (Boyan et al, faces may enhance matrix vesicle production and, hence,
1988; Schwartz et al, 1988b), it appears that surface com- the propensity toward calcification by the growth zone
position has a profound effect on matrix vesicle produc- chondrocytes.
tion and/or activity. The facts that surface composition
can modulate cellular differentiation, as evidenced by MATRIX PRODUCTION
increases in alkaline phosphastase activity, and that The ability of a surface to elicit a desirable wound-heal-
matrix vesicles play a role in primary calcification sug- ing response is much more complicated than its just
gest that surface energy can have significant sequelae being able to increase the number of pluripotent mes-
with respect to the osteointegration process. enchymal cells present at the site and to initiate their
Studies utilizing our surface roughness experimental development along the proper cell lineage pathway.
system suggest that, given essentially the same surface When osteointegration is the desired result, the proper
chemistry, surface roughness may play an important role pathway entails development of cells that are capable of
in cell differentiation. In the MG63 cell cultures (Martin et producing a calcifiable matrix. This may occur via the
al, 1995), as well as the chondrocyte cultures (Schwartz secretion of calcifiable osteoid by osteoblasts or by the
et al, 1996), cellular alkaline phosphatase specific activi- production of a calcifying matrix by cells participating in
ty was decreased. Further, a cell-maturation-dependent endochondral ossification at the wound site.
influence was noted. Resting zone cells appeared to be Regardless of cell type, adequate and appropriate
more sensitive to differences in surface roughness than matrix production is necessary for development of a cal-
were growth zone cells. Cellular alkaline phosphatase cifiable matrix. We have studied the responses of
activity of resting zone chondrocytes grown on all sur- osteoblasts and chondrocytes to various surface compo-
faces, except the smoothest surface (EP), was signifi- sitions and roughness in our two model systems
cantly lower than that of the plastic controls, while cellu- (Windeler et al, 1991; Hambleton et al, 1994; Martin et al,
lar alkaline phosphatase activity of growth zone chon- 1995; Schwartz et al, 1996). Proteoglycan and collagen
drocytes was only significantly different from that of con- production were assessed by [35S)-sulfate incorporation
trols on the two roughest surfaces (CA and TPS). This into the cell layer and [3H]-proline incorporation into
same phenomenon was evident when the alkaline phos- collagenase-digestible protein, respectively. No differ-

336 Crit Rev Oral Biol Ued 7(4):329-345 (1996)

Downloaded from cro.sagepub.com at University of Sydney on November 18, 2015 For personal use only. No other uses without permission.
TABLE 3 tem, the TiO2 layer was flat, while that on the surfaces
utilized in the roughness study was less so, since it cor-
Cellular Responses of Osteoblast-like Cells responds to the underlying surface conformation. With
(MG63s) to Surfaces of Various Roughnesses the exception of increased cell layer alkaline phos-
EP PT FA CA TW" phatase specific activity, the responses of resting zone
Proliferation cells grown on the extremely smooth sputter-coated TiO2
Cell number — i i i ii surfaces were identical to those of cells grown on the tis-
[3H]-thymidine incorporation — — l<l i i<l sue culture plastic. A slight increase in surface roughness
Differentiation (as seen in the EP and PT surfaces), however, had effects
Cell layer ALPase _ _ U — U on the cells that were indicative of a retardation of their
Matrix production ability to develop into more mature phenotypes relative
Collagen synthesis — — T TT TT to cells grown on plastic. A continued increase in surface
Comparisons of the PT, FA, CA, and TPS are relative to the EP sur- roughness (such as seen on the FA, CA, and TPS disks),
face. M G 6 3 cells were cultured on one of the following five titani- however, had the opposite effect. Overall, cells grown on
um surfaces, ranked from smoothest to roughest: electropolished these rougher surfaces appeared to be phenotypically
(EP), pre-treated surface (PT), fine-grit-blasted (FA), coarse-grit-
more mature than those grown on the plastic controls.
blasted (CA), and titanium-plasma-sprayed (TPS). ALPase: alkaline
phosphatase specific activity. The responses to increasing surface roughness by
growth zone cells were different. Initially, as seen with
the resting zone cells, growth zone cells exhibited
ences were evident in collagen production between increased alkaline phosphatase activity relative to the
MG63 cells grown on either the amorphous or crystalline plastic controls on the smooth sputter-coated surface.
TiO2 surfaces and the polystyrene controls (Windeler et An increase in surface roughness to the values seen on
al, 1991). When examined on surfaces of various rough- EP and PT had essentially no effect on cellular response
nesses, however, collagen production by the MG63 cells (Tables 1 and 2). Cells grown on these surfaces had
on the rougher surfaces (TPS, CA), although not signifi- essentially the same biochemical characteristics as
cantly different from that of polystyrene controls, was those grown on the control polystyrene tissue culture
significantly higher than on the smoother cpTi surfaces. plates. Unlike the resting zone cells, growth zone cells
Proteoglycan production, on the other hand, was signifi- grown on the rougher surfaces showed increased prolif-
cantly reduced relative to the plastic control on all of the eration and decreased collagen synthesis.
titanium surfaces. Overall, the effect was most pro-
The osteoblast response to the various surfaces
nounced in cells grown on the CA surface. Cells grown on
(Table 3) was very similar to that of the resting zone cells.
these surfaces produced less proteoglycan than those
grown on PT, FA, and TPS. When the MG63 cells were first characterized, they were
found to exhibit many of the characteristics of immature
Unlike MG63 cells, the chondrocytes were sensitive
or pre-osteoblasts (Franceschi et al, 1985). The similarity
to surface chemistry. With the exception of cells grown
in responses of both immature cell lines supports the
on the amorphous titanium surfaces, all chondrocytes
hypothesis that cell maturation state affects cellular
grown on the smooth sputter-coated surfaces showed a
response to a surface.
marked decrease in both proteoglycan and percent colla-
gen production relative to the plastic controls
(Hambleton et al, 1994; Schwartz et al, 1996). When
Local Factor Production
grown on the titanium surfaces of various roughnesses, Cells at different maturation states are known to produce
the level of collagen production, relative to plastic con- different repertoires of cytokines, growth factors, and
trols, was less; the responses of resting zone and growth proteins (Grigoriadis et al, 1988, 1989; Nakahara et al,
zone chondrocytes to surface roughness, however, were 1991). Studies using chondrocytes in our culture system
markedly different. Collagen production by the less have demonstrated that differences in extracellular
mature resting zone chondrocytes was significantly less matrix production by resting zone and growth zone cells
than that of controls on the two smoother surfaces (EP are very pronounced, particularly with respect to the
and PT). In contrast, collagen production by the more composition of extracellular matrix vesicles (Bonewald et
mature growth zone chondrocytes on the two smooth al, 1992; Boyan et al, 1992b; Sela et al, 1992; Swain et al,
surfaces was equivalent to that of plastic controls, but 1992). For example, although the matrix vesicles of both
was significantly less on the three rougher surfaces. cell types have approximately the same levels of phos-
The notion that surface roughness modulates cell pholipase A2 activity, they differ considerably with
response in a cell-maturation-dependent manner is sup- respect to their alkaline phosphatase specific activity.
ported by studies examining the response of resting zone Since these organelles play a role in calcification, the
and growth zone chondrocytes to the amorphous titani- differences in composition are thought to be related to
um dioxide passivation layers. In the sputter-coated sys- the cells' relative abilities to calcify their matrices. The

7(4):329-345 (1996) Crit Rev Oral Biol Med 337

Downloaded from cro.sagepub.com at University of Sydney on November 18, 2015 For personal use only. No other uses without permission.
 would occur not only from the sur-
rounding tissue toward the implant,
but also from the implant toward the
Bone
outer portions of the defect.
Furthermore, differences in the
cytokine and growth factor profile
Implant
present at the bone-implant interface
may result in differences in the quali-
ty, extent, and rate of bone formation.
Osteoblasts In the event that the cells on the
surface were incapable of producing,
or not sufficiently stimulated to pro-
Mesenchymal cells duce, the appropriate complement of
adherent to implant
autocrine and paracrine factors for
surface
adequate stimulation of osteogene-
Osteocyte in sis by the distal cells, then it would
its lacuna
be expected that the wound-healing
response would culminate in encom-
passing the device in a fibrous cap-
sule and not osteointegration. This
Figure 2. The implant surface plays a pivotal role in osteointegration. Mesenchymal cells same outcome would also be expect-
on the surface must be able to communicate with cells in the surrounding tissue before an ed if the distance between the cells
osteogenic environment can be established. The latent or active local factors produced by on the implant surface and those in
these cells may have autocrine/paracrine effects. Autocrine effects of these factors would the surrounding tissue was sufficient-
serve to induce the mesenchymal cells at the implant surface along the proper differenti-
ation in the osteoblastic pathway, while paracrine effects would serve to decrease the
ly large as to be prohibitive for
extent of resorption by osteoclasts and stimulate osteoid synthesis by osteoblasts. paracrine communication between
the two populations.
Recent studies in our laboratory
production of matrix vesicles with different biochemical have begun to examine how material surface characteris-
compositions is a salient feature of the differentiation tics may affect the complement of cytokines produced by
process that culminates in fully mature growth zone cells cells at the surface. By understanding this initial set of
from the less mature resting zone phenotype. The pro- growth factors and cytokines produced at the implant
gression along this lineage pathway is dependent on the site, and combining this information with current studies
communication between cells that are in the pathway of bone formation, we hope that, in the future, we can
with those in the surrounding tissue. It is our belief that examine the entire cascade of events involved in the
since mesenchymal cells are capable of differentiating mineralization of tissue at the device interface. Currently,
along one of several pathways based upon the concen- our focus is on examining the effects of surface rough-
tration and presence of local factors (Grigoriadis et al., ness on the production of two local factors,
1988, 1989; Nakahara et al., 1991), this principle can be prostaglandin E2 (PGE2) and transforming growth factor-
extended to the interactions between cells adherent to P, (TGFP), that are important in both wound healing and
an implant surface and those in the surrounding vicinity bone formation (Chyun and Raisz, 1984; Bonewald et al.,
(Fig. 2). Our premise, therefore, is that cells on the 1990, 1992; Dworetzky et al, 1990; Joyce et a/., 1990; Raisz
implant surface are capable of affecting those at sites and Fall, 1990; Schwartz et al., 1992b). We have conduct-
distal to the implant by the type, concentration, and acti- ed these studies using the same five surfaces previously
vation state of the growth factors they produce. Likewise, used to investigate the effects of surface roughness on
via the same autocrine/paracrine signals, cells that are proliferation, differentiation, and matrix production of
distal to the implant are also capable of exerting an influ- cells (Kieswetter eta!., 1996).
ence on those at the device surface. It would be expect- The hypothesis that MG63 cells grown on rougher
ed, therefore, that in an osteogenesis-favorable environ- surfaces (FA, CA, and TPS) are more differentiated than
ment, cells on the device surface would produce a reper- those grown on smoother surfaces is supported by
toire of growth factors that would induce differentiation enhanced production of both factors. As has been seen,
of the mesenchymal cells in the surrounding tissue along proliferation, particularly as indicated by cell number, is
a pathway that would preferentially lead to the formation significantly reduced on the rougher surfaces. Although a
of a calcifiable matrix. This being the case, bone resorp- smaller number of cells is present on these surfaces, the
tion in the vicinity would be repressed, and formation total levels of PGE2 and TGFI3 were significantly greater

338 Crit Rev Oral Biol Ued

Downloaded from cro.sagepub.com at University of Sydney on November 18, 2015 For personal use only. No other uses without permission.
7(4):329-345 (1996)
(Kieswetter et al, 1996). The combination of decreased gy can provide us with considerable insights into the
cell number and increased PGE2 and TGFp production on ability of a surface to modulate cell response.
the rougher surfaces indicates that these cells are syn- The various theories with respect to what type of sur-
thesizing considerably more local factor on a per-ce\\ face best promotes osteointegration by the surrounding
basis. tissue, however, can only be properly elucidated by stud-
The effect of surface roughness on the production of ies in vivo. The surgical injury necessary for implantation
local factors is not straightforward, since the exact role of is generally expected to heal according to the normal
PGE2 and TGFp in calcification still remains under active mechanisms of inflammation and wound healing. The
investigation (Chyun and Raisz, 1984; Bonewald et al, presence of an implant, however, affords the opportunity
1990, 1992; Dworetzky etal, 1990; Joyce etal, 1990; Raisz for these normal mechanisms to be altered. In order for
and Fall, 1990; Schwartz et al, 1992b). Furthermore, the proper bone healing and repair to occur, the device must
results indicate that changes in production capacity by direct the inflammatory and wound-healing responses
the osteoblasts may vary with the local factor as well as along normal directions. Perhaps more importantly, it
the surface being examined (Kieswetter et al, 1996). For must not direct the body toward a pronounced, chronic
instance, when normalized to cell number, production of inflammatory response, whereby the implant site
PGE2 on the FA surface was found to be elevated with becomes encapsulated with fibrous tissue, and bone
respect to production on the smoother EP and PT sur- resorption around the implant ensues. The direction of
faces. TGFp production on a per-ce\\ basis, however, was the cellular response is initiated immediately upon the
statistically identical on all three surfaces. This indicates placement of the device into the wound site, since this is
that, relative to cells grown on the EP and PT surfaces, the time at which the material becomes coated with
the cells grown on the FA surface are more prone to a endogenous proteins and other factors.
drastic increase in the up-regulation of PGE2 production, The classic biocompatibility studies (Beder and
but not TGFp production. The differences in the local Eade, 1956; Bechtol et al, 1959; Hall et al, 1966; Homsy et
concentrations of these individual factors, therefore, may al, 1968; Hall, 1970; Homsy, 1970; Homsy and
account for some of the similarities between the FA and Armediades, 1971; Hulbert et al, 1971, 1972, 1973b;
the two other smoother surfaces with respect to cell pro- Natiella et al, 1974, 1977; Nelson et al, 1977) clearly
liferation and differentiation, but not matrix production. addressed the effects of surface composition on in vivo
The amount of TGFp produced by MG63 cells on rough response. The importance of the other factors, however,
surfaces is comparable with levels shown to stimulate was generally not taken into consideration. At the time of
osteoblast-like cells in vitro (Bonewald et al, 1992). While the early studies, some materials were clearly found to
levels of PGE2 are slightly lower than necessary to stim- be unacceptable, and their use in vivo has been aban-
ulate osteoblasts in vitro (Farr et al, 1984; McCarthy et al, doned. The remaining satisfactory materials resulted in
1991), these factors may act in concert with others to the broad classifications of materials discussed at the
enhance osteoblastic differentiation. beginning of this text. Given the advent of many new and
This suggests the possibility that implant surface novel materials such as polyfdesaminotyrosinetyrosyl-
roughness may affect not only the concentration of hexyl) ester carbonate (Pulapuna and Kohn, 1992;
cytokines produced by the cells, but also how the cells Nathan et al, 1993; Ertel et al, 1995), as well as the com-
react to hormonal or paracrine stimuli. While it is unlike- bination of more traditional materials into composites
ly that PGE2 and TGFp alone can account for differences (Guild and Bonfield, 1993; Knowles and Bonfield, 1993),
in bone formation observed adjacent to implants in vivo, the question of how the body interacts with materials of
it is believed that, for the mechanisms of osteointegra- various compositions remains of importance.
tion at the bone-biomaterial interface to be understood, Surface roughness has been shown to affect the abil-
the effects of materials on the surrounding cells and ity of materials to osteointegrate. A variety of studies
their production of cytokines, growth factors, and other appears to indicate that, in general, devices with
local mediators must be understood. smoother surfaces are more likely to result in the forma-
tion of thicker fibrous encapsulation than are those with
Surface Effects on the Wound-healing Response rougher surfaces (Albrektsson et al, 1981, 1985; Linder et
and Subsequent Osteogenesis in vivo al, 1983, 1988; Thomas and Cook, 1985; Albrektsson and
While it is expected that perhaps one of the four surface Hansson, 1986; Chehroudi etaf., 1989; HaddadeU!., 1987;
characteristics may play a more dominant role in some Inoueetfll., 1987; Thomas eta!., 1987; Ricci et al, 1991). In
situations than others, it is important to realize that the a study by Albrektsson and Hanson (1986), the interface
four are interrelated and that their relationship will dic- of a smooth titanium (sputter-coated or vapor-deposit-
tate the course whereby osteogenesis occurs. Carefully- ed) surface was found to consist of a 200-to-400-A-wide
thought-out and controlled in vitro studies on the effects layer of proteoglycan associated with collagen bundles
of surface composition, roughness, texture, and/or ener- that were 1 to 2 nm from the implant surface. The fibrous

7(4):329-345 (1996) Crit Rev Oral Biol Med 339

Downloaded from cro.sagepub.com at University of Sydney on November 18, 2015 For personal use only. No other uses without permission.
capsule surrounding smooth, but not rough, implants surrounding tissue initiates a sequence of events neces-
has been evident to others (Linder et al, 1983, 1988; sary for bone formation. The joining of the two opposing
Thomas and Cook, 1985; Inoue et al, 1987; Thomas et al, fronts (one growing from the implant surface toward the
1987; Chehroudi et al, 1989; Ricci et al, 1991). Different tissue, the other from the bone toward the implant) sig-
surface textures result in different contacts between cells nifies osteointegration. The modeling and remodeling
and the surrounding matrix. This, in turn, affects the cel- that occur in this region, and our subsequent inability to
lular interactions at the surface and produces differences determine the directional nature of the bone growth in
in the types of bonding that can occur at the interface. this area, are further indications of successful wound
Recent studies by Gotfredsen et al. (1995) have fur- repair.
ther confirmed previous evidence of increased osteointe- Successful wound repair is dependent upon inter-
gration in the vicinity of rougher implants (Thomas and communication of cells participating in the healing
Cook, 1985; Carlsson eta/., 1988; Buseretal, 1991; Ricci et response. It is therefore critical that cells at an implant
al, 1991). In these studies, implants of various configu- surface be capable of communicating with cells in the
rations and roughnesses were implanted into rabbit tib- tissue surrounding the device. In order to optimize the
ias. The removal torque of relatively smooth machined possibility of osteointegration, we must understand not
implants (Ra = 31 ±0.12 |im) was compared with that of only how an implant's surface characteristics, as deter-
a rougher titanium surface (Ra = 61 ± 0.03 jim) obtained mined by the composite effect of surface energy, compo-
by grit-blasting with titanium dioxide particles on the sition, roughness, and topography, affect the initial bio-
order of 10 to 53 [im. Both cylindrical and screw-type logical response, but also how these characteristics may
implants were made with each surface treatment. In both direct longer-term biological responses, such as cell pro-
cases, the rougher surfaces had significantly higher liferation, differentiation, matrix production, and, ulti-
removal torques than did the smoother surfaces. These mately, bone formation.
differences were evident as early as three weeks after
implantation and continued to be significant at 12 weeks Summary and Future Directions
post-implantation (Gotfredsen et al, 1995). Surface roughness has been shown to affect osteoblast
The improved response seen at the longer time peri- proliferation, differentiation, matrix synthesis, and local
ods is dependent upon the development of viable bone factor production. The studies described in the current
at this interface. Traditionally, it is thought that the review provide insight into how osteoblasts interact with
development of bone at the site of an implant occurs several different types of implant surfaces. By under-
from the surrounding distal tissue toward the device. The standing how changes in the nature of a surface affect
ability to sustain bidirectional bone formation at an the maturation state and local factor production of cells
implant surface in humans is supported by recent stud- growing on it, investigators can design surfaces that pref-
ies of porous plugs implanted into the medial femoral erentially induce osteointegration at the implant site. A
condyles of older humans (67 ± 9 years; n = 19) major finding, described in this paper, is that osteoblasts
(Bloebaum et al, 1994). Bone apposition and remodeling exhibit a more mature phenotype when grown on
were significantly enhanced at the implant interface rel- rougher surfaces. Surface roughness and other character-
ative to that found in peripheral tissue. One and three istics needed for optimal osteointegration, however,
weeks prior to retrieval, the subjects participated in a fluo- remain to be elucidated. Moreover, studies are needed
rochrome labeling regime. At this cancelous bone site, which more clearly define the relationship between the
the presence of fluorochrome labels was noted within characteristics of an ideal implant surface and the heal-
the porous coating in samples that were in situ between 6 ing process. In particular, the effects of local factors and
and 11 weeks. This suggests that appositional bone hormones on the responses of cells at the implant sur-
growth was occurring (i.e., bone was growing out from the face are necessary. This new information will help us
implant). No bone ingrowth was apparent when gaps of achieve better clinical results.
greater than 50 jum existed between the porous coating In dentistry, the implanted device interfaces with a
and the surrounding bone, suggesting that the intercom- variety of tissues (epithelial, connective tissues, and
munication between cells on the surface and those in the bone). By understanding the implant surface characteris-
surrounding tissue was hindered, possibly due to the tics which are optimal for each one of these tissues,
size of the gap or the presence of micromotion and the investigators will be able to design better implants by
pooling of blood in the interface. customizing specific regions of the implant for each tis-
Cells adherent to, and growing on, rougher surfaces sue type. While currently used devices are generally sat-
appear to be more adept at communicating their isfactory, development of a c'-avice designed to optimize
osteogenic potential via paracrine and signaling mecha- the responses of the cells at its surface may result in
nisms to other cells in the vicinity (Schwartz et al, 1991, faster and more solid integration of the device into the
1992a, 1993). Response to this message from cells in the host tissue.

340 Crit Rev Oral Biol Med 7(4):329-345 (1996)

Downloaded from cro.sagepub.com at University of Sydney on November 18, 2015 For personal use only. No other uses without permission.
 Acknowledgments Boyan BD (1990). Stimulation of plasma membrane
This research was supported by the State-Industry-University and matrix vesicle enzyme activity by transforming
Cooperative Research Center for the Enhancement of the growth factor-beta in osteosarcoma cell cultures. 1 Cell
Biology/Biomaterials Interface at the University of Texas Health Science Physiol 145:200-206.
Center at San Antonio (NSF EEC 9209612), by NIH grants DE- Bonewald LF, Schwartz Z, Swain LD, Boyan BD (1992).
05937 and DE08603, and by grants from the Implant Dentistry Stimulation of matrix vesicle enzyme activity in
Research and Education Foundation, and the Ministry of Health, osteoblast-like cells by 1,25(OH)2D3 and transforming
Israel. growth factor beta (TGF beta). Bone Miner 17:139-144.
Bowers K, Keller JC, Randolph B, Wick D, Michaels C
(1992). Optimization of surface micromorphology for
enhanced osteoblast responses in vitro. \nt ) Oral
REFERENCES Maxillofac Implants 7:302-310.
Albrektsson T, Hansson HA (1986). An ultrastructural Boyan BD, Schwartz Z, Swain LD, Carnes DL, Zislis T
characterization of the interface between bone and (1988). Differential expression of phenotype by rest-
sputtered titanium or stainless steel surfaces. ing zone and growth region costochondral chondro-
Biornaterials 7:201-205. cytes in vitro. Bone 9:185-194.
Albrektsson T, Branemark P-I, Hansson HA, Lindstrom J Boyan BD, Schwartz Z, Swain LD (1992a). \n vitro studies
(1981). Osseointegrated titanium implants. on the regulation of endochondral ossification by vi-
Requirements for ensuring a long-lasting, direct tamin D. Crit Rev Oral Biol Med 3:15-30.
bone-to-implant anchorage in man. Ada Orthop Scand Boyan BD, Schwartz Z, Swain LD (1992b). Cell matura-
52:155-170. tion-specific autocrine/paracrine regulation of matrix
Albrektsson T, Hansson HA, Ivarsson B (1985). Interface vesicles. Bone Miner 17:263-268.
analysis of titanium and zirconium bone implants. Boyan BD, Dean DD, Sylvia VL, Schwartz Z (1994).
Biomaterials 6:97-101. Nongenomic regulation of extracellular matrix events
Aspenberg P, Goodman SB, Toksvig-Larson S, Ryd L, by vitamin D metabolites. I Cell Biochem 56:331-339.
Albrektsson T (1992). Intermittent micromotion Branemark P-I (1985). Tissue integrated prosthesis
inhibits bone ingrowth: Experiment using titanium osteointegration in clinical dentistry. In: Tissue-inte-
implants in rabbits. Ada Orthop Scand 63:141-145. grated prostheses: Osseointegration in clinical den-
Baier RE, Meyer AE (1988). Future directions in surface tistry. Branemark P-I, Zarb GA, Albrektsson T, editors.
preparation of dental implants. I Dent Educ 52:788- Chicago: Quintessence Publishing Co., pp. 11-76.
791. Brown IW, Ring PA (1985). Osteolytic changes in the
Baier RE, Meenaghan MA, Hartman LC, Wirth IE, Flynn upper femoral shaft following porous-coated hip
HE, Meyer AE (1988). Implant surface characteristics replacement. I Bone]tSurg |Br) 67(B)218-221.
and tissue interaction. I Oral Implantol 13:594-606. Buser D, Schenk R, Steinemann S, Fiorellini I, Fox C,
Bechtol CO, Ferguson AB, Laiberte F (1959). Metals and Stich H (1991). Influence of surface characteristics on
engineering in bone and joint surgery. Baltimore: bone integration of titanium implants. A histomor-
Williams &Wilkins. phometric study in miniature pigs. I Biomed Mater Res
Beder OE, Eade G (1956). An investigation of tissue tol- 25:889-902.
erance to titanium metal implants in dogs. Surgery Cameron HU, Pilliar RM, MacNab I (1973). The effect of
39:470-473. movement on the bonding of porous metal to bone. J
Ben-Zeev A (1991). Animal cell shape changes and gene Biomed Mater Res 7:301-311.
expression. Bioessays 13:207-212. Carlsson L, Rostlund T, Albrektsson B, Albrektsson T
Bloebaum RD, Bachus KN, Momberger NG, Hofmann AA (1988). Removal torques for polished and rough tita-
(1994). Mineral apposition rates of human cancellous nium implants. \nt I Oral Maxillofac Implants 3:21-24.
bone at the interface of porous coated implant. ) Chehroudi B, Gould TRL, Brunette DM (1989). Effects of
Biomed Mater Res 28:537-544. grooved titanium-coated implant surface on epithe-
Bobyn JD, Pilliar RM, Cameron HU, Weatherby GC, Kent lial cell behavior in vitro and in vivo. } Biomed Mater Res
GM (1980a). The effect of porous surface configura- 23:1067-1085.
tion on the tensile strength of fixation of implants by Chyun YS, Raisz LG (1984). Stimulation of bone forma-
bone growth. Clin Orthop 149:291-298. tion by prostaglandin E2. Prostaglandins 27:97-103.
Bobyn ID, Pilliar RM, Cameron HU, Weatherby GC Cochran DL, Simpson ], Weber H, Buser D (1994).
(1980b). The optimum pore size for the fixation of Attachment and growth of periodontal cells on
porous coated metal implants by the ingrowth of smooth and rough titanium. \nt ) Oral Maxillofac
bone. Clin Orthop 150:263-270. Implants 9:289-297.
Bonewald LF, Schwartz Z, Swain LD, Ramirez B, Poser), Cook SD (1991). Clinical, radiographic and histologic

7(4).-329-345 (1996) Crit Rev Oral Biol Med 341

Downloaded from cro.sagepub.com at University of Sydney on November 18, 2015 For personal use only. No other uses without permission.
 evaluation of retrieved human noncemented porous- shape in the presence of dynamic shear. ] Biomed Mater
coated implants. I Long-term Effects of Medical Implants Res 27:247-253.
1:11-51. Goodman SB, Wang IS, Doshi A, Aspenberg P (1993b).
Cook SD, Thomas KA, Haddad RJ (1988). Histologic Difference in bone in growth after one versus two daily
analysis of retrieved human porous-coated total joint episodes of micromotion: Experiments with titanium
components. Clin Orthop 234:90-98. chambers in rabbits. J Biomed Mater Res 27:1419-1424.
Davies IE, Lowenberg B, Shiga A (1990). The bone-titani- Goodship AE, Kenwright I (1985). The influence of
um interface in vitro. J Biomed Mater Res 24:1289-1306. induced micromovement upon the healing of experi-
DePalma VA (1976). Correlation of surface electrical mental tibial fractures. J Bone ]t Surg 67(B):650-655.
properties with initial events in bioadhesion (doctor- Gotfredsen K, Wennerberg A, Johansson C, Skovgaard
al dissertation). Buffalo, NY: State University of New LTT, Hjorting-Hansen E (1995). Anchorage of TiO2-
York at Buffalo. blasted, HA-coated, and machined implants: An
DePalma VA, Baier RE (1978). Microfouling of metallic experimental study with rabbits. J Biomed Mater Res
and coated metallic flow surfaces in model heat 29:1223-1231.
exchange cells. In: Proceedings of the Ocean Thermal Grigoriadis AE, Heersche JN, Aubin IE (1988).
Energy Conversion (OTEC) Biofouling and Corrosion Differentiation of muscle, fat, cartilage, and bone
Symposium, US Department of Energy, PNL-SA-7 from progenitor cells present in a bone-derived clo-
115, pp. 89-106. nal cell population: Effect of dexamethasone. 1 Cell
Donley DG, Gillette WB (1991). Titanium endosseous Biol 106:2139-2151.
implant-soft tissue interface. ) Periodontol 62:153-160. Grigoriadis AE, Aubin JE, Heersche IN (1989). Effects of
Dworetzky SI, Fey EG, Penman S, Lian JB, Stein JL, Stein dexamethasone and vitamin D3 on cartilage differen-
GS (1990). Progressive changes in the protein compo- tiation in a clonal chondrogenic cell population.
sition of the nuclear matrix during rat osteoblast dif- Endocrinology 125:2103-2110.
ferentiation. Proc Natl head Sci 87:4605-4609. Guild FJ, Bonfield W (1993). Predictive modeling of
Ericsson T, Ericsson Y (1967). Effect of partial fluorine hydroxyapatite-polyethylene composite. Biomaterials
substitution on the phosphate exchange and protein 14:985-993.
adsorption of hydroxyapatite. Helv Odontol Acta 11:10- Haddad RJ Jr, Cook SD, Thomas KA (1987). Biological fix-
14. ation of porous coated implants. J Bone ]t Surg
Ertel SI, Kohn ], Zimmerman MC, Parsons JR (1995). 69(A):1459-1466.
Evaluation of polyfDTH carbonate), a tyrosine- Hall CW, Liotta D, O'Neal RM, Adams JG, DeBakey ME
derived degradable polymer, for orthopedic applica- (1966). Medical applications of the velour fabrics. Ann
tions. J Biomed Mater Res 29:1337-1348. NY AcadSci 146:314-324.
Farr D, Pochal W, Brown M, Shapiro E, Weinfeld N, Dziak Hall JA, editor (1970). Bioceramics in medicine. New
R (1984). Effects of prostaglandins on rat calvarial York: Wiley.
bone-cell calcium. Arch Oral Biol 29:885-991. Hambleton JF Schwartz Z, Khare A, Windeler SW, Luna M,
Folkman J, Moscona A (1978). Role of cell shape in Brooks BP, et al. (1994). Culture surfaces coated with
growth control. Nature 273:345-349. various implant materials affect chondrocyte growth
Franceschi RT, lames WM, Zerlauth G (1985). 1,Alpha,25- and metabolism. I Orthop Res 12:542-552.
dihydroxyvitamin D3 specific regulation of growth, Hanein D, Sabanay H, Addadi L, Geiger B (1993).
morphology, and fibronectin in a human osteosarco- Selective interactions of cells with crystal surfaces.
ma cell line. I Cell Physiol 123:401-409. Implications for the mechanism of cell adhesion. I Cell
Gabriel BL, Gold J, Gristina AG, Kasemo B, Lausmaa J, Sci 104:275-288.
Harrer C, et al. (1994). Site specific adhesion of Hanein D, Geiger B, Addadi L (1994). Differential adhe-
Staphylococcus epidermidis (RP12) in Ti-Al-V metal sys- sion of cells to enantiomorphous crystal surfaces.
tems. Biomaterials 15:628-634. Science 263:1413-1416.
Galante T, Rostoker W, Luik R, Ray RD (1971). Sintered Hench LL, Ethridge EC (1982). Biomaterials: An interfa-
fiber metal composites as a basis for attachment of cial approach. New York: Academic Press.
implants to bone. Bone]t Surg [Am] 53(A): 101-114. Hench LL, Paschall HA (1973). Direct chemical bond of
Goodman SB, Aspenberg P (1992). The effect of ampli- bioactive glass-ceramic materials to bone and mus-
tude of micromotion on bone ingrowth into titanium cle. I Biomed Mater Res 7:25-42.
chambers implanted in the rabbit tibia. Biomaterials Hirschhorn IS, Reynolds IT (1969). Powder metallurgy
13:944-948. fabrication of cobalt alloy surgical implant materials.
Goodman SB, Toksvig-Larson S, Aspenberg P (1993a). In: Research in dental and medical materials.
Ingrowth of bone into pores in titanium chambers Korostoff E, editor. Proceedings of a symposium on
implanted in rabbits: effect of pore cross-sectional research in dental and medical materials,

342 Crit Rev Oral Biol Med 7(4):329-345 (1996)

Downloaded from cro.sagepub.com at University of Sydney on November 18, 2015 For personal use only. No other uses without permission.
 Metallurgical Society of AIME, Detroit, MI, Oct., 1968. 5:360-367.
New York, London-. Plenum Press, pp. 137-141. Kieswetter K, Schwartz Z, Hummert TW, Cochran DL,
Homsy C (1970). Biocompatibility in selection of materi- Simpson J, Dean DD, et al. (1996). Surface roughness
als for implantation. I Biomed Mater Res 4:341-356. modulates the local production of growth factors and
Homsy CA, Tullos HS, Anderson MS, King JW (1968). cytokines by osteoblast-like MG-63 cells. J Biomed
Prevention of interfacial corrosion of SMO stainless Mater Res (in press).
steel appliances. West Orthop Assoc. Kilpadi DV, Lemons JE (1994). Surface energy characteri-
Homsy CA, Armediades CD, editors (1971). Biomaterials zation of unalloyed titanium implants. I Biomed Mater
for skeletal and cardiovascular applications. New Res 28:1419-1425.
York: Wiley. Knowles JC, Bonfield W (1993). Development of a glass
Howlett CR, Evans MDM, Walsh WR, [ohnson G, Steele reinforced hydroxyapatite with enhanced mechanical
JG (1994). Mechanism of initial attachment of cells properties. The effect of glass composition and
derived from human bone to commonly used pros- mechanical properties and its relationship to phase
thetic materials during cell culture. Biomaterials changes. I Biomed Mater Res 27:1591-1598.
15:213-222. Lanyon LE (1987). Functional strain in bone tissue as an
Hulbert SF, Klawitter )] (1971). Application of porous objective, and controlling stimulus for adaptive bone
ceramics for the attachment of load-bearing internal remodelling. J Biomechan 20:1083-1093.
orthopedic applications. Biomed Mater Symp 2:161-229. Lian JB, Stein GS (1992). Concepts of osteoblast growth
Hulbert SF, Klawitter JJ, Leonard RB (1971). Compatibility and differentiation: Basis for modulation of bone cell
of bioceramics with the physiological environment. development and tissue formation. Crit Rev Oral Biol
In. Ceramics in severe environments. Kriegel WW, Med 3:269-305.
Palmour H III, editors. New York: Plenum Press, pp. Lian JB, Stein GS (1993). The developmental stages of
417-434. osteoblast growth and differentiation exhibit selec-
Hulbert SF, Morrison SI, Klawitter II (1972). Tissue reac- tive responses of genes to growth factors (TGFJ31) and
tion to three ceramics of porous and non-porous hormones (vitamin D and glucocorticoids). I Oral
structures. I Biomed Mater Res 6:347-374. Xmplantol 19:95-105.
Hulbert SF, Cooke FW, Klawitter II, Leonard RB, Sauer BW, Linder L, Albrektsson T, Branemark P-I, Hansson HA,
Moyle DD, et al. (1973a). Attachment of prostheses to Ivarsson B, Jonsson U, et al. (1983). Electron micro-
the musculo-skeletal system by tissue ingrowth and scopic analysis of the bone-titanium interface. Acta
mechanical interlocking. Biomed Mater Symp 4:1-23. Orthop Scand 54:45-52.
Hulbert SF, Levine SN, Moyle DD, editors (1973b). Linder L, Carlsson A, Marsal L, Bjursten LM, Branemark
Prostheses and tissue: The interfacial problem. New P-I (1988). Clinical aspects of osseointegration in
York: Wiley. joint replacement: a histological study of titanium
Hulbert SF, Matthews JR, Klawitter JJ, Sauer BW, Leonard implants. I Bone ]t Surg 70(B):550-555.
RB (1974). Effect of stress on tissue ingrowth into Martin )Y, Schwartz Z, Hummert TW, Schraub DM,
porous aluminum oxide. Biomed Mater Symp 5:85-97. Simpson J, Lankford I Jr, etal. (1995). Effect of titanium
Inoue T Cox JE, Pilliar RM, Melcher AH (1987). Effect of surface roughness on proliferation, differentiation,
the surface geometry of smooth and porous-coated and protein synthesis of human osteoblast-like cells
titanium alloy on the orientation of fibroblasts in vitro. (MG63). I Biomed Mater Res 29:389-401.
I Biomed Mater R^s 21:107-126. McCarthy TL, Centrella M, Raisz LG, Canalis E.(1991).
loyce ME, Roberts AB, Sporn MB, Bolander ME (1990). Prostaglandin E2 stimulates insulin-like growth factor
Transforming growth factor-p and the initiation of synthesis in osteoblast-enriched cultures from fetal
chondrogenesis and osteogenesis in the rat femur. I rat bone. Endocrinology 128:2895-2900.
Cell Bio! 110:2195-2207. Meyer AE, Baier RE, Natiella JR, Meenaghan MA (1988).
Kambara M, Norde W (1995). Influence of fluoride appli- Investigation of tissue/implant interactions during
cations on some physicochemical surface properties the first two hours of implantation. I Oral Xmplantol
of synthetic hydroxyapatite and human dental enam- 14:363-379.
el and its consequences for protein adsorption. Caries Nakahara H, Dennis JE, Bruder SP, Haynesworth SE,
Res 29:210-217. Lennon DP, Caplan Al (1991). In vitro differentiation of
Kenwright J, Goodship AE (1989). Controlled mechanical bone and hypertrophic cartilage from periosteal-
stimulation in the treatment of tibial fractures, din derived cells. Exp Cell Res 195:492-503.
Ortho 241:36-47. Nathan A, Zalipsky S, Ertel SI, Agathos SN, Yarmush ML,
Keller JC, Draughn RA, Wightman IP, Dougherty WJ, Kohn I (1993). Copolymers of lysine and polyethylene
Meletiou SD (1990). Characterization of sterilized CP glycol: a new family of functionalized drug carriers.
titanium implant surfaces. \nt I Oral Maxillofac Implants Bioconjugate Chemistry 4:54-62.

7(4):329-345 (1996) Crit Rev Oral Biol Med 343

Downloaded from cro.sagepub.com at University of Sydney on November 18, 2015 For personal use only. No other uses without permission.
Natiella IR, Armitage IE, Meenaghan MA (1974). Tissue Direct effects of 1,25-dihydroxyvitamin D3 and 24,25-
response to dental implants protruding through dihydroxyvitamin D3 on growth zone and resting zone
mucous membrane. Oral Sci Rev 5:85-105. chondrocyte membrane alkaline phosphatase and
Natiella IR, Meenaghan MA, Flynn HE (1977). Unilateral phospholipase A2 specific activities. Endocrinology
subperiosteal oral implants. II. A light microscopic 123:2878-2884.
study. Biomed Mater Res Syrnp Trans 1:73. Schwartz Z, Knight G, Swain LD, Boyan BD (1988b).
Nelson JF, Stanford HG, Cutright DE (1977). Evaluation Localization of vitamin D3-responsive alkaline phos-
and comparisons of biodegradable substances as phatase in cultured chondrocytes. I Biol Chem
osteogenic agents. Oral Surg Oral Med Oral Pathol 263:6023-6026.
43:836-843. Schwartz Z, Amir D, Boyan BD, Cochavy D, Muller-Mai C,
Pankowsky DA, Ziats NP, Topham NS, Ratnoff OD, Swain LD, etal. (1991). Effect of glass ceramic and tita-
Anderson JM (1990). Morphological characteristics of nium implants on primary calcification during rat tib-
adsorbed plasma proteins on vascular grafts and bio- ial bone healing. Calcif Tissue \nt 49:359-364.
materials. J Vase Surg 11:599-606. Schwartz Z, Swain LD, Marshall T, Sela I, Gross U, Amir D,
Pearson BS; Klebe RJ, Boyan BD, Moskowicz D (1988). et al. (1992a). Modulation of matrix vesicle enzyme
Comments on the clinical application of fibronectin activity and phosphatidylserine content by ceramic
in dentistry. J Dent Res 67:515-517. implant materials during endosteal bone healing.
Pilliar RM, Lee JM, Maniatopoulos C (1986). Observation Calcif Tissue \nt 51:429-437.
on the effect of movement on bone ingrowth into Schwartz Z, Swain LD, Kelly DW, Brooks B, Boyan BD
porous surfaced implants. Clin Orthop 208:108-113. (1992b). Regulation of prostaglandin E2 production
Pulapuna S, Kohn J (1992). Tyrosine-derived polycarbo- by vitamin D metabolites in growth zone and resting
nates: backbone modified "pseudo" polyfamino zone chondrocyte cultures is dependent on cell mat-
acids) designed for biomedical applications. uration. Bone 13:395-401.
Biopolymers 32:411-417. Schwartz Z, Braun G, Kohavi D, Brooks B, Amir D, Sela I,
Raisz LG, Fall PM (1990). Biphasic effects of et al. (1993). Effect of hydroxyapatite implants on pri-
prostaglandin E2 on bone formation in cultured fetal mary calcification during rat tibial bone healing (bio-
rat calvariae: interaction with cortisol. Endocrinology chemical and histomorphometric analysis). I Biomed
126:1654-1659. Mater Res 27:1029-1038.
Ricci JL, Spivak JM, Blumenthal NC, Alexander H (1991). Schwartz Z, Martin JY, Dean DD, Simpson J, Cochran DL,
Modulation of bone ingrowth by surface chemistry Boyan BD (1996). Effect of titanium surface roughness
and roughness. In: The bone-biomaterials interface. on chondrocyte proliferation, matrix production, and
Davies JE, editor. Toronto: University of Toronto, pp. differentiation depends on the state of cell matura-
334-349. tion. J Biomed Mater Res 30.145-155.
Robertson DM, Pierre L, Chahal R (1976). Preliminary Sela J, Schwartz Z, Amir D, Swain LD, Boyan BD (1992).
observations of bone ingrowth into porous materials. The effect of bone injury on extracellular matrix vesi-
J Biomed Mater Res 10:335-344. cle proliferation and mineral formation. Bone Miner
Rolla G, Hsu SD, Bowen WH (1977). The influence of flu- 17.163-167.
oride on the uptake of proteins by hydroxyapatite. Stein GS, Lian IB, Owen TA (1990). Relationship of cell
Caries Res 11:308-312. growth to the regulation of tissue-specific gene
Rubin CT, Lanyon LE (1984). Regulation of bone forma- expression during osteoblast differentiation. FASEB J
tion by applied dynamic loads. J Bone joint Surg Am 4:3111-3123.
66:397-402. Swain LD, Schwartz Z, Boyan BD (1992). Regulation of
Rubin CT, McLeod KJ (1994). Promotion of bone ingrowth matrix vesicle phospholipid metabolism is cell matu-
by frequency-specific, low-amplitude mechanical ration-dependent. Bone Miner 17:192-196.
strain. Clin Orthop Rel Res 298:165-174. Takatsuka K, Yamamuro T, Nakamura T, Kokubo T (1995).
Schneider G, Burridge K (1994). Formation of focal adhe- Bone-bonding behavior of titanium alloy evaluated
sions by osteoblasts adhering to different substrata. mechanically with detaching failure load. I Biomed
Exp Cell Res 214:264-269. Mater Res 29:157-163.
Schwartz Z, Boyan BD (1988). The effects of vitamin D Thomas KA, Cook SD (1985). An evaluation of variables
metabolites on phospholipase A activity on growth influencing implant fixation by direct bone apposi-
zone and resting zone cartilage cells in vitro. tion. J Biomed Mater Res 19:875-901.
Endocrinology 122:2191-2198. Thomas KA, Kay JF, Cook SD, Jarcho M (1987). The effect
Schwartz Z, Boyan BD (1994). Underlying mechanism at the of surface macrotexture and hydroxyapatite coating
bone-biomaterial interface. I Cell Biochem 56:340-347. on the mechanical strengths and histologic profiles of
Schwartz Z, Schlader D, Swain LD, Boyan BD (1988a). titanium implant materials. J Biomed Mater Res

344 Crit Rev Oral Biol Med 7(4):329-345 (1996)

Downloaded from cro.sagepub.com at University of Sydney on November 18, 2015 For personal use only. No other uses without permission.
 21:1395-1414. Williams DF (1981). Titanium and titanium alloys. In: CRC
Vroman L, Adams AL (1969). Identification of rapid fundamental aspects of biocompatibility. Williams DF,
changes at plasma-solid interfaces. ) Biorned Mater Res editor. Boca Raton, FL: CRC Press, pp. 9-44.
3:43-67. Windeler AS, Bonewald L, Khare AG, Boyan BD, Mundy
Vroman L, Adams AL, Fischer CG, Munoz PC (1980). GR (1991). The influence of sputtered bone substi-
Interaction of high molecular weight kininogen, factor tutes on cell growth and phenotypic expression. In:
XII and fibrinogen in plasma at interfaces. Blood The bone-biomaterials interface. Davies JE, editor.
55:156-159. Toronto: University of Toronto, pp. 205-213.
Watt FM, Jordan PW, O'Neill CH (1988). Cell shape con- Ziats NP, Miller KM, Anderson JM (1988). Iw vitro and in
trols terminal differentiation of human epidermal ker- vivo interactions of cells with biomaterials. Biomaterials
atinocytes. Proc Nail head Sri USA 85:5576-5580. 9:5-13.

7(4)-.329-345 (1996) Crit Rev Oral Biol Med 345

Downloaded from cro.sagepub.com at University of Sydney on November 18, 2015 For personal use only. No other uses without permission.