Carbonate-substituted hydroxyapatite (CHA) is more osteoconductive and more resorbable than hydroxyapatite (HA), but the underlying mode of its action is unclear. We hypothesised that increased resorption of the ceramic by osteoclasts might subsequently upregulate osteoblasts by a coupling mechanism, and sought to test this in a large animal model.

Defects were created in both the lateral femoral condyles of 12 adult sheep. Six were implanted with CHA granules bilaterally, and six with HA. Six of the animals in each group received the bisphosphonate zoledronate (0.05 mg/kg), which inhibits the function of osteoclasts, intra-operatively.

After six weeks bony ingrowth was greater in the CHA implants than in HA, but not in the animals given zoledronate. Functional osteoclasts are necessary for the enhanced osteoconduction seen in CHA compared with HA.

Hydroxyapatite (HA) is familiar to orthopaedic surgeons as a biocompatible material which forms a ‘bond’ with host bone and is used as a coating for metallic implants and as a substitute bone graft.1 However, the rate at which it forms such a bond is relatively slow,2 and it can display limited integration in vivo.3 There is a need to improve the performance of synthetic materials for skeletal reconstruction.

Bone mineral itself is a calcium phosphate apatite containing multiple ionic substitutions, among which carbonate ions are the most abundant at 2% to 8% by weight.4,5 This has prompted investigation into the use of synthetic carbonate-substituted hydroxyapatite (CHA) as a skeletal biomaterial, on the basis that a ceramic which more closely resembles the naturally occurring mineral phase of bone may demonstrate improved osteoconduction. Several studies have described improved bone formation in CHA implants compared with HA controls,6,7 as well as enhanced bioresorption by osteoclasts of the carbonate-substituted material.8,9

These studies show that responses to small changes in the physicochemical qualities of a biomaterial can be quite marked,2,10,11 although little is known about the underlying mechanisms. The future development of bio-materials for skeletal applications depends on understanding these interactions. We hypothesised that the increased bone formation seen in response to carbonate substitution might be due to the enhanced osteoclastic activity, analogous to the process of coupling seen in bone remodelling. We have shown in vitro that human osteoblast-like cells respond to ceramic surfaces that have previously been resorbed by osteoclasts by increasing their production of collagen, and that this occurs earlier with CHA than with HA.12 The purpose of this study was to test this hypothesis by implanting HA and CHA in a large animal model, in which the long-acting bisphosphonate zoledronate was used to inhibit osteoclast function in half the animals in each group.

Materials and Methods

HA was synthesised by an aqueous precipitation reaction between calcium hydroxide and orthophosphoric acid, according to the method of Akao, Aoki and Kato.13 CHA was produced as described by Gibson and Bonfield.14 The quantities of the above reagents were calculated such that the Calcium/Phosphate (Ca/P) ratio was 1.78, and carbon dioxide (CO2) gas was used as the source of carbonate. This results in a single-phase material substituted with carbonate alone. The unsintered materials were processed into sintered microporous granules (1 mm to 2 mm in diameter) for implantation by ApaTech Limited (London, United Kingdom). A moist CO2 atmosphere was used to sinter the CHA material in order to preserve phase purity and diminish carbonate loss.14

The materials were characterised by X-ray diffraction, X-ray fluorescence spectroscopy (CERAM Research Limited, Stoke-on-Trent, United Kingdom), and carbon-hydrogen-nitrogen elemental analysis (Medac Limited, Egham, United Kingdom). Microporosity was quantified by image analysis using a Zeiss Axioskop optical microscope linked to a KS300 image analyser (Carl Zeiss Limited, Welwyn Garden City, United Kingdom). The materials were sterilised by gamma irradiation prior to implantation.

All procedures were carried out in accordance with the regulations laid down in the Animals (Scientific Procedures) Act 1986. A total of 12 Texcel × Continental ewes aged two to three years, weighing approximately 70 kg to 80 kg, were used for the implantation procedures. Under general anaesthesia, a 5 cm incision was made over the lateral femoral condyle and the vastus lateralis was elevated to expose the bone. A cylindrical defect 9 mm in diameter and 9 mm deep was created in the middle of the condyle using custom-made drill bits. The defect was washed with sterile 0.9% saline and then filled with granules using a custom-made applicator and an aseptic technique.

HA or CHA granules were placed in six animals each, with the same implant used on each side. Three animals in each group also received intravenous zoledronate at a dose of 0.05 mg/kg by slow intravenous infusion over 15 minutes, after induction of general anaesthesia, but before the surgery had begun.

The vastus lateralis muscle was replaced over the defect and the wound closed in layers using absorbable sutures. The wound was infiltrated with 10 ml 0.5% bupivacaine at closure for analgesia. No dressing was used. All the animals were allowed immediate full weight-bearing and unrestricted activity.

In order to quantify dynamic measures of bone formation, all animals were administered intravenous fluorochromes in the post-operative period, following cannulation of the cephalic vein. Calcein green (Sigma-Aldrich, Poole, United Kingdom) (15 mg/kg) was given 21 days and alizarin red (Sigma-Aldrich) (40 mg/kg) 14 days before the animals were killed by an intravenous overdose of barbiturate at six weeks. The distal femora were then excised and the soft-tissues stripped. A high-powered band saw was used to remove the bone containing the implants, which was immediately immersed in fixative (4% formaldehyde in 0.1M HEPES buffer, pH 7.4) prior to transportation to the laboratory for processing.

In order to quantify the baseline measurements of bone turnover and to assess the systemic effects of the intravenous zoledronate, full-thickness bicortical transiliac biopsies of 5 mm were also harvested unilaterally from the iliac crest of each animal using a trephine. These specimens were also immersed in fixative as above.

After fixation for 24 hours, the specimens were dehydrated in an alcohol series under vacuum and defatted in acetone. They were then embedded in methylmethacrylate. Thin (10 μm) sections were cut from the centre of the specimens. Serial sections were either stained with 1% toluidine blue (Sigma-Aldrich) pH 4.5 in sodium acetate buffer, or left unstained for fluorochrome analysis. Serial sections approximately 250 μ m thick were also cut from the centre of the implants, the resin dissolved in acetone, and the sections rehydrated in an alcohol series. They were then stained for tartrate-resistant acid phosphatase using 1 mmol naphthol AS-BI phosphate (Sigma-Aldrich; Poole, United Kingdom) as substrate in the presence of 10 mmol sodium tartrate (Sigma-Aldrich) at pH 4.5. The product was reacted with the post-coupler mixture, consisting of 0.1 mmol acetate buffer containing 2.2 mmol Fast Garnet GBC (Sigma-Aldrich) at pH 6.2.15

Digital composite images of the sections of the implant stained with toluidine blue were analysed using Scion Image analysis software (Scion Corporation, Frederick, Maryland). Bony ingrowth within the specimens was quantified. Unstained sections were used to measure the rate of bone formation (BFR). Digital images captured using fluorescence microscopy with appropriate filters were analysed using Summasketch software for histomorphometric analysis (Summagraphics Corp., Fairfield, Connecticut). The mineral apposition rate (MAR) and the extent of the mineralised surface (MS) were measured, and the rate of bone formation calculated according to the formula BFR = MAR (MS/BS), where MS = (dLS + sLS/2), where dLS = double labelled surface and sLS = single labelled surface, and where BS = bone surface.16 Tartrate-resistant acid phosphatase-stained images were analysed semi-quantitatively by a subjectively scored scale from 1 to 5, where 1 is occasional and 5, a very dense population of tartrate-resistant acid phosphatase-positive cells. Two observers (SP and RB) scored the sections independently and were blinded to the material and to the treatment with or without zoledronate. The weighted κ score between the observers was 0.55, indicating moderate agreement.

Bicortical transiliac biopsies with toluidine blue were analysed using Summasketch software for a static measure of bone resorption, and the BFR of the eroded surface16 was also measured using unstained sections from the transcortical biopsies, as outlined above.

Statistical analysis was carried out using the software package SPSS 10.0 (SPSS Inc., Chicago, Illinois). According to the distribution and the homogeneity of the variables, the following tests were used: one-way analysis of variance (ANOVA) for the comparison of four or more groups, with post hoc testing using Bonferroni’s correction, and either Student’s two-tailed t-test for independent variables or the Mann-Whitney U test for comparisons between two groups. The significance level was set at p ≤ 0.05.


Synthesis resulted in two single-phase ceramics after sintering with no evidence of phase decomposition (Fig. 1). X-rayfluorescence spectroscopy showed that the levels of ionic impurities in the samples were very low (< 0.03 wt%). These data were used to calculate the calcium/phosphorus ratio of the two materials. The carbonate contents after sintering were derived from carbon-hydrogen-nitrogen elemental analysis. The results are shown in Table I. The microporosities of the two ceramics were comparable (28.5% (sd 2.4) for HA vs 29.7% (sd 1.8) for CHA; p = 0.43)).

View this table:
Table I.

Carbonate contents of the green and sintered ceramics (carbon-hydrogen-nitrogen elemental analysis), and calcium/phosphorus molar ratio (X-ray fluorescence analysis) of the sintered ceramics (CHA, carbonate-substituted hydroxyapatite; HA, hydroxyapatite)

Fig. 1

X-ray diffractograms of the sintered ceramics.

The extent of the eroded surfaces in the transiliac biopsies from animals given zoledronate was limited to occasional individual pits. The mean eroded surface was 4.6% (sd 0.98) for the control animals and 0.48% (sd 0.46) for those given zoledronate (p < 0.01). The indices of bone formation are summarised in Table II.

View this table:
Table II.

Indices of bone formation in the transiliac ceramics

Within the implants, bony ingrowth was higher in control CHA samples than in HA samples. The administration of zoledronate was associated with significantly reduced bony ingrowth in the CHA samples but not in the HA samples (Fig. 2). The rate of bone formation was greater in the control CHA samples than in the control HA specimens, although this did not reach statistical significance. However, the administration of zoledronate was associated with a significant reduction in the rate of bone formation in the CHA samples but not in the HA samples (Fig. 3).

Fig. 2

Graph showing bone ingrowth at six weeks as a percentage of the total area of the implant for the four experimental conditions. Values are mean sd (n = 6). * indicates p < 0.05 (CHA, carbonate-substituted hydroxyapatite; HA, hydroxyapatite).

Fig. 3

Graph showing bone formation rate (BFR/BS) during the fourth postoperative week for the four experimental conditions. Values are mean sd (n = 6). * indicates p < 0.05 (CHA, carbonate-substituted hydroxyapatite; HA, hydroxyapatite).

Sections stained for tartrate-resistant acid phosphatase showed positive multinucleated cells within the new bone and in large numbers adjacent to the granules. The number of tartrate-resistant acid phosphatase-positive cells within the implants was variable. The cells were generally clustered in the central area of some implants, with a more scanty distribution in the periphery. In the implants containing CHA the density of the tartrate-resistant acid phosphatase-positive cells was particularly marked, with cells forming in large numbers and in extensive ‘colonies’. In places these had the appearances of cutting cones (tunnel in cortical bone caused by osteoclast resorption), but with many more tartrate-resistant acid phosphatase-positive cells than are usually seen during normal remodelling. Occasional normal cutting cones were seen in the bone adjacent to the implants, except in those animals that had received zoledronate. An example is shown in Figure 4a. Figure 4b shows a typical colony of tartrate-resistant acid phosphatase-positive cells within an implant at the same magnification for comparison.

Fig. 4a, Fig. 4b

Photomicrographs of thick undecalcified sections stained for tartrate-resistant acid phosphatase (TRAP). Note the difference between a) the scanty TRAP-positive cells in a normal cutting cone in bone around the implant and b) the large numbers adjacent to the granules (scale bar = 1 mm).

It was difficult to discern the areas of the granules that had been subjected to lacunar osteoclastic resorption owing to the relatively rough surface topography. However, some areas showed possible sites of resorption on the CHA ceramic which were not seen on HA (Fig. 5). They appeared as large pits, 100 μm to 200 μm across, and were associated with colonies of tartrate-resistant acid phosphatase-positive cells. Semi-quantitative scores for the number of tartrate-resistant acid phosphatase-positive cells within the implants were significantly higher on CHA than on HA (p < 0.01). Although the scores for animals who had received zoledronate were slightly lower than those of controls, this was not statistically significant (p = 0.525).

Fig. 5a, Fig. 5b

Photomicrographs of thick undecalcified sections of implants containing granules of hydroxyapatite and carbon-substituted hydroxyapatite stained for tartrate-resistant acid phosphatase (TRAP). Note colonies of TRAP-positive multinucleated cells, with possible areas of resorption seen in b) (white chevrons) (scale bars = 100 μm)


Hydroxyapatite ceramics are considered to be ‘osteoconductive’ materials, and previous studies in vivo have shown that this property can be enhanced by the use of carbonate substitution.68 The results of this study using microporous materials also show that at six weeks more bone had grown into the CHA implants than into the HA specimens. Although this is a useful property in a clinical product, it describes an ‘end-result’ rather than indicating how carbonate substitution brings about its effects.

The results of this study suggest that the advantage conferred on HA by carbonate substitution in terms of bone ingrowth is lost with the administration of zoledronate, a powerful anti-resorptive agent. This implies that the enhanced bone formation depends on the presence of functional osteoclasts. The results of the analysis of the transiliac biopsies suggest that zoledronate has had the expected effect of reducing osteoclast activity within the skeleton as a whole, and of reducing the rate of bone formation, presumably as a result of the arrest of remodelling. Similar results have been demonstrated in sheep.17 As the rate of bone formation in the CHA implants, but not in the HA, is also reduced by the administration of zoledronate, this suggests that similarities exist between peri-implant bone growth on CHA and the process of remodelling.

Bone ingrowth in this model appears to commence at the periphery of the defect, from the sides and from the base, and is laid down sequentially in a centripetal fashion. The presence of tartrate-resistant acid phosphatase-positive multinucleated cells in the central part of the implants suggests that they may be on the surface of the granules prior to the arrival of the vanguard of migratory osteogenic cells which are destined to commence bone formation. The early appearance of osteoclasts during bone healing, and their importance in coupling over and above their resorptive role, has recently been described in a sheep model.18 If this is the time sequence of events, then the tartrate-resistant acid phosphatase-positive multinucleated cells would be in a position to influence subsequent bone formation.

Carbonate substitution appears to increase the genesis of tartrate-resistant acid phosphatase-positive multinucleated cells on the ceramic surface in vivo, as we have previously noted using in vitro studies of the development of osteoclasts from human mononuclear precursors.19 Although the histomorphometric analysis of the transiliac biopsies and the assessment of the bone surrounding the implants showed that the administration of zoledronate reduced both the numbers of osteoclasts and their resorption in bone, the number of tartrate-resistant acid phosphatase-positive cells within the implants themselves, particularly in the central regions, appeared to be affected to a lesser extent by the administration of zoledronate. It is possible that the tartrate-resistant acid phosphatase-positive multi-nucleated cells are osteoclasts which have formed on the ceramic surfaces and are able to survive despite being unable to resorb the surface. There is recent evidence that a similar bisphosphonate, pamidronate, can affect osteoclasts by inhibiting their resorptive activity without affecting their numbers.20

Several recent studies have addressed the use of bisphosphonates to enhance bone formation around skeletal implants. Little et al21 have shown that the administration of zoledronate can favourably affect the quantity and mechanical properties of callus in a rat model, and that the drug can act in synergy with the anabolic agent osteogenic protein-1 (OP-1). Zoledronate has also been shown to increase bone formation in a porous tantalum implant in dogs.22 It is perhaps not unexpected that inhibiting osteoclastic activity with bisphosphonate might have different effects depending on whether the implant is a resorbable ceramic or a metal. However, the synergy between zoledronate and anabolic agents identified by Little et al21 may be species dependent. In sheep, the bisphosphonate tiludronate inhibited the rate of bone formation in trans-iliac biopsies,17 in which the extent of fluorescent labelling was almost completely abolished, an effect also seen in this study. Of perhaps greater relevance is the fact that tiludronate also blocked the normally anabolic effects of human parathyroid hormone when they were co-administered.17 This was in direct contrast to the results of a similar study in rats.23 The differences between the effects of bisphosphonates on bone formation in rats and other animals may be explained by the fact that rats, unusually among vertebrates, do not form lamellar bone even during remodelling.24

Comparison of the bone growth in a large animal model within implants consisting of granules of HA and CHA has shown significantly greater bony ingrowth in the CHA implants, an advantage lost when the bisphosponate zoledronate was administered. This suggests that carbonate substitution affects osteoclast function, which in turn affects bone formation via a coupling process. If aspects of osteoclastic conditioning of implant surfaces favour subsequent bone formation, these might be strategies for the improvement of biomaterials in the future.


  • Although none of the authors has received or will receive benefits for personal or professional use from a commercial party related directly or indirectly to the subject of this article, benefits have been or will be received but will be directed solely to a research fund, foundation, educational institution, or other nonprofit organisation with which one or more of the authors are associated.

  • Received February 18, 2008.
  • Accepted June 26, 2008.


View Abstract