Biomaterials that enhance bone regeneration following spinal fusion and at other osseous sites are in increasing demand. The challenge of repairing or regenerating defects with good-quality bone can be a difficult one, and the results are often unpredictable. Recently, research interest has focused on the development of synthetic bone-graft substitutes. In addition to osteoconductive, osteogenic, and osteoinductive capabilities, the ideal properties of such a material include consistency, limited immunogenicity, low risk of disease transmission, accessibility, and cost-effectiveness1. The mineral in bone is a nonstoichiometric hydroxyapatite in which various ions including F–, Cl–, SiO44–, CO32–, Na+, Mg2+, K+, Mn2+, Sr2+, Zn2+, Ba2+, Cu2+, and Al3+ have been substituted2.
The involvement of some trace minerals in bone metabolism has been studied in animals3. The importance of silicon (Si) has been demonstrated in chickens and rats, which developed abnormalities in bone mineralization and in articular cartilage and connective tissue development when they were placed on a silicon-deficient diet4-6. The importance of low levels of silicon in stimulating bone metabolism in humans has also been suggested by the association of increased dietary silicon intake with increased bone mass7. Silicon may be involved in an early stage of bone calcification, as localization of silicon in active growth areas in young bone has been observed in mice and rats4. In vivo studies of bone graft substitutes have indicated that the incorporation of comparable levels of silicate ions into a calcium phosphate scaffold can enhance osteogenesis8-11.
The aim of the current study was to investigate the effect of silicate substitution on osteoinduction by comparing two bone substitute materials of identical morphology. Our hypotheses were (1) that an osteoinductive response in the absence of any exogenously applied growth factors (e.g., a bone morphogenetic protein [BMP]) could be achieved at an ectopic site with use of either stoichiometric calcium phosphate or silicate-substituted calcium phosphate and (2) that silicate-substituted calcium phosphate would result in greater osteoinduction than stoichiometric calcium phosphate would.
Osteoinductivity was studied with use of an ovine model of ectopic bone formation. Calcium phosphate-based implants were inserted into isolated regions of the left and right paraspinal muscles of six skeletally mature female commercially crossbred sheep (weight, 65 to 80 kg; age, two to five years). Each sheep received a stoichiometric calcium phosphate implant (CaP) on one randomly chosen side and a silicate-substituted calcium phosphate implant (SiCaP-30) on the contralateral side. All procedures were carried out in compliance with U.K. Home Office regulations as stated in the Animal Scientific Procedures Act (1986).
The SiCaP-30 contained 2.6 wt% silicate (0.8 wt% Si). Both the SiCaP-30 and the CaP granules were 1 to 2 mm in size and had a total intragranular porosity (macroporosity) of 80% and a strut porosity of 30%. Strut porosity is defined as the porosity at the micrometer level within the struts of the material. The typical diameter of the macropores was 100 to 200 μm. The micropores within the struts were <50 μm in diameter and had a typical diameter of 1 to 10 μm. The implantation of the material into the defect sites was performed in a manner similar to that in which calcium phosphate has been used clinically12. However, neither the intergranular porosity nor the packing density between granules was formally measured. Both materials were manufactured to possess a high degree of interconnectivity between the macropores and the strut micropores. The porosity within each type of granule was measured with use of a mercury-intrusion porosimetry technique, and the size and distribution of the pores were compared with use of scanning electron microscopy. No significant differences in morphology between the stoichiometric and the silicate-substituted material were found.
Sintered granules of stoichiometric CaP and of SiCaP-30 were prepared by ApaTech (Elstree, United Kingdom). The CaP was prepared by means of an aqueous precipitation reaction between calcium hydroxide (Ca[OH]2) and orthophosphoric acid (H3PO4) as previously described13. The SiCaP-30 was prepared by the addition of silicon tetraacetate (Si[CH3CO2]4) to calcium hydroxide and orthophosphoric acid as previously described14 (see Appendix). Porous granules of each material were produced by means of a slip foaming process15. X-ray diffraction analysis verified the phase purity of each material. X-ray fluorescence confirmed that the intended level of silicate substitution had been achieved. Image analysis of polished samples, combined with analysis of total porosity and density with use of Archimedes’ principle, demonstrated that the stoichiometric and silicate-substituted granules had similar levels of open strut porosity and of macroporosity (Table I). The only significant difference was the presence of Si in the silicate-substituted implants.
Surgery
With the animal in a prone position, an incision approximately 3 cm in length was made over the right sacrospinalis muscle, approximately 3 cm lateral to the spine and just caudal to the shoulders. The muscle was exposed and split by blunt dissection. A syringe was used to measure and insert 7 mL of either stoichiometric CaP granules or SiCaP-30 granules into a plastic cylinder (40 mm in length and 15 mm in diameter) that contained a removable stainless-steel wire mesh. Five milliliters of venous blood was drawn from the jugular vein and poured onto the granules. Granules were allowed to coagulate within the mesh and inside the tube. (Coagulation with blood was necessary to ensure that granules did not migrate from the implant site.) After clotting had occurred, the implant was removed from the tube and placed within the split muscle bed. The wound was then closed. An incision was then made on the contralateral side, and the procedure was repeated with the other material.
Animals were allowed immediate postoperative mobilization as tolerated. Antibiotic and analgesic prophylaxis consisting of subcutaneous injections of 5 mg/kg of enrofloxacin (Baytril; Bayer, Leverkusen, Germany) and 2 mg/45 kg of flunixin meglumine (Finadyne; Schering-Plough, Hertfordshire, United Kingdom) was administered daily for three days following surgery. Animals were kept in individual pens for one week postoperatively and then in group housing. Implants remained in vivo for twelve weeks.
Histological Analysis
After twelve weeks in vivo, the implant and surrounding tissue were removed and immediately placed in a 4% paraformaldehyde solution for three days before being prepared for undecalcified histology. After dehydration in alcohol and defatting in chloroform, specimens were embedded in hard-grade acrylic resin (LR White; London Resin Company, Reading, United Kingdom). A longitudinal thin section (∼60 μm thick) through the center of each implant was prepared with use of grinding and polishing equipment (EXAKT, Norderstedt, Germany). The thin section was stained with toluidine blue (for soft tissue) and Paragon stain (for bone).
Seven images spanning the thin section in a longitudinal direction (i.e., imaging the center of the implant from end to end) were made. Each image was captured with use of a 5× objective lens and analyzed with use of image analysis software (AxioVision 4.5; Carl Zeiss, Jena, Germany). The line intersection method was used to quantify the percentage of bone within the implant, of bone in contact with the exterior of the implant, and of implant material in the image. A 10 × 12-mm grid of lines was placed over the image. The type of material (bone in the interior of the implant, bone in contact with the exterior of the implant, soft tissue, or implant material) present at each of the 225 intersection points of the grid lines was determined. The mean proportion of each material was then calculated in the set of seven images of each implant, and statistical comparisons were performed with use of these mean values.
Scanning Electron Microscopy
The thin sections were sputter-coated with a layer of gold and palladium and viewed with use of backscattered scanning electron microscopy (JSM-35C; Jeol, Welwyn Garden City, United Kingdom). Elemental maps of the ceramic and surrounding tissue were generated with use of energy-dispersive x-ray analysis.
Statistical Analysis
Analysis of the data was performed with use of SPSS software (version 10.1; SPSS, Chicago, Illinois). Since the Kolmogorov-Smirnov test showed that all variables were nonparametric, the results for the two implant materials were compared with use of the Mann-Whitney U test. A p value of <0.05 was considered significant. Results are reported as the mean and the standard error.
Source of Funding
ApaTech, Ltd. UK supplied the graft materials and half of the animals investigated in this study. All remaining costs, including histological costs and the remaining animals, were funded by the John Scales Centre for Biomedical Engineering. One of the authors is an employee of ApaTech.
All animals in the study healed and recovered well postoperatively. All of the retrieved implants were surrounded by well-vascularized muscle tissue, with no obvious evidence of an inflammatory reaction or encapsulation within fibrous tissue. Significantly more bone had formed within and between the granules in the SiCaP-30 group (7.65% ± 3.2% of the cross-sectional area) than in the stoichiometric CaP group (0.99% ± 0.9%, p = 0.01). Bone growth was evident within all six of the retrieved SiCaP-30 implants compared with only two of the six CaP implants. The amount of bone on the implant surface was significantly greater in the SiCaP-30 group (26.00% ± 7.8%) than in the CaP group (2.2% ± 2.0%, p = 0.01). Implant resorption was significantly greater in the SiCaP-30 group (with 27.5% ± 0.7% of the imaged area representing implant material) than in the CaP group (33.5% ± 1.4%, p < 0.05).
Qualitative comparison of all of the specimens in each group with use of backscattered scanning electron microscopy revealed greater bone formation adjacent to the silicate-containing SiCaP-30 biomaterial than adjacent to the CaP scaffold (Fig. 1), and this finding was confirmed by quantitative histological examination. Material with the characteristics of bone was observed in direct contact with the implant surface and within strut pores <5 μm in diameter in both groups (Fig. 2). The material within the struts stained in a manner similar to bone on the histological sections. Energy-dispersive x-ray analysis showed that silicon was present in the newly formed bone within the micropores of the SiCaP-30 implants and that the silicon concentration within the bone was higher than that in the adjacent soft tissue (Fig. 3).
Little osteogenic activity was apparent in the cells on the surface of most of the CaP implant material. In the two specimens in which bone had formed, localized regions of large, densely stained osteoblasts and osteoid deposition adjacent to the implant surface were evident. In many of the areas where osteogenesis was noted, bone formation was occurring by an intramembranous process, with no evidence of endochondral ossification. In other regions, multinucleated giant cells were seen in contact with the implant surface. The soft tissue was fibrous in nature in some regions close to the implant, but good vascularization was present throughout the implant structure and within interconnecting pores. The soft tissue was noteworthy for its lack of cellular activity. The CaP material had fragmented in a few discrete locations, but this was not necessarily associated with increased cellular activity. Soft tissue within the macropores in several regions in each sample in the CaP group had thickened and formed areas of “mesenchymal condensation” incorporating osteoblast-like cells. This may have been an indication of imminent calcium phosphate precipitation and the formation of woven bone (Fig. 4).
In both groups, bone formed preferentially on the concave surfaces within the macropores of the implant structure. In all cases, bone was initially deposited directly on the collagen fibers through an intramembranous ossification mechanism (Fig. 5). In most cases, bone formation occurred by the formation of an osteoid seam, which led to the formation of lamellar bone. Within smaller macropores, bone formed in a series of concentric layers, often resulting in a lumen containing a blood vessel surrounded by lamellar bone.
No evidence of cartilage formation was seen in any of the areas of osteogenesis. Bridging of new bone between granules was observed in some areas in both the SiCaP-30 and the CaP group. Bone formation was evident within the micropores and macropores throughout each CaP or SiCaP-30 implant. However, there appeared to be substantially more bone within the strut pores of the SiCaP-30 samples compared with the CaP samples. In both groups, the greatest amount of bone formed within the strut pores, with a smaller amount forming within the macropores and the least amount between the granules. In several localized regions within both the CaP and the SiCaP-30 implants, osteoblasts were seen actively depositing osteoid within macropores, with the bone being laid down centripetally (Fig. 6). In all specimens in both groups, mineralization was associated with osteoblastic activity wherever osteocytes were seen within newly formed bone. De novo mineralization without cellular activity was rarely seen. In the SiCaP-30 group, giant cells were present in localized regions on the implant surface. In other areas, the surface of the SiCaP-30 appeared scalloped, indicating osteoclast activity. The SiCaP-30 material had fragmented in some areas, and many macrophages and osteoclast-like cells in these areas were observed to be filled with SiCaP-30 fragments (Fig. 7). The soft tissue within both types of implants was generally well vascularized.
The aim of the current study was to determine the effect of silicate substitution on the osteoinductive potential of calcium phosphate in the absence of any exogenously applied growth factors. The results for silicate-substituted calcium phosphate were compared with those for morphologically matched stoichiometric calcium phosphate. Habibovic et al. previously suggested that osteoinductive ceramics perform better in critical-sized bone defects than similar non-osteoinductive ceramics do16. More recently, Habibovic et al. also reported that the amount of bone formation induced by such a material at an ectopic site could be predictive of the extent of osteoinduction of orthotopic bone17. These two studies suggest the value of investigating a potential bone substitute material at an ectopic site, as the results may be predictive of the response at an osseous site.
The number of animals used in the present study was based on a prospective power calculation of the number of specimens in each group that would be needed to detect a 20% difference in the mean with a power of 0.8. Although our study demonstrated that the amount of bone formation differed significantly between the two groups, it remains to be seen whether this will translate into a clinically relevant difference in humans.
Osteoinduction refers to the process in which “primitive, undifferentiated and pluripotent cells are somehow stimulated to develop into the bone-forming cell lineage.”18 The ability of calcium phosphate materials to induce bone formation at an ectopic location in vivo has been widely reported during the past decade, but the mechanism of induction remains unclear. It is currently unknown whether this is an effect of the biomaterial itself or a consequence of a reaction between the biomaterial and appropriate proteins. It has been suggested that implant chemistry and geometry are critical factors in the osteoinductivity of such materials19,20. The presence of macropores and/or concavities is a known prerequisite for osteoinduction within scaffolds in the absence of exogenous cytokines such as BMPs. The presence of a well-interconnected macroporous structure also plays an important role in allowing the development of a vascular supply and the convection of nutrients throughout the implant construct20-22. The total porosity, strut porosity, macroporosity, and pore size distribution of the two materials in the present study were all matched, thus leaving the substitution of silicate as the only variable. Bone ingrowth occurred in the voids, which were often <5 μm in size, within the struts in both materials. Staining confirmed the presence of collagen at these locations, and energy-dispersive x-ray analysis confirmed the presence of calcium and phosphate. Osteoblasts are too large to migrate within these small struts, and therefore remote bone formation may be associated with osteoid deposition onto the scaffold surface followed by infiltration within the porous scaffold prior to its calcification.
Both CaP and SiCaP-30 had an osteoinductive effect during the twelve-week period, as demonstrated by the formation of measurable quantities of bone in both groups. However, the substitution of silicate significantly increased the induction of bone at this ectopic site. The bone formation in every sample occurred only at the implant surface or within the construct itself; there was no evidence of heterotopic bone formation in adjacent muscle. Light microscopy also showed that the granules of both bone-graft substitute materials formed a scaffold that became knitted together by well-organized mature bone in some locations.
Although both BMPs and biomaterials induce ectopic bone formation, it is unknown whether they do so by the same process. BMP-triggered osteoinduction is generally endochondral in origin23, with cartilage formed as a precursor to mineralized bone. However, in some situations intramembranous ossification can also occur21. In contrast, the present study and many others have shown that in the absence of BMPs, bone formation within biomaterials is always intramembranous rather than endochondral21. An analysis of collagen type, which was not performed in this study, could have more objectively confirmed the type of ossification that occurred within the SiCaP-30 and CaP scaffolds. At the twelve-week time point, osteogenesis was still occurring both within the macropores and adjacent to the scaffold surface, but none of the bone graft substitutes showed evidence of cartilage matrix deposition. Although bone formation by endochondral ossification at an earlier time point could not be ruled out, the many localized areas of osteogenesis were all intramembranous in origin.
Angiogenesis must precede formation of new bone. This implies that adequate vascularization of the central region and periphery of the implant material must have taken place in both groups to allow the observed ectopic bone formation throughout the scaffold. Qualitative observation with use of light microscopy showed good vascularization in both types of implants, with blood vessels evident within both the strut pores and the macropores. However, no obvious differences between the two groups with respect to the amount of vascularization that had occurred were apparent on qualitative examination. A study by Hing et al. reported increased and more advanced cellular organization and neovascularization adjacent to the pores of silicate-substituted calcium phosphate scaffolds compared with pure calcium phosphate of the same porosity one week after implantation into bone24. Those investigations suggest that the incorporation of silicate ions within the lattice may have promoted these differences by influencing surface charge and wettability24 parameters, which have been demonstrated to be highly sensitive to silicate addition25.
The limitations of our study include those associated with use of an animal rather than human model and the fact that the implants were not under any imposed load while in situ. Further work is also needed to investigate bone formation within these implants over shorter and longer periods of time in order to ascertain the precise mechanism of bone formation.
In conclusion, this ovine model study has shown that both stoichiometric calcium phosphate and silicate-substituted calcium phosphate materials result in osteoinduction at an ectopic site and that silicate substitution enhances the bone formation. The use of a silicate-substituted calcium phosphate material instead of stoichiometric calcium phosphate ceramic during orthopaedic surgery may substantially augment repair and regeneration of bone.