Biomaterials 20 (1999) 2287}2303

Bioactive ceramics: the e!ect of surface reactivity

on bone formation and bone cell function
P. Ducheyne*, Q. Qiu
Center for Bioactive Materials and Tissue Engineering, Department of Bioengineering, University of Pennsylvania,
Philadelphia, PA 19104, USA

Abstract

Surface reactivity is one of the common characteristics of bone bioactive ceramics. It contributes to their bone bonding ability and
their enhancing e!ect on bone tissue formation. During implantation, reactions occur at the material}tissue interface that lead to
time-dependent changes in the surface characteristics of the implant material and the tissues at the interface. This review describes
some of the current concepts regarding the surface reactivity of bone bioactive materials and its e!ect on attachment, proliferation,
di!erentiation and mineralization of bone cells. ( 1999 Elsevier Science Ltd. All rights reserved.

Keywords: Bioactivity; Bioactive materials; Tissue engineering

1. Introduction the risk of inducing transmissible diseases [4,5]. There-

Musculoskeletal conditions are among the most fre-
quently occurring medical conditions, and they have
a substantial impact on the health and quality of life of
the population. In 1988, musculoskeletal conditions im-
posed a $128 billion burden on the United States eco-
nomy [1]. In reconstructive surgery, repair of large bone
defects is a major reparative problem. About 1 230 000
fractures are treated with osteosynthesis material each
year in the USA [1]. Approximately 80% of these require
adjuvant grafting [1]. In 1988, there were about 400 000
total hip and knee arthroplasties performed and, current-
ly, an increasing number of these procedures are revision
surgeries with their concomitant need for bone grafting
[1]. The approaches to these di$cult bone repair prob-
lems include utilization of autografts or allografts [2].
While the use of autograft material is the preferred tech-
nique, there are limitations such as donor site morbidity,
limited donor bone supply, anatomical and structural
problems, and elevated levels of resorption during heal-
ing [3]. Allografts have the disadvantage of eliciting an
immunological response due to genetic di!erences and
* Corresponding author. Department of Bioengineering, Centre for
Bioactive Materials and Tissue Enginnering, Univeristy of Pennsyl-
vania, 3320 Smith Walk, Philadelphia, PA 19104, USA.
fore, considerable attention has been directed towards
the use of synthetic grafts, including hydroxyapatite
(HA), tricalcium phosphate (TCP) and bioactive glass
and glass ceramics (BG) [6}10]. However, since it is often
desirable to have graft material that will eventually re-
sorb, leaving space for new bone formation, the low
resorbability of stoichiometric HA is considered to be
a limiting factor. Furthermore, despite the fact that bone
growth can occur in porous and dense HA particulates,
the bone conductive e!ect is limited [11].
HA, TCP and BG are usually considered bone bioac-
tive ceramics. These are materials which, generally, bond
to surrounding osseous tissue and enhance bone tissue
formation. The chemical, physical and mechanical prop-
erties of these materials have been summarized in pre-
vious reviews [12}14]. Since direct bone bonding to
bioactive glasses was "rst observed [15], considerable
progress has been made in understanding the basic mech-
anisms of the formation of bone}biomaterial bond and
its e!ect on bone formation. This progress resulted main-
ly from two approaches. One focused on studying the
bone}biomaterials interface that developed in vivo. The
examination of bonding zone revealed the consistent
presence of an interfacial hydroxyapatite layer [15,16].
The other approach used in vitro immersions in
simulated physiological #uids or cell containing media
[17,18]. These analyses revealed that reactions occurred

0142-9612/99/$ - see front matter ( 1999 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 4 2 - 9 6 1 2 ( 9 9 ) 0 0 1 8 1 - 7
2288 P. Ducheyne, Q. Qiu / Biomaterials 20 (1999) 2287}2303
at the implant material surfaces such as dissolution,
precipitation and ion exchange. These reactions were
accompanied by adsorption and incorporation of biolo-
gical molecules [19,20]. The combination of in vivo and
in vitro studies led to a better understanding of surface
reactions of bioactive ceramics in the body and their
e!ects on bone formation and cell function.
In this paper we will primarily review the data from
our own laboratory with the goal to addressing the e!ect
of synthetic graft materials on bone tissue formation and
bone cell function. Since the calcium-phosphate-based
ceramics and glasses are reactive materials, we will "rst
review their physicochemical properties, especially those
leading to surface reactions in vitro and in vivo. In the
second half of the paper, we will review the e!ects of
physicochemical properties of the bioactive ceramics on
cellular functions, and we will use this information to
suggest applications of bioactive ceramics in tissue engi-
neering.
2. Material characteristics a4ect bone tissue formation
In the 1970s, studies of bioactive glasses and hydroxy-
apatite focused "rst on the analysis of the bonding inter-
face between these materials and surrounding bone tissue
[15,21]. These studies were followed by data underscor-
ing the bene"cial e!ect of a bioactive ceramic lining on
the rate of bonding of porous-coated materials [22].
When a porous stainless-steel "ber network was coated
with a slip cast hydroxyapatite (HA) lining, a marked
increase in bone ingrowth was observed in comparison to
the ingrowth in the same porous metal without HA
lining. This e!ect was pronounced at 2 and 4 weeks, but it
disappeared at 12 weeks [23]. This phenomenon allowed
achieving the distinct therapeutic bene"t of faster reha-
bilitation for patients with HA-coated devices.
The enhancement e!ect on bone tissue ingrowth
by calcium-phosphate-based ceramic linings was sub-
sequently con"rmed in other experiments [24}29]. How-
ever, the magnitude of the e!ect varied greatly among
these studies. The variation was likely caused by the
di!erences in experimental parameters used. Thus, the
hypothesis was formulated that structural and composi-
tional changes of the Ca}P ceramic occur due to process-
ing of the coatings, and that these variations of the
material characteristics a!ect the rate of enhancement of
the bioactive ceramic coating. Using an experimental
protocol that extensively controlled factors such as ani-
mal model, pore size, pore morphology and properties of
the metallic substrate, it was found that there were sta-
tistically signi"cant variations in bonding strength in the
"rst few weeks after implantation arising from di!erences
in ceramic characteristics [30]. Fig. 1 shows that cal-
cium-phosphate coatings signi"cantly enhanced bone tis-
sue growth "xation, but that the e!ect di!ered among
Fig. 1. Shear strength as a function of implantation time for controls
(porous surfaced titanium) and porous specimens with various cal-
cium-phosphate coatings. Note that the greatest e!ect on bone tissue
ingrowth "xation is associated with coating CAP3, which is the coating
with the highest in vitro dissolution rate (with permission [30]).
coatings. In addition, these data show that the intensity
of the e!ect was consistent among the three, early-time
points studied.
3. Mechanisms of bone bioactivity
The ability to bond to bone tissue is a unique property
of bioactive ceramics. Analyses of the bone}implant in-
terface revealed that the presence of hydroxyapatite is
one of the key features in the bonding zone. Another
consistent "nding is that the larger the solubility rate of
the ceramic, the more pronounced is the enhancement
e!ect of bone tissue growth [30}34]. A few examples will
illustrate this. Among the bioactive ceramics, dense,
stoichiometric hydroxyapatite has a limited reactivity
in vitro [35]. Thus, in in vivo experiments, a lesser e!ect
on bone tissue formation was found [11]. Considering
more reactive calcium-phosphate ceramics, such as bi-
phasic calcium-phosphate ceramic consisting of equal
amounts of b-tricalcium phosphate and hydroxyapatite,
a decrease in average crystal size and an increase in
microporosity were observed after implantation in oss-
eous defects in dogs for 6 months [36]. These observa-
tions indicated that dissolution and precipitation
occurred in vivo. At the same time, excellent bone growth
was found in the implantation sites. Yet another illustra-
tion is related to machinable bioactive glass}ceramic.
X-ray intensity pro"les along the implant}bone interface
found a Ca}P rich layer after 16 weeks of implantation
[37].
In a previous review paper we summarized a variety
of events that were reported to occur at the bioactive
ceramic}tissue interface [38]. Fig. 2 schematically shows
these phenomena. The list does not imply a ranking in
 P. Ducheyne, Q. Qiu / Biomaterials 20 (1999) 2287}2303
Fig. 2. Schematic diagram representing the events which take place at
the interface between bioactive ceramics and the surrounding biological
environment (with permission [38]).
terms of time sequence or importance: (1) dissolution
from the ceramic [15,16,33,35,39}41]; (2) precipitation
from solution onto the ceramic [16,32,36,40,42]; (3) ion
exchange and structural rearrangement at the ceramic}
tissue interface [11,15,16,32,36,43]; (4) interdi!usion from
the surface boundary layer into the ceramic [44]; (5)
solution-mediated e!ects on cellular activity [11,19,
45}47]; (6) deposition of either the mineral phase (a), or
the organic phase (b), without integration into the ce-
ramic surface [32,36,40,48,49]; (7) deposition with integ-
ration into the ceramic [11,36,40]; (8) chemotaxis to the
ceramic surface [11]; (9) cell attachment and prolifer-
ation [19,46,50,51]; (10) cell di!erentiation [11]; and (11)
extracellular matrix formation [48,49,51].
The observation of what transpires at the interface,
however, does not represent a mechanistic explanation
for the e!ect which bioceramics have on bone tissue
formation. We can focus on mechanisms by relying on an
increasing body of evidence which suggests that bone
bonding and bone tissue ingrowth enhancement are the
result of multiple, parallel and sequential reactions at the
material}tissue interface. The interaction is being de-
scribed in broad terms here, but will be elaborated upon
in the next sections. These interactions are related to
either physicochemical phenomena that occur in the
presence or absence of cells, or are related to reactions
a!ected by cellular activity. An important aspect of the
overall reaction sequence between these materials and
tissues is that, in the absence of a biologically equivalent,
calcium de"cient, carbonate containing hydroxyapatite
(c-Ap) surface upon implantation, dissolution, precipita-
tion and ion exchange reactions lead to a biologically
equivalent apatitic surface on the implanted material.
This reaction does not proceed by itself, but is accom-
panied by parallel reactions, such as adsorption and
incorporation of biological molecules and attachment of
surrounding cells. Furthermore, cells that have adhered
to the ever-reacting material surface interact with the
material and produce some of the surface changes. In the
inverse direction, i.e. from material to environment, there
2289
is both a solution-mediated as well as surface-controlled
e!ect on cellular activity, organic matrix deposition and
mineralization. The gradual change of the ceramic sur-
face to become a biologically equivalent HA with small
crystal dimensions is a rate determining step in the cas-
cade of events underlying bioactive behavior. All phe-
nomena, collectively, lead to the gradual incorporation of
the bioactive implant into developing bone tissue.
To understand the bone}material bonding mecha-
nisms, the atomic and molecular phenomena occurring
at the material surface and their e!ects on relevant reac-
tion pathways of cells and tissues must be elucidated.
Furthermore, for mechanisms to be uniquely identi"ed, it
is important to establish the extent of each of these
reactions and the sequence in which they occur. Thus, we
will summarize some of the current insight on these issues
in the following sections.
4. Reaction-related surface modi5cations
4.1. Dissolution, precipitation and ion exchange reactions
Formation of a biologically equivalent apatitic surface,
a common characteristic of bioactive materials, can be
reproduced in vitro by immersion experiments using
a simulated physiologic solution that mimics the typical
ion concentrations in body #uids. These in vitro physico-
chemical analyses are useful to explain the observations
on ex vivo specimens [35,42]. Such experiments have
shown that the materials with high solubility also readily
induce the precipitation of a biologically equivalent
apatite on their surface. Using this in vitro immersion
methodology, considerable evidence has been obtained
revealing the mechanisms of surface reactions
[17,40,42,52,53]. The reactions include dissolution, pre-
cipitation and ion exchange accompanied by absorption
and incorporation of biological molecules [15}19,31].
Daculsi et al. used transmission electron microscopy to
determine the in vivo degradation of biphasic implant
materials which were various mixtures of hydroxyapatite
and tricalcium phosphate [36]. Implants were inserted in
surgically created periodontal defects. At six months,
a direct correlation was established between the rate of
dissolution of the material and the quantity of newly
formed microcrystals with Ca/P ratios similar to those of
bone apatite. The transmission electron microscopical
studies carried out by Jarcho et al. [21], and Tracy and
Doremus [54], who each used dense, stoichiometric hy-
droxyapatite, showed direct apposition of bone crystals
onto the synthetic apatite lattice. Whereas in these
studies solid}solution ion exchange was not addressed,
Davies et al. [55] documented in a more recent study
a limited in vivo reactivity of this material.
A similar result of limited reactivity of dense, stoi-
chiometric hydroxyapatite was also found by Schepers
2290 P. Ducheyne, Q. Qiu / Biomaterials 20 (1999) 2287}2303

et al. [11] who implanted granules of this material in

various bone tissue sites of the beagle mandible. This
species and site were chosen to compare the in vivo
behavior of several calcium-phosphate ceramics and
glasses in view of their possible or actual use in clinical
dentistry. Dense hydroxyapatite, porous hydroxyapatite
and bioactive glass 45S5 (nominal composition by
weight: 45% SiO , 24.5% Na O, 24.5% CaO, 6% P O )
2 2 2 5
were implanted [11]. Four types of bone defects were
created and "lled at the time of surgery: tooth extrac-
tions, occlusal and buccal cavities in the mandible, and
periodontal defects. The materials were evaluated after 1,
2, 3, 6, 12 and 24 months after implantation. On dense
hydroxyapatite granules, bone tissue conduction was
consistently observed at the early time points. Bone tis-
sue grew from the preexisting bone along the particles
that were closest to the defect wall. However, bone tissue
did not grow further than about 1 mm into the defect
after about 12 weeks. In addition, the presence of multi-
nucleated giant cells was continuously observed from
that time on and the ceramic surface exhibited a moth
eaten appearance resulting from the local resorption of
the material. In the areas where giant cells were present,
there was no new bone tissue formation. Osteoconduc-
tion and giant cell-mediated resorption frequently had
taken place on separate sides of a single particle. These
phenomena are shown in Fig. 3a and b.
This experiment showed beyond doubt that dense,
stoichiometric hydroxyapatite is osteoconductive in vivo.
However, it also revealed that this material only displays
a limited bone bioactivity. This behavior can be ex-
plained by the mechanisms of bioactivity formulated
above. In this instance, it su$ces to focus on physico- Fig. 3. Histological section of dense, stoichiometric hydroxyapatite.
chemical phenomena of dissolution, and dissolution} (a) After three months, bone tissue has grown along the particle sur-
precipitation, without need to address cellular activity face, but only over a distance of about 1 mm. (b) From six months on,
per se. Among all calcium-phosphate ceramic (CPCs) the particle surfaces have a moth eaten appearance caused by multinuc-
studied, dense, stoichiometric hydroxyapatite is the one leated cell resorption. The resorption is not paralleled by a bone
formation process (with permission [11]).
that dissolved the slowest [35]. In addition to its slow
dissolution reaction kinetics, in vitro experiments simula-
ting precipitation reactions in vivo, showed that the pre-
cipitation rate is also amongst the slowest. Furthermore, not unequivocally support this hypothesis. The tissue
the precipitate that forms on its surface is not a biolo- response to a porous carbonated apatite material im-
gically equivalent HA, but is a precursor phase with planted in beagle, mandibular bone was very similar to
greatly lower Ca/P ratio than 1.67 [42]. Thus, even that with dense, stoichiometric hydroxyapatite [11]. Os-
though this material can intrinsically react with sur- teoconduction from the bone defect wall occurred over
rounding bone tissue, its reaction rate is very slow. We a limited distance and bone tissue conduction into the
suggest that competing reactions take over quickly and center of the defects was not observed. In addition,
that, from that time forward, bone tissue formation does a chronic, mild multinucleated cell resorption of the
no longer take place on the hydroxyapatite surface. In material started as early as 3 months after implantation.
clinical dentistry this sequence of events is well known, These observations are shown in Fig. 4a and b [11,56].
since most sites treated with dense, stoichiometric hy- Our in vitro experiments suggest that the characteristic
droxyapatite never fully heal. at the basis of this limited osteoconductivity is the
If a key element for bone bioactive behavior is the considerable induction time for precipitation reactions
development of a carbonated apatite surface, one might on the material to get underway, impeded as they are by
expect an excellent response by using a carbonated apa- protein adsorption to the material surface. Transposing
tite as implant material. However, implantation data do these data to the in vivo situation, protein molecules
 P. Ducheyne, Q. Qiu / Biomaterials 20 (1999) 2287}2303
Fig. 4. Tissue response to porous, carbonated hydroxyapatite. (a) At
24 months, bone does not make contact with the particles. (b) At
3 months, a particle lined with multinucleated cells (with permission
[11]).
adsorb onto the implant material upon implantation,
and signi"cantly slow down the formation of a bone
mineral surface layer [53,57]. As a corollary, bone bioac-
tive behavior is then not strongly expressed without this
carbonated apatitic surface layer.
Among the studies on the partially crystallized Kyoto
glass formulations, which are mixtures of glass and apa-
tite or wollastonite, two reports are noteworthy. Two
separate bioactive glass ceramic plates, implanted in soft
tissue sites in close proximity to one another, were found
to be "rmly attached to each other after 3 months of
implantation [58]. Such a phenomenon can be explained
by assuming that there was a precipitation and sub-
sequent growth of crystals eventually bridging the pre-
existing gap between the two plates. Although this
experiment did not address the surface dissolution, the
same glass ceramic, immersed in a simulated physiolo-
gical solution, caused an increase in calcium and phos-
2291
phate concentration in solution prior to the subsequent
drop associated with the calcium-phosphate precipita-
tion [17].
When discussing reaction-induced surface modi"ca-
tions of bioactive ceramics, it is also useful to review that
materials with excellent tissue response to bone generally
undergo a change of their surface properties. In many
instances, a calcium-phosphate layer can be found at the
interface with the tissues. This observation was also made
for other ceramics than the ones discussed so far (sol}gel
processed silica and titania], metallic materials such as
titanium, and various polymers such as polyethylene
oxide/polybutylene terephthalate copolymers and
tyrosine derived polycarbonates [59}67]. Using in vitro
modeling results, it was suggested that this phenomenon
was related to the presence of hydroxyl (Si}OH and
Ti}OH) or carbonyl groups (C}OH) at the material sur-
faces [62]. But, as is the case for stoichiometric
hydroxyapatite, the conditions under which a calcium-
phosphate layer forms in vitro do not necessarily mimic
the in vivo conditions well, nor are the reaction kinetics
such that competing reactions (not modeled in vitro)
dominate the events at the biomaterial surface in vivo. If
the previous suggestion would be correct, the implication
would be that materials do not need to contain calcium
and phosphate in order to bond with bone tissue. How-
ever, given that the in vitro modeling studies did not
address the mechanisms by which Ca and P containing
phases form on hydroxylated surfaces, de"nitive word as
to whether this question can be reduced to a purely
inorganic event, is still needed. In fact, it would appear
that an event without intervening biological components
is a strained concept for, essentially, an organic reaction
microcosm. This observation may draw credence from
an analogous case, i.e. the issue of the reaction
sequence at the surface of bioactive glasses. In what
follows we will provide evidence to suggest that this
surface reaction layer does not form as a result of inor-
ganic reactions "rst, followed by biologically driven
events next, as has been proposed through the 1970s and
1980s [68]. In fact, serum proteins do have a major e!ect
on the properties of the surface reaction layer, as will be
described next.
4.2. Ewect of serum proteins on the surface reactions
of bioactive glass
To explore the e!ect of proteins on the formation of
surface reaction layers on bioactive glass, we analyzed
the atomic and molecular changes in bioactive glass 45S5
surface during immersion in protein-free bu!er solution
and protein-containing culture medium [69]. When im-
mersed in serum-free solutions, atomic force microscopy
(AFM) showed that the surface of bioactive glass reacted
non-uniformly. After 5 min of immersion, the glass sur-
face showed a rough texture due to the formation of
2292 P. Ducheyne, Q. Qiu / Biomaterials 20 (1999) 2287}2303
particles on the surface. With increasing immersion time,
the particles increased in size and number. Unlike the
reaction on the substrates immersed in serum-free solu-
tion, a uniform layer of globules, presumably proteins,
covered the substrates immersed in serum-containing
solutions. As the immersion time increased, the density of
globules increased and the layer thickened. Rutherford
backscattering spectroscopy (RBS) of the outermost sur-
face layers revealed that substrates formed Si-rich surface
layers in the "rst hour of immersion in the absence or
presence of serum proteins. However, the RBS spectra
also showed that the formation of the calcium phosphate
layer was di!erent in serum-containing solutions relative
to the serum-free solutions. FTIR analysis showed that,
in the serum-free solution, crystalline hydroxyapatite was
formed by transformation from the initially formed
amorphous calcium phosphate. In contrast, immersion in
serum-containing solutions only produced an amor-
phous calcium phosphate.
Prior to this experiment, Radin et al. already demon-
strated, using one-parametric variations in solution com-
position, that the surface reactions on bioactive glass
45S5 were exquisitely dependent upon the modeling con-
ditions [70]. The solutions used were tris bu!er, tris
bu!er complemented with plasma electrolyte and/or
serum, and serum. After a short immersion time (3 h)
a crystalline, carbonated hydroxyapatite (c-HA) layer
formed only in tris. Reaction surfaces of di!erent struc-
ture, morphology and composition were observed in all
other solutions after either short- and long-term immer-
sions. They comprised two layers with the layer in con-
tact with the bulk consisting mainly of Si; the outer layer
was composed of Si, Ca and P, was amorphous and had
a Ca/P ratio of about 1. Serum proteins adsorbed on the
BG surfaces at the early stages of the solution-mediated
BG reactions. Formation of a crystalline carbonated HA
layer was delayed up to three or more days in solution
with plasma ions. And again, in the presence of serum,
only amorphous surfaces composed of Si, Ca and P were
observed for any time up to seven days of immersion
[70].
Given the repeated observation of the absence of
crystalline hydroxyapatite when proteins co-adsorb, it
is unlikely that, in vivo, adsorption of biological
molecules will take place subsequent to the transforma-
tion of an amorphous calcium-phosphate rich layer to
carbonated apatite, as was previously suggested [68].
The present data suggest that serum proteins adsorb in
tandem with the occurrence of solution-mediated reac-
tions leading to formation of a silica-gel. Amorphous
calcium-phosphate phases accumulate in this Si-rich
matrix.
A signi"cantly di!erent behavior between immersion
in serum protein-free and serum protein-containing solu-
tions was also observed in experiments intended to
model silicon dissolution from bioactive glass particles
[71]. As will be described below, bioactive glass granules
can be internally hollowed out leading to shells of cal-
cium-phosphate in which osteogenesis can be observed
[11]. Bioactive glass granules were immersed under inte-
gral (no solution exchange during the experiment, there-
by simulating stagnant #uid conditions) or di!erential
conditions (conditions simulating continuous #uid #ow
in vivo) in tris bu!ered solution complemented with
either plasma electrolyte (TE) or with electrolyte and
10% serum (TES-10). Only when the solution was con-
tinuously replenished, thereby avoiding Si saturation in
solution, and only when the solution contained serum
proteins, was full Si dissolution from the core of the
granules observed [71]. This result was ascribed to the
di!erent structure and composition of the surface reac-
tion layer that formed in the presence of serum. In
the serum-free solution, the carbonated hydroxyapatite
surface protects the glass from further corrosion. In
contrast, the reaction layer formed in serum-containing
solutions does not o!er su$cient protection from con-
tinued corrosion. This surface reaction layer is slightly
porous and comprises proteins, silica and amorphous
calcium phosphate [71]. In the presence of serum pro-
teins, the apatite formation is extensively delayed, a "nd-
ing which is consistent with other data on the e!ect of
proteins on nucleation and crystal growth of apatite
[57,72}74].
In the present section, data were summarized docu-
menting that in solutions more closely approaching the
physiological state, the maturation of amorphous cal-
cium-phosphate to crystalline hydroxyapatite does not
take place readily. The proteinaceous layer which ad-
sorbs onto the glass interferes with the solid to liquid
interaction of the amorphous calcium-phosphate layer.
Whereas amorphous calcium phosphate can form exclus-
ively as the result of physicochemical phenomena in the
solid glass phase, it is suggested here that adsorbed
serum proteins impede the nucleation and growth
reactions by which it would transform to carbonated
apatite. The reactions in the glass are not blocked, how-
ever, as Ca and P di!usion leads to a continuously
thickening of the Ca}P rich zone under the adsorbed
protein layer. The adsorption of the protein layer may
also be critical in terms of providing attachment sites for
bone cells such as osteoblasts and their progenitors. The
e!ect of adsorbed molecules is the subject of the next
sections.
5. Stimulation of cellular function0e4ect of porosity
5.1. Marrow-derived osteoprogenitor cells
in porous matrices
In vivo studies have clearly documented that bioactive
ceramics a!ect cellular function. A series of experiments
 P. Ducheyne, Q. Qiu / Biomaterials 20 (1999) 2287}2303
using porous hydroxyapatite and bone marrow cells
showed that the osteoprogenitor nature of cells from
bone marrow was activated more readily in heterotopic
sites when they were cultured with porous, carbonated
hydroxyapatite than when implanted by themselves
[45,75}77]. With these studies, Caplan and his associates
were "rst to describe the idea of culturing cells capable of
expressing the osteoblastic phenotype with the intent to
synthesize arti"cial bone grafts. These authors focused
primarily on issues related to stem cell preparation.
When cultured on appropriate templates, the cell cul-
tures produced extracellular matrix. Investigators from
this laboratory also used periosteal-derived cells.
Nakahara et al. [78] showed that when the cells were
combined with porous calcium-phosphate ceramics and
implanted in a subcutaneous site in athymic mice, bone
tissue was formed.
In these studies using marrow stroma-derived osteo-
progenitor cells, it is important to realize that it is pri-
marily the intrinsic capacity of the osteoprogenitor cells
that produces the upregulation to cells of the osteoblast
lineage. Okumura et al. [79] demonstrated that stem cell
di!erentiation to cells expressing the osteoblastic pheno-
type also occurred in porous titanium. However, the
pattern of bone tissue formation was di!erent. Whereas
in porous hydroxyapatite it started at the ceramic sur-
face, it started in the middle of the pore and was much
slower in porous titanium. These data suggested that
even though the intrinsic capability of pluripotential cells
led to di!erentiation along the osteoblast pathway, the
osteogenic potential of these cells was also stimulated by
the bioactive material surface.
5.2. Bioactive glass granules of narrow size range
Another example of stimulation of cellular function is
the phenomenon of bone tissue formation with bioactive
glass granules of narrow size range. In the aforemen-
tioned beagle study with porous HA, bioactive glass
particles with a well-controlled particle size were also
implanted. Bioactive glass granules of narrow size range
(300}355 lm, as determined by sieving) elicited bone tis-
sue formation throughout 5 mm defects in the beagle
mandible as soon as 1 month after implantation [11].
The critical phenomenon taking place "rst was the
formation of internal pouches in each of the particles.
This was followed by di!erentiation of osteoprogenitor
cells to osteoblasts actively laying down bone tissue in
these pouches. Subsequently, bone tissue grew out of the
excavations, surrounded the particles and connected
with bone tissue formed around the neighboring par-
ticles. As a result, defects were quickly repaired in about
3}6 months, as each particle served as a nucleus for
repair [11]. Fig. 5 shows bone tissue formation in an
excavated particle. This "nding of upregulation of os-
teoprogenitor cells to bone tissue-forming osteoblasts in
2293
Fig. 5. Bone tissue formation in excavated bioactive glass particles of
narrow size range. Bone tissue is stained red. Also note the channels
connecting the interior of the particle with the surrounding milieu (with
permission [116]).
excavated bioactive glass particles was independently
veri"ed by Gatti et al. [80]. Bone biopsies taken from
sinus lift treatment sites at the time of implant insertion
con"rmed the excavation process and bone tissue forma-
tion within the excavated particles in a clinical setting
[81].
The excavation of the particles itself is related to the
well-documented surface reactions on bioactive glass, as
well as to an extensive control of the glass granule size.
Interfacial reactions on bioactive glass lead to the forma-
tion of two surface layers: an inner silica gel layer and an
outer calcium-phosphate rich layer [82]. In the dog man-
dible, the thickness of a reaction layer on bioactive glass
45S5 was found to be about 150}170 lm at 3 months
[83]. Thus, with a particle size that is about double the
reaction layer thickness, the particles fully react. These
reactions produce phase transformations, which in turn
may provoke dimensional variations and cause cracking.
Macrophages move into the particles through these
cracks and assist with the removal of the interior silica gel
by phagocytosis [11,84]. What remains is a shell of
calcium-phosphate, which was previously formed as
a biological reaction product. Osteoprogenitor cells mi-
grate into the protective space of these in situ formed
hollow chambers and undergo di!erentiation to osteo-
blasts. This di!erentiation does not occur outside these
biologically formed calcium-phosphate chambers. The
proposed mechanism for the excavation process also
allows to explain why large particles do not react
throughout and, as such, do not cause the di!erentiation
of precursor cells. For smaller particles, it was shown that
they elicited an in#ammatory response in maxillofacial
applications [85,86]. Considering that in particle prep-
arations with a large variation of particle size, there is
always a much larger number of smaller particles than
larger ones, a restricted size range is important to achiev-
ing this unique phenomenon of promotion of cell
2294 P. Ducheyne, Q. Qiu / Biomaterials 20 (1999) 2287}2303
di!erentiation throughout the treated defect. The re-
stricted size range is also important to allow optimum
vascularity development. When a wide size range of par-
ticles is installed, the space between the larger particles is
more readily "lled up, thereby obstructing the tissue
repair [10].
5.3. Bone tissue induction in porous calcium phosphates
Ripamonti [87] and his coworkers observed a bone
induction phenomenon in porous hydroxyapatite. This
"nding was unusual in that the porous specimens were
not preloaded with cells. In addition, the implantation
sites were soft tissue sites, unlike in the Schepers et al.
study [11] where di!erentiation of progenitors to osteo-
blasts caused by a synthetic material were observed in
a bony site. The "nding of induction by Ripamonti et al.,
however, did not occur until after longer implantation
durations. In addition, it was not consistently observed in
all animal species. Porous hydroxyapatite obtained after
hydrothermal conversion of the calcium carbonate exo-
skeleton of coral was implanted in the rectus abdominis
of adult rabbits, dogs, and baboons. Minimal amounts of
bone formed in specimens harvested from rabbits and
dogs. In contrast, 10.4 and 25.9% of the pore space was
"lled in specimens retrieved from baboons after 3 and
6 months, respectively [87,88]. Other workers have
con"rmed the phenomenon of bone induction in porous
bioactive ceramics implanted in soft tissue sites [89}91].
5.4. The ewect of protected spaces on bone formation
As early as in 1965, Urist observed that &2new bone
was deposited in pockets or excavation chambers "lled
with proliferating osteoprogenitor cells. The process be-
gan within a few weeks and was complete in a few
months. How much new bone could be attributed to the
osteoconduction of cells growing in from the walls of the
host bone, and how much could be accounted for by
osteogenic induction, was not apparent from experiments
on normal, healthy bones2' [92]. Clearly, pores, and
pockets which resemble pores, have an e!ect beyond the
e!ect produced by growth factors and other signaling
molecules. In every study summarized in this section the
protective space in excavations or inside pores was essen-
tial to stimulate di!erentiation of precursors to osteo-
blasts.
Addressing the phenomenon of di!erentiation in the
excavated bioactive glass granules, we hypothesized that
the e!ect could be due to two main sets of reasons [93].
First, the bioactive glass is a reactive material and, as
a result, solid to solution interaction in the limited space
of the pouch will be modi"ed from the reactions without
transport limitations outside the particles. Although we
speci"cally focused on the ionic species of the glass, given
that growth factors and signaling molecules can be
drawn into the developing surface reaction layers, di!er-
ences in reaction kinetics can also lead to di!erences in
factor concentrations at the glass surface. Second, di!er-
ences in #uid #ow can produce di!erences in shear stress
between inside and outside.
We used "nite element analysis to model numerically
compositional variations of the extracellular #uids, spe-
ci"cally the Na`, Ca`2, and HPO~2 concentrations. We
4
also modeled the #uid #ow around and within porous
bioactive glass [93]. For the range of interstitial #uid
velocities examined, the analysis revealed several signi"-
cant "ndings. First, the creeping #ow of interstitial #uid
around the bioactive glass granules was dominated by
viscous forces [low Reynolds number #ow), producing
shear stresses at the granule}#uid interface. Second, the
#uid inside the granules remained stagnant, resulting in
zero shear stress at the granule}#uid interface. Di!usion
was the dominant mass transport mechanism within the
porous granules. Third, the governing mass transport
mechanism around the granules depended on the #ow
"eld. At low velocities, di!usion dominated the transport
of chemical species. At intermediate velocities, di!usion
and convection governed the mass transport, while at
high velocities, convection was the controlling transport
mechanism [93]. These numerical analyses documented
variations in #ow and stress patterns, reaction rates and
solute concentrations between the interior and exterior of
the granules. As an outcome of these analyses, the hy-
pothesis was formulated that variations in material sur-
face composition, medium composition and shear stress
caused the observed di!erences in cellular function. Spe-
ci"cally, any of the cellular activities leading up to tissue
formation, including attachment, proliferation, di!erenti-
ation and extracellular matrix formation, could be im-
plied. This broad hypothesis was the basis for a number
of subsequent experiments.
6. E4ects of surface reactions on cellular function
6.1. Surface modixed bioactive glass stimulates
the osteoblast phenotype in vitro
In an e!ort to reproduce the e!ect of BG in that it
stimulated the osteoblast phenotype in excavated glass
pouches [11], we cultured neonatal rat calvaria osteo-
blasts on porous BG templates. Brighton and coworkers
documented that these cells expressed the osteoblast
phenotype, but their osteoblastic markers were not ex-
pressed till after 17}21 days [94}97]. In our experiments,
the osteoblast phenotype was already expressed after
7 days. Fig. 6 is a scanning electron micrograph of the
cross section of porous BG showing the formation of
extracellular matrix after 7 days in cell culture. The cells
also exhibited a high alkaline phosphatase activity and
synthesized collagen type I and osteocalcin [49].
 P. Ducheyne, Q. Qiu / Biomaterials 20 (1999) 2287}2303
Fig. 6. SEM micrograph of the fractured surface of the porous modi"ed
bioactive glass template seeded with neonatal rat calvaria osteoblasts
for 7 days. Note the uniform formation of extracellular matrix and
bone-like tissue throughout the porous disk (with permission [49]).
In this study we also documented that the expression
of the osteoblastic phenotype critically depended on the
preparation of the glass surface. Prior to our experi-
ments, a limited e!ect of the BG substrates in cell culture
was reported [98,99]. A con#uent monolayer was not
observed until 12 days of culture using fetal rat calvaria
osteoblasts. Additionally, only a &weak' expression of the
osteocalcin was observed at this time point. A better
cellular response in vitro was achieved by these authors
by adding various oxides which rendered the glass sur-
face less reactive. However, the oxides selected were
among those that produced an adverse e!ect on the rate
of mineralized bone formation in vivo, as reported by
Gross et al. [100]. We overcame the issue of limited
reactivity by treating the glass surface prior to cell seed-
ing. The bioactive glass was transformed into a calcium
phosphate surface by immersing it in serum-free, physio-
logical solution in a "rst step (step 1), followed by immer-
sion in serum-containing media (step 2). Fig. 7 shows that
with this 2-step surface preparation, the cells grown on
the modi"ed glass surface maintained critical features of
the osteoblast phenotype, such as a high alkaline phos-
phatase activity, among others [101]. This phenotype
was not expressed if the cells were seeded on BG tem-
plates conditioned with only one of the steps (immersion
in either simulated physiological saline to produce a cal-
cium-phosphate surface without pre-adsorbed proteins,
or immersion in tissue culture medium to produce a glass
surface with adsorbed proteins). The cells exhibited a low
alkaline phosphatase activity. It is also noteworthy that
equivalent amounts of protein were produced by cells
inoculated on the templates regardless of the speci"c
conditioning treatment, 2-step or either one of the
2295
Fig. 7. Alkaline phosphatase (AP) activity of 7-day cell cultures on
porous bioactive glass conditioned by three procedures. &BG-0' is the
glass surface onto which serum proteins were adsorbed for 1 h; &BG-48'
is the glass surface transformed to a crystalline hydroxyapatite, but not
subjected to a protein adsorption treatment; &BG 2-step' is the glass
treated sequentially such that it was transformed to a crystalline hy-
droxyapatite onto which serum proteins adsorbed. Note that cells
grown on &BG 2-step' showed high AP activity, and in contrast, the
other two conditions not (with permission [18]).
single-step immersions [101]. Furthermore, the DNA
content did not indicate a loss of cellular viability on
those surfaces that were either calcium phosphate with-
out adsorbed proteins, or the glass surface with adsorbed
proteins only.
6.2. Serum protein adsorption
The data summarized in the previous section evoked
the hypothesis that concentration of serum proteins at
the glass surface transformed to calcium-phosphate, ser-
ved to enhance the expression of the osteoblast pheno-
type. Thus, we measured the amount of protein that
absorbed from the tissue culture medium. A greater
quantity of serum proteins absorbed onto the calcium-
phosphate-coated glass than onto the untreated BG.
However, a yet larger amount absorbed onto
stoichiometric HA, which was used concurrently [20].
This was surprising, as in vivo results clearly documented
a more intense bioactive e!ect for BG than HA [11,102].
Thus, we reasoned that the e!ect of the serum protein
absorption treatment was related to the species that
absorbed and found, in fact, signi"cantly di!erent
absorption pro"les for untreated BG, calcium-phos-
phate-coated BG and stoichiometric HA [20]. Fig. 8
shows the protein absorption spectra on each of these
three materials. Whereas on HA the spectrum of ab-
sorbed proteins re#ected the composition of the medium,
there was a preferential absorption pattern to each of the
two types of BG specimens. Especially, the broad band of
proteins at molecular weights in excess of 200 kDa which
2296 P. Ducheyne, Q. Qiu / Biomaterials 20 (1999) 2287}2303
Fig. 8. (a) Densitometry of di!erent serum protein bands present in the
polyacrylamide gel of the protein fractions that are present in the tissue
culture medium (TCM), adsorbed onto synthetic hydroxyapatite (HA),
untreated BG (BG-0) and BG of which the surface was transformed to
crystalline calcium phosphate (BG 2-step). The intensity of the protein
bands was visualized using Image Quant software and the Molecular
Dynamics Personal Densitometer. Noteworthy is that the crystalline
calcium phosphate layer on BG (BG 2-step) extensively adsorbed
proteins in a molecular weight range 210}250 KDa; this protein was
minimally adsorbed on BG-0. Furthermore, while &BG 2-step' appeared
to be very selective, HA adsorbed a wide range of proteins. (b) Western
blot analysis of serum proteins adsorbed on BG 2-step (lane 1), BG-0
(lane 2), HA (lane 3), tissue culture plastic dish (lane 5). Bovine serum
"bronectin and TCM are present in (lane 4) and (lane 6), respectively.
BG 2-step selectively adsorbed higher amounts of "bronectin than HA
or BG-0. The intensity of "bronectin bands decreased in the order BG
2-step'BG-0'HA (with permission [20]).
absorbed onto calcium-phosphate-coated BG was note-
worthy. Western blot analysis showed that this band
arose from extensive, preferential absorption of "bronec-
tin onto the calcium-phosphate-coated BG [20]. By vir-
tue of "bronectin being an ubiquitous molecule, which,
among other functions, also serves as an attachment
molecule between substrate and cell membrane of an-
chorage-dependent cells, an experiment was formulated
to assess cellular attachment. Four di!erent materials or
material surfaces were prepared prior to cell seeding:
(1) BG immersed in the electrolyte containing, serum free,
simulated physiological solution (BG coated with a Ca}P
surface, but without pre-absorbed proteins), (2) BG im-
mersed in tissue culture medium (BG with a layer of
proteins), (3) BG treated consecutively in the electrolyte
solution and the tissue culture medium (BG coated with
a Ca}P surface onto which "bronectin was extensively
adsorbed), (4) HA immersed in tissue culture medium
(HA extensively coated with a broad spectrum of pro-
teins). The results shown in Fig. 9 reveal a much greater
cellular attachment to the Ca}P and protein-coated BG
than to any other surface. These data demonstrate that
the preferential absorption of "bronectin is a "rst impor-
tant observation to explain BG's e!ect on bone cell
function, and as a corollary, bone tissue formation [20].
6.3. The ewect of xbronectin on osteoblast cell
attachment to bioactive glass
Having found that "bronectin preferentially absorbed
onto calcium-phosphate-coated BG, it is possible that
the e!ect on bone cell function of adsorption of the ligand
was due exclusively to the quantity of "bronectin ab-
sorbed, and was not dependent upon the e$cacy of the
ligand's attachment function. Thus, we also addressed the
question whether, for the same concentration of adsor-
bed "bronectin onto various material surfaces, there
would be di!erences in cell attachment. We quantita-
tively determined cellular attachment to bioactive glass
as a function of bioactive surface reaction stage and
quantity of adsorbed "bronectin. It was found that for
Fig. 9. DNA content of MC3T3-E1 cells attached on BG with di!erent
surface modi"cations and HA ceramic. &BG-0' is the glass surface onto
which serum proteins were adsorbed for 1 h; &BG-48' is the glass surface
transformed to a crystalline hydroxyapatite, but not subjected to a pro-
tein adsorption treatment; &BG 2-step' is the glass treated sequentially
such that it was transformed to a crystalline hydroxyapatite onto which
serum proteins adsorbed. Note that a signi"cantly larger number of
cells adhered to the glass surface that had been transformed to a cal-
cium-phosphate layer prior to adsorbing serum proteins and seeding
the cells (with permission [20]).
 P. Ducheyne, Q. Qiu / Biomaterials 20 (1999) 2287}2303 2297

the same amounts of adsorbed "bronectin, the attach-

ment strength to treated bioactive glass was signi"cantly
greater than for untreated glass, control glass or stoichio-
metric hydroxyapatite [103]. The treated BG surfaces
were either amorphous calcium phosphate (obtained by
immersion for one day in serum free, simulated physiolo-
gical solution: BG1d) or a crystalline calcium phosphate
(obtained by immersion in the same solution for 7 days:
BG7d).
The osteoblast cell attachment experiments were per-
formed on substrates that were incubated in "bronectin
solutions for 30 min, then in 1% bovine serum albumin
solution for 30 min to block non-speci"c adhesion. Cells
were seeded onto the disks (34 000 cells/cm2) and allowed
to attach for 15 min. The disks were spun in a rotating
disk device and the fraction of adherent cells at a given
stress level were counted and normalized to the number
of cells not experiencing any stress (cells at the center of
the disk). These results were plotted as a function of shear
stress and "tted to a logistic equation to estimate the
critical shear stress (q ) for 50% detachment (Fig. 10). The
# Fig. 10. Cell detachment pro"les for substrates coated with equal
q was used as a measure of attachment strength. The amounts of "bronectin using a 0.1 lg/ml "bronectin solution
# (10 dyn/cm2"1 N/m2). The specimens used included non-reactive
adherent fraction decreased sigmoidally as the shear
stress increased. s increased as the "bronectin concen- borosilicate glass as control glass (CG); stoichiometric hydroxyapatite
# (sHA); untreated bioactive glass (BG0); and bioactive glass of which the
tration increased. For "bronectin-treated substrates, surface was transformed to an amorphous calcium phosphate (BG1d)
s was 50% higher for treated bioactive glasses (BG1d,
# or crystalline carbonated hydroxyapatite (BG7d) by a 1 or 7 day
BG7d) than for untreated glass (BG0), a control non- immersion treatment respectively. Cell adhesion was signi"cantly high-
bioactive glass and HA. er on BG of which the surface had been transformed to calcium
The data clearly documented that the surface treat- phosphate (BG1d, BG7d), than on unreacted bioactive glass (BG0),
control glass (CG) or stoichiometric hydroxyapatite (sHA) (with per-
ment of bioactive glass resulted in enhanced "bronectin- mission [103]).
mediated cell attachment. Since the surface treatment did
not a!ect the amount of adsorbed "bronectin, these data
suggested that the observed increase in attachment di!erentiation. On HA, proliferation is slow to start in
strength resulted from di!erences in "bronectin confor- comparison to the proliferative activity on the calcium-
mation. Previous studies documenting "bronectin con- phosphate-coated BG. In addition, it continues through-
formational changes resulting from adsorption onto out the two week duration of the experiment without
material surfaces can be invoked to explain our own slowing down, all while di!erentiation activity is not
results [104]. Thus, an enhanced biological response activated. Thus, the activity of this cell line was stimu-
resulted from bioactive glass surface modi"cations which lated towards extracellular matrix formation on BG, but
mimicked the changes in vivo [20]. not on HA [105].
It is known that the ligand "bronectin connects to the
cell membrane via integrin receptors, which are also 6.4. Synergism between the calcium-phosphate
activated by this interaction. The activation of integrins surface on bioactive glass and growth factors
then triggers cytoplasmic reactions, and thereby, stimu-
lates the intracellular signaling pathway. The evidence As the previous sections documented, concentration of
that the attachment function of "bronectin was better "bronectin at the surface of bioactive glass transformed
expressed when it was absorbed to calcium-phosphate- to calcium-phosphate-serves to enhance osteoblast func-
coated BG suggests the possibility that this molecule may tion. These data raise a tempting idea, namely that other
also activate the integrins' intracellular signaling much biologically active molecules present in solution may
more readily than when adsorbed to other material surfa- also adsorb onto bioactive calcium-phosphates, and
ces. This implies that the cellular functions such as prolif- especially the calcium-phosphate layer grown on bio-
eration and di!erentiation will be stimulated. The data active glass. Such adsorbed molecules, together with
shown in Fig. 11a and b support this hypothesis an optimally bioactive calcium-phosphate would then
[105]. On treated BG, MC3T3-E1 cell proliferation is extensively a!ect cellular function. This idea of con-
extensive in the "rst few days after cell seeding. After centrated adsorption is the inverse of using cal-
about 7 days, proliferation slows down at the expense of cium-phosphate ceramics and glasses as a vehicle for
2298 P. Ducheyne, Q. Qiu / Biomaterials 20 (1999) 2287}2303
Fig. 11. DNA content (a) and alkaline phosphatase activity (b) of
MC3T3-E1 cells grown on bioactive glass transformed into crystalline
hydroxyapatite (BG-180H) or synthetic hydroxyapatite (HA). Both ma-
terials were subjected to an immersion in tissue culture medium prior to
cell seeding in order to pre-adsorb proteins. Note that proliferation
starts much sooner on BG-180H than on HA. Around day 7 it slows
down to express the di!erentiated function. In contrast on HA prolifer-
ation continued throughout the experiment in a similar manner as in
control conditions (no substrates*not shown here) (with permission
[105]).
controlled release of incorporated biologically active
molecules, such as, e.g. for the release of growth factors
[106}110]. Thus, in analogy to the observation of en-
hanced function of "bronectin, we addressed the question
whether bone morphogenetic protein adsorbed into the
calcium-phosphate surface on BG would a!ect di!erenti-
ation of osteoblast precursor cells more extensively than
when added to the culture medium by itself [111]. In this
experiment we also included two other conditions for
combining BMP-2 (BMP-2, Genetics Institute, Cam-
bridge, MA) with bioactive glass. First, cells were seeded
on the surface treated glass and BMP-2 was added to the
medium. Second, since we synthesized the glass by
a sol}gel technique, we incorporated BMP-2 throughout
the glass matrix. The xerogel matrix allows di!usion of
the molecule through its porous network and, as such,
the molecule is released in a controlled, continuous way
[108]. Control conditions were the rat marrow stromal
cells by themselves and treated glass disks with cells, but
without addition of BMP-2 to the medium.
Fig. 12 summarizes the alkaline phosphatase activity
of the cells cultured under these various conditions [111].
It is evident that the glass treated to simulate in vivo
conditions leading to a calcium-phosphate surface stimu-
lated the di!erentiation of the osteoprogenitor cells. In
fact, the alkaline phosphatase activity of these cells was
higher than that of the cells seeded on the tissue culture
plastic dishes. Under the conditions of the experiment,
the alkaline phosphatase activity of cells grown on the
glass was also higher than that of cells cultured under the
same conditions, but in medium with BMP-2 added. All
data taken together suggested that the e!ect of the glass
by itself followed from the adsorption of growth factors
that are naturally present in the medium. This could be
readily deduced from the results of the three conditions
in which the sol}gel glass was combined with BMP-2.
When the molecule was added to the medium, it
adsorbed onto the Ca}P surface and a!ected cells in
a more extensive way than under any other condition at
the early, i.e. 6 day, time point. In the other cases the
molecule had to di!use to the surface, and therefore it can
be assumed that less was available to stimulate plu-
ripotential precursor cells to the di!erentiated function at
Fig. 12. Average normalized alkaline phosphatase (AP) activity of rat
marrow stromal cells in controls, in which cells were cultured on tissue
culture plastic without and with BMP added to culture medium (&C'
and &C-BT' respectively), and in experimental groups. Cells were grown
on room-temperature processed silica sol}gel materials with a cal-
cium-phosphate surface; BMP was added to the culture medium (SG-
BT), or the BMP was incorporated in the calcium-phosphate surface
(SG-BP), or the BMP was incorporated throughout the glass specimen
(SG-BI). All showed an elevated AP activity over the corresponding
control groups. Cells grown on the glass without added factor (SG) also
showed an increased AP activity when compared with cells on tissue
culture plastic (C) (with permission [111]).
 P. Ducheyne, Q. Qiu / Biomaterials 20 (1999) 2287}2303
this "rst time point. At the later time point, i.e. 10 days, this
disadvantage of di!usion through the solid disappeared
and all conditions wherein the glass and BMP-2 were
combined, stimulated the expression of the osteoblast
phenotype in a similar, major way (assays of other pheno-
typic markers were performed to con"rm this) [111].
Another interesting "nding is the synergistic e!ect be-
tween the calcium-phosphate surface reaction layer on
the glass, and the growth factor used. In fact, there was
a sevenfold increase in expression of the di!erentiated
function by osteoprogenitor cells in comparison to cells
to which BMP-2 was added without the glass [111].
Summing up, the data suggest that concentration of
osteoinductive proteins which are naturally present in
bone tissue repair sites or the addition of a recombinant
form of these growth factors lead to a biologically very
potent state of the molecules. The results also suggest
that the calcium-phosphate surface layers formed on this
Si-containing, biologically reactive glass material, play
an important role in stimulating osteoblastic di!erenti-
ation of osteoprogenitor cells from the bone marrow.
7. Applications in tissue engineering
7.1. Bioactive scawolds for bone tissue engineering
The use of in vitro synthesized bone tissue with mar-
row cells obtained from the patient is an appealing idea
to avoid the profound limitations of biological and syn-
thetic bone grafts. Advantages of in vitro synthesized
bone constructs include that large segments can be pro-
duced when allografts would appear to be unavoidable
because of large defect dimensions, and that cells and
tissue cultured outside the body could aid in the repair of
defects at sites with limited growth potential.
The main components of tissue-engineered bone are
signaling molecules, responding cells and biocompatible
or bioactive matrix materials [112]. It is expected that
ideal matrix materials stimulate osteoprogenitor cell
function and the expression of the osteoblastic pheno-
type. They should also integrate with tissues, gradually
be degraded and fully replaced by bone tissue.
The premise of the bone tissue engineering program in
our laboratory is that for bone cell function and extracel-
lular matrix formation to be optimized, the critical ma-
terial properties that stimulate signaling molecule and
bone cell function must be identi"ed. It is in this context
that the data related to the e!ect of bioactive ceramics and
glasses on cell function are very useful. This is especially
true, given that in an arti"cial or in vitro system, osteo-
progenitor cells, as well as bone cells themselves can rap-
idly change their phenotype. Hence, the substrate needs to
promote the expression of the bone cell phenotype.
Surface modi"ed bioactive glass templates combine
several requirements of the ideal template for in vitro
2299
synthesis of bone tissue. Especially when made in porous,
resorbable form, and conditioned to develop a bone-like
surface prior to being seeded with pluripotential cells
capable of expressing the osteoblastic phenotype, these
templates lead to expeditious and abundant in vitro
synthesis of extracellular matrix with most important
characteristics of bone tissue [49]. In addition, when
growth factors with established bene"cial e!ect on bone
tissue formation would be incorporated in the templates
and be released in a time and dose controlled fashion, it is
possible to further optimize in vitro bone tissue engineer-
ing concepts.
Surface modi"ed bioactive glass templates are resorb-
able and are gradually replaced by bone tissue. This
follows from an implantation experiment in which the
typical femoral bone defects encountered in revision hip
arthroplasties were simulated [113]. In this study, a uni-
cortical window defect measuring 1.1]4.5 mm (width
over length) was created bilaterally in the femoral dia-
physis of adult, male syngeneic rats. Porous, surface
modi"ed bioactive glass ceramics (pSMC) sca!olds were
prepared from 45S5 bioactive glass granules. Two tis-
sue-engineered constructs were synthesized, namely the
sca!old with bone marrow stromal cells seeded at the
time of surgery (&primary'), or these sca!olds with culture
expanded cells producing extracellular bone-like matrix
(&hybrid'). Defects were treated randomly with pSMC,
primary, hybrid, or left untreated (&sham') to compare
healing rates at 2, 4, and 12 weeks.
At 2 weeks, the long bones treated with the hybrid and
primary constructs had 40% more bone in the defect
than bones with the pSMC. Signi"cant increases of bone
tissue formation in the defect occurred over time for all
groups, resulting in approximately 40% bone in the
defect for all treatment groups by 12 weeks. By 12 weeks,
the surface modi"ed bioactive glass sca!old was fully
replaced by in situ formed calcium-phosphate. The ratio
of transformed sca!old to defect size decreased signi"-
cantly for all groups over time. The ultimate torque to
failure and sti!ness were a!ected by treatment and by
time. By 2 weeks, defects treated with hybrid had the
highest sti!ness and this property was not signi"cantly
di!erent from the one of intact bone. Between 2 and
4 weeks, there were signi"cant increases in sti!ness for
primary and sham, resulting in no signi"cant di!erence
among these groups and intact bone by 4 weeks. By 12
weeks, all treatments had comparable sti!ness and
torque to intact bone.
Porous, surface modi"ed bioactive ceramic integrated
well with bone tissue in the healing of a skeletal defect site
and resorbed in concert with bone formation, which
allows for improvement of the long bone's structural
integrity over time. The osteogenic activity of both tis-
sue-engineered constructs, osteoprogenitor cells seeded
onto sca!olds at the time of surgery, or cells expanded to
form in vitro synthesized bone}sca!old, was similar.
2300 P. Ducheyne, Q. Qiu / Biomaterials 20 (1999) 2287}2303
Bone formation and the return of normal torsional prop-
erties were enhanced for the tissue-engineered constructs
as compared to the sca!old alone.
7.2. Intervertebral disc tissue engineering
Bone tissue engineering using surface modi"ed bioac-
tive glass sca!olds relies on the e!ect of the calcium-
phosphate that forms at the glass surface on bone cell
function. When osteoblasts were cultured on glass, the
apatite layer was related to the formation of calci"ed
extracellular matrix [105]. It is of value to note, though,
that the calcium-phosphate layer, which can be formed in
various ways on the glass surface, can also perform other
functions. Especially, the e!ect on non-mineralizing cells
is interesting to explore.
We focused on studying the e!ect on nucleus pulposus
cells by virtue of the signi"cance of the underlying clinical
problem that was thereby addressed. Spinal fusion is
frequently used as a treatment for degenerative disc dis-
ease, especially in cases where instability is a concern.
Fusion provides load bearing and passive restraint to the
motion segment. However, joint motion, which is the
primary function of the disc, is eliminated. As a result,
other motion segments become overloaded, leading to
accelerated degeneration. Other undesirable long-term
e!ects associated with spinal fusions have also been re-
ported [114]. Thus, we pursued tissue engineering prin-
ciples for treating the intervertebral disc and regenerating
its nucleus pulposus (NP) tissue.
When studying cultures with nucleus pulposus cells,
we observed the formation of an apatite layer on the
bioactive glass surface with and without cells by day 21
[115]. However, the layer formed in the cultures with
nucleus pulposus cells di!ered from the one formed in
cultures with osteoblasts. Speci"cally, the layer was thin-
ner and only poorly crystallized. Furthermore, the min-
eral was restricted to a layer of globules on the glass
surface. With nucleus pulposus cells the function of this
layer as an attachment substratum is probably the preva-
lent function. Accordingly, this would imply that the
layer binds "bronectin and other serum molecules, which
then mediate the initial attachment of the nucleus pul-
posus cells and facilitate their subsequent proliferation.
In fact, FTIR analysis revealed carbonyl, ester and amide
group bands in the spectrum. Nucleus pulposus cells
were found to rapidly attach to the surface-treated glass
substrate, colonizing it within 12 h. They proliferated
forming a lawn of cells and matrix by day 21. DNA
analysis showed an increase in cell number with time.
Among other markers, these cells expressed CD-44, a cell
surface glycoprotein that binds hyaluronic acid, which in
turn can bind to "bronectin [115]. All of these observa-
tions point to the qualities of surface modi"ed bioactive
glass as a carrier for nucleus pulposus cells in vitro, and
suggest that glass-cell hybrids provide an exciting novel
approach to repairing and regenerating diseased inter-
vertebral discs.
8. Conclusions
Within excavated bioactive glass granules, osteo-
progenitor cells di!erentiate and lay down bone tissue. In
vitro experiments, summarized in this paper, have re-
vealed some of the critical events for bioactive glass to
stimulate the expression of the osteoblast phenotype. We
suggest that the events that were modeled in vitro are
also likely to take place in vivo, although other phe-
nomena, which have yet to be analyzed, may also con-
tribute to the observed di!erentiation in the excavated
glass particles.
Whereas tissue engineering has found its way into the
clinical practice of restoring skin tissue in severely burnt
patients and patients with chronic decubitus wounds,
methods to regenerate tissues of the musculoskeletal sys-
tem are still in the preclinical phase. Numerous are the
statements emphasizing the importance of carrier mate-
rials for optimal delivery of cells and biological signaling.
In this context, resorbable, bioactive glass is an excellent
material for the stimulation and delivery of cells and
molecules to achieve repair of bone tissue.
The studies on the e!ect of reactivity of bioactive
materials on bone cell function have shown their unique
role in bone bonding and bone formation processes.
Investigations into the mechanisms of bone}bioactive
ceramic interaction provide guidance for the design of
future biomaterials that enhance cell attachment, stimu-
late bone growth and ultimately integrate with the newly
developed tissue bed. However, not every question re-
garding the successful use of bioactive ceramics as tissue
engineering sca!old materials has yet been answered. It
would appear that it is most pressing to elucidate the
composition and structure of the mineral phase that
exquisitely a!ects the function of various biological mol-
ecules, and the relationship between this inorganic phase
and the spatial structure of the adsorbing proteins.
Acknowledgements
Support over the years from various agencies is grate-
fully acknowledged. Part of the work was supported
by the NIH (AR-40194, DE-10639, DE-13051), NSF
(BCS-9202314, BCS-9309053), NASA (NAG9-817,
NAG8-1483) and VA (1189-RA).
References
[1] Praemer A, Furner S, Rice DP. Musculoskeletal conditions in
the United States. Park Ridge: Am Acad Orthop Surg 1992.
p. 1}85.
 P. Ducheyne, Q. Qiu / Biomaterials 20 (1999) 2287}2303
[2] Damien CJ, Parsons JR. Bone graft and bone graft substitutes:
a review of current technology and applications. J Appl Bio-
mater 1992;2:187}208.
[3] Glowacki J, Mulliken JB. Demineralized bone implants. Clin
Plast Surg 1985;12(2):233}41.
[4] Buck BE, Malinin TI, Brown MD. Bone transplantation and
human immunode"ciency virus. An estimate of risk of acquired
immunode"ciency syndrome (AIDS). Clin Orthop 1989;
240:129}36.
[5] Binderman I, Fin N. Bone substitutes-organic, inorganic, and
polymeric: cell material interactions. In: Yamamuro T, Hench
LL, Wilson J, editors. CRC handbook of bioactive biomaterials.
Boca Raton: CRC Press, 1990. p. 45}61.
[6] Passuti N, Daculsi G, Martin S. Macroporous calcium-phos-
phate ceramics for long bone surgery in human and dogs*clini-
cal and histological studies. In: Heimke G, Soltesz U, Lee A,
editors. Clinical implant materials. Amsterdam: Elsevier, 1990.
p. 255}8.
[7] LeGeros RZ. Calcium phosphate materials in restorative den-
tistry: a review. Adv Dent Res 1988;2(1):164}80.
[8] Ricci JL, Blumenthal NC, Spivak JM, Alexander H. Evaluation
of a low-temperature calcium-phosphate particulate implant
material: physical}chemical properties and in vivo bone re-
sponse. J Oral and Maxillofacial Surg 1992;50:969}78.
[9] de Groot K, Tencer A, Waite P, Nichols J, Kay J. Signi"cance of
the porosity and physical chemistry of calcium phosphate cer-
amics. Dental and other head and neck uses. In: Ducheyne P,
Lemons J, editors. Bioceramics: materials characteristics versus
in vivo behavior. New York: NY: Academic Science, 1988.
p. 272}7.
[10] Schepers EJG, Ducheyne P, Barbier L, Schepers S. Bioactive
glass particles of narrow size range: a new material for the repair
of bone defects. Implant Dent 1993;2:151}6.
[11] Schepers E, Declercq M, Ducheyne P, Kempeneers R. Bioactive
glass particulate material as a "ller for bone lesions. J Oral
Rehab 1991;18:439}52.
[12] de Groot K. Ceramics of calcium phosphates: preparation and
properties. In: de Groot K, editor. Bioceramics of calcium-phos-
phate. Boca Raton: CRC Press, 1983. p. 99}114.
[13] Van Raemdonck W, Ducheyne P, De Meester P. Calcium-
phosphate ceramics. In: Ducheyne P, Hastings GW, editors.
Metal and ceramic biomaterials, vol. II. Strength and surface.
Boca Raton: CRC Press, 1984. p. 143}66.
[14] Ducheyne P, Lemons JE, editors. In: Bioceramics: materials
characteristics versus in vivo behavior. New York: NY
Acadamic Science, 1988.
[15] Hench L, Splinter R, Greenlee T, Allen W. Bonding mechanisms
at the interface of ceramic prosthetic materials. J Biomed Eng
1971;2:117}41.
[16] Neo M, Nakaruma T, Yamamuro T, Ohtsuki C, Kokubo T.
Transmission microscopic study of apatite formation on bioac-
tive ceramics in vivo. In: Ducheyne P, Kokubo T, van Blitter-
swijk CA, editors. Bone-bonding biomaterials. Leiderdorp,
The Netherlands: Reed Healthcare Communications, 1993.
p. 111}20.
[17] Kokubo T, Kushitani H, Sakka S, Kitsugi T, Yamamuro T.
Solutions able to reproduce in vivo surface-structure changes in
bioactive glass}ceramic A-W. J Biomed Mater Res 1990;
24:721}34.
[18] El-Ghannam A, Ducheyne P, Shapiro IM. Formation of surface
reaction products on bioactive glass and their e!ects on the
expression of the osteoblastic phenotype and the deposition of
mineralized extracellular matrix. Biomaterials 1997; 18:295}303.
[19] Hyakuna K, Yamamuro T, Kotoura Y, Kakutani Y, Kitsugi T,
Takagi H, Oka M, Kokubo T. The in#uence of calcium-phos-
phate ceramics and glass}ceramics on cultured cells and their
surrounding media. J Biomed Mater Res 1989; 23:1049}66.
2301
[20] El-Ghannam A, Ducheyne P, Shapiro IM. E!ect of serum pro-
tein adsorption on osteoblast adhesion to bioactive glass and
hydroxyapatite. J Orthop Res 1999;17:340}5.
[21] Jarcho M, Kay JF, Gumaer KI, Doremus RN, Drobeck HP.
Tissue, cellular and subcellular events at a bone}ceramic hy-
droxylapatite interface. J Bioeng 1977;1:79}92.
[22] Ducheyne P, Hench L. Comparison of the skeletal "xation of
porous and bioreactive materials. Trans 1st Mtg Europ Soc
Biomater 1977:2PS}P0.
[23] Ducheyne P, Hench LL, Kagan A, Martens M, Bursens A,
Mulier JC. E!ect of hydroxyapatite impregnation on skeletal
bonding of porous coated implants. J Biomed Mater Res 1980;
14:225}37.
[24] Berry JL, Geiger JM, Moran JM, Skraba JS, Greenwald AS. Use
of tricalcium phosphate or electrical stimulation to enhance the
bone-porous implant interface. J Biomed Mater Res 1986;
20:65}77.
[25] Eschenroeder Jr HC, McLaughlin RE, Reger SI. Enhanced sta-
bilization of porous-coated metal implants with tricalcium phos-
phate granules. Clin Orthop Rel Res 1987;216:234}46.
[26] Rivero DP, Fox J, Skipor AK, Urban RM, Galante JO. Calcium
phosphate-coated porous titanium implants for enhanced skel-
etal "xation. J Biomed Mater Res 1988;22:191}201.
[27] Mayor MD, Collier JB, Hanes CK. Enhanced early "xation of
porous-coated implants using tricalcium phosphate. Trans 32nd
ORS 1986. p. 340 (Abstract).
[28] Cook SD, Thomas KA, Kay JF, Jarcho M. Hydroxyapatite-
coated porous titanium for use as an orthopedic biologic attach-
ment system. Clin Orthop Rel Res 1988;230:303}12.
[29] Oonishi H, Yamamoto M, Ishimaru H, Tsuji E, Kushitani S,
Aono M, Ukon Y. The e!ect of hydroxyapatite coating on bone
growth into porous titanium alloy implants. J Bone Jt Surg
1989;71:213}6.
[30] Ducheyne P, Beight J, Cuckler J, Evans B, Radin S. E!ect
of calcium-phosphate coating characteristics on early post-
operative bone tissue ingrowth. Biomaterials 1990;11:
531}40.
[31] Ducheyne P, Cuckler JM. Bioactive ceramic prosthetic coatings
[review]. Clin Orthop Rel Res 1992;276:102}14.
[32] de Bruijn JD, Bovell YP, van Blitterswijk CA. Structural ar-
rangements at the interface between plasma sprayed calcium-
phosphates and bone. Biomaterials 1994;15:543}50.
[33] de Bruijn JD, Flach TS, Leenders H, van de Brink J, van
Blitterswijk CA. Degradation and interface characteristics
of plasma sprayed hydroxyapatite coatings with di!erent
crystallinities. In: Yamamuro T, Kokubo T, Nakamura T, edi-
tors. Bioceramics, vol. 5. Kyoto: Kobunshi Kankokai, 1992.
p. 291}8.
[34] Maxian SH, Zawadski JP, Dunn MG. In vitro evaluation of
amorphous calcium-phosphate and poorly crystallized hy-
droxyapatite coating on titanium implants. J Biomed Mater Res
1993;27:111}7.
[35] Ducheyne P, Radin S, King L. The e!ect of calcium-phosphate
ceramic composition and structure on in vitro behavior.
I. Dissolution. J Biomed Mater Res 1993;27:25}34.
[36] Daculsi G, LeGeros RZ, Nery E, Lynch K, Kerebel B. Trans-
formation of biphasic calcium phosphate ceramics in vivo: ultra-
structural and physicochemical characterization. J Biomed
Mater Res 1989;23:883}94.
[37] Holand W, Vogel W, Naumann K, Gummel J. Interface reac-
tions between machinable bioactive glass}ceramics and bone.
J Biomed Mater Res 1985;19:303}12.
[38] Ducheyne P, Bianco P, Radin S, Schepers E. Bioactive materials:
mechanisms and bioengineering considerations. In: Ducheyne
P, Kokubo T, van Blitterswijk CA, editors. Bone-bioactive bio-
materials. Leiderdorp, The Netherlands: Reed Healthcare Com-
munications, 1993. p. 1}12.
2302 P. Ducheyne, Q. Qiu / Biomaterials 20 (1999) 2287}2303
[39] Klein CP, Driessen AA, de Groot K, van den Hoo! A. Biode-
gradation behavior of various calcium phosphate materials in
bone tissue. J Biomed Mater Res 1983;17:769}84.
[40] LeGeros RZ, Daculsi G, Orly I, Gregoire M, Heughebaert M,
Gineste M, Kijkowska R. Formation of carbonate apatite on
calcium phosphate materials: dissolution/precipitation pro-
cesses. In: Ducheyne P, Kokubo T, van Blitterswijk CA, editors.
Bone-bonding biomaterials. Netherlands: Reed Healthcare
Communications, 1993. p. 201}12.
[41] Patka P, den Hollander W, Klein CPAT, de Groot K. Behavior
of ceramics, hydroxyapatite, and tricalcium phosphate in bone.
Trans Mtg SFB 1989;15:103 (Abstract).
[42] Radin SR, Ducheyne P. The e!ect of calcium phosphate ceramic
composition and structure on in vitro behavior. II. Precipita-
tion. J Biomed Mater Res 1993;27:35}45.
[43] Andersson OH, Karlsson KH, Kangasniemi K. Calcium-phos-
phate formation at the surface of bioactive glass in vivo. J Non-
Cryst Solids 1990;119:290}6.
[44] Ducheyne P, Kim CS, Pollack SR. The e!ect of phase di!erences
on the time-dependent variation of the zeta potential of hy-
droxyapatite. J Biomed Mater Res 1992;26:147}68.
[45] Ohgushi H, Goldberg VM, Caplan AI. Heterotopic osteogenesis
in porous ceramics induced by marrow cells. J Orthop Res
1989;7:568}78.
[46] Gregoire M, Orly I, Manankau J. The in#uence of calcium-
phosphate biomaterials on human bone cell activites. An in vitro
approach. J Biomed Mater Res 1990;24:165}77.
[47] Vrouwenvelder WC, Groot CG, de Groot K. Histological and
biochemical evaluation of osteoblasts cultured on bioactive
glass, hydroxylapatite, titanium alloy, and stainless steel. J Bio-
med Mater Res 1993;27:465}75.
[48] de Bruijn JD, Davies JE, Klein CPAT, de Groot K, van Blitter-
swijk CA. Biological responses to calcium-phosphate ceramics.
In: Ducheyne P, Kokubo T, van Blitterswijk CA, editors. Bone-
bonding biomaterials. Leiderdorp, The Netherlands: Reed
Health-care Communications, 1993. p. 57}72.
[49] El-Ghannam A, Ducheyne P, Shapiro IM. Bioactive material
template for in vitro synthesis of bone. J Biomed Mater Res
1995;29:359}70.
[50] Keeting PE, Oursler MJ, Weigand KE, Bonde SK, Spelsberg
TC, Riggs BL. Zeolite A increases proliferation, di!erentiation,
and transforming growth factor beta production in normal adult
human osteoblast-like cells in vitro. J Bone Miner Res
1992;7:1281}9.
[51] Matsuda T, Davies JE. The in vitro response of osteoblasts to
bioactive glass. Biomaterials 1987;8:275}84.
[52] Kokubo T, Ito S, Haung ZT, Hayashi T, Sakka S, Kitsugi T,
Yamamuro T. Ca}P-rich layer formed on high strength bioac-
tive glass}ceramic A-W. J Biomed Mater Res 1990;24:331}43.
[53] Radin S, Ducheyne P. The e!ect of bioactive ceramic composi-
tion and structure on in vitro behavior: III porous versus dense
ceramics. J Biomed Mater Res 1994;28:1303}9.
[54] Tracy BM, Doremus RH. Direct electron microscopy studies of
the bone}hydroxylapatite interface. J Biomed Mater Res
1984;18:719}26.
[55] Davies JE, Pilliar RM, Smith DC, Chernecky R. Bone interfaces
with retrieved alumina and hydroxyapatite ceramics. In: Bon-
"eld W, Towner KE, Hastings G, editors. Bioceramics, vol. 4.
Oxford: Butterworth, 1991. p. 199}204.
[56] Schepers E, Ducheyne P. The application of bioactive glass
particles of narrow size range as a "ller materials for bone
lesions: a 24 month animal experiment. In: Christiansen D,
Ducheyne P, editors. Bioceramics, vol. 6. Oxford: Butterworth-
Heinemann, 1993. p. 401}4.
[57] Radin S, Ducheyne P. The e!ect of serum proteins on solution
induced surface transformations of bioactive ceramics. J Biomed
Mater Res 1996;30:273}80.
[58] Kitsugi T, Yamamuro T, Nakamura T, Kokubo T, Takagi M,
Shibuya T, Takeuchi H, Ono M. Bonding behavior between two
bioactive ceramics in vivo. J Biomed Mater Res 1987;21:1109}23.
[59] Li P, Bakker D, van Blitterswijk CA. A bone-bonding polymer
(Polyactive(R) 80/20) induces hydroxycarbonate apatite forma-
tion in vitro. J Biomed Mater Res 1997;34:79}86.
[60] Levene H, Phuvanartnuruks V, Abramson S, James K, Kohn J.
Chelation of calcium ions by some tyrosine-derived polymers
may be related to improved bone biocompatibility. Trans 25th
Mtg SFB 1999. p. 13 (Abstract).
[61] Li P, Ducheyne P. Quasi-biological apatite "lm induced on tita-
nium in a simulated body #uid. J Biomed Mater Res 1998;41:341}8.
[62] Li P, Ohtsuki C, Kokubo T, Nakanishi K, Soga N, de Groot K.
The role of hydrated silica, titania, and alumina in inducing
apatite on implants. J Biomed Mater Res 1994;28:7}15.
[63] Cho S-B, Nakanishi K, Kokubo T, Soga N, Ohtsuki C,
Nakamura T. Apatite formation on silica gel in simulated body
#uid: its depedence on structures of silica gels prepared in
di!erent media. J Biomed Mater Res 1996;33:145}51.
[64] Li P, Nakanishi K, Kokubo T, de Groot K. Induction and
morphology of hydroxyapatite, precipitated from metastable
simulated body #uids on sol}gel prepared silica. Biomaterials
1993;14:963}8.
[65] Li P, de Groot K. Calcium phosphate formation within sol}gel
prepared titania in vitro and in vivo. J Biomed Mater Res
1993;27:1495}500.
[66] Pereira MM, Clark AE, Hench LL. Calcium-phosphate forma-
tion on sol}gel-derived bioactive glasses in vitro. J Biomed
Mater Res 1994;28:693}8.
[67] Li R, Clark AE, Hench LL. An investigation of bioactive
glass powders by sol}gel processing. J Appl Biomater 1991;
2:231}9.
[68] Hench LL. Bioceramics: from concept to clinic. J Am Ceram Soc
1991;74(7):1487}510.
[69] E!ah-Kaufman EAB, Ducheyne P, Radin S, Bonnell DA, Com-
posto RJ. Immersion-induced changes of bioactive glass surfaces
in the presence or absence of serum proteins. J Biomed Mater
Res 1999, submitted for publication.
[70] Radin S, Ducheyne P, Rothman B, Conti A. The e!ect of in vitro
modeling conditions on the surface reactions on bioactive glass.
J Biomed Mater Res 1997;37:363}75.
[71] Radin S, Ducheyne P, Falaize S, Hammond A. Transformation
of bioactive glass granules into Ca}P shells in vitro. J Biomed
Mater Res 1999, in press.
[72] Boutin PB, Testerman MC, Wheeler DL, Miller GJ, Schultz GS,
Strates BS, Petty RW. Osteogenic e!ects of transforming
growth-factor-b (TGF-b) with Hydroxyapatite. J Bone Miner
Res 1993;8:S237.
[73] Johnsson MS-A, Paschalis E, Nancollas GH. Kinetics of min-
eralization, demineralization, and transformation of calcium-
phosphates at mineral and protein surfaces. In: Davies JE,
editor. Bone}biomaterial interface. Toronto: University of
Toronto Press, 1991. p. 62}75.
[74] Martin RI, Brown PW. Formation of hydroxyapatite in serum.
Mater Med 1994;5:96}102.
[75] Ohgushi H, Okumura M, Tamai S, Shors EC, Caplan AI. Mar-
row cell induced osteogenesis in porous hydroxyapatite and
tricalcium phosphate: a comparative histomorphometric study
of ectopic bone formation. J Biomed Mater Res 1990;24:1563}70.
[76] Goshima J, Goldberg VM, Caplan AI. The origin of bone
formed in composite grafts of porous calcium-phosphate ce-
ramic loaded with marrow cells. Clin Orthop 1991;269:274}83.
[77] Dennis JE, Haynesworth SE, Young RG, Caplan AI. Osteogen-
esis in marrow-derived mesenchymal cell porous ceramic com-
posites transplanted subcutaneously: e!ect of "bronectin and
laminin on cell retention and rate of osteogenic expression. Cell
Transplant 1992;1:23}32.
 P. Ducheyne, Q. Qiu / Biomaterials 20 (1999) 2287}2303
[78] Nakahara H, Goldberg VM, Caplan AI. Culture-expanded per-
iosteal-derived cells exhibit osteochondrogenic potential in por-
ous calcium-phosphate ceramics in vivo. Clin Orthop Rel Res
1992:291}8.
[79] Okumura J, Ohgushi H, Ducheyne P. Porous titanium felt as
a carrier of osteogenic cells. In: Ducheyne P, Christiansen D,
editors. Bioceramics, vol. 6. Oxford: Butterworths-Heinemann,
1993. p. 305.
[80] Gatti AM, Ducheyne P, Piattelli A, Schepers E, Trisi L, Chiarini
L, Monari E. Glass corrosion layers on bioactive glass granules
of uniform size a!ect cellular function. In: Ducheyne P, Chris-
tiansen D, editors. Bioceramics, vol. 6. Oxford: Butterworth-
Heinemann, 1993. p. 395}400.
[81] Furusawa T, Mizunuma K. Osteoconductive properties and
e$cacy of resorbable bioactive glass as a bone-grafting material.
Implant Dent 1997;6:93}101.
[82] Hench LL, Ethridge EC. Biomaterials: an interfacial approach.
New York: Plenum Press, 1982.
[83] Schepers E, Ducheyne P, Declercq M. Interfacial analysis of
"ber-reinforced bioactive glass dental root implants. J Biomed
Mater Res 1989;23:735}52.
[84] Schepers EJ, Ducheyne P. Bioactive glass particles of narrow
size range for the treatment of oral bone defects: a 1}24 month
experiment with several materials and particle sizes and size
ranges. J Oral Rehab 1997;24(3):171}81.
[85] Low SB, Fetner AE, Clark AE, Hench LL, Wilson-Hench J.
Periodontal osseous defect repair. US Patent 1989, Patent
d 4815046.
[86] Wilson J, Low SB. Bioactive ceramics for periodontal treatment:
comparative studies in the patus monkey. J Appl Biomater
1992;3:123}9.
[87] Ripamonti U. The morphogenesis of bone in replicas of porous
hydroxyapatite obtained from conversion of calcium carbonate
exoskeletons of coral. J Bone Jt Surg 1991;73A:692}703.
[88] Ripamonti U. Osteoinduction in porous hydroxyapatite im-
planted in heterotopic sites of di!erent animal models. Bio-
materials 1996;17:31}5.
[89] Klein C, de Groot K, Chen W, Li Y, Zhang X. Osseous sub-
stance formation induced in porous calcium-phosphate ceram-
ics in soft tissues. Biomaterials 1994;15:31}4.
[90] de Bruijn JD, Dalmeijer R, de Groot K. Osteoinduction by
microstructured calcium-phosphates. Trans 25th Mtg SFB
1999;22:235 (Abstract).
[91] Toth JM, Lynch KL, Hackbarth DA. Ceramic-induced osteogen-
esis following subcutaneous implantation of calcium-phosphates.
In: Ducheyne P, Christiansen D, editors. Bioceramics, vol. 6.
Philadelphia: Butterworth-Heinemann, 1993. p. 9}14.
[92] Urist MR. Bone: formation by autoinduction. Science 1965;
150:893}9.
[93] GarcmH a AJ, Ducheyne P. Numerical analysis of extracellular #uid
#ow and chemical species transport around and within porous
bioactive glass. J Biomed Mater Res 1994;28:947}60.
[94] Brighton CT, Scha!er JL, Shapiro DB, Tang JS, Clark CC.
Proliferation and macromolecular synthesis by rat calvaria bone
cells grown in various oxygen tension. J Orthop Res 1991;9:847}54.
[95] Brighton CT, McCluskey WP. Response of cultured bone cells
to a capacitively coupled electric "eld: inhibition of cAMP
response to parathyroid hormone. J Orthop Res 1987;6:567}71.
[96] Scha!er JL, Brighton CT, Clark CC, Brinly FJ. In vitro growth
of rat calvaria bone cells at various oxygen tensions. Trans
Orthop Res Soc 1987;12:176.
[97] Spindler KP, Shapiro DB, Gross SB, Brighton CT, Clark CC.
The e!ect of ascorbic acid on the metabolism of rat calvaria
bone cells in vitro. J Orthop Res 1989;7:696}701.
[98] Vrouwenvelder WC, Groot CG, de Groot K. Behaviour of fetal
rat osteoblasts cultured in vitro on bioactive glass and nonreac-
tive glasses. Biomaterials 1992;13(6):382.
2303
[99] Vrouwenvelder WC, Groot CG, de Groot K. Better histology
and biochemistry for osteoblasts cultured on titanium-doped
bioactive glass: bioglass 45S5 compared with iron-, titanium-,
#uorine- and boron-containing bioactive glasses. Biomaterials
1994;15:97}106.
[100] Gross U, Brandes J, Strunz V, Bab I, Sela J. The ultrastructure of
the interface between a glass}ceramic and bone. J Biomed Mater
Res 1981;15:291}305.
[101] El-Ghannam A, Ducheyne P, Shapiro I. Bioactive glass tem-
plates for the synthesis of bone-like tissue in vitro. In: Mikos
AG, Murphy RM, Bernstein H, Peppas NA, editors. Bio-
materials for cell and drug delivery. Pittsburgh: Materials Re-
search Society, 1994. p. 257}62.
[102] Hench LL. Bioactive ceramics. In: Ducheyne P, Lemons JE,
editors. Bioceramics: material characteristics versus in vivo be-
havior. New York: The New York Academy of Sciences, 1988.
p. 54}71.
[103] Garcia AJ, Ducheyne P, Boettiger D. E!ect of surface reaction
stage on "bronection-mediated adhesion of osteoblast-like cells
to bioactive glass. J Biomed Mater Res 1998;40:48}56.
[104] Grinnell F, Feld MK. Adsorption characteristics of plasma
"bronectin in relationship to biological activity. J Biomed Mater
Res 1981;15:363}81.
[105] El-Ghannam A, Ducheyne P, Shapiro IM. Porous bioactive
glass and hydroxyapatite ceramic a!ect bone cell function in vitro
along di!erent time lines. J Biomed Mater Res 1997;36:167}80.
[106] Chesmel K, Beight J, Rothman R, Tuan R. TGF-beta enhances
osseointegration in vivo. In: Ducheyne P, Christiansen D, edi-
tors. Bioceramics, vol. 6. Oxford: Butterworth-Heinemann Ltd,
1993. p. 21}6.
[107] Beck LS, Deguzman L, Lee WP, Zioncheck TF, Osaka G,
Nguyen T, Ongpipattanakul B, Gorrel J, Aufdemorte TB,
Plouhar PL. TGF-b1 bound to tricalcium phosphate persists at
segmental radial defects and induces bone formation. Trans 41st
Ann Meet ORS 1995;20:593 (Abstract).
[108] Santos EM, Radin S, Ducheyne P. Sol}gel derived carrier for
the controlled release of proteins. Biomaterials 1999;
20:1695}700.
[109] Nicoll SB, Radin S, Santos EM, Tuan RS, Ducheyne P. In vitro
release kinetics of biologically active transforming growth factor
b1 from a novel xero-gel carrier. Biomaterials 1997;18:853}9.
[110] Downes S, DiSilvio L, Klein CPAT, Kayser MV. Growth hor-
mone loaded bioactive ceramics. J Mater Sci: Mater Med
1991;2:176}80.
[111] Santos EM, Radin S, Shenker BJ, Shapiro IM, Ducheyne P.
Si}Ca}P xerogels and bone morphogenetic protein act synergis-
tically on rat stromal marrow cell di!erentiation in vitro. J Bio-
med Mater Res 1998;41:87}94.
[112] Ripamonti U, Reddi AH. Tissue engineering, morphogenesis,
and regeneration of the periodontal tissues by bone mor-
phogenetic proteins. Crit Rev Oral Biol Med 1997;8:154}63.
[113] Livingston T, Ducheyne P, Garino J. Mechanical and histologi-
cal evaluation of a tissue engineered bioactive sca!old in vivo.
In: Ohgushi H, Hastings GW, Yoshikawa T, editors. Bio-
ceramics, vol. 12. Singapore: World Scienti"c Publishing Co.,
1999. p. 245}8.
[114] Lehmann TR, Spratt KF, Tozzi JE, Weinstein JN, Reinarz SJ,
el-Khoury GY, Colby H. Long-term follow-up of lower lumbar
fusion patients. Spine 1987;12(2):97}104.
[115] Gan JC, Ducheyne P, Vresilovic E, Shapiro IM. Bioactive glass
serves as a substrate for maintenance of phenotype of nucleus
pulposus cells of the intervertebral disk. J Biomed Mater Res
1999, submitted for publication.
[116] Schepers E, Pinruethai P. A comparative study of bioactive glass
and hydroxyapatite particles in periodontal bone lesions. In:
Ducheyne P, Christiansen D, editors. Bioceramics, vol. 6. Ox-
ford: Butterworth-Heinemann, 1993. p. 113}6.