Orthopaedic and dental surgeons are fully aware of the need for implants to bond well with the surrounding living bone if long-lasting clinical success is to be achieved. For example, well-bonded hip implants have a 10 year failure rate, which is lowered fivefold if bonding is poor or absent. The techniques that are currently available to impart implant surfaces with the desired osteoconductive properties are essentially limited. To overcome the inherent difficulties, we have developed a ‘biomimetic’ coating process. By this means, implants with complex surface geometries, such as porous spinal implants, can be furnished with a bone-bonding surface. Furthermore, these coatings can be rendered osteoinductive as well as osteoconductive (by incorporating osteogenic agents). Using this facility, we have induced bone formation at an ectopic site in rats, and have accelerated osseointegration (bone bonding) at an orthotopic dental site in adult miniature pigs. Our preliminary results indicated that these osteoinductive dental implants bond with surrounding bone within one week instead of the usual three weeks. We believe that surfaces coated with biomimetic coatings into which osteogenic growth factors are incorporated hold great potential for use in clinical orthopaedics and dentistry.
Current clinical research in the field of orthopaedic and dental implantology is focused on the development of novel tools and techniques to improve the regeneration of bone tissue. Titanium implants are widely used in orthopaedic surgery and dentistry, and numerous attempts have been made to improve their osteoconductivity. These measures include modifications in their surface properties and coating with a layer of calcium phosphate to act as an artificial bone substitute. The technique whereby such layers are produced has recently undergone a revolutionary change, which has had profound consequences for their potential to serve as drug-carrier systems. Formerly, calcium phosphate layers were deposited upon the surfaces of metal implants under highly unphysiological physical conditions, which precluded the incorporation of biologically active osteoinductive agents. These growth factors could only be adsorbed, superficially, upon preformed inorganic layers. But such superficially adsorbed molecules are released too rapidly within a biological milieu to be effective in their osteoinductive capacity. Now, it is possible to deposit calcium phosphate layers under physiological conditions of temperature and pH by the so-called biomimetic process, during which bioactive agents can be co-precipitated. Since these molecules are integrated into the inorganic latticework, they are released gradually in vivo as the layer undergoes degradation, which enhances the potential of these coatings to serve as a slow-release carrier system for the delivery of osteogenic agents to the implantation site.
2. Clinical background and requirements
Worldwide, more than one million patients need to be treated annually for skeletal affections. These problems occur in the fields of plastic and reconstructive surgery, orthopaedic surgery and dental implantology. Surgical handling includes the treatment of bony defects created during tumour surgery or caused by trauma, the reconstruction of congenital skeletal abnormalities, the promotion of fracture healing, spinal arthrodesis, and joint and tooth replacement (Langer & Lobkovsky 1999; Hutmacher et al. 2000). Unfortunately, treatment does not always solve the problem, due to inadequate local bone conditions and impaired bone healing. Complicated fractures may fail to heal, resulting in so-called delayed unions or non-unions. The treatment of bone tumours or congenital syndromes frequently involves the artificial creation of large bony defects, which need to be filled with autogenic or allogenic bone. Autogenic bone is of limited availability for grafting purposes and its excavation is associated with donor site morbidity. Hence, suitable and biocompatible substitutes for bone grafts have been actively sought (Maxson et al. 1990; Schepers et al. 1991). These bone substitutes can be divided into three classes: (i) osteoconductive, (ii) directly osteogenic and (iii) osteoinductive. By definition an osteoconductive material facilitates the spontaneous formation of bone. In other words, it furnishes a microenvironment that supports the ingrowth of blood vessels, perivascular tissue and osteoprogenitor cells into the site where it is deposited. The limitation of an osteoconductive material is that it needs to be surrounded by bone tissue on all sides. Moreover, osteoconductive bone substitutes, such as ceramic materials, do not actively stimulate the bone-formation process. Directly osteogenic and osteoinductive bone substitutes do actively stimulate bone-formation processes. These latter materials can be generated by combining a porous scaffold with either osteogenic cells (osteogenic bone substitutes) or osteoinductive growth factors (osteoinductive bone substitutes). Osteoinduction actively triggers the formation of bone. It promotes the recruitment of osteoprogenitor cells from an ectopic or an orthotopic site and stimulates their proliferation and differentiation into bone-forming cells, i.e. osteoblasts (within a time period of a few days to maximally two weeks). The various routes whereby these different types of bone substitute promote bone formation are illustrated in figure 1. Material scientists often confuse osteoconduction and osteoinduction processes with dystrophic calcification at an ectopic site. An example of this is ossifying myositis, which, following implantation of a foreign material, occurs at a later time than osteoinduction, i.e. after several weeks to months rather than after several days to weeks.
3. Titanium or titanium alloy implants
Metal-based implants or endoprostheses have been used for many decades in clinical dentistry and orthopaedic surgery. Titanium and its alloys are especially popular, due to their excellent mechanical properties and the ease with which they can be handled during surgery. Furthermore, they are highly biocompatible with the bony tissue compartment (Meffert 1997; Albrektsson & Sennerby 2000). In orthopaedics, titanium or titanium-based materials are utilized not only for prosthetic devices (Dearnley 1999) but also for internal fixation during fracture healing (Disegi 2000).
For a successful long-term performance, devices implanted in load-bearing regions must be firmly anchored within the host tissue. The advantage of porous materials in this respect is that they can become permeated with ingrowing bony tissue (Murray & Semple 1981; Maniatopoulos et al. 1986, 1988a,b). There exist two types of porous implants: one has a porous surface, whereas the other is porous throughout its bulk.
4. Porous titanium
During the past 30 years, porous metallic orthopaedic implants have been used for fixation purposes. In 1968, Hirschhorn was the first to report on the fabrication of a porous metal (a cobalt–chromium alloy) for use as an implant material (Hirschhorn et al. 1969). One year later, Lueck et al. reported on the commercial manufacture and implantation of a porous, pure titanium fibrous composite (Lueck et al. 1969). During the early 1970s, research studies relating to the production of porous metal coatings were published by Welsh, Galante, Pilliar and Cameron (Galante et al. 1971; Welsh et al. 1971; Cameron et al. 1972, 1978; Galante & Rostoker 1973; Pilliar et al. 1975, 1979). These materials were further developed by Bobyn et al. (1980a,b, 1982), and by others. These dual-structured implants were prepared using a particulate method (sintered metal powders or fibres), which yielded a porous surface around a solid-machined or cast metallic core. On the basis of clinical results and of histological evidence gleaned from retrieved implants, these porous implants were revealed to be biologically fixed by the ingrowth of bony or other tissue. The ingrowth of bone into the porous structure ensures a good transfer of skeletal forces. Porous Ti6Al4V systems are particularly popular due to their high corrosion resistance and good mechanical properties.
(a) Conventional methods for producing porous titanium
A number of processes have been applied to produce porous titanium and titanium alloy implants. Generally these different methods can be divided into three categories. (i) Porous coatings can be produced by sintering uniformly sized beads or fibres onto the substrate by isostatic pressing or loose packing (Liao & Zhang 1978; Pilliar 1998a,b). These methods are associated with numerous drawbacks. These include: (a) a low porosity (generally less than 50%), which limits the interfacial strength that can be achieved by the ingrowth of bone; (b) uncontrollable pore size, which is determined by the size of the particles; and (c) incomplete interconnections between the pores. (ii) Porous coatings can also be fabricated by mixing titanium powder with an organic spacer holder. This mixture is applied to the surface of the implant and becomes permanently affixed under controlled conditions of temperature and pressure. The organic material is subsequently removed, leaving a porous, metallic coating on the implant surface. The problem with this method is that it is difficult to control the interconnectivity of the pores, some of which remain closed. (iii) The third methodological category includes plasma spraying. Titanium powder is sprayed onto the surface of an implant under reduced pressure conditions in an inert gas chamber at a high temperature. This yields a coating with an irregularly shaped porous surface. The drawbacks of this method are the poor interconnectivity of the pores and a small pore size.
(b) A new technique to produce porous titanium
Recently, a novel type of porous Ti6Al4V implant was developed by our group using a positive replica method (Li & De Groot 2003). Briefly, a polyurethane foam is dipped into a Ti6Al4V slurry. The excess slurry is then removed by a press roller. The dipping–pressing process is repeated until the struts of the polyurethane foam are coated with the titanium alloy slurry. The superfluous slurry is again removed by press-rolling under pressure, which yields an evenly distributed coating on the foam. After drying, the samples are heated in an argon atmosphere to 500 °C to burn out the foam. Finally, the metal bodies are sintered in a vacuum furnace (10−5 mbar) at 1250 °C with a holding time of 2 h.
The advantages of fabricating porous Ti6Al4V by this method include a closely controlled size of both the pore diameters and the diameters of the pore interconnections, and a virtually complete interconnection of the uniformly spaced pores. The sample exhibits the basic structural characteristics of continuous open polyurethane foam, with a three-dimensional interpenetrating network of struts and pores (figure 2a). The porous structure is similar to that of human cancellous bone (figure 2b). The pore size, which is several hundred micrometres in diameter, and an interconnecting pore structure such as that shown in figure 2a, provide sufficient space for vessels (Joschek et al. 2000) and bone. Besides the interconnected macropore structure, micropores exist on the walls of the macropores (figure 3a). These small pores are formed by sintered powder and are interconnected, which is crucial for the circulation of body fluids supplying the necessary nutrients and mineral ions for biological functions. The micropores range from 1 to 13 μm in diameter. Several studies have shown that this microarchitecture plays an important role in bone formation and growth (Schwartz & Boyan 1994; Cerroni et al. 2002).
5. The coating of titanium implants with calcium phosphate
Although commercially manufactured porous metallic coatings yield good clinical results, bonding between the implant and bone is not always optimal, due to an inadequate porous structure. Hence, the implants ultimately tend to fail. The geometry of porous metals produced by coating beads and fibres is limited to simple spherical structures and wires, and the pore size cannot be precisely controlled. Pore obstruction and an uneven covering with calcium phosphate are associated with coated implants in the form of beads and fibres (Hayashi et al. 1994; Nakashima et al. 1997). Although pore obstruction does not occur on microtextured surfaces, such as grit-blasted or metal arc-sprayed ones, a firm anchorage cannot be achieved using such a microtextured surface. Interdigitating trabecular bone has a strong anchoring effect and acts as a buffer between the weight-bearing rigid central shaft of the prosthesis and the femur (Galante & Rostoker 1973). To meet the practical requirements of bone ingrowth and long-term clinical success, an implant should ideally mimic the architecture of natural bone, thereby maximizing contact with living osseous tissue, facilitating bone growth into the pore spaces, and providing a biological interlock between the implant and the surrounding bone.
The microtopographic profile of a metal implant surface is known to influence its osteoconductivity (Cochran et al. 1996; Brunette 1999; Brunette & Chehroudi 1999). Modifications in surface geometry have been effected by blasting with corundum or sand (Pilliar 1998a,b; Esenwein et al. 2001), or by etching with acid (Klokkevold et al. 1997). Such treatment generates small pits within the metal surface, which correspond in size to the resorption pits excavated by osteoclasts in bone. This pitted microtopographic profile is conducive to osteogenesis. It also enhances osseointegration, by facilitating interlocking between the implant substrate and ongrown bone, thereby improving the long-term mechanical stability of the prosthesis (de Groot et al. 1987; de Groot 1989; Wolfe & Cook 1994). Investigators realized that if an implant surface could be coated with a layer whose characteristics mimicked those of bone matrix, particularly the mineralized components, then its osteoconductive features could be still further enhanced. And such was found to be the case. Metal implants have been coated (by plasma spraying or other methodologies) with layers of hydroxyapatite (de Groot et al. 1987), calcium phosphate (de Groot 1989; Gondolph-Zink 1998) or mixtures of the two (Klein et al. 1994a–c). These coated implants are, moreover, characterized by a rough surface profile, which further improves osteoconduction and osseointegration.
6. The biomimetic coating procedure
Biomimetics is a relatively new field, which literally means the mimicry of biology. It is a branch of science in which biologists and engineers jointly endeavour to produce bioinspired materials that can be used for tissue engineering. We wish to develop an osteoinductive implant by incorporating osteogenic growth factors into calcium phosphate coatings deposited biomimetically upon titanium implants. The same principle would also apply to the incorporation of other drugs into such implants. By this means, we would be provided with implants that are osteoinductive (growth factors) as well as osteoconductive (calcium phosphate layer). A few years ago, attempts were made to coat metal implants with layers of calcium phosphate under more physiological or ‘biomimetic’ conditions of temperature and pH (de Groot 1989; Kokubo et al. 1990; Kokubo 1991; de Groot et al. 1998; Wen et al. 1999; Liu et al. 2001), primarily to improve their biocompatibility and biodegradability. The mineral layers generated by existing methods, being composed of large, partially molten hydroxyapatite particles, were not only prone to delamination but also poorly degraded in a biological milieu. An additional advantage of the biomimetic method is that biologically active molecules, such as osteogenic agents, can be co-precipitated with the inorganic components. As a consequence, the proteins are truly incorporated into the crystal latticework and not merely deposited upon its surface. In forming an integral part of the calcium phosphate coatings, the protein molecules are liberated not in a single burst (as when superficially adsorbed), but gradually, which bodes well for an enduring osteogenic effect at the implantation site.
The biomimetic coating technique involves the nucleation and growth of bone-like crystals (figure 4) upon a pretreated substrate by immersing this in a supersaturated solution of calcium phosphate under physiological conditions of temperature (37 °C) and pH (7.4). The method, originally developed by Kokubo (Kokubo et al. 1990), has since undergone improvement and refinement by several groups of investigators (Gondolph-Zink 1998; Barrere et al. 1999, 2001, 2002a,b; Liu et al. 2001; Habibovic et al. 2002). It is simple to perform, is cost-effective and can be applied even to heat-sensitive, non-conductive and porous materials of large dimensions and with complex surface geometries.
7. Biomimetic coating of porous titanium
In vivo testing of porous implants coated with a layer of calcium phosphate has revealed an early ingrowth of bone into pores and an increase in the strength of the attachment at six weeks. But at 32 weeks, no advantage over other systems was demonstrated. Most studies have indeed reported a positive effect on bone ingrowth and implant fixation strength in the short term (Spivak et al. 1990; Soballe et al. 1991a,b; Greis et al. 1992; Soballe 1993). This positive effect has been demonstrated not only for porous surface structures but also for grooved structures (Stephenson & Freeman 1991). In one study with a longer follow-up time, porous metallic implants coated with calcium phosphate enhanced the strength of the attachment to bone for as long as 1 year after implantation, and even a 1 mm gap between the implant and the host surface was bridged. One major problem with calcium phosphate coatings is that they partially obliterate the original metal surface geometry by obstructing the pores. When measures are taken to prevent this obstruction by using metal spheres, bonding with bone is compromised (Oonishi et al. 1989).
8. Incorporation of growth factors into biomimetic calcium phosphate coatings deposited on titanium implants
Titanium alloy implants bearing a fine, dense, amorphous layer of calcium phosphate (to promote crystal growth) were immersed in a supersaturated solution of calcium phosphate containing growth factors such as BMP-2 for 48 h under physiological conditions of temperature (37 °C) and pH (7.4). The implants became coated with a crystalline latticework of the inorganic components into which BMP-2 was incorporated. The osteogenic potency of BMP-2 thus incorporated was assessed by monitoring the alkaline phosphatase activity of rat bone-marrow stromal cells cultured directly upon the coated implants. BMP-2 incorporated into calcium phosphate coatings was more potent in stimulating the alkaline phosphatase activity of bone-marrow stromal cells cultured thereupon than was the freely suspended drug in stimulating that of the same cell population cultured on a plastic surface. Hence, the osteogenic potency of BMP-2 incorporated biomimetically into calcium phosphate layers is not only retained but enhanced in this situation, probably by virtue of a localized concentration effect. In consequence, this type of coating now assumes great potential value as an osteoinductive system in orthopaedic and dental implantology (Liu et al. 2001, 2003, 2004a, in press).
9. Bone formation at an ectopic site in rats
In this study, we investigated the potential of BMP-2 incorporated biomimetically into calcium phosphate coatings to induce and sustain bone formation at an ectopic (subcutaneous) site in rats over a course of five weeks. For this purpose, one experimental and three control groups were set up: titanium alloy implants bearing a co-precipitated layer of calcium phosphate and BMP-2 (experimental group); naked titanium alloy discs (negative control for the effects of a calcium phosphate layer and of BMP-2); titanium alloy discs bearing a layer of calcium phosphate (negative control for the effects of BMP-2); and titanium alloy discs bearing a layer of calcium phosphate and superficially adsorbed BMP-2 (positive control for superficially adsorbed BMP-2). Six implants per group were implanted subcutaneously in the dorsal region of rats, two per animal. Samples were retrieved at 7 day intervals over a period of five weeks for histological and histomorphometric analyses. Ectopic bone formation occurred only in the experimental group of animals. Bone tissue first became apparent two weeks after implantation and thereafter was deposited continually until the end of the five week follow-up period. Ossification occurred by a process of intramembranous (direct) growth, there being virtually no evidence of enchondral bone formation. The total volume of bone formed increased significantly (p<0.005) from 5.8 mm3 per implant after the second week to 10.2 mm3 per implant after the third, when the net daily rate of bone formation was highest (0.8 mm3 per implant per day). Bone tissue was deposited not only upon the implant surfaces but also at some distance therefrom, maximally at 340 μm by the end of the third week. This distance had decreased to 200 μm by the end of the fifth week, indicating that bone-remodelling processes were actively underway, which would be a normal occurrence under physiological conditions. The biomimetic coating was degraded gradually during the course of the five week follow-up period, by the end of which time one-third of its volume remained. The implication is that one-third of the initially incorporated amount of BMP-2 (1.7 μg per implant) likewise remained, and that bone formation could have continued for several more weeks after the end of the experimental period. The findings of this study demonstrate that BMP-2 incorporated biomimetically into calcium phosphate coatings is capable not only of inducing, at a low pharmacological level, the formation of bone at an ectopic site, but of sustaining osteogenic activity for a considerable period of time (Liu et al. 2004a, in press).
The grafting of autogenous osseous tissue has long been held as the ‘gold standard’ for enhanced bone formation at an injured site. Although the mechanisms involved are ill-defined, autogenous bone appears to combine osteoconductive and osteoinductive properties with the stimulation of an inflammatory response and an accompanying release of cytokines.
Tissue ingrowth into porous metallic orthopaedic and dental implants is commonly used as a means of achieving the long-term fixation of these prostheses. Although calcium phosphate coatings facilitate bone formation on the porous metallic implants, the degree of tissue ingrowth is often insufficient and unpredictable. If the pores of these implants were superficially coated with a layer of calcium phosphate into which an osteogenic agent was incorporated, then the ingrowth of bone tissue could be enhanced and implant osseointegration accelerated.
(a) Basic requirements for bone ingrowth into a porous structure
Most investigators deem bone ingrowth to be necessary for the biological bonding of porous implants. The principal requirements for bone ingrowth are as follows:
Biocompatibility of the implant materials: in the history of total joint replacement, various metallic, ceramic and polymeric implant materials have been tested. Titanium alloys are now preferred to stainless steel and cobalt-based alloys, due to their lower modulus, superior biocompatibility and enhanced corrosion resistance.
Cellular and matrix response: the adherence of cells to a biomaterial is a prerequisite for tissue integration. New bone formation on biomaterials depends on whether the surface structure promotes cell proliferation and production of an extracellular matrix. Using human bone-marrow cell cultures, Wilke et al. (1998) have demonstrated clear differences in cell proliferation, cell differentiation and the production of an extracellular matrix between hydroxyapatite, titanium and a CrMoCo alloy. Hydroxyapatite surfaces yielded the best results, followed by titanium and CrCoMb. Lowenberg et al. (1988) have reported that surface geometry can affect the attachment and orientation of cells in vitro, and they suggested that this parameter could be modified by the application of biological molecules.
(b) Implant stability/micromotion
Local contaminants within the surface that could invoke an undesirable inflammatory response should be avoided, and an initial degree of implant stability within the host bone must be achieved (Cameron et al. 1973; Pilliar et al. 1986). There is evidence that too much relative motion between the implant and the host bone leads to the ingrowth of fibrous connective tissue rather than bone. Some studies have suggested that relative movements greater than 50 μm at the implant–bone interface inhibit the osteogenic process in the peri-implant region (Burke et al. 1991). Metallic implants used in orthopaedics and dentistry should ideally combine great mechanical strength with biocompatibility, osteoconductivity and osteoinductivity. Our studies have furnished evidence that titanium alloy implants bearing a biomimetically co-precipitated layer of calcium phosphate and an osteogenic growth factor (BMP-2) do indeed combine these properties.
(c) Osteoinductive titanium implants
That calcium phosphate layers are biocompatible and osteoconductive is well known. Indeed, it was for these attributes that this type of coating was chosen to act as a carrier for BMP-2. Although BMP-2 is itself osteoinductive, its osteogenic potency is markedly enhanced after incorporation into biomimetic calcium phosphate coatings (Liu et al. 2004a, in press). This finding is accounted for by the drug's slow rate of release from these coatings over a protracted period of time. BMP-2 is liberated gradually because it forms an integral part of the inorganic latticework. BMP-2 deposited superficially upon the surfaces of preformed calcium phosphate coatings is released in a single rapid burst, and the drug, being highly water soluble, diffuses away from the implantation site too quickly to be able to exert an osteogenic effect (Liu et al. 2004a, in press). The forces governing protein adsorption onto the surface of a preformed calcium phosphate coating are believed to be chiefly electrostatic in nature (Becker & Becker 1993), but the mechanisms underlying the incorporation of protein molecules during their co-precipitation with calcium phosphate are more complex and poorly understood. Protein molecules possibly participate in the formation of the three-dimensional crystal latticework, or at least influence its growth. They may act as tied molecules, in which case they would be effectively bonded, or serve as seeds for crystallization, possibly influencing the number of nucleation sites generated (Coombes & Heckman 1992; Shirkhanzadeh & Liu 1994). Protein molecules may also be aggregated and penetrated by mineral ions during the formation of a coating. Although bovine serum albumin influences the structure of the calcium phosphate latticework into which it is incorporated, BMP-2 does not have this effect. Hence, the mechanism of incorporation may be different in each case. And in addition to these physical considerations, the release of BMP-2 in vivo will of course be influenced by cell-mediated coating degradation (Liu et al. 2004a, in press). The physical properties of biomimetic coatings are also modified by the presence of incorporated BMP-2. In vivo, proteins are known to strengthen mineralized tissue and to stabilize its mineral contents. And in our studies, the rate of release of calcium ions from biomimetic coatings was indeed slowed down by the presence of incorporated BMP-2, which may improve the coating's attachment to the underlying titanium alloy implant. That such is the case was demonstrated by performing the micro-scratch test, which revealed BMP-2-containing coatings to be mechanically more cohesive than those composed purely of calcium phosphate. BMP-2 incorporated into the body of the inorganic latticework possibly reinforces the coating by surrounding the brittle calcium phosphate crystals, in analogy to the situation occurring in dental enamel. This property of mechanical cohesiveness is important during the surgical implantation procedure, which generates high shear forces that could rupture the coatings. That the coated implants furnished a mechanically stable environment in vivo was evidenced by the finding that ectopic bone formation in the rat occurred by intramembranous growth. In previous studies involving this model, ossification has followed a predominantly enchondral course, which requires a less stable mechanical environment. Our studies with this model (Liu et al. 2004b) also yielded evidence of the high biocompatibility of BMP-2-containing coatings, in that bone-marrow tissue was sometimes observed to contact these directly, without intervening bone. Bone-marrow tissue contains immunocompetent cells, which are very sensitive to foreign material.
Dental implants bearing a biomimetic BMP-2-containing calcium phosphate coating osseointegrated more rapidly at an orthotopic site in adult miniature pigs than did uncoated ones (unpublished data). We hope that planned tests with porous implants will yield even better results.
Implants should be rapidly osseointegrated to expedite patient recovery and rehabilitation and to cut down on health-care costs. We have recently shown that BMP-2 can be incorporated in a dose-dependent manner into biomimetic calcium phosphate coatings deposited on titanium implants. The incorporated BMP-2 underwent gradual release (over a period of weeks) into the surrounding milieu wherein it retained its biological activity. At an ectopic site in rats, BMP-2 was released from implant coatings and induced the formation of bone tissue not only upon the implant surface but also within its immediate surroundings. When such bioactive implants were implanted at an orthotopic dental site in adult miniature pigs, they enhanced the osseointegration process. A porous titanium surface possesses advantages over a non-porous one, and we now wish to investigate the coating of such a surface with a biomimetic BMP-2-containing calcium phosphate layer. We are optimistic that this system will further expedite the implant osseointegration process.
The authors would like to thank Wyeth (Cambridge, Mass. USA) for providing them with rhBMP-2, Dr P. Layrolle and Dr J. de Bruijn for their general guidance, M. Stigter for his technical expertise, C. Van de Valk for her surgical assistance, and S. Nüssli and V. Gaschen for their help in the processing of tissue for histological evaluation. We would like to thank the ITI Foundation and the Swiss National Science Foundation for their financial support of our ongoing studies.
One contribution of 18 to a Discussion Meeting Issue ‘Engineered foams and porous materials’.
- © 2005 The Royal Society