Bioactive ceramic-reinforced composites for bone augmentation

Abstract

Biomaterials have been used to repair the human body for millennia, but it is only since the 1970s that man-made composites have been used. Hydroxyapatite (HA)-reinforced polyethylene (PE) is the first of the ‘second-generation’ biomaterials that have been developed to be bioactive rather than bioinert. The mechanical properties have been characterized using quasi-static, fatigue, creep and fracture toughness testing, and these studies have allowed optimization of the production method. The in vitro and in vivo biological properties have been investigated with a range of filler content and have shown that the presence of sufficient bioactive filler leads to a bioactive composite. Finally, the material has been applied clinically, initially in the orbital floor and later in the middle ear. From this initial combination of HA in PE other bioactive ceramic polymer composites have been developed.

1. Introduction

Biomaterials have been used since prehistoric times, such as the use of nacre (mother of pearl) by the Mayans to integrate dental implants into bone (Westbroek & Marin 1998). Also, in the nineteenth century silver-coated steel fracture fixation plates were used to repair patella fractures in the USA (Parkhill 1897). However, in the 1970s, the early hip and knee replacements started to fail in significant numbers. Prior to this time, joint replacements had been confined to an elderly and very low activity patient cohort and the implants were able to outlive their recipients. The implants were loosening with large areas of osteolytic (bone) resorption, so much so that at revision surgery the implants could be removed with minimal effort. The rationale for these failures was considered in detail and the concepts of ‘cement disease’ and ‘stress shielding’ were developed (Engh et al. 1987; Harrigan et al. 1992).

‘Cement disease’ was considered a biological response to the long-term application of bone cement (Charnley 1975; Jones & Hungerford 1987). Fatigue of bone cement leads to the production of cement particles and the release of the zirconia or barium sulphate particles used as radiographic opacifiers. These opacifier particles then become entrapped in the articulating surfaces of the prosthesis and accelerate the production of wear particles of the polyethylene (PE) and metal components of the joint replacement. Cement, cement opacifier, PE and metal particles have all been found in the osteolytic lesions surrounding loose joint replacements. A dose-dependent response has been seen with over 10¹⁰ polymer particles per gram of wet tissue being found in osteolytic lesions (Kadoya et al. 1997; Kobayashi et al. 1997). Other studies have shown that these wear particles can cause macrophages to initiate bone resorption (Quinn et al. 1992; Sabokbar et al. 1996).

Stress shielding occurs because of the mismatch between the stiffness of the bone, which has a Young modulus of 7–25 GPa (e.g. Bonfield & Grynpas 1977; Currey 1998), and that of the stainless steel implant stems, which have a Young modulus of 210 GPa. Various methods of solving this problem have been considered, including changing the size and shape of the stem to reduce the differences in the structural stiffness of the implant and the surrounding bone and changing the implant material from steel to commercially pure titanium or Ti-6%Al-4%V alloy (Sarmiento et al. 1979).

At this time, Katz and Lakes (Lakes & Katz 1979a,b; Lakes et al. 1979), Currey (1969, 1998) and Bonfield (Bonfield & Li 1966, 1967; Bonfield & Grynpas 1977) were all investigating the mechanical properties of bone, its structure and anisotropy. Katz presented the idea that bone itself is ‘a composite of a composite of a composite’ working from the macroscale to the nanoscale (Katz et al. 2007). Thus, Bonfield and colleagues at Queen Mary College, University of London, developed the concept of manufacturing an analogue of bone itself of PE reinforced with hydroxyapatite (HA) (Bonfield et al. 1980, 1981; Bonfield 1988). All the biocompatible metals and bioceramics available at that time are stiffer than bone, while the polymers are more flexible, with both cases leading to stress concentrations at the implant–bone interface. Bonfield and colleagues also realized that, if they used PE and HA, the material could potentially be bioactive owing to the presence of the HA and that approval for clinical use should be simplified as the individual materials were already approved for clinical application. The composite with 40 vol% HA was given the trade name HAPEX; thus, in this paper HA/PE is used to describe materials with different filler contents and HAPEX the 40 vol% HA composite used clinically.

Bonfield (1988) described bioinert materials such as stainless steel and PE as the ‘first-generation’ biomaterials and the materials aiming to interact beneficially with the body, such as HAPEX and Bioglass, as ‘second-generation’ biomaterials, while Hench & Polak (2002) have described resorbable and bioactive materials as ‘third-generation’ biomaterials, as they ‘help the body heal itself’.

2. Material production

The mechanical properties are strongly controlled by the production method in composites. In collaboration with Brunel University, twin screw compounding extrusion was used to add HA into high-density PE (HDPE). Through all the studies HDPE was used, as it can be processed by compounding and extrusion unlike ultra-high-molecular weight PE, which can only be compression moulded from powder. The initial studies used calcined bovine bone ash with a mean particle size of 10 µm (Bonfield et al. 1980, 1981). The bone ash was replaced with synthetically produced HA with a smaller mean particle size of 4 µm, which did, to some extent, improve the mechanical properties, but also allowed more reproducible and controllable starting HAs to be used. Twin screw compounding–extrusion generates high shear forces ensuring that up to 50 vol% HA could be compounded into the PE. Tensile testing showed that the increase in HA content led to increases in the Young modulus from 1.3 ± 0.2 GPa for the unfilled PE to 7.7 ± 1.3 GPa for 50 vol%; however, at 45 vol%, the tensile strength started to drop and the failure mode changed from ductile to brittle, as, instead of the material consisting of HA particles totally surrounded by PE, which acted as a matrix, HA to HA particle contact occurred (table 1). Thus, from this time all studies were confined to a maximum of 40 vol% HA. In comparison, the volume content of bone mineral of human bone is between 45 and 50 vol% and the bone mineral particles are of nanometre size with various estimates for particle size including 5 × 5 × 20 and 5 × 10 × 40 nm (Currey 1998).

Table 1.

Mechanical properties and bioactivity of hydroxyapatite reinforced polyethylene composites (from Bonfield 1988). (× indicates not bioactive, ✓ indicates bioactive and ?? indicates no data as this formulation was not tested in vivo.)

hydroxyapatite volume fraction	Young's modulus (GPa)	strain to fracture (%)	fracture mode	bioactivity (bone apposition around implant)
0	1.3 ± 0.2	>90	ductile	×
10	1.4 ± 0.2	79 ± 10	ductile	??
20	2.0 ± 0.1	50 ± 4	ductile	??
25	2.5 ± 0.2	43 ± 3	ductile	??
30	3.0 ± 0.2	34 ± 5	ductile	??
35	3.7 ± 0.4	32 ± 8	ductile	??
40	4.4 ± 0.7	29 ± 5	ductile	✓
45	5.9 ± 0.5	7 ± 3	brittlea	??
50	7.7 ± 1.3	3 ± 1	brittleb	✓

^aK_IC = 2.9 ± 0.3 MN m^−3/2.

^bK_IC = 2.4 ± 0.2 MN m^−3/2.

Various studies have investigated the effect of altering particle size and particle morphology. Wang et al. (1994, 1998a) compared two grades of spray-dried HA: one with a mean particle size of 4.14 µm and a specific surface area of 8.27 m² g⁻¹ and the other with a mean particle size of 7.32 µm and a specific surface area of 7.61 m² g⁻¹, but both produced by spray-drying leading to each particle being an agglomerate of nanocrystallites. Minor differences in the tensile strength were seen for the different filler particles and the filler contents investigated; however, the tensile and torsional stiffnesses were higher with the smaller filler particles. Joseph et al. (2001a,b, 2002a,b) investigated the effect of HA particle shape and used a spray-dried and a sintered grade of HA powder with similar particle size distributions and mean particle sizes of 4.02 and 3.58 µm, respectively, but a factor of over 10 difference in the specific surface areas. They compound extruded the composite and then characterized the rheological properties. The HA/PE manufactured with spray-dried HA had substantially higher viscosity than that manufactured using sintered HA (figure 1). They suggested that the difference was produced by the amount of PE involved in coating the HA particles and thus how much was free to act as a matrix between the coated particles (figure 2). To improve the impact properties of the composites, Zhang & Tanner (2008) produced composites using sintered particles of HA for comparison with the spray-dried HA-filled composites. These two HAs were again chosen to have similar particle size and particle size distribution with mean values of 3.80 µm and 4.46 µm, but different specific surface areas of 13.536 and 0.965 m² g⁻¹ for the spray-dried and sintered HAs, respectively. Composites with 20, 30 and 40 vol% fillers were manufactured and subjected to impact testing, using an instrumented falling weight impact tester. For all volume fractions, both the initiation and propagation energies were higher with the sintered HA with a lower surface area. Scanning electron microscopy (SEM) of the fracture surfaces showed that the smoother particles allowed more PE to be drawn during the fracture process (figure 3), in agreement with the concept of Joseph et al. (2002a) of the effect of specific surface area.

Figure 1.

Apparent shear viscosity of 48 650 molecular weight PE reinforced with 40 vol% spray-dried HA (P205) and sintered HA particles (P215) at temperatures between 200 and 250°C (adapted from Joseph et al. 2002a).

Figure 2.

Schematic showing the different amounts of matrix used to coat high specific surface area spray-dried HA (P205) versus sintered low specific surface area HA (P215S) (adapted from Joseph et al. 2002a).

Figure 3.

Impact fracture surfaces of 30 vol% (a) spray-dried HA and (b) sintered HA in polyethylene, showing the longer draw fibrils with the lower surface area filler particles (marker bars = 10 µm) (adapted from Zhang & Tanner 2008).

The effect of HDPE molecular weight was investigated by both Joseph et al. (2002a) and Zhang & Tanner (2003). Joseph et al. (2002a) were interested in producing an injection moulding grade of HA/PE; thus, they used three grades of HDPE with melt flow indices between 7.6 and 26 g (10 min)⁻¹, i.e. molecular weights between 72 350 and 48 650, respectively, and filled these PEs with 0, 20 and 40 vol% of spray-dried HA. The effect of the increasing filler content or the molecular weight was to increase the shear stress at any shear rate during injection moulding. Zhang & Tanner (2003) used two different HDPEs with melt flow indices of 6 and 26 g (10 min)⁻¹, i.e. molecular weights of 249 233 and 48 650, respectively. Under drop weight impact testing, the effect of molecular weight was substantial, with the energy absorbed being 5.1 J for the non-filled lower molecular weight HDPE compared with 93.6 J for the non-filled high molecular weight HDPE. The impact of energy absorbed decreased to 7.6 J for the high molecular weight HDPE and to less than 2 J for the lower molecular weight HDPE when these were filled with 40 vol% HA. As with the studies changing the HA filler morphology, the tougher material showed longer drawn polymer fibrils on the fracture surfaces.

3. Material modifications

Failure of such composites is initiated by debonding of the filler matrix interface (Friedrich & Karsch 1983; Bonfield et al. 1998). Chemically coupling the HA and the PE phases to increase the mechanical properties was investigated by Deb and colleagues (Deb et al. 1996; Wang et al. 2000a; Wang & Bonfield 2001), who silanated the HA phase and acrylic acid coupled the PE. Silane coupling without acrylic acid grafting reduced the Young modulus and did not affect the strength; however, the Young modulus and tensile strength were increased when both silane coupling and grafting were combined. The most substantial difference was the increase in the strain to failure at both 20 and 40 vol% filler produced by silane coupling with or without acrylic acid grafting (table 2).

Table 2.

Mechanical properties of HA/PE composites at 20 and 40 vol% HA showing the effect of silane-coupling and acrylic acid grafting, indicating significant differences when compared with the non-coupled, non-grafted material of the same volume fraction (from Deb et al. 1996).

filler volume content	silane coupling	acrylic acid grafting	tensile strength (MPa)	Young's modulus (GPa)	strain to fracture (%)
20	no	no	17.77 ± 0.09	1.60 ± 0.02	34.0 ± 9.5
20	yes	no	17.01 ± 0.19**	1.54 ± 0.02*	>100
20	yes	yes	19.97 ± 0.07**	1.81 ± 0.05**	39.7 ± 1.5
30	no	no	22.67 ± 0.17	4.29 ± 0.17	2.6 ± 0.4
30	yes	no	22.08 ± 0.05	3.66 ± 0.20**	7.8 ± 0.6**
30	yes	yes	23.16 ± 0.40*	3.87 ± 0.21*	6.8 ± 0.6**

*Significant difference at p < 0.05.

**Significant difference at p < 0.01.

A different approach to increase the mechanical properties was used by Ladizesky et al. (1997), McGregor et al. (2000), Wang et al. (2000b) and Bonner et al. (2002). They hydrostatically extruded HAPEX to orient the polymer chains and thus produce an anisotropic material. Extrusion ratios of up to 11 : 1 were used to increase the stiffness and flexural strength. While the stiffness continued to increase with increases in the extrusion ratio, the flexural strength and ductility peaked at an extrusion ratio of 8 : 1 (figure 4). All the materials were ductile, and at the highest extrusion ratios the failure was due to fibrillation parallel with the extrusion direction, rather than actual fracture of the material.

Figure 4.

Effect of the extrusion ratio on the flexural strength (crosses), flexural modulus (open squares) and strain to failure (filled triangles) of hydrostatically extruded 40 vol% HA/PE.

More recently, selective laser sintering (SLS) was used to manufacture porous scaffolds of HA/PE and HA/polyamide (HA/PA) composites (Hao et al. 2006a,b, 2007, 2009; Savalini et al. 2006, 2007; Zhang et al. 2008). Twin screw extrusion was used to process the composites, which were then ground down into particles of the order of 40–300 µm across. These composite particles were then sieved into two groups—those with particle size greater than and those with particle size less than 165 µm—and these were selectively laser sintered. A purpose-built SLS system was manufactured to allow infrared heating of the powder to just below the polymer melt temperature and then low laser power (less than 10 W) could be used to perform the sintering. By optimizing the particle size and laser power, scaffolds of 20 vol% HA/PE and 30 vol% HA/PA could be sintered successfully. Infiltration of the samples with resin and subsequent sectioning and polishing for microscopy showed that the material was all open celled, which is essential for tissue ingrowth (figure 5). Optimization studies showed that 20 vol%HA in PE and 30 vol% HA in PA produced the greatest consolidation. Dynamic mechanical analysis (DMA) was used to measure the mechanical properties as this allowed much smaller test samples to be used (Zhang et al. 2008). The stiffness was lower than that of the fully dense material, but the materials still had moduli of 0.7 GPa at 37°C when compared with 2 GPa for the fully dense compression-moulded HA/PE composite.

Figure 5.

Optical micrographs of selectively laser-sintered 20 vol% HA in PE showing the infiltration of the resin and thus that it is an open-celled material with each particle composed of HA/PE composite (marker bars are 50, 50 and 10 µm, respectively) (adapted from Zhang et al. 2008).

4. Mechanical testing

The initial mechanical testing concentrated on quasi-static testing to optimize the production methods using compression-moulded samples, although some injection-moulded samples were tested. Implants are subjected to both creep and fatigue loading in vivo. Suwanprateeb et al. (1995) investigated the creep behaviour at 37°C in saline and started with isochronous testing, where a small stress is applied for 100 s and the strain is measured at the end of the loading time, followed by unloading for 400 s and then applying a higher stress for 100 s and continuing with gradually increasing stress levels to produce isochronous stress–strain curves (figure 6). These isochronous tests allowed basic comparison of the creep behaviour of the difference volume fracture fillers and the optimization of the load levels for conventional long-term creep testing. These authors found that increasing the filler content from 0 to 20 vol% or from 20 to 40 vol% halved the creep strain for a given applied load; however, these increases in filler content also decreased the time to creep rupture. In subsequent studies Suwanprateeb et al. (1997, 1998) showed that pre-soaking the composite in saline at 37°C increased the strain deformation owing to the plasticizing effect of the absorbed liquid, but that gamma irradiation decreased the creep owing to the cross-linking of the polymer, and finally that thermal annealing also decreased the creep and increased the time to creep rupture (figure 7).

Figure 6.

Isochronous creep testing of HAPEX showing (a) the loading regime and (b) the results obtained for high-density polyethylene (HDPE) and reinforced with 20 vol% (20 HA/PE) and 40 vol% HA (40 HA/PE) (adapted from Suwanprateeb et al. 1995).

Figure 7.

(a) Influence of immersion upon the creep behaviour of 2.5 Mrad γ-irradiated 40 HA/PE at 6 MPa applied stress at 37°C in Ringer's solution (continuous line, non-immersed; dash–dot line, 1 day; short-dashed line, 7 days; dotted line, 90 days; long-dashed line, 150 days; adapted from Suwanprateeb et al. 1997) and (b) effect of thermal annealing on the creep behaviour (continuous lines, irradiated; dashed lines, annealed + irradiated; adapted from Suwanprateeb et al. 1998).

Ton That et al. (2000a,b) studied the fatigue behaviour of HAPEX under both uniaxial tension–compression and when combined with fully reversed torsional loading. As to be expected for a filled polymer, the tensile phase of the loading caused more damage to the material than the compressive phase as the interface between the polymer and the HA opened up. This behaviour led to non-symmetrical stress–strain loops with dynamic creep and increased the energy absorbed per load cycle. However, fully reversed torsional loading produced symmetrical loops, which reduced in the secant modulus and increased in the energy absorbed as the material failed (figure 8). Failure was seen first as a reduction in the secant modulus followed by an increase in the energy absorbed per load cycle. Increasing the phase angle between the axial and torsional loading increased the fatigue life and, as with the quasi-static and creep testing, failure was by debonding of the HA from PE and drawing of PE fibrils between the filler particles. Joseph & Tanner (2005) performed uniaxial tensile only fatigue tests, in saline at 37°C, with the maximum stress being 75 per cent of the ultimate tensile strength on 40 vol% of HA/PE manufactured using both spray-dried and sintered HA. They found that the energy absorbed per load cycle was higher and the secant modulus was lower with the sintered filler particles. These differences led to earlier initiation of failure with the sintered filler particle composite.

Figure 8.

Stress–strain loops for fully reversed (a) tension compression at ±50% of ultimate tensile strength and (b) torsion at ±50% of ultimate shear strength (adapted from Ton That et al. 2000a).

Nazhat et al. (2000) used DMA at 1 Hz and a temperature scan between 20 and 100°C of different volume fraction HA/PE using the same spray-dried HAs with different particle sizes and the same PE as used by Wang et al. (1998a). Similar to other studies, they found that increasing the filler content increased the stiffness and that the smaller particle size led to a higher modulus for a given volume fraction. The use of DMA allowed the damping to be measured as tanδ, the ratio of the loss modulus (E′) to the storage modulus (E′), and this was found to increase with increasing temperature, but to be reduced by higher filler content with minimal differences between the two types of filler particles.

Eniwumide et al. (2004) used compact tension testing to measure the fracture toughness of composites manufactured with two different molecular weight PEs and both sintered and spray-dried HAs. They showed that testing the material at 37°C rather than room temperature increased the fracture toughness from 1.30 ± 0.04 to 1.34 ± 0.07 MN m^−3/2 for 40 vol% HA/PE manufactured with 240 000 molecular weight PE. When the molecular weight was reduced to 50 000, the fracture toughness decreased to 0.5 MN m^−3/2, whereas decreasing the specific surface area of the filler particles increased the fracture toughness.

To elucidate the failure mechanisms, Guild & Bonfield (1993, 1998) used microscale finite element analysis (FEA) modelling of spherical filler particles in a polymeric matrix. The tensile modulus estimated from the FEA analysis is slightly lower than that measured experimentally at intermediate filler contents. These authors showed that the highest stresses occurred at the poles of the filler particles; thus, the failure started by debonding of the filler from the polymer and that the filler particles are seen between drawn fibrils, similar to the experimental mechanical testing results. They also showed an increase in stress concentration at about 40 vol% filler in agreement with the experimental results presented in Bonfield (1988; figure 9).

Figure 9.

(a) Comparison of the Young modulus of HA in PE measured experimentally (filled squares) and predicted using FEA modelling with the PE bulk modulus at 5 GPa (circles) and 10 GPa (open squares); (b) contours of stress concentration of the von Mises stress in the polyethylene matrix viewed from the centre of the HA sphere, with the sphere removed and (c) effect of volume fraction of HA spheres on the stress concentration factor (adapted from Guild & Bonfield 1998).

5. in vitro testing of bioactivity

When HAPEX was initially developed, cell culture on biomaterials was insufficiently developed as a science to be used to assess the biological response to materials. Thus, the initial biological assessment was performed in vivo (Doyle et al. 1990). As cell culture studies on biomaterials developed, in vitro studies of HAPEX were performed. Di Silvio and colleagues (Huang et al. 1997a,b; Di Silvio et al. 1998, 2002a,b; Dalby et al. 1999, 2000, 2002) cultured osteoblasts obtained at elective surgery from human femoral heads on various HA/PE composites and showed that the material was biocompatible. Huang et al. (1997b) and Di Silvio et al. (1998) also showed that the osteoblasts proliferated and matured faster on HAPEX than on plain PE, although the response was slower on both the composites than on Thermanox (PE terephthalate) coverslips used as a control (figure 10a). The most interesting finding was that the osteoblast cell processes attached down onto the HA particles in obvious preference to the surrounding PE (figure 10b).

Figure 10.

(a) Alkaline phosphatase activity on samples of HAPEX (filled bars) and PE (striped bars) with Thermanox (TMX) (unfilled bars) as a control surface and (b) SEM of osteoblasts attaching down onto the HA particles in HAPEX (marker bar = 10 µm) (adapted from Huang et al. 1997b).

In a later study (Di Silvio et al. 2002a,b), 20 and 40 vol% of HA/PE were compared with Thermanox controls. Cell proliferation was measured using tritiated thymidine incorporation and cell differentiation using alkaline phosphatase activity. The cells proliferated faster on 40 vol% HA/PE than on 20 vol% HA/PE and then differentiated more, the difference peaking at day 7 of culture. SEM showed the osteoblasts attaching down on to the HA particles and showed cells in the process of division. Confocal laser microscopy showed actin stress fibres in the cells and with vinculin staining adhesion plaques could be seen, although the number of adhesion plaques and actin fibres were higher and more organized on the 40 vol% HA/PE than on the 20 vol% HA/PE. Transmission electron microscopy (TEM) of the cell culture showed that more collagen was formed between the cells on the higher HA content material.

Dalby et al. (2000) showed the importance of surface features by culturing osteoblasts on machined rather than polished surfaces of HAPEX. The osteoblasts aligned themselves with the grooves on the surface. Rea et al. (2004a) also investigated the effect of surface texture and produced a series of different grooves, pillars and pits on the surface of HAPEX (figure 11). They found that compared with a polished surface there was not much difference in the cell number on the textured surfaces with the possible exception of 0.3 mm wide grooves. However, cell differentiation, as measured by alkaline phosphatase activity, was increased on the textured surfaces.

Figure 11.

Effect of surface texture on osteoblast attachment to (a) grooved with grooves 50 µm deep and 50 µm wide and (b) pitted with pits 50 µm deep and 50 µm wide and polished areas of HAPEX (marker bars = 100 µm) (adapted from Rea et al. 2004c).

The importance of the HA morphology was shown by Zhang et al. (2007), who cultured osteoblasts on polished sections of 30 vol% HA/PE manufactured using the sintered and spray-dried HA particles investigated in Zhang & Tanner (2008). Three control materials were used—100% PE and 100% HA in addition to Thermanox. When the composite surface was polished, the spray-dried particles were seen to be slightly porous and thus textured while the polished sintered particles were solid and thus smooth (figure 12a,b). The osteoblasts were from both human femoral heads removed at hip replacement and human calveria, again taken at elective surgery. The human calveria origin cells reacted slightly more slowly than those from the femoral heads, although with similar trends. Both origin cells attached with more, but finer, cell processes onto the spray-dried HA containing composite than onto the sintered HA composite with higher cell proliferation followed by higher cell differentiation (figure 10c). However, these responses were higher and faster on the composite than on either 100% PE or 100% HA.

Figure 12.

The interaction of femoral head origin osteoblasts with 30 vol% HA/PE manufactured using (a) spray-dried HA, (b) sintered HA particles (marker bars = 10 µm) and (c) the alkaline phosphatase activity of osteoblasts from femoral heads (HOBf) and calvaris (HOBc) cultured on Thermanox controls (TMX, open bars), non-filled PE (PE, filled bars), 30 vol% spray-dried HA in PE (30 vol% HA1–PE, striped bars), 30 vol% sintered HA in PE (30 vol% HA2–PE, divided bars) and HA (HA, checked bars) (adapted from Zhang et al. 2007).

The cellular response to selectively laser-sintered HA/PE and HA in polyamide (HA/PA) composites was investigated by Zhang et al. (2009). They used 20 vol% HA and 30 vol% HA in PE, as processing constraints had shown these to be the highest HA content possible in these two matrix materials (Hao et al. 2006a,b, 2007, 2009; Savalini et al. 2006, 2007; Zhang et al. 2008). In this study, two controls were used—Thermanox, which should lead to a beneficial cellular response, and tin-doped polyvinylchloride (SnPVC), a known toxic material. The cell proliferation and differentiation were higher on the 30 vol% HA/PA than on the 20 vol% HA/PE, but the production of osteocalcin, a marker for bone mineralization, was higher on both composites than on the Thermanox (figure 13). Cell culture was progressed through to 28 days and, without the addition of dexamethasone or any other additional factors to drive bone mineralization, on both the composites mineralized nodules were seen, indicating the strong drive for the osteoblasts to progress through to bone formation.

Figure 13.

The interaction of human osteoblasts on selectively laser-sintered (a) 20 vol% HA/PA; (b) 30 vol% HA/PA (marker bars = 10 µm) and (c) osteocalcin levels on Thermanox control (TMX, unfilled bars), electively laser-sintered 20 vol% HA/PA (SLSHAPE, grey bars) and 30 vol% HA/PA (SLSHAPA, black bars) with differences by Student's t-test: *p < 0.05; ***p < 0.001 (adapted from Zhang et al. 2009).

6. in vivo testing of bioactivity

The in vivo biocompatibility and bioactivity testing was initiated before the in vitro process, as the development of this material pre-dated the use of in vitro cell culture studies for biocompatibility testing. In the first of these studies, Bonfield et al. (1986) implanted rivet-shaped implants of HAPEX, with non-filled PE as a control, into the femoral shaft of rabbits, passing through both cortical and cancellous bone. While the PE was surrounded with a fibrous layer, direct bone contact onto the HAPEX was seen. In later studies, high-resolution TEM was used to investigate the crystallographic detail of this interface and showed that the mineral phase in the newly formed bone was aligned with that of the HA in the composite, indicating that the newly formed bone appeared to be using the HA on the composite surface to nucleate mineralization (Doyle et al. 1990).

Tanner et al. (1990) performed push-out testing to measure the strength of the interface-produced bone between HAPEX and apatite–wollastonite (A–W) glass ceramics provided by Prof. Kokubo or non-filled PE. Even after three months of implantation, the non-filled PE–bone interface failed with a shear strength of less than 0.6 MPa, while the A–W glass ceramic–bone interface reached 20 MPa and the HAPEX–bone interfacial strength gradually increased up to 10 MPa at three months.

Revell et al. (1997) implanted rods into the distal femora of rabbits, with each rod having a slot cut into it 2 mm deep and 2 mm across. Two materials were used, HAPEX and titanium, where the inside of the slot was plasma sprayed with HA (Ti + HA) to model plasma-sprayed titanium used for the stems of non-cemented joint replacements. After five weeks, the new bone formation was down to the bottom of the HAPEX slot and filled 80 per cent of the available space, unlike the Ti + HA, where the new bone filled only 25 per cent of the available space (figure 14).

Figure 14.

Section through the base of the slot machined into a HAPEX cylinder and implanted in a rabbit knee for five weeks.

7. Clinical applications

HAPEX was the first of the bioactive composites implanted into patients and was used by Downes and co-workers (Downes et al. 1991; Tanner et al. 1994) in orbital floor applications. The two designs of implants either filled the space left after an eye was removed to allow the artificial eye to fit better or restored eye alignment in patients with orbital floor fractures. These were low load-bearing applications, but showed that the material encouraged the supporting bone to bond to the implant material. Previous implants were manufactured of silicone and manual palpation had shown them lying loose in a fibrous capsule with some extruded. None of the HAPEX implants moved on palpation or were extruded. Computerized tomography images (figure 15) showed no gaps between the implant and the supporting bone.

Figure 15.

A HAPEX implant in the orbital floor of a patient who has lost an eye. The HAPEX implant is shown on the right of the image with the same radiographic density as the bone to which it is bonded while the spherical black object is a glass ball implanted in an earlier attempt to increase the volume of material in the orbital socket (adapted from Downes et al. 1991).

This initial study had only 18 patients, but allowed the material to be used by Smith & Nephew ENT to manufacture the shaft of a range of middle ear implants including the Goldenberg (1994) and Dornhoffer (1998) implants. These implants consist of a monolithic HA head on a shaft and are designed to replace the mechanical function of sound transmission through the three bones of the middle ear. The HA head is placed against the tympanic membrane (ear drum) and the surgeon cuts the shaft to fit the individual patient. The range of implants allows for patients in whom all three bones of the middle ear have been lost or damaged or in whom only two bones are affected and the stapes are still present. Originally, the shafts were manufactured of PE and the surgeon trimmed them to shape with a scapel blade; when the PE shafts were replaced with HAPEX shafts, the ability of the surgeon to trim the implant to shape was retained and indeed improved (figure 16). One study by Meijer et al. (2002) has reported the clinical application of these implants. They used SEM to review the response to 11 HAPEX implants which had been removed owing to recurrence of the original clinical problem, rather than for device failure. In most of the cases, the implants were covered with fibrous tissue, and in half of these there was a thin epithelial outer layer. The response to implantation was considered to be good even after 30 months of implantation. Also, in no cases were macrophages or other signs of inflammation seen, unlike similar retrieved implants manufactured of Proplast or Plastipore where macrophage activity was obvious and some inflammation was seen.

Figure 16.

Middle ear implant in situ showing the notch cut in the HAPEX shaft to allow the implant to sit over the arch of the stapes (top) and the HA head resting on the tympanic membrane (bottom).

8. Other composites

This work with HA-reinforced PE has led to a range of other potentially bioactive composites being investigated, both by Bonfield and his colleagues and by other groups. These composites can be degradable or non-degradable depending on the matrix polymer. For non-degradable materials, PE remains a popular matrix material, while a range of degradable polymers have been used as matrix materials in composites, including polylactic acid (PLA) and polyhydroxybutyrate (PHB).

Among the non-degradable composites, Harper (1998) reinforced PMMA bone cement with 17 wt% (6 vol%) HA in an attempt to confer bioactivity without reducing the mechanical properties. Dalby et al. (1999, 2001) investigated the cellular response to this composite. The addition of HA increased the cellular response to the PMMA with again the osteoblast processes attaching down onto the HA particles and higher actin organization in the cells grown on the filled PMMA than in those grown on the non-filled (figure 17). Bonner et al. (2001) reported the manufacture and hydrostatic extrusion of polypropylene reinforced with 40 vol% HA (HA/PP). Hydrostatic extrusion increased the modulus from 7.5 to 9.3 GPa and the flexural strength from 25 to over 80 MPa. Neither of these materials has progressed beyond the initial testing.

Figure 17.

Actin cytoskeleton (green)/vinculin (red) interaction surrounding the cell nucleus (blue) on (a) PMMA (marker bar = 10 µm) and (b) PMMA reinforced with 6 vol% HA (marker bar = 10 µm) (adapted from Dalby et al. 2001).

In the 1980s, PHB and its co-polymer polyhydroxyvalerate (PHV) were being developed as biodegradable polymers with similar mechanical properties to polypropylene. Doyle et al. (1991) manufactured HA-reinforced PHB and characterized the mechanical properties and performed in vivo analysis of the material, including push-out testing similar to that performed by Tanner et al. (1990). The non-filled PHB had a Young modulus of 4 GPa, which was increased to 11 GPa with the addition of 40 vol% HA; however, the tensile strength dropped from 37 to 22 MPa. The biological response to implantation of 40 vol% HA/PHB was bone ongrowth (figure 18), but in push-out testing failure occurred though the material rather than at the implant–bone interface, as occurred with HA/PE composites. Luklinska & Bonfield (1997), using high-resolution TEM, showed that the bone had bonded across the interface. Knowles & Hastings (1993) manufactured composites by dry blending phosphate-based glasses of 20–40 wt% into PHB with 7 vol% PHV and then injection-moulded samples. The presence of the glass accelerated the degradation. However, all these studies stopped owing to the ending of production of medical grade PHB for commercial reasons. More recently, PHB has again been used in bioactive composites such as PHB reinforced with 5 and 15 wt% HA short fibres produced by Coskun et al. (2005) and PHB 12% PHV reinforced with up to 14 vol% nano-HA produced by Chen et al. (2007). As a matrix material, PHB has the advantages of higher modulus than most other degradable polymers; furthermore, by including some PHV co-polymer, the ductility can be increased, although with reductions in the stiffness and strength.

Figure 18.

Interface developed between cortical bone and 40 vol% HA/PHB after one month (marker bar = 100 µm) (adapted from Luklinska & Bonfield 1997).

In contrast to these composites with different polymers reinforced with HA, Huang et al. (1997a,c, 1998) and Wang et al. (1998b) reinforced PE with Bioglass. Again, increases in the modulus were seen with the addition of the Bioglass, but the sharp corners of the Bioglass led to stress concentrations and thus significant reductions in the strength (figure 19). However, the response to soaking in simulated body fluid (SBF) was excellent with dissolution of the Bioglass acting to initiate the development of a hydrated carbonate layer which started from the Bioglass particles and then progressed across the composite surface (figure 20). The material was seen to be biocompatible with increased cell viability when cultured in cell culture medium that had been exposed to the composite for 24 h. This increase in cell viability was presumably due to the ions released during the dissolution of the Bioglass in the cell culture medium.

Figure 19.

Comparison of the reinforcing effects of hydroxyapatite (HA) (filled diamonds), Bioglass (filled squares) and A–W glass (filled triangles) ceramic in polyethylene on (a) Young's modulus and (b) tensile strength at a range of volume fractions.

Figure 20.

Polished sections of 40 vol% Bioglass-reinforced PE. (a) Polished section (marker bar = 100 µm) and (b) after 7 days in simulated body fluid (marker bar = 10 µm) (adapted from Huang et al. 1997a).

Juhasz et al. (2003, 2004) reinforced PE with A–W glass ceramic and again found that the addition of the glass ceramic increased the stiffness of the composite, but, owing to the shape of the A–W glass ceramic particles, stress concentrations led to low-strength materials. However, the bioactivity, as measured by time to produce a surface layer after soaking in SBF (Kokubo et al. 1990), was excellent. With even as little as 10 vol% A–W glass ceramic, 100 per cent of the surface of samples was covered in 3 days and with 50 vol% total coverage took less than 6 h. Rea et al. (2004a–c) compared the response of osteoblast-like cells on HAPEX and A–W glass ceramic–PE composites and showed that, as with HAPEX, when the cells were cultured on A–W glass ceramic–PE they preferentially attached down on the filler particles via filopodia. The response increased with increasing A–W glass ceramic content but for all surface textures, be they grooves, pits or pillars, the response, in terms of cell proliferation and differentiation, was higher on HAPEX than on A–W glass ceramic–PE (figure 21).

Figure 21.

(a) An osteoblast growing on A–W glass ceramic in PE composite showing the cell attaching preferentially onto the A–W glass ceramic particles (marker bar = 10 µm) and (b) the increases in cell number with time for 30 and 50 vol% A–W glass ceramic in PE with mean particle sizes of 4.5 and 7.7 µm (adapted from Rea et al. 2004b).

The idea of reinforcing a polymer with a bioactive filler was taken further by Törmälä et al. (1997) and Bleach et al. (2001, 2002), who reinforced PLA with HA or tricalcium phosphate (TCP) particles and then used this composite to act as the matrix around drawn PLA fibres, thus creating a three-phase fibre-reinforced composite. The stiffness and strength were at the lower bounds of those for cortical bone at over 7 GPa and 70 MPa, respectively. An interesting effect of adding HA or TCP was to chemically buffer the degradation of the PLA so that the start of loss of strength was delayed from eight to 12 weeks. DMA performed by Nazhat et al. (2001) at 1 Hz test frequency and scanning from −50 to +100°C showed the glass transition temperatures for both the PLAs used in the manufacture of the composite. The drawn fibres increased the storage modulus from 3.5 GPa to over 7 GPa and the loss modulus was relatively unaffected by temperature up to the glass transition temperature.

Other composites containing polymer fibres have been developed. For example, Waris et al. (1994) manufactured plates of PLA and reinforced PLA and used these in the repair of skull defects in growing children, as the degradation of the plates allowed skull growth to continue once the plate had disappeared.

9. Conclusions

The idea of adding HA to polymers has led from the bioinert monolithic materials used in the 1970s and 1980s through to the bioactive biodegradable composites used now as modulus matched materials and more recently as the scaffolds for tissue engineering. The work has led to the application of conventional composite manufacturing techniques in the field of biomaterials, with materials that have similar mechanical properties to bone. The addition of the bioactive phases has led to materials that have been shown to be bioactive in both in vitro studies, using soaking in SBF, and in vitro cell culture, where the interaction of the cells with the filler phases has indicated the manner of the biological and cellular response in vivo. The production methods, filler content and filler particle morphology all affect the tensile strength and other mechanical properties, so must be optimized before in vitro and in vivo biological testing is performed. Finally, the materials have been used clinically, leading to improvements in the quality of life of patients worldwide. While the materials described here are no longer available for clinical use, their production and development have led to other newer materials being developed for clinical applications. Furthermore, this work has shown that composites can be produced with both good mechanical and biological properties, rather than either good mechanical or good biological properties, using engineering composite manufacturing techniques.