Biological organisms have evolved to produce hierarchical three-dimensional structures with dimensions ranging from nanometres to metres. Replicating these complex living hierarchical structures for the purpose of repair or replacement of degenerating tissues is one of the great challenges of chemistry, physics, biology and materials science. This paper describes how the use of hierarchical porous materials in tissue engineering applications has the potential to shift treatments from tissue replacement to tissue regeneration. The criteria that a porous material must fulfil to be considered ideal for bone tissue engineering applications are listed. Bioactive glass foam scaffolds have the potential to fulfil all the criteria, as they have a hierarchical porous structure similar to that of trabecular bone, they can bond to bone and soft tissue and they release silicon and calcium ions that have been found to up-regulate seven families of genes in osteogenic cells. Their hierarchical structure can be tailored for the required rate of tissue bonding, resorption and delivery of dissolution products. This paper describes how the structure and properties of the scaffolds are being optimized with respect to cell response and that tissue culture techniques must be optimized to enable growth of new bone in vitro.
The average age of the UK population is increasing as life expectancy increases and birth rate decreases, with 20% of the population being over the age of 60. The life expectancy of men is now 78.5 years, while that of women is 82.5 years. Unfortunately, 60% of those over 60 are chronically ill. In the UK alone one in four people will die from respiratory disease, while three million people suffer from osteoporosis. Osteoporosis reduces bone density and affects everyone to some degree as they age. The density and strength of bones decrease because bone resorption occurs faster than new bone is produced. The disease eventually leads to collapse or fracture of bones, especially in the hip, wrist, knee and spine.
A common consequence of osteoporosis, arthritis and trauma is the need for skeletal replacements. Current surgical procedures for bone repair are transplantation or implantation.
The gold standard in reconstructive surgery is the autograft, which is the harvesting of the patient's tissue from a donor site and transplantation to the damaged site. Alternatives are homografts (transplantation from another patient) and xenografts (tissue from a different species, e.g. freeze-dried bovine bone). There are many limitations to using these techniques; autografts have low availability and can cause morbidity at the donor site. Homografts carry the risk of disease transmission, bone resorption and rejection, requiring indefinite administration of immunosuppressant drugs to the patient. Xenografts are in large supply but they have even larger risks of immune rejection, in situ degeneration and disease transmission (Jones & Hench 2001).
The treatment for advanced stage arthritis of the hip joint is a total joint replacement. However, all orthopaedic implants have a limited lifespan as they lack three of the most critical characteristics of living tissues: (i) the ability to self-repair; (ii) the ability to maintain a blood supply; and (iii) the ability to modify in response to stimuli such as mechanical load.
As life expectancy increases there is a growing need for an artificial alternative to an autograft. A paradigm shift from replacement to regeneration of tissues may provide a solution (Hench & Polak 2002).
The aim of regenerative medicine is to restore diseased or damaged tissue to its original state and function, reducing the need for transplants and joint replacements. There are two strategies to achieve this aim; tissue engineering and tissue regeneration. Both of these strategies use scaffolds to guide and stimulate growth and differentiation of cells and form tissues (Langer & Vacanti 1993; Davies 2000).
2. Bone tissue engineering
An ideal strategy for the tissue engineering of bone is the harvesting of osteogenic cells from the patient, which are then expanded in culture and seeded on a scaffold that acts as guide and stimulus for tissue growth in three dimensions (Ohgushi & Caplan 1999; Takezawa 2003). The osteogenic cells lay down bone extracellular matrix in the shape of the scaffold as woven (immature) bone. The tissue engineered construct can then be implanted into the patient. Over time, the synthetic scaffold should resorb into the body as non-toxic degradation products, allowing the bone to remodel itself into mature bone structure.
(a) The hierarchical structure of bone
To be able to grow new bone it is important to understand its structure. Figure 1 shows the hierarchical structure of bone. Bone is a natural composite of collagen (polymer) and bone mineral (ceramic). Collagen is a triple helix of protein chains, a complex structure that has high tensile and flexural strength and provides a framework for the bone. The helical chains are of the order of 10 nm in length and they are arranged into orientated collagen fibres that are of the order of 500 nm in length. Bone mineral is a crystalline calcium phosphate ceramic (carbonated hydroxyapatite, HCA) that provides the stiffness and high compressive strength of bone. The mechanism of osteogenesis is complex, but simplistically the extracellular matrix of mineralizable collagen is laid down by osteoblasts (osteogenic cells), which develop (differentiate) from stem cells and are 20–50 μm in diameter. They secrete type I collagen which then mineralizes to form an HCA-collagen structure. An osteoblast that becomes surrounded by concentric rings of mineralized tissue is called an osteocyte (figure 1). The two most important types of bone are cortical and cancellous bone. Cortical bone is a dense structure with high mechanical strength and is also known as compact bone. Cancellous or trabecular bone, also called spongy bone, is an internal porous supporting structure of a network of struts (trabeculae) enclosing large voids (macropores). These two types of bone then grow together to form a bone, which in the case of a femur is approximately 0.5–1 m long.
Bone is remodelled in response to its local loading environment by the body. Osteoclasts are cells that resorb old bone and bone that is not required (i.e. not under any load), while osteoblasts lay down new bone. Osteoporosis occurs as osteoblasts become less active but bone is still removed by osteoclasts. The struts of the trabecular bone are most affected by osteoporosis. An aim of regenerative medicine is to stimulate the body to reactivate osteogenic cells to re-create the natural three-dimensional architecture of bone.
When minor damage is done to a bone, it can repair itself by the biochemical activity of the osteoblasts, called osteogenesis. However, if the defect exceeds a critical diameter or volume, the bone cannot repair itself. Such defects can result from trauma or from the removal of diseased tissue. Graft implants (transplants) or synthetic bone filler materials are currently used to repair critical size bone defects. The use of a regenerative scaffold would guide and stimulate bone growth. A tissue engineered construct could also be used to fill critical sized defects.
3. An ideal scaffold
The general criteria for an ideal scaffold for bone tissue engineering are that it:
acts as template for in vitro and eventually in vivo bone growth in three dimensions;
resorbs at the same rate as the bone is repaired, with degradation products that are non-toxic and that can be easily be excreted by the body;
is biocompatible (not toxic) and promotes cell adhesion and activity, stimulating osteogenesis at the genetic level;
bonds to the host bone without the formation of scar tissue, creating a stable interface;
exhibits mechanical properties matching that of the host bone after in vitro tissue culture;
is made from a processing technique that can produce irregular shapes to match that of the bone defect; and
has the potential to be commercially producible and sterilizable to the required international standards for clinical use.
Each of these criteria will be discussed in this paper. In summary, a material is required that has properties either very similar to trabecular bone or that can be used to stimulate new bone growth and create a biocomposite that has structure and properties similar to trabecular bone.
To fulfil criterion (i), the scaffold must have an open porous structure to allow cell penetration, tissue ingrowth and eventually vascularization on implantation. There has been much debate regarding the minimum interconnected pore diameter to achieve this, and 100 μm is recognized to be the minimum interconnected pore aperture diameter for an in vivo scaffold as it has been found to encourage the vascularization that is required for complete regeneration of bone (Okii et al. 2001). The criteria for an optimized pore network for in vitro bone growth are less clear, especially if the scaffold resorbs in vitro and the structure changes before implantation. There have been few investigations of pore orientation, how the pores link to each other to form channels or how fluid flows through the pores to provide a preferential route of cell migration. This paper will show how these issues can be addressed.
To achieve criterion (ii), resorbable porous polymeric scaffolds have been developed. Bone cells may initially attach to polymers in vitro, especially if attachment specific proteins are incorporated on their surface. However, polymers do not bond to bone and do not stimulate cells at the genetic level. Commonly used polymers, such as polyglycolic acid, have the Young modulus much lower than bone and many biodegradable polymers degrade rapidly, reducing the strength of the scaffolds before tissue can regenerate. They, therefore, do not fulfil criteria (iii), (iv) and (v) and there is also concern over the acidic degradation products of the biodegradable polymer scaffolds (Holy et al. 2003). Bioactive materials have the potential to fulfil criteria (ii) and (iii). When implanted into the body, bioactive materials stimulate a biological response from the body such as bonding to tissue (Hench & Wilson 1993). There are two classes of bioactive material; class B bioactive materials bond to hard tissue (bone) and stimulate bone growth along the surface of the bioactive material (osteoconduction). Examples of class B bioactive materials are synthetic hydroxyapatite (HA), tri-calcium phosphate ceramics (β-TCP) and HA-coated porous titanium oxide (titania). Class A bioactive materials not only bond to bone and are osteoconductive but they are also osteoproductive, i.e. they stimulate the growth of new bone on the material away from the bone/implant interface and can bond to soft tissues such as gingival (gum) and cartilage. Examples of class A bioactive materials are bioactive glasses.
4. Bioactive glasses
Bioactive glasses are based on a random network of silica tetrahedra containing Si–O–Si bonds. The network can be modified by the addition of network modifiers such as Ca, Na and P, which are bonded to the network via non-bridging oxygen bonds. The mechanism of bone bonding to bioactive glasses is due to the formation of a carbonate substituted hydroxycarbonate apatite layer (HCA) on the surface of the materials after immersion in body fluid. This layer is similar to the apatite layer in bone and, therefore, a strong bond can form (Hench 1991). Bioactive glasses are osteoproductive, which means they stimulate new bone growth on their surface, even away from the glass/bone interface (osteoproduction) and can be resorbable (Hench & Polak 2002).
Importantly for criterion (iii), the dissolution products of bioactive glasses (soluble silicon and calcium) have been found to up-regulate seven families of genes in osteoblasts (Xynos et al. 2000a–,c) and to have an effect on the cell cycle whereby more cells express biochemical markers for new bone formation and cells that are not capable of forming new bone are eliminated by apoptosis (programmed cell death; Hench et al. 2000). Synthetic HA does not release such dissolution products. However, chemical substitution of silicon for calcium in synthetic HA shows improved bone ingrowth in vivo over phase pure HA granules (Patel et al. 2002).
There are two types of bioactive glasses; melt-derived and sol-gel derived. A certain composition of melt-derived bioactive glass (46.1 mol% SiO2, 24.4 mol% Na2O, 26.9 mol% CaO and 2.6 mol% P2O5), called Bioglass is used in the clinic as a treatment for periodontal disease (Perioglas) and as a bone filling material (Novabone; Hench 1991; Fetner et al. 1994). Bioglass implants have also been used to replace damaged middle ear bones, restoring the hearing to thousands of patients and as tooth root replacements (Wilson et al. 1995).
Sol-gel derived bioactive glasses are synthesized by the hydrolysis of alkoxide precursors to form a sol, which is a colloidal silica solution. The sol then undergoes polycondensation to form a silica network (gel). The gel is then heat treated to form a glass (Brinker & Scherer 1990; Hench & West 1990). Sol-gel derived bioactive glasses tend to be more bioactive and resorb quicker than melt-derived glasses of similar compositions. This is because sol-gel glasses have a nanometre scale textural porosity that is inherent to the sol-gel process, which increases the specific surface area by two orders of magnitude compared to a melt-derived glass of a similar composition (Sepulveda et al. 2002a). The textural porosity not only increases the surface area for cation exchange and network dissolution by two orders of magnitude, but it also exposes many silanol groups to the solution, which act as nucleation sites for HCA layer formation.
(a) Bioactive glass scaffolds
Pores have been introduced into melt-derived bioactive glasses but the pores were few in number and were in the form of orientated channels of irregular diameter running through the glass so interconnectivity was poor (Yuan 2001). By foaming sol-gel derived bioactive glasses, our team at Imperial College London, produced scaffolds with a hierarchical pore structure similar to trabecular bone (Sepulveda et al. 2002b).
Figure 2 shows a flow chart of the sol-gel foaming process. In the first step a sol is synthesized from a silica-based alkoxide precursor, such as tetraethyloxysilane (TEOS).
After hydrolysis is completed, the sol is foamed by vigorous agitation in air. The viscosity increases rapidly due to the addition of a gelling agent (hydrofluoric acid, HF) and a surfactant is added, which lowers the surface tension and stabilizes the air bubbles on initial foaming (Rosen 1989). The bubbles are permanently stabilized by the gelation reaction (polycondensation). Tertiary (SiO2, CaO, P2O5), binary (SiO2, CaO) and unary systems (SiO2) can all be successfully foamed as scaffolds (Sepulveda et al. 2002b).
(b) Hierarchical structure of bioactive glass scaffolds
Figure 3 shows a scanning electron micrograph (SEM) of a bioactive glass foam of the 70S30C (70 mol% SiO2, 30 mol% CaO) composition. Figure 3 shows that the foam is comprised of large macropores with diameters in the region of 200–600 μm that are highly interconnected (dark areas). Many of the apertures have diameters in excess of the 100 μm required for tissue engineering applications. As the foam is made from sol-gel derived bioactive glass, the three-dimensional interconnected solid network has a textural porosity with diameters in the range 2–20 nm, termed mesoporosity.
Figure 4 shows an X-ray micro-computed tomography (XMT) image of a similar scaffold, to that shown in figure 3, obtained using a commercial XMT unit (Phoenix X-ray Systems and Services GmbH). The XMT unit is based on the same principles as a CAT scan (computed axial tomography) where series of two-dimensional transmission X-ray images are reconstructed to form a three-dimensional image. The key difference is that geometric enlargement is used to magnify the image by placing the object close to a micron sized spot source, producing a magnified image which is projected onto a solid-state detector a large distance from the object (relative to the source–object distance). Like a CAT scan, it provides quantitative data on the integrated density and atomic number of the matter in each voxel (volume pixel). Reconstructed images consisting of 512×512×512 voxels, each of 4.7 μm on a side, were collected from each sample set and cropped digitally to remove the edge artefacts. Figure 4 shows that the macropore network is very highly interconnected and is very similar to the XMT image of trabecular bone shown in Stock (1999). The pore shape and size appear to be very homogeneous. This is because the pores form in a liquid that is well mixed with an evenly dispersed surfactant content. In figure 5, an isolated pore that is representative of the entire sample, which is connected to other pores on 4–6 sides. Connectivity occurs because the spherical air bubbles are all in contact with each other immediately prior to gelation, separated only by a thin film of silica-based sol that is stabilized by the surfactant. Upon gelation and subsequent thermal processes the thin film drains, shrinkage occurs and the surfactant is combusted, leaving the apertures.
(c) Tailoring of the structure
We have found that variables in each stage of the foaming process (figure 2) have an effect on the pore structure. Such variables include the sol (glass) composition and surfactant concentration (Jones & Hench 2003), gelling agent concentration, the temperature at which the process is carried out and whether additional water is added with the surfactant to improve its efficiency (Jones et al. 2004a). For specific applications it may be necessary to select a particular pore diameter and interconnected pore size. For tissue engineering applications the macropore diameter has little importance, but the modal interconnected pore diameter should be greater than 100 μm. Changing the surfactant concentration while keeping all other variables constant is the most efficient method to control the aperture diameter (Jones & Hench 2003).
Figure 6 shows the dissolution profiles of silicon, calcium and phosphate ions from powders, foams and monolithic discs of bioactive glasses of the 58S (60 mol% SiO2, 36 mol% CaO, 4 mol% P2O5) composition. The same mass of each glass was immersed in simulated body fluid (SBF) and the ions released were quantified by inductive coupled plasma analysis (ICP). Figure 6 shows that the release of the gene activating ions Si and Ca was highest for powders and lowest for monoliths. This is because dissolution is more rapid as the ratio of surface area to solution volume ratio increases. The amount of phosphate in solution decreases because of formation of a calcium phosphate (HCA) layer on the surface of the glass. The rate of formation of this layer increased as the dissolution rate increased. The morphology of a potential scaffold will, therefore, have a large effect on rate of delivery and concentration of gene stimulating ions and therefore rate of bone bonding and regeneration of bone.
None of the processing variables listed above have a large effect on the textural mesoporosity. However, this can be controlled by changing the final sintering temperature of the scaffolds (Jones et al. 2004b). When calcium nitrate is used in the sol-gel process to introduce calcium into the glass composition, the residual nitrates must be removed to chemically stabilize the glass and make it biocompatible (non-toxic to cells). Nitrates are burnt off at approximately 550 °C, therefore 600 °C is the minimum sintering temperature for these glasses. Figure 7 shows textural pore size distributions obtained from foams sintered at 600, 700, 800 and 1000 °C for 2 h, using the Barrett Joyney Halenda (BJH) method on data obtained from nitrogen adsorption analysis (Barrett et al. 1951). The vertical axis is a derivative of the volume of nitrogen desorbed from the foam relative to the pore diameter.
Figure 7 shows that scaffolds sintered at 600 and 700 °C exhibit narrow pore size distributions with narrow pore diameters and a modal pore diameter of approximately 17 nm. The shape of the pores is difficult to ascertain, but the isotherms (plots of volume of nitrogen adsorbed and desorbed as a function of pressure at constant temperature, 77 K) were type IV isotherms according to IUPAC classification (Sing), with type II hysteresis loops, which implies ink-bottle shaped pores in the mesopore range (Saravanapavan & Hench 2003). However, this shape of isotherm would also be obtained for cylindrical pores that contain bulbous portions. Figure 6 shows that as the sintering temperature increased from 700 to 800 °C, the modal pore diameter dropped to approximately 12 nm and the textural porosity appeared to have been removed after sintering at 1000 °C. The decrease in textural porosity coincides with an increase in compressive strength. Foams sintered at 600 °C have a compressive strength of approximately 0.25 MPa (Instron parallel plate, diameter to height ratio of foam discs was 3 : 1) while similar foams sintered at 800 and 1000 °C have a compressive strength of approximately 2.5 MPa (Jones et al. 2004b), similar to that of trabecular bone (Hench & Wilson 1993). Figure 8 shows XMT images of sections of scaffolds sintered at 700, 800 and 900 °C. The images show that the macropore diameter also decreased with sintering, but it is important to quantify how the interconnected pore diameter was affected. However, obtaining quantitative data from XMT images has required developing new interpretive software (Atwood et al. 2004).
5. Quantitative three-dimensional image analysis
Although the XMT images provide quantitative data of the integrated density and composition of the scaffolds at evenly spaced points in three dimensions, converting these pictures from greyscale images to quantified descriptors of the structures requires the development and use of appropriate mathematical morphological operators. As an example, a single two-dimensional slice from a three-dimensional scan of a 70S30C scaffold is shown in figure 9a. The apertures which provide the interconnectivity between the pores (black) at areas in the scaffold walls (whitish) are visible; however, quantifying the size of these apertures requires classification of the image into individual pores. This is simple if the pores are closed, but for this open cell porosity a new algorithm was developed (as described in detail in Atwood et al. 2004):
threshold the image, classifying each voxel as either scaffold or empty space;
apply a dilation algorithm to grow from the scaffold walls into the centre of the pores, noting the number of steps it has taken to grow to each voxel. Centroids of each pore will fill last;
using the centroids, together with the number of steps grown as a distance map, a three-dimensional watershed algorithm (Mangan & Whitaker 1999) was applied to divide the image into individual pores. (Watershed algorithms find the set of points in a function, considered as a height map, that divide regions in which water flows to the same final point; analogous to the watersheds of a river basin in geography.);
voxels with neighbours on the same two pores were then grouped and defined as apertures; and
the individual pores and apertures objects were then quantified to determine their volume/area and maximum diameter.
Using this algorithm, the scaffolds shown in figure 8 were quantified and the pore and aperture size distributions are plotted in figure 10. The pore size decreases with increasing firing temperature; however, the median interpore aperture diameter does not change significantly.
Therefore, after sintering at 800 °C for 2 h, the scaffold has a modal interconnected pore diameter in excess of 100 μm and a maximum compressive strength of 2.4 MPa.
During in vitro cell culture, the flow of culture medium containing cells during cell seeding is critical to developing an evenly populated scaffold. The flow in porous media has been shown to be described at a macroscopic level by the generalized tensor form of Darcy's law, allowing the bulk velocity to be related to the change in pressure using what is called the permeability tensor, . Therefore, provides a quantitative descriptor of the ease at which seeded cells will penetrate the scaffold, as well as the ease of getting nutrient fluid into the scaffold during tissue growth. Measuring the permeability can be difficult since flow will occur around the edge of the sample as well as through it, and preference flow channels may form as material dissolves under the flow rates required for measurement. However, an alternative method for measuring the permeability exists; using the three-dimensional geometry of the scaffold obtained via XMT in a microscale flow simulation. The flow within porous medium obeys Stokes equations at the local scale (Sahimi 1995), hence the permeability can be calculated using the geometry and by numerically solving Stokes equations (Papathanasiou & Lee 1997). For this study the permeability was calculated using the code developed by Prof. Bernard (CNRS Bordeaux) to study water flow in reservoir rocks (Anguy et al. 1995).
The resulting flow predictions on a local scale within the complex three-dimensional structure are shown in figure 11a. The plotted streaklines illustrate that the flow is dominated by the size of the apertures. The calculated macroscopic components of the permeability tensor are plotted in figure 11b as a function of the size of the representative volume element (RVE) used in the simulation. For a RVE of the size of the pores (350 μm), the x, y and z components of are very different. However, once the RVE is larger then three pores across (1 mm), the different components converge. This illustrates that the 70S30C composition scaffolds fired at 800 °C have an isotropic structure and that only a volume of approximately 1 mm3 need be simulated to determine the flow characteristics to design the seeding and nutrient flow techniques.
7. Mechanical properties
Measuring the mechanical properties of the scaffolds as a function of the scaffold pore structure and soaking time is laborious, and it cannot provide a tool for extrapolation to design an optimal scaffold structure. An alternative is to extract the internal structure of the scaffolds as surfaces from the XMT images and mesh the volumes created by these surfaces. This was performed using scaffolds of the 70S30C composition fired 800 °C, with the resulting mesh shown in figure 12a. Using bulk properties for the scaffold material from the literature (Amaral et al. 2002), the resulting structure was compressed by displacing the nodes along the top face downwards whilst fixing the nodes on the bottom face. The simulated load versus displacement graph was then converted into an effective stiffness for the porous structure, predicting an effective modulus of 3.8 GPa, in reasonable agreement with a modulus of 3.2 GPa measured on a similar sample. Although only a single scaffold was analysed, it illustrates the viability of implementing such a methodology which can be used to design scaffold materials (the bulk stiffness can be tuned by either altering the composition or the mesoporosity) and structures which exhibit mechanical properties matching that of the host bone, at least after in vitro tissue culture.
8. Molecular level characterization
A collaboration with the teams of Professors Mark Smith (University of Warwick) and Bob Newport (University of Kent) has allowed the characterization of the glass network structure at the atomic level using magic angle spinning (MAS), nuclear magnetic resonance (NMR) and X-ray diffraction. Solid-state NMR is an element specific probe technique with high sensitivity to the local structural environment around the nucleus under investigation (MacKenzie & Smith 2002). 43Ca MAS NMR spectra have been obtained for unstabilized samples of sol-gel derived calcium silicates (70S30C) heated at 120 and 350 °C, with the spectra suggesting that at this stage the calcium remained in an environment similar to the initial calcium nitrate (Lin et al. 2004). As the temperature was increased the samples became more disordered and no calcium signal was observed.
Extended X-ray absorption fine structure spectroscopy (EXAFS) and X-ray absorption near edge structure (XANES), X-ray fluorescence spectroscopy (XFS) and X-ray powder diffraction (XRD) were also used to study the local calcium environment in sol-gel-derived bioactive calcium silicate glasses (Skipper et al. 2004). The calcium oxygen environment was found to be six-coordinate across a range of binary compositions. The formation of the HCA layer on the 70S30C composition in SBF was also investigated. Both the EXAFS and XANES showed a gradual increase in coordination number and Ca–O bond distance as immersion time in SBF increased. XFS showed that calcium was quickly lost from the glass on exposure to SBF. XRD showed that the formation of the crystalline HCA layer was preceded by formation of a non-crystalline calcium phosphate phase after 1 h of immersion in SBF.
These techniques provide the basis for relating the mechanisms of dissolution and bioactivity to the molecular structure of the porous bioactive glass scaffolds. The results show that it is important to be able to monitor the molecular structure of the scaffolds as well as the hierarchical pore structure in order to optimize the scaffolds from the molecular to macro scale.
9. Cell response
Primary human osteoblasts, harvested from the tops of femurs removed during total hip replacements, have been cultured on bioactive glass foams of both the 58S and 70S30C compositions. Gough et al. (2004) seeded the cells (passage two or three) on to foams of the 58S composition at seeding density of 80 000 cells cm−3. Prior to culture the foams were mounted in agar, sterilized with UV light and soaked in culture medium for 3 days. During cell culture the media was changed every 2 days. The cells attached, proliferated and formed secreted bone extracellular matrix (mainly collagen type I), which mineralized after 10 days of culture. Mineralization is the development of HCA, bone mineral. Figure 13 shows an SEM image of primary human osteoblasts cultured on a 58S foam for 10 days. The seeded scaffolds were fixed, dehydrated and gold coated and observed in an SEM at 15 kV. Figure 13 shows an SEM image of a mineralized bone nodule inside a macropore. A bone nodule is a group of cells that have laid down some extracellular bone matrix. Mineralized bone nodules can form on many biocompatible materials in vitro, if supplementary growth factors, such as dexamethasone, are added to the medium. When cultured on the bioactive glass scaffold, the nodules mineralized without the addition of mineralization supplements to the culture media, which indicates great potential of the bioactive glass foam for use as an osseous tissue scaffold. The release of calcium and silicon ions from the glass are thought to stimulate the rapid mineralization.
The role of phosphate in the glass scaffold composition has much less effect. When cells were cultured on 70S30C scaffolds, mineralized bone nodules were observed after two weeks of culture without supplements. This result implies that combinations of silicon and calcium ions released from the scaffold stimulate the cells.
The effect of the surface texture of porous silica (100S) on the proliferation of lung cells was investigated by culturing cells from the A549 (human lung carcinoma) cell line on sol-gel derived monoliths with different mean pore diameters in the range 25–740 Å (2.5–74 nm). A cell seeding density of 20 000 cells cm−2 was used. After 48 h in culture the cells were fixed with paraformaldehyde. A primary antibody for vinculin conjugated to the fluorochrome Fluorescein was used for immunoflorescence staining. Positively stained cells were counted from five randomly selected fields of view at 10×10 magnification using a florescence microscope. Figure 14 shows a graph of relative metabolic activity as a function of mean pore diameter of silica discs, where the relative metabolic activity is the percentage number of cells positively stained for vinculin within a field of view in the florescence microscope compared to the number on the control. The control is dense fused silica glass (pore diameter of zero). Figure 14 shows that the proliferation of the lung cells increased as mean pore diameter of the substrate increased, up to maximum at a mean pore diameter of 75 Å (7.5 nm). It is not clear why 7.5 nm is the optimum pore diameter for lung cell attachment and growth, but the results show that optimizing the textural porosity of sol-gel derived bioactive glass scaffolds is important for optimal cell response. The optimal mesoporosity is likely to be correlated with the biochemistry of the cell membrane cytokines that are involved in cell attachment.
10. A hybrid scaffold for in situ bone regeneration
An optimized bioactive glass foam has high potential to be used as a scaffold to grow a bone/scaffold biocomposite with tissue engineering techniques; however, it would have too low a strength in tension to be used as a scaffold for in situ bone regeneration. In such an application a scaffold (with or without cells seeded on it) would be implanted directly into a defect site without culturing new tissue on the scaffold first. In this case the mechanical properties of the scaffold are highly important. First, the majority of defect sites in bone are load-bearing sites. Secondly, in this case the Young modulus of the scaffold should match that of the host bone to prevent stress shielding. It is therefore necessary to mimic the hierarchical structure of trabecular bone as a whole, rather than just the bone mineral, i.e. a polymer must be introduced in the ceramic scaffold structure to mimic the collagen/mineral composite of bone, but the interconnected pore network and bioactivity of the scaffold must be maintained.
The foaming process was modified, to create bioactive glass/polymer hybrid scaffolds, by reacting poly(vinyl alcohol) (Acros Organics, USA, average molecular weight of 16 000) in acidic solution with tetraethylorthosilicate. The inorganic phase was also modified by incorporating a calcium compound. Hydrated calcium chloride was used as precursor. The polymer/sol was then processed in the same way as in figure 2, except that the gelled foam hybrids were aged at 40 °C and vacuum dried at 40 °C. Mechanical behaviour of the hybrid materials produced was determined by compression test (parallel plate method) using dynamic mechanical analysis (DMA) equipment. From SEM images, the pore network of the hybrid foams was very similar to that of the glass foams.
Hybrid foams of the glass composition 70S30C containing 20 wt% PVA were compression tested using DMA parallel plate compression. Sample dimensions were 5×5×5 mm at a loading rate of 0.5 N min−1. The hybrid foams exhibited higher compressive strength and higher deformation to failure than the glass foams with similar porosity, therefore hydrid glass/polymer foams have the potential to have mechanical properties suitable for implantation into load bearing defect sites.
A bioactive glass scaffold of the 70S30C composition has the potential to serve as a scaffold for bone tissue engineering applications. X-ray microcomputer tomography can be used in conjunction with three-dimensional image analysis to quantify the macropore network and to non-destructively predict fluid flow within the scaffold and its mechanical properties. A compressive strength of 2.4 MPa can be attained by sintering the macroporous scaffold at 800 °C for 2 h and the modal interconnected pore diameter remains to be in excess of 100 μm. The techniques of NMR, EXAFS, XANES, XFS and XRD provide the basis for relating the mechanisms of dissolution and bioactivity to the molecular structure of the glass scaffolds. Optimization of the scaffolds from the molecular to the macro scale with respect to cell response is vital if an ideal scaffold is to be developed. When primary human osteoblasts are cultured on the bioactive glass foam scaffolds mineralized bone nodules form within 10 days of culture without the addition of supplementary growth factors. Bioactive glass/polymer hybrid scaffolds have the potential to fulfil all the criteria for an ideal scaffold for both bone tissue engineering and in situ bone regeneration.
The authors thank US Defence Advanced Research Projects (Contract no. N66001-C-8041), EPSRC, MRC, Lloyds Tercetenary Foundation and the Royal Academy of Engineering for financial support. The authors would also like to thank Dr Robert Atwood (3D image analysis), Dr Dominique Bernard (permeability model), Dr Daan Maijer (mechanical property model), Olga Tsigkou, Papy Embanga and Dr Molly Stevens (all cell biology) for their valuable assistance.
One contribution of 18 to a Discussion Meeting Issue ‘Engineered foams and porous materials’.
- © 2005 The Royal Society