Anda di halaman 1dari 11


Biomaterials 26 (2005) 3565–3575

3D microenvironment as essential element for osteoinduction by

Pamela Habibovica, Huipin Yuanb, Chantal M. van der Valkb, Gert Meijerc,
Clemens A. van Blitterswijka,, Klaas de Groota
Institute for Biomedical Technology, Twente University, Department Bilthoven, Professor Bronkhorstlaan 10-D, Bilthoven, 3723 MB, The Netherlands
IsoTis SA, Professor Bronkhorstlaan 10-D, Bilthoven, 3723 MB, The Netherlands
University Medical Center Utrecht, P.O. Box 85500, Utrecht, 3508 GA, The Netherlands
Received 14 July 2004; accepted 22 September 2004


In order to unravel the mechanism of osteoinduction by biomaterials, in this study we investigated the influence of the specific
surface area on osteoinductive properties of two types of calcium phosphate ceramics. Different surface areas of the ceramics were
obtained by varying their sintering temperatures.
Hydroxyapatite (HA) ceramic was sintered at 1150 and 1250 1C. Biphasic calcium phosphate (BCP) ceramic, consisting of HA
and beta-tricalcium phosphate (b-TCP), was sintered at 1100, 1150 and 1200 1C.
Changes in sintering temperature did not influence the chemistry of the ceramics; HA remained pure after sintering at different
temperatures and the weight ratio of HA and b-TCP in the BCP was independent of the temperature as well. Similarly,
macroporosity of the ceramics was unaffected by the changes of the sintering temperature. However, microporosity (pore diameter
o10 mm) significantly decreased with increasing sintering temperature. In addition to the decrease of the microporosity, the crystal
size increased with increasing sintering temperature. These two effects resulted in a significant decrease of the specific surface area of
the ceramics with increasing sintering temperatures.
Samples of HA1150, HA1250, BCP1100, BCP1150 and BCP1200 were implanted in the back muscles of Dutch milk goats and
harvested at 6 and 12 weeks post implantation. After explantation, histomorphometrical analysis was performed on all implants.
All implanted materials except HA1250 induced bone. However, large variations in the amounts of induced bone were observed
between different materials and between individual animals.
Histomorphometrical results showed that the presence of micropores within macropore walls is necessary to make a material
osteoinductive. We postulate that introduction of microporosity within macropores, and consequent increase of the specific surface
area, affects the interface dynamics of the ceramic in such a way that relevant cells are triggered to differentiate into the osteogenic
r 2004 Elsevier Ltd. All rights reserved.

Keywords: Hydroxyapatite (HA); Biphasic calcium phosphate (BCP); Sintering temperature; Macrostructure; Microstructure; Specific surface area;

1. Introduction

Osteoinduction can be defined as the ‘‘induction of

Corresponding author. Tel.: +31 30 2295 231; fax: +31 30 undifferentiated inducible osteoprogenitor cells that are
not yet committed to the ostogenic lineage to form
E-mail addresses:, osteoprogenitor cells’’ [1]. In order to prove the (C.A. van Blitterswijk). osteoinductive potential of a biomaterial, soft tissue

0142-9612/$ - see front matter r 2004 Elsevier Ltd. All rights reserved.
3566 P. Habibovic et al. / Biomaterials 26 (2005) 3565–3575

implantation, i.e. implantation in the absence of cells indicator of the influence of sintering temperature on
with direct bone forming capacity, can be used. One of ceramic osteoconductive properties is that the expres-
the first evidence of osteoinduction was given by Urist in sions of relevant markers for osteogenic differentiation,
1965 [2], after implantation of demineralized bone such as alkaline phosphatase and osteocalcin, have been
matrix (DBM) in soft tissue of rabbits, rats, mice and reported to be different for varying sintering tempera-
guinea pigs. Later studies [3–5] suggested that DBM tures [28].
contained morphogenetic factors capable of The goal of the current in vivo study was to try to
inducing the differentiation of resident extraskeletal unravel the possible mechanism of osteoinduction by
mesenchymal cells firstly into chondrocytes and then investigating the influence of the microenvironment
into osteoblasts, i.e. osteoinduction. At present, highly around the undifferentiated inducible cells on the
purified native bone morphogenetic proteins (BMPs) osteoinductive capacity of the CaP ceramics. The
and recombinant human BMPs are available and in specific surface area was used to quantify the micro-
many studies their osteoinductive potential has been environment. The tool used to vary the specific surface
shown [6–8]. area was the sintering temperature of the ceramics. In
The general idea that BMPs are always necessary for order to avoid possible influence of the chemical
triggering bone induction was challenged by Winter’s composition, not one but two types of CaP ceramics,
and Simpson’s discovery [9] of osteoinduction by a namely HA and BCP, were chosen. Osteoinduction was
polymeric sponge. In this study, it was observed that evaluated by implantation in the back muscles of adult
prior to the process of bone formation, the calcification Dutch milk goats.
of a polymeric sponge had taken place, suggesting
the importance of the in vivo calcification, and thereby
calcium phosphates (CaPs) in the process of osteoin- 2. Materials and methods
The importance of CaPs in osteoinduction has been 2.1. Implants
supported by the reports of various groups in the last
decade. Some examples are studies that showed bone Porous HA implants were produced by using the
induction in synthetic hydroxyapatite (HA) ceramic in dual-phase mixing method described earlier [21]. In this
dogs [10–14], in coral-derived HA ceramic in dogs, method, commercially available HA powder (Merck,
monkeys and baboons [14–16], in a-TCP-, b-TCP-, Amsterdam, The Netherlands) was used. The processing
BCP-, a-pyrophosphate- and b-pyrophosphate ceramics route consisted of three steps. In the first step, HA slurry
[16–22]. In addition to the CaP-containing biomaterials, was prepared by mixing 2/3 wt% of calcined HA powder
there have also been a few reports showing osteoinduc- with 1/3 wt% water containing deflocculant (Dolapix
tion by alumina ceramic [23] and titanium [24] in dogs. CE 64, Germany) and binder (carboxyl-methyl cellulose,
Similar to Winter’s and Simpson’s observation [9], the in Pomosin BC, The Netherlands). In the second step, two
vivo calcification of these materials is believed to be the immiscible phases were mixed: water-based HA slurry
precursor of bone induction. and polymethyl methacrylate (PMMA) resin with a
Because the mechanism of osteoinduction by bioma- volume ratio of 1:1. The PMMA resin consisted of
terials is not completely understood, it is unknown PMMA powder, MMA monomer and naphthalene
whether it is the biomaterial, or possibly an interaction (o10 v/v%) as an additional fugitive pore maker. In
between the biomaterial and the relevant proteins the final step, the mixture was mould-shaped, polymer-
present in body that is responsible for the process of ized, dried and pyrolized. Finally, the materials were
bone induction. Since most implants do not induce divided into two groups and sintered in air at 1150 and
bone, specific material properties are apparently needed 1250 1C, respectively, for 8 h. A lathe was then used to
for starting the process of bone induction. To start the produce cylinders (+5  10 mm3).
differentiation of the undifferentiated inducible osteo- Ultrastructures of the porous HA ceramics were
progenitor cells into bone-forming cells, it has been characterized by an environmental scanning electron
suggested that not only the chemistry, but also geometry microscope (ESEM; XL30, ESEM-FEG, Philips, The
of the biomaterial in contact with these cells are critical Netherlands) in the secondary electron mode. Macro-
factors [20–24]. In other words, the microenvironment porosities and average pore sizes of the ceramics were
around the cells is crucial. As reported earlier, changing determined by using histological slides (10 cross-sections
the sintering temperature of a CaP ceramic has a for 6-week- and 10 cross-sections for 12-week-implanta-
consequence on its microstructure and crystal size, i.e. tion). First, the labels of the sections were covered.
its specific surface area. This in turn not only influences Then, high-resolution (300 dpi), low-magnification
mechanical strength [25] and initial bone bonding of the (10  ) digital micrographs were made of these sections.
ceramic [26], but also the microenvironment and thus Using Adobe Photoshop 7.0, bone and material were
possibly the ceramic osteoconductive properties [27]. An pseudocoloured, red and green, respectively. Image
P. Habibovic et al. / Biomaterials 26 (2005) 3565–3575 3567

analyses were carried out with a computer-based system 2.2. Animals and implantation
equipped with KS400 version 3.0 software (Carl Zeiss
Vision, Oberkochen, Germany). Prior to measurements, This study was approved by the Dutch Animal Care
the system was geometrically calibrated with an image and Use Committee. In total, ten adult Dutch milk
of a block of known dimensions. A program was goats were used. The animals were housed in Central
developed in KS400 to quantitate the pore diameter for Animal Laboratory Institute (GDL), Utrecht, The
individual pores and the total macroporosities of the Netherlands, at least 4 weeks prior to surgery.
ceramics. The macroporosity was determined as: [(total Before surgical procedures, a dose of 0.1 mL in 5 mL
implant surfacescaffold surface)/total implant of physiologic saline solution (71 mL/25 kg body
surface]  100%. Microstructures of the ceramics were weight) of Domosedan (Pfizer Animal Health BV,
analyzed by using a mercury intrusion porosimeter (MP, Capelle a/d Ijssel, The Netherlands) was administered
AutoPore IV 9500, Micromeritics Instrument Coopera- by intravenous injection. Surgical procedures were
tion, Norcross, Georgia). For a good comparison, the performed under general inhalation anesthesia of the
microporosity of each ceramic was expressed as total animals. Thiopental (Nesdonal, 7400 mg/70 kg of body
incremental pore volume (mL) per gram of the material weight, on indication, Rhone Merieux, Amstelveen, The
for the pores with a diameter lower than 10 mm. Specific Netherlands) was injected intravenously, and anesthesia
surface areas of the ceramics were determined from the was maintained with a gas mixture of nitrous oxide
mercury porosimeter results, as the cumulative surface (66 v/v%), oxygen (33 v/v%) and Halothane (1 v/v%)
area (m2/g) when all the pores were filled with mercury. (ICI-Farma, Rotterdam, The Netherlands).
Pore interconnectivity was visually analyzed by an Besides the implantations described in this study, the
ESEM on the material cross-sections. animals were used for a different study, to be published
Compositions and crystal structures of the ceramics separately. Different groups of implants were assumed
were determined by using fourier transform infra red not to influence each other’s behavior, as they were
spectroscopy (FTIR; Spectrum100, Perkin Elmer Ana- implanted either at a different implantation site or at a
lytical Instruments, Norwalk, CT) and X-ray diffraction sufficient distance from each other as approved by the
(XRD; Miniflex, Rigaku, Japan). Dutch Animal Care and Use Committee.
Porous BCP implants were prepared by using the After shaving the lumbar areas and disinfection with
H2O2 method as published earlier [21]. For the iodine, left muscle fascias were exposed and cut. Using
preparation of the ceramic, in-house made BCP powder blunt dissection, intramuscular pockets were created,
was used. Porous green bodies were produced by mixing and filled with one of each above-mentioned implants:
this powder with 2% H2O2 solution (1.0 g powder/ HA1150, HA1250, BCP1100, BCP1150 and BCP1200.
1.270.05 ml solution) and naphthalene (Fluka Chemie, Subsequently, fascias were closed with nonresorbable
The Netherlands) particles (710–1400 mm; 100 g powder/ sutures to facilitate implant localization at explantation.
30 g particles) at 60 1C. The naphthalene was then The skin was closed in two layers. After 6 weeks, the
evaporated at 80 1C and the green porous bodies were same procedure was repeated in the right back muscle of
dried. They were divided into three groups and sintered each of the ten goats. Table 1 shows the total amounts
at 1100, 1150 1C and 1200 1C respectively for 8 h. of implanted materials for both time points.
Finally, a lathe was used to produce cylinders with a Immediately after surgery, pain relief was given by
size of +5  10 mm3 for BCP1200 and +7  7 mm3 for buprenofine (Temgesic; Schering-Plough, Kenilworth,
BCP1100 and BCP1150. Due to the differences in the NJ).
production techniques, BCP implants sintered at 1100 Twelve weeks after the first implantations (i.e.
and 1150 1C had a different size from the BCP sintered implantation times 6 and 12 weeks), each animal
at 1200 1C and HA implants. Although the influence of was sacrificed by an overdose of pentobarbital
the implant size cannot be completely excluded, the (Euthesaat, Organon, Oss, The Netherlands) and
results of this study suggest that we can reliably make potassium chloride.
the comparison between all implanted ceramics. Ultra-
structures of the porous BCP ceramics were character-
ized by an ESEM. Porosities, pore sizes, pore Table 1
interconnectivity and specific surface areas were ana- Overview of the total amounts of all implanted ceramics
lyzed by the same techniques as described for the HA
Intramuscular implantation 6 weeks 12 weeks
implants. Ceramic chemical compositions and crystal
structures were determined by using an FTIR and XRD. HA1150 10 10
HA/b-TCP weight ratios in the BCP ceramics were HA1250 10 10
calculated by comparing the BCP XRD patterns to the BCP1100 10 10
BCP1150 10 10
calibration patterns prepared from the powders with the
BCP1200 10 10
known HA/b-TCP weight ratios.
3568 P. Habibovic et al. / Biomaterials 26 (2005) 3565–3575

2.3. Retrieval of the implants, histology and The Wilcoxon signed rank test [32] was used to
histomorphometry analyze the difference in bone formation per material
between the two time points and the difference in bone
The implants with surrounding tissue were explanted contact per material between the total implant and the
by sharp dissection and fixed in Karnovsky’s fixative. peripheral zone.
They were dehydrated in a graded ethanol series In both tests, the significance level was set at p ¼ 0:05:
(70–100%) and transferred into a methylmethacrylate
(MMA) solution that polymerized at 37 1C within 1
week. Longitudinal sections (10–15 mm in thickness) 3. Results
were made by using the modified interlocked diamond
saw (Leica Microtome, Nussloch, Germany). Sections 3.1. Material characterization
were stained with 1% methylene blue and 0.3% basic
fuchsin after etching with HCl/ethanol mixture. HA ceramics, sintered at both 1150 and 1250 1C
For histomorphometry, the same image analysis consisted of pure HA, as shown by the FTIR spectra
method was used as previously described for the and XRD patterns (Fig. 1a and b, respectively). No
measurement of material macroporosities. In the com- differences could be found in FTIR spectra and XRD
puter-based image analysis system, a program was patterns between the HA ceramics sintered at either
developed to quantify different parameters concerning temperatures. Observations by the stereomicroscope
bone formation: and by the ESEM showed that both ceramics had
similar macrostructures, consisting of well-intercon-
(a) percentage of bone occupying available pore area
nected macropores. Macropore diameters for both
ceramics were similar ranging between 200 and
(b) percentage of available scaffold outline (in 3D
400 mm, with an average of 249738 mm (Table 2).
surface of the scaffold) in contact with bone:
However, microstructures of the two ceramics differed:
[%contact=(bonescaffold contact length/scaffold
HA1250 had a rough microstructure, with only few
outline length)  100%] in the total area of the
micropores (pore diametero10 mm) (Fig. 2a), while
implant (%b.cont. in implant), and
macropore walls of HA1150 contained many micro-
(c) percentage of available scaffold outline in contact
with bone in the peripheral area that was defined by
drawing a line 40 pixels central from the outer
contour (corresponds to a distance of 350 mm)
(%b.cont. in outer zone).

In addition to the measurements of the amounts of

induced bone in the available pore areas, the bone
contact measurements were performed as this parameter
can give additional information on the bone growth
direction in time [29,30]. Measurements of the bone
contact in both total area of the implant and in the
peripheral area were performed in order to get more
insight into the preferable location for bone induction
and consequently, into the possible mechanism of this

2.4. Statistical analysis

Statistical calculations were done with the SPSS

(Chicago, IL) 9.0 software. We found large variances
between the individual animals. However, the trend
within each animal was similar (material A induced
more bone than material B in all goats). Because the
distribution of the data was not normal, we chose non-
parametric tests to perform the statistical analysis.
Friedman rank test, followed by a post hoc test [31]
was chosen to make the comparisons between all Fig. 1. FTIR spectra (a) and XRD patterns (b) of HA1250 and
ceramics at both time points. HA1150.
P. Habibovic et al. / Biomaterials 26 (2005) 3565–3575 3569

Table 2
Macroporosity and specific surface area overview

Material Macroporosity Diameter Specific surface

(%) macropores area (m2/g)

HA1150 47.571.2 243.9738.7 1.32

HA1250 46.574.9 248.4751.0 0.07
BCP1100 50.973.0 367.1744.7 1.60
BCP1150 54.378.0 380.2760.6 1.02
BCP1200 53.774.2 371.1763.9 0.71

Fig. 3. Total incremental micropore volume for the implanted

ceramics. Decrease in the sintering temperature results in an increase
of the total micropore volume, i.e. microporosity of a ceramic.

Fig. 2. ESEM photographs (magnification 5000  ) of HA1250 (a) and

HA1150 (b). Decrease in the sintering temperature results in an
increase of the amount of micropores and a decrease in the crystal size
of a ceramic.
Fig. 4. FTIR spectra (a) and XRD patterns (b) of BCP1200, BCP1150
and BCP1100.
pores (Fig. 2b). Furthermore, the average crystal size of
the HA1250 was higher as compared to HA1150. These
observations were confirmed by the porosity measure- mental volume of micropores, i.e. the microporosity is
ments. Macroporosities of ceramics were 4773% as much higher for HA1150 as compared to HA1250. The
determined from the material cross-sections (Table 2). specific surface area of HA1150 was 1.32 m2/g and of
For the pores with the pore diameter lower than 10 mm, HA1250 0.07 m2/g, respectively (Table 2).
large differences in the total pore volume between the BCP ceramics sintered at 1100, 1150 and 1200 1C
HA1150 and HA1250 were observed. As can be seen in exhibited similar FTIR spectra and XRD patterns as is
Fig. 3, for both temperatures, most micropores had a shown in the Figs. 4a and b, respectively. As can be seen
diameter of around 0.5 mm. However, the total incre- from the XRD pattern, there is a peak at 2y ¼ 31:21
3570 P. Habibovic et al. / Biomaterials 26 (2005) 3565–3575

typical of b-TCP, and absent on the XRD pattern of  Specific surface areas: BCP1100 (1.60 m2/g)4HA1150
pure HA (Fig. 1b). Compositions of the three BCP (1.32 m2/g)4BCP1150 (1.02 m2/g)4BCP1200 (0.71 m2/
ceramics were the same, all consisting of 88 wt% HA g)4HA1250 (0.07 m2/g).
and 12 wt% b-TCP. Similar to the two HA ceramics,  Chemical composition: 88 wt% HA/12 wt% b-TCP in
BCP ceramics also consisted of highly interconnected BCP versus 100 wt% HA.
macropores ranging in size between 320 and 470 mm,
with an average of 373756 mm (Table 2). LM and
ESEM observations showed similar macrostructures for
3.2. Intramuscular implantation
BCP1100, BCP1150 and BCP1200. Microstructures of
the BCP ceramics sintered at the three temperatures did
There were no surgical complications and all implants
differ. As can be seen from the high magnification
were retrieved. At retrieval, all implants were sur-
ESEM photographs (Figs. 5a–c), lowering the sintering
rounded by well-vascularized muscle tissue. Histology
temperature increased the amount of micropores within
showed no evidence for toxicity of the implants, or signs
macropore walls and decreased ceramic’s crystal size.
of an inflammatory tissue response directly related to
Measurements on cross-sections showed similar macro-
the implants.
porosities for the three ceramics of 5375% (Table 2).
Table 3 gives an overview of the bone incidence after 6
As illustrated by Fig. 3, most micropores of the three
and 12 weeks of implantation. In some animals, bone
BCPs had a diameter of around 1 mm. Fig. 3 also shows
was not induced at all, while in the others (2 out of 10
that the increase in the sintering temperature resulted in
after 6 weeks, and 5 out of 10 after 12 weeks) bone
a decreased total incremental micropore volume per
induction was considerable (42% bone contact) as
gram of the material (microporosity), resulting, in turn,
compared to the results of similar studies in the same
in a decreased specific surface area: BCP1100=1.60 m2/
g, BCP1150=1.02 m2/g, and BCP1200=0.71 m2/g Table 3
(Table 2). Bone incidence after 6 and 12 weeks of implantation
Material characterization results of this study can be
Material 6 weeks 12 weeks
summarized as follows:
HA1150 5/10 7/10
 Macroporosities of all implanted ceramics were HA1250 0/10 0/10
BCP1100 6/10 8/10
similar (50%). Average macropore size of HAs
BCP1150 7/10 6/10
was around 245 mm and that of BCPs around 370 mm BCP1200 3/10 6/10
in diameter.

Fig. 5. ESEM photographs (magnification 5000x) of BCP1200 (a), BCP1150 (b) and BCP1100 (c). Decrease in the sintering temperature results in an
increase of the amount of micropores and a decrease in the crystal size of the ceramic.
P. Habibovic et al. / Biomaterials 26 (2005) 3565–3575 3571

model. Because of the large variations in the amounts of Significant differences (po0:05) between 6 and 12
induced bone between individual animals (non-normal weeks of implantation were only found for BCP1100
data distribution), calculations of the average values and and HA1150, for all three measured parameters.
the standard errors of the mean for bone contact and Significant differences (po0:05) between %b.cont. in
bone area were not possible. To illustrate the large total implant and %b.cont. in peripheral zone were
differences, we plotted the data found for %b.cont. in found for two ceramics: %b.cont. in total implant was
total implant for goats 1 and 3 (Fig. 6). significantly higher than %b.cont. in peripheral zone
HA1250 was the only ceramic that did not induce after both 6 and 12 weeks of implantation for HA1150
bone formation in any of the animals. Table 4 gives an and only after 6 weeks of implantation for BCP1150.
overview of the significant differences (po0:05) between No bone was observed on the outside of the implants.
the ceramics for each of the measured parameters and Bone formation was limited to inside the pores. Fig. 7a
for both time points. illustrates new bone formation in BCP1150 after 12
weeks of implantation. The formed bone was normal in
appearance as illustrated by Fig. 7b. Fig. 7c is a
histological slide of HA1250 after 12 weeks of implanta-
tion, the ceramic that did not induce bone in any of the
Bone induction results of this study can be summar-
ized as follows:
 Bone induction: (BCP1100, BCP1150)4(BCP1200,
HA1150)4HA1250 (no induction at all).
 Bone induction in individual animals: 42% b. cont.
in goats no. 1 and 3 after 6 weeks and in goats no.
1–4, and 6 after 12 weeks of implantation. Hardly any
bone induction in goats no. 7, 8 and 10.

Fig. 6. Histomorphometrical results: % bone contact with the implant

surface in the total area of the implant in goats 1 and 3 after 6 weeks of
implantation. Variations in the amount of formed bone are large 4. Discussion
between the individual animals, but the order of osteoinductive
potential of the materials remains similar. Winter’s and Simpson’s observation of bone induc-
tion by a polymeric sponge [9] after soft tissue
implantation in pigs could not be explained by the
Urist’s BMP theory [33], as the implanted polymeric
Table 4 sponge initially neither contained nor produced BMPs.
Overview of the implanted ceramics between which a significant An interesting finding of the Winter’s and Simpson’s
differences (po0:05) was observed after 6 and 12 weeks of study was that the implanted polymer showed in vivo
implantation for the three measured parameters according to the calcification prior to the process of bone formation,
Friedman’s rank test
suggesting the significance of CaPs in the process of
6 weeks 12 weeks osteoinduction. In a large number of publications,
osteoinduction by CaP biomaterials in the form of
%b. BCP11004HA1250 BCP11004HA1250
ceramics [10,12–18,20,22,34,35], cements [18,22,36],
BCP11504HA1250 BCP11504HA1250
BCP11004BCP1200 coatings [37–40] and glass ceramics [41] has been shown.
BCP11504BCP1200 In addition to CaP-containing biomaterials, osteoinduc-
HA11504=HA1250* tion has also been observed in alumina ceramic [23] and
%b.cont. BCP11004HA1250 BCP11004HA1250 titanium [24].
implant BCP11504HA1250 BCP11504HA1250 Although osteoinduction by biomaterials is obviously
HA11504=HA1250* BCP11004BCP1200 a real phenomenon, its mechanism is still largely
BCP11504BCP1200 unknown. This and previous work indicate some of
%b.cont. BCP11004HA1150 BCP11004HA1250 the parameters necessary for making a biomaterial
peripheral BCP11504HA1150 osteoinductive. In turn, these parameters suggest the
zone BCP11004HA1250 mechanisms involved in the phenomenon of osteoin-
BCP11504HA1250 duction.
The presence of macropores, or concavities [42] is
*p=0.05. shown to be a prerequisite for osteoinduction by
3572 P. Habibovic et al. / Biomaterials 26 (2005) 3565–3575

Fig. 7. LM photographs of BCP1150 (magnification 4  ) (a), BCP1150 (magnification 20  ) (b), and HA1250 (magnification 4x) (c) after 12 weeks
of implantation. B=bone, ST=soft tissue BCP=BCP1150 ceramic, HA=HA1250 ceramic. The induced bone is formed in the pores of the implant,
aligning its surface (a). The bone is normal in appearance, aligned with osteoblasts and with osteocytes and osteoid clearly visible (b). The non-
inductive ceramic is filled with fibrous tissue, but no signs of bone formation are observed (c).

biomaterials. The presence of a well-interconnected the amount of bone induced by BCP1200 was signifi-
macroporous structure is important for a good supply cantly lower than the amount of bone induced by both
of the body fluids with nutrients throughout the BCP1100 and BCP1150. However, no significant differ-
implant. Accompanied with this supply, the release of ences in the amounts of bone induced between BCP1100
calcium and phosphate ions, that is believed to be the and BCP1150 were found, although they had different
origin of the bioactivity of CaP biomaterials [42–45], specific surface areas, which suggest the existence of an
took place, followed by the precipitation of a biological optimal specific surface area. Materials with a specific
apatite layer [46]. The precipitation of this apatite layer surface area below the optimum will degrade slower,
took place when the concentration of calcium and and will finally induce less bone. Materials with a
phosphate ions had reached the supersaturation level in specific surface area above the optimum might degrade
the material vicinity. This might explain the fact that too fast, losing thereby the shape and stability that is
bone induction always takes place in the pores in the necessary to facilitate bone formation. Finally, there is a
center of the implant and not on the implant periphery minimum threshold in the amount of micropores and
[30]. The diffusion of the released calcium and phos- hence specific surface area below which bone will not be
phate ions might occur too fast at the implant periphery, induced.
and therefore not allow for ion concentration increase Also, the changes of the material chemical composi-
required for the biological apatite formation. tion can influence its in vivo degradation. Both
It is expected that a material with a higher dissolution BCP1100 and BCP1150 performed significantly better
rate will release calcium and phosphate ions faster, than HA1250. Moreover BCP1150 induced more bone
followed by a faster formation of the biological apatite than HA1150. Although they were sintered at the same
layer. One way to influence the in vivo dissolution rate temperature, the specific surface area of the HA ceramic
of a material is by changing its specific surface area. This was slightly higher than that of BCP. From the
study showed, for the first time, that HA1250, which conclusion on the effect of specific surface area drawn
possessed very few micropores, and hence a low specific above, one would expect more bone in HA1150 than in
surface area, did not induce any bone, while HA1150, BCP1150. Therefore it is evident that there are other
having the same chemical composition but a much factors influencing osteoinductive capacity of a ceramic.
higher specific surface area, did induce bone. Similarly, In addition to the difference in the average macropore
P. Habibovic et al. / Biomaterials 26 (2005) 3565–3575 3573

Fig. 8. Flow chart of the proposed mechanism of osteoinduction by biomaterials.

diameter of about 100 mm between BCP1150 and between the higher animals and rodents [47], there are
HA1150, the most important difference between the reports of differences in the response to BMPs between
two materials was the presence of b-TCP in BCP. the individuals of the same species, probably due to
Although we cannot exclude the importance of the genetic factors [48]. Similar differences were also
macropore size, based on the earlier research we observed in humans [49]. Thus, it seems plausible that
hypothesize that the effect of the presence of the more endogenous BMPs, or other relevant proteins play a role
soluble b-TCP in BCP, and subsequently faster dissolu- in osteoinduction by biomaterials. This means that
tion of BCP is more relevant than the difference in precipitation of the biological apatite layer in vivo could
macropore size between the two materials. be accompanied by the coprecipitation of the relevant
In the biomaterials that initially do not contain CaPs, proteins (e.g. BMPs), which in turn trigger the recruited
like alumina ceramic and titanium, the formation of the cells to differentiate into the osteogenic lineage. The
biological apatite layer will take longer than in amounts of coprecipitated proteins, that vary with the
biomaterials that do contain CaPs. In the non-CaP amounts of endogenous proteins, in conjunction with
biomaterials, calcium and phosphate ions from the body the variations in the body response to these proteins,
fluids will precipitate onto the microporous surface of could explain the observed intra-species differences. We
the material. The microporous surface acts as the have summarized our ideas in Fig. 8.
collection of nucleation sites for this precipitation. In conclusion, from the results of this study, we
The formed bone-like biological CaP layer could be a postulate that the undifferentiated inducible osteopro-
physico-chemical trigger for the cells to differentiate into genitor cells are triggered to start differentiating into the
the osteogenic lineage similar to being in a bony osteogenic lineage by the BMPs or other relevant
environment. However, there is another important proteins that are coprecipitated in the newly formed
observation from this study: the large variation in biological apatite layer, and that, consequently, this
osteoinductive potential of individual animals, in addi- biological apatite layer acts as the conductor of the bone
tion to the inter-species variation [14]. If the process of formed by the induced cells.
osteoinduction by biomaterials is only based on the
above-described dissolution–reprecipitation process of
calcium and phosphate ions, it would not be likely to 5. Conclusion
observe such a large difference between individual
animals. No reports describing large variations among This study shows that, in addition to macropores, a
individuals’ response to CaP biomaterials in an ortho- minimum amount of micropores within the macropore
topic context could be found. On the other hand, in walls is necessary for a material to be osteoinductive.
addition to the difference in the response to BMPs These micropores are probably required for acceleration
3574 P. Habibovic et al. / Biomaterials 26 (2005) 3565–3575

of the dissolution–reprecipitation process of CaPs on the [17] Yuan H, Yang Z, Zou P, Li Y, Zhang X. An investigation of the
material interface. The large variation in the amounts of osteoinduction of synthetic porous phase-pure hydroxyapatite
induced bone that is observed between the individual ceramic. Biomed Eng Appl Basis Commun 1997;9:274–8.
[18] Yuan H, Yang Z, Li Y, Zhang X, de Bruijn JD, de Groot .
animals suggests that some endogenous proteins such as Osteoinduction by calcium phosphate biomaterials. J Mater Sci:
BMPs might also play a role in the mechanism of Mater Med 1998;9:723–6.
osteoinduction by biomaterials. [19] Yuan H, Kurashina K, de Bruijn JD, Li Y, de Groot K, Zhang X.
A preliminary study on osteoinduction of two kinds calcium
phosphate ceramics. Biomaterials 1999;20:1799–806.
[20] Yuan H, de Bruijn JD, Li Y, Feng K, Yang Z, de Groot K, Zhang
Acknowledgements X. Bone formation induced by calcium phosphate ceramics in soft
tissue of dogs: a comparative study between a-TCP and b-TCP.
The authors would like to thank Dr. Maarten Terlou J Mater Sci: Mater Med 2001;12:7–13.
from the Image Analysis Department of the University [21] Yuan H, van den Doel M, Li S, van Blitterswijk CA, De Groot K,
Utrecht for developing the software used for the de Bruijn JD. A comparison of the osteoinductive potential of two
calcium phosphate ceramics implanted intramuscularly in goats.
histomorphometry and Dr. Paul Westers and Dr. Edwin J Mater Sci: Mater Med 2002;13:1271–5.
Martens from the Biostatistics Department for their [22] Gosain AK, Song L, Riordan P, Amarante MT, Nagy PG,
generous help with statistical analysis. Wilson CR, Toth JM, Ricci JL. A 1-year-study of osteoinduction
A part of this study was financially supported by the in hydroxyapatite-derived biomaterials in an adult sheep model:
EU ‘‘Intelliscaf’’ Project (G5RD-CT-2002-00697). Part I. Experimental 2002;109(2):619–30.
[23] Yuan H, de Bruijn JD, Zhang X, van Blitterswijk CA, de Groot
K. Osteoinduction by porous alumina ceramic, abstract book of
16th European conference on biomaterials, London 2001 p.209.
[24] Fujibayashi S, Neo M, Kim HM, Kokubo T, Nakamura T.
Osteoinduction of porous bioactive titanium metal. Biomaterials
[1] Friedenstein AY. Induction of bone tissue by transitional 2004;25(3):443–50.
epithelium. Clin Orthop 1968;59:21–37.
[25] Ruys AJ, Wei M, Sorrell CC, Dickson MR, Brandwood A,
[2] Urist MR. Bone: formation by autoinduction. Science 1965;
Milthorpe BK. Sintering effect on the strength of hydroxyapatite.
Biomaterials 1995;16:409–15.
[3] Reddi AH, Huggins CB. Biochemical sequences in the transfor-
[26] Kitsugi T, Yamamuro T, Nakamura T, Kokubo T, Tagagi M,
mation of normal fibroblasts in adolescent rats. Proc Nat Acad
Shibuya T, Takeuchi H, Ono M. Bonding behavior between two
Sci USA 1972;69:1601–5.
bioactive ceramics in vivo. J Biomed Mater Res 1987;21(9):
[4] Reddi AH. Cell biology and biochemistry of endochondral bone
development. Collagen Relat Res 1981;1:209–26.
[27] Wang FR. Experimental study of osteogenic activity of sintered
[5] Urist MR, DeLange RJ, Finerman GAM. Bone cell differentia-
hydroxyapatite-on the relationship of sintering temperature and
tion and growth factors. Science 1983;220:680–6.
pore size. Nippon Seikeigeka Gakkai Zasshi 1990;64(9):847–59.
[6] Cook SD, Rueger DC. Osteogenic protein-1. Clin Orthop 1996;
[28] Villarreal DR, Sogal A, Ong JL. Protein adsorption and
osteoblast response to different calcium phosphate surfaces.
[7] Riley EH, Lane JM, Urist MR, Lyons KM, Lieberman JR. Bone
morphogenetic protein-2. Clin Orthop 1996;324:39–46. J Oral Implantol 1998;24(2):67–73.
[8] Reddi AH. Bone Morphogenetic protein, bone marrow stromal [29] Kruyt MC, de Bruijn JD, Wilson CE, Oner FC, van Blitterswijk
cells and mesenchymal stem cells. Clin Orthop 1995;313:115–9. CA, Verbout AJ, Dhert WJA. Viable osteogenic cells are
[9] Winter GD, Simpson BJ. Heterotopic bone formation in a obligatory for tissue-engineered ectopic bone formation in goats.
synthetic sponge in the skin of young pigs. Nature 1969; Tissue Eng 2003;9(2):327–36.
223(201):88–90. [30] Kruyt MC, de Bruijn JD, Wilson CE, Yuan H, van Blitterswijk
[10] Yamasaki H. Heterotopic bone formation around porous CA, Verbout AJ, Oner FC, Dhert WJA. Optimization of bone
hydroxyapatite ceramics in the subcutis of dogs. Jpn J Oral Biol tissue engineering in goats: a peroperative seeding method using
1990;32:190–2. cryopreserved cells and localized bone formation in calcium
[11] Zhang X. A study of porous block HA ceramics and its phosphate scaffolds. Transplantation 2004;77(3):359–65.
osteogenisis. In: Ravaglioli A, Krajewski A, editors. Bioceramics [31] Hollander M, Wolfe DA. Nonparametric statistical inference.
and the human body. Amsterdam: Elsevier; 1991. p. 692–703. New York: Wiley; 1973. p. 139–146.
[12] Klein CPAT, de Groot K, Chen W, Li Y, Zhang X. Osseous [32] Mendenhall W. Nonparametric statistics: introduction to prob-
substance formation induced in porous calcium phosphate ability and statistics, 3rd ed. Belmont: Duxbury Press; 1971.
ceramics in soft tissues. Biomaterials 1994;15:31–4. p. 379–382.
[13] Yamasaki H, Saki H. Osteogenic response to porous hydro- [33] Urist MR, Strates BS. Bone morphogenetic protein. J Dent Res
xyapatite ceramics under the skin of dogs. Biomaterials 1992; 1971;50:1392–404.
13:308–12. [34] Yuan H, Yang Z, de Bruijn JD, Li Y, de Groot K, Zhang X.
[14] Ripamonti U. Osteoinduction in porous hydroxyapatite im- Material-dependent bone induction by calcium phosphate cera-
planted in heterotopic sites of different animal models. Biomater- mics: a 2.5-year study in dogs. Biomaterials 2001;22(19):
ials 1996;17:31–5. 2617–23.
[15] Ripamonti U. The morphogenisis of bone in replicas of porous [35] Toth JM, Lynch KL, Hackbarth DA. Ceramic-induced osteogen-
hydroxyapatite obtained from conversion of calcium carbonate esis following subcutaneous implantation of calcium phosphates.
exoskeletons of coral. J Bone Joint Surg 1991;73A:692–703. Bioceramics 1993(6):9–13.
[16] Vargervik K. Critical sites for new bone formation. In: Habal [36] Yuan H, Li Y, de Bruijn JD, de Groot K, Zhang X. Tissue
MB, Reddi AH, editors. Bone grafts and bone substitutes. response of calcium phosphate cement: a study in dogs.
Philadelphia: WB Saunders; 1992. p. 112–20. Biomaterials 2000;21:1283–90.
P. Habibovic et al. / Biomaterials 26 (2005) 3565–3575 3575

[37] Barrère F, van der Valk CM, Dalmeijer RAJ, Meijer G, van [46] Le Huec JC, Clement D, Brouillaud B, Barthe N, Dupuy B,
Blitterswijk CA, de Groot K, Layrolle P. Osteogenicity of Foliguet B, Basse-Cathalinat B. Evolution of the local calcium
octacalcium phosphate coatings applied on porous metallic content around irradiated b-tricalcium phosphate ceramic im-
implants. J. Biomed. Mater. Res. 2003;66A(4):779–88. plant: in vivo study in the rabbit. Biomaterials 1998;19:733–8.
[38] de Bruijn JD, Yuan H, Dekker R, Layrolle P, de Groot K, van [47] Miyamoto S, Takaoka K, Ono K. Bone induction in monkeys by
Blitterswijk CA. Osteoinductive biomimetic calcium-phosphate bone morphogenetic protein: a trans-filter technique. J Bone Joint
coatings and their potential use as tissue-engineering scaffolds. In: Surg 1993;75(1):107–10.
Davies JE, editor. Bone engineering. Toronto: University of [48] Marusic A, Katavic V, Grcevic D, Lukic IK. Genetic variability
Toronto Press; 2000. p. 421–31. of new bone induction in mice. Bone 1999;25:25–32.
[39] Habibovic P, van der Valk CM, Meijer G, van Blitterswijk CA, de [49] Geesink RGT, Hoefnagels NHM, Bulstra SK. Osteogenic activity
Groot K. Influence of octacalcium phosphate coating on of OP-1 bone morphogenetic protein (BMP-7) in a human fibular
osteoinductive properties of biomaterials. J Mater Sci: mater defect. J Bone Joint Surg (Br) 1999;81-B:710–8.
Med 2004;15:373–80.
[40] Habibovic P, Li JP, van der Valk CM, Meijer G, Layrolle P, van
Blitterswijk CA, de Groot K. biological performance of uncoated
and octacalcium phosphate-coated Ti6Al4V. Biomaterials 2004;
26(1):23–36. Further reading
[41] Yuan H, de Bruijn JD, Zhang X, van Blitterswijk CA, de Groot
K. Bone induction by porous glass ceramic made from bioglass [50] Heughebaert M, LeGeros RZ, Gineste M, Guilhem A, Bonel G.
(45S5). J Biomed Mater Res 2001;58(3):270–6. Physicochemical characterization of deposits associated with HA
[42] Ripamonti U, Crooks J, Kirkbride AN. Sintered porous ceramics implanted in nonosseous sites. J Biomed Mater Res:
hydroxyapatites with intrinsic osteoinductive activity: geometric Appl Biomater 1988;22(A3):257–68.
induction of bone formation. S Afr J Sci 1999;95:335–43. [51] Elliot JC. Studies in inorganic chemistry: structure and chemistry
[43] Geesink RG, de Groot K, Klein CP. Bonding of bone to apatite- of the apatites and other calcium orthophosphates. Amsterdam:
coated implants. Bone Joint Surg Br 1988;70B:17–22. Elsevier; 1994. p. 4.
[44] Hanawa T, Kamira Y, Yamamoto S, Kohgo T, Amemyia A, [52] Kohri M, Miki K, Waite DE, Nakajima H, Okabe T. In vitro
Ukai H, Murakami K, Asaoka K. Early bone formation around stability of biphasic calcium phosphate ceramics. Biomaterials
calcium-ion-implanted titanium insterted into rat tibia. J Biomed 1993;14(4):299–304.
Mater Res 1997;36:131–6. [53] Koerten HK, van der Meulen J, Verhoeven MCH. Degradation
[45] Kay JF, Cook SD. Biological profile of calcium phosphate of calcium phosphate ceramics: a comparative study on three
coatings. In: Geesink RGT, Manley MT, editors. Hydroxyapatite types of calcium phosphate. In: Ducheyne P, Kokubo T, van
coatings in orthopaedic surgery. New York: Raven Press Ltd.; Blitterswijk CA, editors. Bone-bonding materials. Leiderdorp:
1993. p. 89–106. Reed Healthcare Communications; 1992. p. 85–94.