Materials Science and Engineering C 42 (2014) 130–136
Contents lists available at ScienceDirect
Materials Science and Engineering C journal homepage: www.elsevier.com/locate/msec
Fabrication and cytocompatibility of spherical magnesium ammonium phosphate granules Theresa Christel a, Martha Geffers a, Uwe Klammert b, Berthold Nies c, Andreas Höß c, Jürgen Groll a, Alexander C. Kübler b, Uwe Gbureck a,⁎ a b c
Department for Functional Materials in Medicine and Dentistry, University of Würzburg, Pleicherwall 2, 97070 Würzburg, Germany Department of Cranio-Maxillo-Facial Surgery, University of Würzburg, Pleicherwall 2, 97070 Würzburg, Germany InnoTERE GmbH, Pharmapark Radebeul, Meissner Straße 191, 01455 Radebeul, Germany
a r t i c l e
i n f o
Article history: Received 14 January 2014 Received in revised form 31 March 2014 Accepted 6 May 2014 Available online 22 May 2014 Keywords: Calcium magnesium phosphate cement Struvite Granules
a b s t r a c t Magnesium phosphate compounds, as for example struvite (MgNH4PO4·6H2O), have comparable characteristics to calcium phosphate bone substitutes, but degrade faster under physiological conditions. In the present work, we used a struvite forming calcium doped magnesium phosphate cement with the formulation Ca0.75Mg2.25(PO4)2 and an ammonium phosphate containing aqueous solution to produce round-shaped granules. For the fabrication of spherical granules, the cement paste was dispersed in a lipophilic liquid and stabilized by surfactants. The granules were characterized with respect to morphology, size distribution, phase composition, compressive strength, biocompatibility and solubility. In general, it was seen that small granules can hardly be produced by means of emulsification, when the raw material is a hydraulic paste, because long setting times promote coalescence of initially small unhardened cement droplets. Here, this problem was solved by using an aqueous solution containing both the secondary (NH4)2HPO4 and primary ammonium phosphates NH4H2PO4 to accelerate the setting reaction. This resulted in granules with 97 wt.% having a size in the range between 200 and 1000 μm. The novel solution composition doubled the compressive strength of the cement to 37 ± 5 MPa without affecting either the conversion to struvite or the cytocompatibility using human fetal osteoblasts. © 2014 Elsevier B.V. All rights reserved.
1. Introduction The use of autografts for the repair of bone defects is considered as “gold standard”, but has several drawbacks such as a lack of availability and its donor site morbidity [1,2]. Alternative synthetic bone replacement materials are mostly based on calcium phosphate chemistry. Material approaches cover both the use of hydroxyapatite derived from natural resources (e.g. BioOss®) [3] as well as fully synthetic hydroxyapatites (HA) or tricalcium phosphates (TCP) prepared by sintering, precipitation or a cement setting reaction [4]. Depending on the chemical composition and crystal size, these materials are either stable (HA) or are able to slowly degrade in vivo (TCP) by active and passive mechanisms such that host autologous bone can grow into the bone defect [5]. Novel material approaches to enhance the degradation ability of bioceramics for bone replacement involve the use of magnesium phosphate compounds [6]. Such materials were found to be as
⁎ Corresponding author. Tel.: +49 931 20173550. E-mail addresses: [email protected] (T. Christel), [email protected] (M. Geffers), [email protected] (U. Klammert), [email protected] (B. Nies), [email protected] (A. Höß), [email protected] (J. Groll), [email protected] (A.C. Kübler), [email protected] (U. Gbureck).
http://dx.doi.org/10.1016/j.msec.2014.05.023 0928-4931/© 2014 Elsevier B.V. All rights reserved.
cytocompatible as calcium phosphates, but offer a faster chemical degradation due to their higher solubility under physiological conditions [7]. Various magnesium phosphate compounds have been investigated in recent studies, e.g. struvite (MgNH4PO4·6H2O) [8–11], newberyite (MgHPO4·3H2O) [12], amorphous magnesium phosphates [13] or magnesium doped calcium phosphates [7,14,15]. The formation of these compounds is often based on a cementitious reaction of either MgO or Mg3(PO4)2 with aqueous sources of phosphate and ammonium salts. The solubility profiles of these cements may be altered by using biphasic calcium magnesium phosphate mixtures. This was already demonstrated by Vorndran et al. [11], who used compounds with the general formula CaxMg(3−x)(PO4)2 allowing to adjust the setting properties as well as the mechanical performance and solubility. By using the formulation Ca0.75Mg2.25(PO4)2 together with a 3.5 M (NH4)2HPO4 solution, a good biocompatibility in vitro [11] and in vivo [12], a high mechanical strength of ~ 80 MPa and an adequate setting time of 14 min were obtained [11]. In the present work, a cement based on Ca0.75Mg2.25(PO4)2 was used to fabricate pre-hardened cement granules since in dental surgery (e.g. sinus lift operation, socket preservation, filling of jaw cysts) the application of bone substitutes in granular form is a typical approach [16]. An easy way to generate granules on the basis of a hydraulic cement paste is the hardening of a cement monolith followed by
T. Christel et al. / Materials Science and Engineering C 42 (2014) 130–136
mechanical grinding [3]. However, this technique leads to irregular sharp-edged granules that can promote irritation and inflammation in living tissues [16]. Another approach used a mixture of precipitated HA and chitosan, whereas the granules were hardened after starting the emulsification process by the addition of a cross-linker leading to granules with sizes in the range of 212 to 1000 μm [17]. However, in this case HA was not formed on the basis of a calcium phosphate cement (CPC) setting reaction but added as an inert filler to the biopolymer matrix prior to emulsification. Other techniques such as precipitation reactions would actually result in the fabrication of nanoscaled granules [18]. For example, Singh et al. [19] produced brushite nanoparticles by the combination of micro-emulsification with the process of precipitation. In the present study, a more sophisticated approach was applied. It is based on dispersing the hydrophilic cement paste in a hydrophobic liquid (e.g. oil) by stirring while stabilizing the formed cement-oil dispersion using surfactants. The influence of different emulsification parameters (e.g. stirring energy, oil type, surfactant type and concentration, and aqueous solution composition) on the granules' properties (morphology, size distribution, degree of conversion, phase composition, cytocompatibility, and solubility) were tested. The main objective was the fabrication of spherical granules with clinically suitable diameters of 200 to 1000 μm. 2. Materials and methods 2.1. Fabrication of cement raw materials Ca0.75Mg2.25(PO4)2 was prepared by sintering 1.5 mol MgHPO4·3H2O (Sigma Aldrich, Steinheim, Germany, Lot# BCBF1937V) with 0.5 mol CaHPO4 (J.T. Baker, Griesheim, Germany, Lot# K06 587), 0.25 mol CaCO3 (Merck, Darmstadt, Germany, Lot# A0074720 927) and 0.75 mol Mg(OH)2 (Fluka, Steinheim, Germany, Lot# 10H040006) at 1100 °C for 5 h in a sintering furnace (Oyten Thermotechnic, Oyten, Germany). The product was crushed with a pestle and mortar, passed through a 355 μm sieve and milled in a planetary ball mill (Retsch, Haan, Germany) at 200 rpm with 500 mL agate jars, four agate balls (30 mm) and a load of 125 g powder per jar for up to 4 h. α-Ca3(PO4)2 was synthesized by heating a mixture of 2 mol CaHPO4 and 1 mol CaCO3 to 1400 °C for 5 h. The product was crushed and sieved using a 355 μm sieve. 125 g of the powder was milled for 8 h under the above mentioned conditions, respectively. 2.2. Fabrication of spherical granules by emulsification process Magnesium phosphate cement pastes were obtained by mixing 10 g Ca0.75Mg2.25(PO4)2 with a (3.5 − x) M (NH4)2HPO4/x M NH4H2PO4 solution (X = 0 to 2.0, Merck, Darmstadt, Germany, Lot# A0367907 314/A564126 728) in a powder to liquid ratio of 2.0 or 3.0 g/mL. The pH values of the different ammonium phosphate solutions were measured using a pH meter (inoLab pH Level 1, WTW, Weilheim, Germany). The cement paste was dispersed either in 300 mL Mygliol 812 (Caesar&Loretz, Hilden, Germany, Lot# 12356906) or safflower oil (Brökelmann&Co, Hamm, Germany), which was modified with 0.2 to 3.0% of either Tween 80 (Merck, Darmstadt, Germany, Lot# K37657061 746) or Span 80 (Sigma Aldrich, Steinheim, Germany, Lot# MKBD4327V) as surfactants. In case of using Brij 35 (Merck, Hohenbrunn, Germany, Lot# S42930762 744) as a surfactant, the compound was directly added to the cement paste as 3% aqueous solution. Dispersion was carried out in a 400 mL glass beaker using a mechanical stirrer RW16 basic (IKA-Werke, Staufen, Germany). Two differently sized half moon stirring blades (dimensions: 60 or 70 × 28 × 3 mm) and three different stirring rates (level 3, 4 or 5 corresponding to a rotational speed of 289, 427 and 556 rpm according to the manufacturer) were chosen for the emulsification process. After ~ 90 min, the cement was set and the granules were washed several times with
131
water and acetone to remove oil and surfactant residues and dried in an oven at 37 °C. Some granules were additionally post-cured in 3.5 M (NH4)2HPO4 solution for 24 h. 2.3. Characterization 2.3.1. Scanning electron microscopy The granules' morphology was analyzed with scanning electron microscopy (DSM 940, Zeiss, Oberkochen, Germany) by using an acceleration voltage of 10 kV. Prior to this, a film of gold was deposited on the samples to avoid electrical charging. 2.3.2. Size distribution The size distribution (wt.%) of the granules was determined by sieving them into fractions of N 2000, 1000, 710, 500, 355 and 200 μm. 2.3.3. X-ray diffraction analysis The degree of conversion to struvite was determined by the addition of 20 wt.% of CaF2 (Merck, Darmstadt, Germany, Lot# 308F652840) as an internal standard according to Alexander et al. [20]. X-ray diffraction patterns of the powder mixtures were recorded on a D5005 diffractometer (Siemens, Karlsruhe, Germany) with Cu-Kα radiation, a voltage of 40 kV and a current of 40 mA to analyze the qualitative and quantitative composition of the granules. Data were collected in a 2θ range from 10 to 40° with a step size of 0.02° and a scan rate of 1.5 s/step. The evaluation of the struvite integral peak intensity at 16.5° and the CaF2 integral peak intensity at 28.3° revealed the degree of conversion. 2.3.4. Mechanical testing For mechanical testing, the above mentioned cement pastes were transferred into silicon rubber molds (dimensions: 12 × 6 × 6 mm). These cuboids were hardened at 37 °C and 100% humidity for 24 h. The compressive strength measurements were performed using the universal testing machine Z010 (Zwick, Ulm, Germany) with a crosshead speed of 1 mm/min. 2.4. Biological testing 2.4.1. Cell culture Human fetal osteoblast cell line hFOB (LGC Standards, Wesel, Germany) was cultured in Dulbecco's Modified Eagle's Medium (DMEM), supplemented with 1% penicillin and streptomycin, 0.3 mg/mL geneticin (G-418 sulfate) and 10% fetal calf serum (FCS, all from Invitrogen Life Technologies, Karlsruhe, Germany). The cells were incubated in a humidified 5% CO2 incubator at 34 °C according to the supplier since they show a faster cell division at this temperature compared with body temperature. 2.4.2. Sample preparation For cytocompatibility measurements, disk-shaped (dimensions: 15 × 2 mm) samples were made from the above mentioned cement paste. Hydroxyapatite (HA) cement disks were taken as control samples and prepared by mixing α-Ca3(PO4)2 and a 2.5% Na2HPO4 (Merck, Darmstadt, Germany, Lot# F1485086 732) solution in a powder to liquid ratio of 3.0 g/mL. All the samples were hardened at 37 °C and 100% humidity for 24 h, washed in phosphate buffered saline for 14 days, disinfected in 70% ethanol and placed in quadruplicate in the wells of a 24-well plate (Nunc, Wiesbaden, Germany). 2.4.3. Cell number and proliferation The cells were seeded on the cement samples with an initial density of 44,000 cells/cm2 and on the polystyrene surfaces with 23,000 cells/cm2, respectively. Because of their higher proliferation rate on polystyrene, less cells were seeded on this material. The cell culture medium was changed on days 3, 5 and 7, while after 4, 6 and 10 days of culture, the cells were both visually counted by means of a 0.4% Trypan blue
132
T. Christel et al. / Materials Science and Engineering C 42 (2014) 130–136
solution (Sigma Chemical, St. Louis, USA) and using the cell counter Cellometer Auto T4 (Nexcellom Bioscience, Lawrence, Massachusetts USA). The cell activity was determined by the cell proliferation reagent WST-1 (Roche Diagnostics, Mannheim, Germany). This reagent is reduced to a soluble orange formazan dye in the cell mitochondria and indicates the amount of produced adenosine triphosphate in the cells and therefore their activity. The cells were incubated with WST-1 supplemented cell culture medium (1:10 mixture) for 30 min at 34 °C and 5% CO2. The adsorption of the supernatant was quantified by a Tecan spectra fluor plus photometer (Tecan, Crailsheim, Germany). The ratio of cell activity and cell number was calculated. 2.4.4. ICP-mass spectrometry To analyze their passive solubility the washed and sterilized diskshaped HA and struvite samples were put in triplicate in a 24-well plate, covered with 10% FCS supplemented DMEM and incubated for 10 days at 37 °C. The medium was changed everyday to avoid supersaturation of ions. The accumulated phosphate, calcium and magnesium content of the used media was measured by mass spectrometer with inductively coupled plasma (ICP-MS, Varian, Darmstadt, Germany) against standard solutions (5 mg/L, 10 mg/L, ICP standard solutions and MgCl2 solution by Merck, Darmstadt/Hohenbrunn, Germany). 2.5. Statistics Analysis of variance was examined by applying the Dunett's test by means of the excel sheet inerSTAT-a (M.H. Vargas, Instituto Nacional de Enfermedades, Mexiko) to compare the results to a control group. 3. Results 3.1. Impact of several processing parameters on the granule size Spherical struvite granules were obtained by the dispersion of a hydraulic cement paste of 10 g Ca0.75Mg2.25(PO4)2 and 5 mL 3.5 M (NH4)2HPO4 solution in safflower oil with 1.0% Tween 80 as a surfactant (Fig. 1a). The granule microstructure showed irregular clinker-like shaped crystals with a size of 5 to 15 μm (Fig. 1b). By varying several processing parameters such as the stirring energy (e.g. by changing the stirring level or blade size), the surfactant type and concentration, and the oil viscosity by altering the oil type (Mygliol 812, dynamic viscosity of 30 mPas or safflower oil, dynamic viscosity of 60 mPas), only a marginal decrease of the granules' size could be realized (Fig. 2). A higher stirring level led to smaller granule diameters (Fig. 2a), but simultaneously promoted a turbulent flow. Also, the use of a bigger stirring blade (Fig. 2b), higher surfactant concentration (Fig. 2d), and increased oil viscosity (Fig. 2e) was not successful in
obtaining smaller granule sizes. Compared to Span 80 and Brij 35, the use of Tween 80 as surfactant led to relatively small granules (Fig. 2c). However, still most of the fabricated granules had a diameter of more than 1000 μm. 3.2. Impact of the aqueous phase composition on the granule size, compressive strength and phase composition The relatively long setting time of the used cement system was identified as the main reason for the unsuccessful attempts to reduce the granule size. This promoted coalescence of the initially small cement droplets to larger granules during the emulsification process. In further experiments, a decrease of the granules' size was achieved by reducing the setting time of the cement system by partially substituting (NH4)2HPO4 in the aqueous solution by NH4H2PO4. With an increasing concentration of NH4H2PO4 in the cement liquid, both the pH value (Table 1) decreased and 97 wt.% of all the resulting granules were now in the size range between 200 and 1000 μm (Fig. 3a). At the same time, the compressive strength of corresponding samples could be doubled without affecting the conversion degree of the raw powder to struvite (Fig. 3b). The stress–strain curves of both a cement sample with 1.5 M and without NH4H2PO4 display a typical course for a brittle material (Fig. 3c). They provide a steep increase with slight strain, but the sample without NH4H2PO4 yet breaks after the half stress. The cement raw material Ca0.75Mg2.25(PO4)2 consisted of farringtonite (Mg3(PO4)2) and stanfieldite (Ca4Mg5(PO4)6) (Fig. 3d iii). The setting reaction with the 2.0 M (NH4)2HPO4/1.5 M NH4H2PO4 solution led to the partial formation of struvite (NH4MgPO4·6H2O) (Fig. 3d ii) which was enhanced by 52% when granules were immersed in 3.5 M (NH4)2HPO4 solution for 24 h (Fig. 3d i). 3.3. Biocompatibility of the optimized material composition and its ion release The biocompatibility of the cement composition leading to an optimum granule size distribution was tested with human fetal osteoblasts (hFOB, Fig. 4). During the whole culture period the cell number was the highest on polystyrene surfaces, whereas it was lower but comparable on HA and struvite surfaces on days 4 and 6 (Fig. 4a). After 10 days, a further loss of osteoblasts was observed on the struvite samples compared to the other materials. In contrast, a non significant fluctuation of cell numbers was observed on HA surfaces. However, within the range of the standard deviation, the normalized activity per cell for the struvite samples remained constant during the culture period of 10 days (Fig. 4b). Cell activities were comparable to those measured on polystyrene on day 4 and day 6 and even higher than the activity on HA on day 6 and day 10 (Fig. 4b). During the same period, a more
Fig. 1. Scanning electron micrographs of granules N710 μm (a) and their microstructure (b). Granules were prepared by dispersing a cement paste of 10 g Ca0.75Mg2.25(PO4)2 and 5 mL 3.5 M (NH4)2HPO4 solution in safflower oil with 1.0% Tween 80. Further parameters were a stirring level of 4 and the 60 mm stirring blade.
T. Christel et al. / Materials Science and Engineering C 42 (2014) 130–136
133
Fig. 2. Struvite granule sizes prepared by dispersing a cement paste of 10 g Ca0.75Mg2.25(PO4)2 and 5 mL 3.5 M (NH4)2HPO4 solution in oil using different (a) stirring rates (with 1.0% Tween 80 in safflower oil), (b) stirring blades (with 3.0% Tween 80 in Mygliol 812), (c) surfactant types (with 3.0% surfactant in safflower oil), (d) Tween 80 concentrations (in safflower oil) or (e) oil types (with 3.0% Tween 80). Further parameters were a stirring level of 4 (b–e) and the 60 mm stirring blade (a, c–e).
than three times higher phosphate content was detected in the cell culture medium (without cells) released by struvite samples compared to HA (Table 2). Simultaneously no Mg2+ ions were released by HA, but more Ca2+ ions were detracted from the medium by this material.
Table 1 pH value of a (3.5−x) M (NH4)2HPO4/x M NH4H2PO4 solution as a function of the NH4H2PO4 concentration (X). NH4H2PO4 concentration X [M]
pH value
0.00 0.50 1.50 1.75 2.00
8.19 7.24 6.31 6.09 5.87
4. Discussion 4.1. Fabrication of granules 4.1.1. Common materials and methods Ceramic granules are commonly used for bone augmentation applications in maxillofacial surgery. Commercially available synthetic granules consist of high-temperature phases such as β-tricalcium phosphate or hydroxyapatite [18], which are synthesized by either sintering pre-granulated powder or by crushing and sieving sintered monoliths [3]. The present study introduces the granulation of struvite ceramics as a low-temperature phase with improved solubility properties. Struvite is naturally found as pathological calcification in kidney stones and is chemically degradable under physiological conditions similar to brushite [11]. A key benefit of the produced granules is their spherical morphology. Irregularly sharp-edged formed granules are supposed to promote irritation and inflammation of surrounding tissues [16,21],
134
T. Christel et al. / Materials Science and Engineering C 42 (2014) 130–136
Fig. 3. (a) Struvite granule sizes prepared by dispersing a cement paste of 10 g Ca0.75Mg2.25(PO4)2 and 5 mL (3.5−x) M (NH4)2HPO4/x M NH4H2PO4 solution in Mygliol 812 with 3.0% Tween 80 using different concentrations of NH4H2PO4. Further parameters were a stirring level of 4 and the 70 mm stirring blade. (b) compressive strength of samples with appropriate compositions after 24 h in the water bath at 37 °C and conversion degree of the granules using 20 wt.% internal standard CaF2. (c) compressive stress–strain curves of exemplary samples with appropriate compositions after 24 h in the water bath at 37 °C (d) X-ray diffraction pattern of Ca0.75Mg2.25(PO4)2 (iii) and of granules that were made of the raw material and a 2.0 M (NH4)2HPO4/1.5 M NH4H2PO4 solution with (i) and without (ii) post-curing in a 3.5 M (NH4)2HPO4 solution for 24 h. The peaks correspond to raw material farringtonite (* PDF-No. 33-0876) and stanfieldite (# PDF-No. 11-0231) or to struvite (° PDF-No. 15-0762). The strong peak at 28.3° belongs to the internal standard CaF2.
even if the shape of granules seemed to have no negative influence on the function of osteoblasts in cell culture [22]. In the case of roundshaped granules, Misiek et al. [21] proved a faster healing in vivo. Another benefit related to their morphology is their improved injectability as demonstrated by Oliveira et al. [23]. 4.1.2. Parameterization of the emulsification process Granulation of cement pastes by using cement-oil suspensions has been recently described by our group for a brushite forming cement
system [16]. However, the adaption of the process to the struvite forming cements of the current study was difficult, since it was practically impossible to dissolve the used surfactants in the highly concentrated ammonium phosphate cement liquid. This problem could be solved by either adding a highly concentrated surfactant solution to the cement paste after mixing, or more efficiently by adding the surfactant directly to the oil phase. Although this worked well in producing spherical struvite granules, it was difficult to obtain granules within a clinically relevant size range of 200 to 1000 μm. By varying different
Fig. 4. (a) Cell number and (b) activity per human fetal osteoblast cell on different surfaces for 4, 6 and 10 days. Struvite samples were made of Ca0.75Mg2.25(PO4)2 and 2.0 M (NH4)2HPO4/1.5 M NH4H2PO4 with a powder to liquid ratio of 2.0 g/mL. Polystyrene and HA surfaces were used as controls.
T. Christel et al. / Materials Science and Engineering C 42 (2014) 130–136 Table 2 Released ions from different samples placed in DMEM + 10% FCS for 10 days. Struvite samples were made of Ca0.75Mg2.25(PO4)2 and 3.5 M (NH4)2HPO4 solution with a powder to liquid ratio of 3.0 g/mL.
DMEM Hydroxyapatite Struvite a
Phosphate as PO3− 4
Mg2+
Ca2+
35 ± 2 ppm 70 ± 2 ppm 157 ± 22 ppma
15 ± 3 ppm 12 ± 1 ppm 108 ± 17 ppma
79 ± 11 ppm 25 ± 2 ppm 39 ± 9 ppm
p b 0.01 compared to hydroxyapatite.
process parameters it was tried to optimize the emulsification procedure in order to preferably produce smaller granules. The application of Mygliol 812 instead of safflower oil showed a shift of the granules' size to larger diameters, because the decrease in viscosity promotes coalescence of the non-set cement droplets. This result is in agreement both with a study of Moseke et al. [16] about the granulation of a brushite forming cement system and with an investigation of Perez et al. [24] about HA granules. In contrast, an increase in application of energy (i.e. by using a bigger stirring blade or higher stirring rate) led to a slight reduction of the granules' diameter. However, caution has to be exercised to avoid turbulences in the oil, which will ultimately lead to big granules with deviating shape according to a study of Perez et al. [24]. This process can by described by the so-called Reynolds number R [24]: 2 R ¼ ρ N D =μ where ρ is the oil density, N the number of rotations, D the diameter of the vessel and μ the oil viscosity. The flow is laminar when R b 2300, otherwise it is turbulent [24]. A high number of rotation and a low oil viscosity results in high Reynolds numbers and promote turbulent flow. The analysis of different surfactants revealed that Tween 80 matches most with the applied cement system, although Tween 80 has an hydrophilic lipophilic balance (HLB) value of 15 [25]. Therefore, it should actually be more suitable as a surfactant for an oil-in-water emulsion [26]. The HLB value describes the ratio of water soluble and oil soluble compartments and indicates the type of resulting emulsion. However, the surfactants were added to the oil phase of the emulsion. Hence, the predominant hydrophilic molecules most likely interacted more with the hydrophilic cement paste to stabilize cement droplets in oil rather than oil droplets in the cement phase. 4.1.3. Fabrication of possibly small granules So far, the presented attempts to fabricate granules smaller than 1000 μm had only little success. Although initially smaller cement paste granules were formed when the paste was dropped into the oil, these small droplets showed a coalescence effect to form much larger granules due to the relatively long setting time of the cement. To avoid this problem, the setting time of the struvite cement had to be decreased. This was recently demonstrated by Moseke et al. [16] who showed that a brushite forming CPC with shorter setting time leads to the fabrication of smaller granules. In contrast to brushite forming cements, where simply the absence of setting retarders leads to setting times of a few seconds [4], struvite cements are generally slow-setting cement systems. A successful reduction of the setting time was achieved by partially replacing the secondary ammonium phosphate (NH4)2HPO4 with primary phosphate NH4H2PO4. By this modification of the liquid cement phase, a significant reduction of the granules' size was realized and most of the fabricated granules exhibited sizes in the range between 200 and 1000 μm. The impact of this compositional change of the cement on other material characteristics is also fundamental with respect to its application. The compressive strength and degree of conversion to struvite can have an important influence on the stability and solubility of the implanted granules. The combination of Ca0.75Mg2.25(PO4)2 with a 2.0 M (NH4)2HPO4/1.5 M
135
NH4H2PO4 solution not only led to the fabrication of granules within the desired size range, but also to a material with improved mechanical properties without affecting the struvite conversion. Besides, the latter can be improved by a post-curing process. 4.2. Evaluation of the biological performance On the analyzed struvite surfaces, the activity of single human fetal osteoblasts (hFOB) was, considering the standard deviations, similar to osteoblasts on polystyrene and partially better than on HA surfaces. The lower cell number and activity per cell on the HA surfaces at day 6 may be a result of a decreased calcium ion concentration in the medium, as it is known that calcium deficient HA will remove these ions from solution and has therefore a detrimental effect on the cell viability grown on these surfaces. It is a well-known fact that both Mg2 + and Ca2+ promote proliferation and differentiation of osteoblasts and play an important role concerning bone deposition and mineralization [7]. Witte et al. [27] showed that implants consisting of Mg stimulated the formation of new bone in vivo. In contrast to HA cement, a high release rate of Mg2 + ions by struvite samples was observed in our study. In addition, less Ca2+ ions were detracted from the cell culture medium and probably adsorbed by struvite. The removal of Ca2+ ions from the surrounding medium is also described in literature for the case of HA and is linked to its recrystallization into stoichiometric HA [5]. In spite of the high activity on struvite surfaces, only few cells were counted on this material. The low cell number can be explained by the high solubility of the material as shown by mass spectrometric analysis. Gradually, the struvite surfaces consist more and more of loose cement particles. If a cell adheres on such a particle that separates from the bulk, the cell will be removed during change of medium and washing procedures. In vivo, this effect is of minor importance, because the cells are in their native environment and are steadily recruited to the defect site. Thus, the bone healing can progress simultaneously to the resorption of the implant by passive and active mechanisms. Furthermore, the biocompatibility of struvite based mineral systems was not only proved in many cases by means of MG36 cell line [6,7,11,28], but also verified in vivo [12]. Klammert et al. [12] already showed that an implant on the basis of Ca0.75Mg2.25(PO4)2 and 3.5 M (NH4)2HPO4 solution stays in the animal model for at least 15 months without the appearance of infection, inflammation or rejection. 5. Conclusions The presented emulsion technique was proven to be an effective method to produce spherical granules on the basis of a hydraulic struvite forming cement paste. Among the many parameters investigated, the setting time of the cements was found to be the key factor controlling the final size of the cement granules. Granules with a clinically relevant size range could be fabricated by adjusting the setting time. Acknowledgments The authors would like to acknowledge financial support from the Federal Ministry of Education and Research (BMBF) under the grant number 13EZ1208B (NAKKRO). References [1] P.V. Giannoudis, H. Dinopoulos, E. Tsiridis, Injury 36 (2005) S20–S27. [2] C.J. Damien, J.R. Parsons, J. Appl. Biomater. 2 (1991) 187–208. [3] F.M. Tamimi, J. Torres, I. Tresguerres, C. Clemente, E. López-Cabarcos, L.J. Blanco, J. Clin. Periodontol. 33 (2006) 922–928. [4] M. Bohner, Injury 31 (Supplement 4) (2000) D37–D47. [5] C. Großardt, A. Ewald, L.M. Grover, J.E. Barralet, U. Gbureck, Tissue Eng. A 16 (2010) 3687–3695. [6] J. Wei, J. Jia, F. Wu, S. Wei, H. Zhou, H. Zhang, J.-W. Shin, C. Liu, Biomaterials 31 (2010) 1260–1269. [7] F. Wu, J. Wei, H. Guo, F. Chen, H. Hong, C. Liu, Acta Biomater. 4 (2008) 1873–1884.
136 [8] [9] [10] [11] [12] [13] [14] [15] [16] [17]
T. Christel et al. / Materials Science and Engineering C 42 (2014) 130–136 G. Mestres, M.-P. Ginebra, Acta Biomater. 7 (2011) 1853–1861. K. Kowalczyk, G. Palavit, J. Mater. Sci. Lett. 12 (1993) 1276–1278. D.V. Ribeiro, M.R. Morelli, Mater. Res. 12 (2009) 51–56. E. Vorndran, A. Ewald, F. Müller, K. Zorn, A. Kufner, U. Gbureck, J. Mater. Sci. Mater. Med. 22 (2011) 429–436. U. Klammert, A. Ignatius, U. Wolfram, T. Reuther, U. Gbureck, Acta Biomater. 7 (2011) 3469–3475. H. Zhou, T.F. Luchini, S. Bhaduri, J. Mater. Sci. Mater. Med. 23 (2012) 2831– 2837. K.J. Lilley, U. Gbureck, J.C. Knowles, D.F. Farrar, J.E. Barralet, J. Mater. Sci. Mater. Med. 16 (2005) 455–460. J. Lu, J. Wei, Y. Yan, H. Li, J. Jia, S. Wei, H. Guo, T. Xiao, C. Liu, J. Mater. Sci. Mater. Med. 22 (2011) 607–615. C. Moseke, C. Bayer, E. Vorndran, J. Barralet, J. Groll, U. Gbureck, J. Mater. Sci. Mater. Med. 23 (2012) 2631–2637. W. Paul, C.P. Sharma, J. Mater. Sci. Mater. Med. 10 (1999) 383–388.
[18] M. Bohner, S. Tadier, N. van Garderen, A. de Gasparo, N. Döbelin, G. Baroud, Biomatter 3 (2013) e25103. [19] S. Singh, V. Singh, S. Aggarwal, U. Mandal, Chem. Pap. 64 (2010) 491–498. [20] L. Alexander, H.P. Klug, Anal. Chem. 20 (1948) 886–889. [21] D.J. Misiek, J.N. Kent, R.F. Carr, J. Oral Maxillofac. Surg. 42 (1984) 150–160. [22] D.P. Pioletti, H. Takei, T. Lin, P. Van Landuyt, Q. Jun Ma, S. Yong Kwon, K.L. Paul Sung, Biomaterials 21 (2000) 1103–1114. [23] S.M. Oliveira, C.C. Barrias, I.F. Almeida, P.C. Costa, M.R.P. Ferreira, M.F. Bahia, M.A. Barbosa, J. Biomed. Mater. Res. B Appl. Biomater. 87B (2008) 49–58. [24] R.A. Perez, S. Del Valle, G. Altankov, M.-P. Ginebra, J. Biomed. Mater. Res. B Appl. Biomater. 97B (2011) 156–166. [25] G. Ben-Et, D. Tatarsky, J. Am. Oil Chem. Soc. 49 (1972) 499–500. [26] W.C. Griffin, J. Soc. Cosmet. Chem. 1 (1949) 311–326. [27] F. Witte, H. Ulrich, C. Palm, E. Willbold, J. Biomed. Mater. Res. A 81A (2007) 757–765. [28] A. Ewald, K. Helmschrott, G. Knebl, N. Mehrban, L.M. Grover, U. Gbureck, J. Biomed. Mater. Res. B Appl. Biomater. 96B (2011) 326–332.
Contents lists available at ScienceDirect
Materials Science and Engineering C journal homepage: www.elsevier.com/locate/msec
Fabrication and cytocompatibility of spherical magnesium ammonium phosphate granules Theresa Christel a, Martha Geffers a, Uwe Klammert b, Berthold Nies c, Andreas Höß c, Jürgen Groll a, Alexander C. Kübler b, Uwe Gbureck a,⁎ a b c
Department for Functional Materials in Medicine and Dentistry, University of Würzburg, Pleicherwall 2, 97070 Würzburg, Germany Department of Cranio-Maxillo-Facial Surgery, University of Würzburg, Pleicherwall 2, 97070 Würzburg, Germany InnoTERE GmbH, Pharmapark Radebeul, Meissner Straße 191, 01455 Radebeul, Germany
a r t i c l e
i n f o
Article history: Received 14 January 2014 Received in revised form 31 March 2014 Accepted 6 May 2014 Available online 22 May 2014 Keywords: Calcium magnesium phosphate cement Struvite Granules
a b s t r a c t Magnesium phosphate compounds, as for example struvite (MgNH4PO4·6H2O), have comparable characteristics to calcium phosphate bone substitutes, but degrade faster under physiological conditions. In the present work, we used a struvite forming calcium doped magnesium phosphate cement with the formulation Ca0.75Mg2.25(PO4)2 and an ammonium phosphate containing aqueous solution to produce round-shaped granules. For the fabrication of spherical granules, the cement paste was dispersed in a lipophilic liquid and stabilized by surfactants. The granules were characterized with respect to morphology, size distribution, phase composition, compressive strength, biocompatibility and solubility. In general, it was seen that small granules can hardly be produced by means of emulsification, when the raw material is a hydraulic paste, because long setting times promote coalescence of initially small unhardened cement droplets. Here, this problem was solved by using an aqueous solution containing both the secondary (NH4)2HPO4 and primary ammonium phosphates NH4H2PO4 to accelerate the setting reaction. This resulted in granules with 97 wt.% having a size in the range between 200 and 1000 μm. The novel solution composition doubled the compressive strength of the cement to 37 ± 5 MPa without affecting either the conversion to struvite or the cytocompatibility using human fetal osteoblasts. © 2014 Elsevier B.V. All rights reserved.
1. Introduction The use of autografts for the repair of bone defects is considered as “gold standard”, but has several drawbacks such as a lack of availability and its donor site morbidity [1,2]. Alternative synthetic bone replacement materials are mostly based on calcium phosphate chemistry. Material approaches cover both the use of hydroxyapatite derived from natural resources (e.g. BioOss®) [3] as well as fully synthetic hydroxyapatites (HA) or tricalcium phosphates (TCP) prepared by sintering, precipitation or a cement setting reaction [4]. Depending on the chemical composition and crystal size, these materials are either stable (HA) or are able to slowly degrade in vivo (TCP) by active and passive mechanisms such that host autologous bone can grow into the bone defect [5]. Novel material approaches to enhance the degradation ability of bioceramics for bone replacement involve the use of magnesium phosphate compounds [6]. Such materials were found to be as
⁎ Corresponding author. Tel.: +49 931 20173550. E-mail addresses: [email protected] (T. Christel), [email protected] (M. Geffers), [email protected] (U. Klammert), [email protected] (B. Nies), [email protected] (A. Höß), [email protected] (J. Groll), [email protected] (A.C. Kübler), [email protected] (U. Gbureck).
http://dx.doi.org/10.1016/j.msec.2014.05.023 0928-4931/© 2014 Elsevier B.V. All rights reserved.
cytocompatible as calcium phosphates, but offer a faster chemical degradation due to their higher solubility under physiological conditions [7]. Various magnesium phosphate compounds have been investigated in recent studies, e.g. struvite (MgNH4PO4·6H2O) [8–11], newberyite (MgHPO4·3H2O) [12], amorphous magnesium phosphates [13] or magnesium doped calcium phosphates [7,14,15]. The formation of these compounds is often based on a cementitious reaction of either MgO or Mg3(PO4)2 with aqueous sources of phosphate and ammonium salts. The solubility profiles of these cements may be altered by using biphasic calcium magnesium phosphate mixtures. This was already demonstrated by Vorndran et al. [11], who used compounds with the general formula CaxMg(3−x)(PO4)2 allowing to adjust the setting properties as well as the mechanical performance and solubility. By using the formulation Ca0.75Mg2.25(PO4)2 together with a 3.5 M (NH4)2HPO4 solution, a good biocompatibility in vitro [11] and in vivo [12], a high mechanical strength of ~ 80 MPa and an adequate setting time of 14 min were obtained [11]. In the present work, a cement based on Ca0.75Mg2.25(PO4)2 was used to fabricate pre-hardened cement granules since in dental surgery (e.g. sinus lift operation, socket preservation, filling of jaw cysts) the application of bone substitutes in granular form is a typical approach [16]. An easy way to generate granules on the basis of a hydraulic cement paste is the hardening of a cement monolith followed by
T. Christel et al. / Materials Science and Engineering C 42 (2014) 130–136
mechanical grinding [3]. However, this technique leads to irregular sharp-edged granules that can promote irritation and inflammation in living tissues [16]. Another approach used a mixture of precipitated HA and chitosan, whereas the granules were hardened after starting the emulsification process by the addition of a cross-linker leading to granules with sizes in the range of 212 to 1000 μm [17]. However, in this case HA was not formed on the basis of a calcium phosphate cement (CPC) setting reaction but added as an inert filler to the biopolymer matrix prior to emulsification. Other techniques such as precipitation reactions would actually result in the fabrication of nanoscaled granules [18]. For example, Singh et al. [19] produced brushite nanoparticles by the combination of micro-emulsification with the process of precipitation. In the present study, a more sophisticated approach was applied. It is based on dispersing the hydrophilic cement paste in a hydrophobic liquid (e.g. oil) by stirring while stabilizing the formed cement-oil dispersion using surfactants. The influence of different emulsification parameters (e.g. stirring energy, oil type, surfactant type and concentration, and aqueous solution composition) on the granules' properties (morphology, size distribution, degree of conversion, phase composition, cytocompatibility, and solubility) were tested. The main objective was the fabrication of spherical granules with clinically suitable diameters of 200 to 1000 μm. 2. Materials and methods 2.1. Fabrication of cement raw materials Ca0.75Mg2.25(PO4)2 was prepared by sintering 1.5 mol MgHPO4·3H2O (Sigma Aldrich, Steinheim, Germany, Lot# BCBF1937V) with 0.5 mol CaHPO4 (J.T. Baker, Griesheim, Germany, Lot# K06 587), 0.25 mol CaCO3 (Merck, Darmstadt, Germany, Lot# A0074720 927) and 0.75 mol Mg(OH)2 (Fluka, Steinheim, Germany, Lot# 10H040006) at 1100 °C for 5 h in a sintering furnace (Oyten Thermotechnic, Oyten, Germany). The product was crushed with a pestle and mortar, passed through a 355 μm sieve and milled in a planetary ball mill (Retsch, Haan, Germany) at 200 rpm with 500 mL agate jars, four agate balls (30 mm) and a load of 125 g powder per jar for up to 4 h. α-Ca3(PO4)2 was synthesized by heating a mixture of 2 mol CaHPO4 and 1 mol CaCO3 to 1400 °C for 5 h. The product was crushed and sieved using a 355 μm sieve. 125 g of the powder was milled for 8 h under the above mentioned conditions, respectively. 2.2. Fabrication of spherical granules by emulsification process Magnesium phosphate cement pastes were obtained by mixing 10 g Ca0.75Mg2.25(PO4)2 with a (3.5 − x) M (NH4)2HPO4/x M NH4H2PO4 solution (X = 0 to 2.0, Merck, Darmstadt, Germany, Lot# A0367907 314/A564126 728) in a powder to liquid ratio of 2.0 or 3.0 g/mL. The pH values of the different ammonium phosphate solutions were measured using a pH meter (inoLab pH Level 1, WTW, Weilheim, Germany). The cement paste was dispersed either in 300 mL Mygliol 812 (Caesar&Loretz, Hilden, Germany, Lot# 12356906) or safflower oil (Brökelmann&Co, Hamm, Germany), which was modified with 0.2 to 3.0% of either Tween 80 (Merck, Darmstadt, Germany, Lot# K37657061 746) or Span 80 (Sigma Aldrich, Steinheim, Germany, Lot# MKBD4327V) as surfactants. In case of using Brij 35 (Merck, Hohenbrunn, Germany, Lot# S42930762 744) as a surfactant, the compound was directly added to the cement paste as 3% aqueous solution. Dispersion was carried out in a 400 mL glass beaker using a mechanical stirrer RW16 basic (IKA-Werke, Staufen, Germany). Two differently sized half moon stirring blades (dimensions: 60 or 70 × 28 × 3 mm) and three different stirring rates (level 3, 4 or 5 corresponding to a rotational speed of 289, 427 and 556 rpm according to the manufacturer) were chosen for the emulsification process. After ~ 90 min, the cement was set and the granules were washed several times with
131
water and acetone to remove oil and surfactant residues and dried in an oven at 37 °C. Some granules were additionally post-cured in 3.5 M (NH4)2HPO4 solution for 24 h. 2.3. Characterization 2.3.1. Scanning electron microscopy The granules' morphology was analyzed with scanning electron microscopy (DSM 940, Zeiss, Oberkochen, Germany) by using an acceleration voltage of 10 kV. Prior to this, a film of gold was deposited on the samples to avoid electrical charging. 2.3.2. Size distribution The size distribution (wt.%) of the granules was determined by sieving them into fractions of N 2000, 1000, 710, 500, 355 and 200 μm. 2.3.3. X-ray diffraction analysis The degree of conversion to struvite was determined by the addition of 20 wt.% of CaF2 (Merck, Darmstadt, Germany, Lot# 308F652840) as an internal standard according to Alexander et al. [20]. X-ray diffraction patterns of the powder mixtures were recorded on a D5005 diffractometer (Siemens, Karlsruhe, Germany) with Cu-Kα radiation, a voltage of 40 kV and a current of 40 mA to analyze the qualitative and quantitative composition of the granules. Data were collected in a 2θ range from 10 to 40° with a step size of 0.02° and a scan rate of 1.5 s/step. The evaluation of the struvite integral peak intensity at 16.5° and the CaF2 integral peak intensity at 28.3° revealed the degree of conversion. 2.3.4. Mechanical testing For mechanical testing, the above mentioned cement pastes were transferred into silicon rubber molds (dimensions: 12 × 6 × 6 mm). These cuboids were hardened at 37 °C and 100% humidity for 24 h. The compressive strength measurements were performed using the universal testing machine Z010 (Zwick, Ulm, Germany) with a crosshead speed of 1 mm/min. 2.4. Biological testing 2.4.1. Cell culture Human fetal osteoblast cell line hFOB (LGC Standards, Wesel, Germany) was cultured in Dulbecco's Modified Eagle's Medium (DMEM), supplemented with 1% penicillin and streptomycin, 0.3 mg/mL geneticin (G-418 sulfate) and 10% fetal calf serum (FCS, all from Invitrogen Life Technologies, Karlsruhe, Germany). The cells were incubated in a humidified 5% CO2 incubator at 34 °C according to the supplier since they show a faster cell division at this temperature compared with body temperature. 2.4.2. Sample preparation For cytocompatibility measurements, disk-shaped (dimensions: 15 × 2 mm) samples were made from the above mentioned cement paste. Hydroxyapatite (HA) cement disks were taken as control samples and prepared by mixing α-Ca3(PO4)2 and a 2.5% Na2HPO4 (Merck, Darmstadt, Germany, Lot# F1485086 732) solution in a powder to liquid ratio of 3.0 g/mL. All the samples were hardened at 37 °C and 100% humidity for 24 h, washed in phosphate buffered saline for 14 days, disinfected in 70% ethanol and placed in quadruplicate in the wells of a 24-well plate (Nunc, Wiesbaden, Germany). 2.4.3. Cell number and proliferation The cells were seeded on the cement samples with an initial density of 44,000 cells/cm2 and on the polystyrene surfaces with 23,000 cells/cm2, respectively. Because of their higher proliferation rate on polystyrene, less cells were seeded on this material. The cell culture medium was changed on days 3, 5 and 7, while after 4, 6 and 10 days of culture, the cells were both visually counted by means of a 0.4% Trypan blue
132
T. Christel et al. / Materials Science and Engineering C 42 (2014) 130–136
solution (Sigma Chemical, St. Louis, USA) and using the cell counter Cellometer Auto T4 (Nexcellom Bioscience, Lawrence, Massachusetts USA). The cell activity was determined by the cell proliferation reagent WST-1 (Roche Diagnostics, Mannheim, Germany). This reagent is reduced to a soluble orange formazan dye in the cell mitochondria and indicates the amount of produced adenosine triphosphate in the cells and therefore their activity. The cells were incubated with WST-1 supplemented cell culture medium (1:10 mixture) for 30 min at 34 °C and 5% CO2. The adsorption of the supernatant was quantified by a Tecan spectra fluor plus photometer (Tecan, Crailsheim, Germany). The ratio of cell activity and cell number was calculated. 2.4.4. ICP-mass spectrometry To analyze their passive solubility the washed and sterilized diskshaped HA and struvite samples were put in triplicate in a 24-well plate, covered with 10% FCS supplemented DMEM and incubated for 10 days at 37 °C. The medium was changed everyday to avoid supersaturation of ions. The accumulated phosphate, calcium and magnesium content of the used media was measured by mass spectrometer with inductively coupled plasma (ICP-MS, Varian, Darmstadt, Germany) against standard solutions (5 mg/L, 10 mg/L, ICP standard solutions and MgCl2 solution by Merck, Darmstadt/Hohenbrunn, Germany). 2.5. Statistics Analysis of variance was examined by applying the Dunett's test by means of the excel sheet inerSTAT-a (M.H. Vargas, Instituto Nacional de Enfermedades, Mexiko) to compare the results to a control group. 3. Results 3.1. Impact of several processing parameters on the granule size Spherical struvite granules were obtained by the dispersion of a hydraulic cement paste of 10 g Ca0.75Mg2.25(PO4)2 and 5 mL 3.5 M (NH4)2HPO4 solution in safflower oil with 1.0% Tween 80 as a surfactant (Fig. 1a). The granule microstructure showed irregular clinker-like shaped crystals with a size of 5 to 15 μm (Fig. 1b). By varying several processing parameters such as the stirring energy (e.g. by changing the stirring level or blade size), the surfactant type and concentration, and the oil viscosity by altering the oil type (Mygliol 812, dynamic viscosity of 30 mPas or safflower oil, dynamic viscosity of 60 mPas), only a marginal decrease of the granules' size could be realized (Fig. 2). A higher stirring level led to smaller granule diameters (Fig. 2a), but simultaneously promoted a turbulent flow. Also, the use of a bigger stirring blade (Fig. 2b), higher surfactant concentration (Fig. 2d), and increased oil viscosity (Fig. 2e) was not successful in
obtaining smaller granule sizes. Compared to Span 80 and Brij 35, the use of Tween 80 as surfactant led to relatively small granules (Fig. 2c). However, still most of the fabricated granules had a diameter of more than 1000 μm. 3.2. Impact of the aqueous phase composition on the granule size, compressive strength and phase composition The relatively long setting time of the used cement system was identified as the main reason for the unsuccessful attempts to reduce the granule size. This promoted coalescence of the initially small cement droplets to larger granules during the emulsification process. In further experiments, a decrease of the granules' size was achieved by reducing the setting time of the cement system by partially substituting (NH4)2HPO4 in the aqueous solution by NH4H2PO4. With an increasing concentration of NH4H2PO4 in the cement liquid, both the pH value (Table 1) decreased and 97 wt.% of all the resulting granules were now in the size range between 200 and 1000 μm (Fig. 3a). At the same time, the compressive strength of corresponding samples could be doubled without affecting the conversion degree of the raw powder to struvite (Fig. 3b). The stress–strain curves of both a cement sample with 1.5 M and without NH4H2PO4 display a typical course for a brittle material (Fig. 3c). They provide a steep increase with slight strain, but the sample without NH4H2PO4 yet breaks after the half stress. The cement raw material Ca0.75Mg2.25(PO4)2 consisted of farringtonite (Mg3(PO4)2) and stanfieldite (Ca4Mg5(PO4)6) (Fig. 3d iii). The setting reaction with the 2.0 M (NH4)2HPO4/1.5 M NH4H2PO4 solution led to the partial formation of struvite (NH4MgPO4·6H2O) (Fig. 3d ii) which was enhanced by 52% when granules were immersed in 3.5 M (NH4)2HPO4 solution for 24 h (Fig. 3d i). 3.3. Biocompatibility of the optimized material composition and its ion release The biocompatibility of the cement composition leading to an optimum granule size distribution was tested with human fetal osteoblasts (hFOB, Fig. 4). During the whole culture period the cell number was the highest on polystyrene surfaces, whereas it was lower but comparable on HA and struvite surfaces on days 4 and 6 (Fig. 4a). After 10 days, a further loss of osteoblasts was observed on the struvite samples compared to the other materials. In contrast, a non significant fluctuation of cell numbers was observed on HA surfaces. However, within the range of the standard deviation, the normalized activity per cell for the struvite samples remained constant during the culture period of 10 days (Fig. 4b). Cell activities were comparable to those measured on polystyrene on day 4 and day 6 and even higher than the activity on HA on day 6 and day 10 (Fig. 4b). During the same period, a more
Fig. 1. Scanning electron micrographs of granules N710 μm (a) and their microstructure (b). Granules were prepared by dispersing a cement paste of 10 g Ca0.75Mg2.25(PO4)2 and 5 mL 3.5 M (NH4)2HPO4 solution in safflower oil with 1.0% Tween 80. Further parameters were a stirring level of 4 and the 60 mm stirring blade.
T. Christel et al. / Materials Science and Engineering C 42 (2014) 130–136
133
Fig. 2. Struvite granule sizes prepared by dispersing a cement paste of 10 g Ca0.75Mg2.25(PO4)2 and 5 mL 3.5 M (NH4)2HPO4 solution in oil using different (a) stirring rates (with 1.0% Tween 80 in safflower oil), (b) stirring blades (with 3.0% Tween 80 in Mygliol 812), (c) surfactant types (with 3.0% surfactant in safflower oil), (d) Tween 80 concentrations (in safflower oil) or (e) oil types (with 3.0% Tween 80). Further parameters were a stirring level of 4 (b–e) and the 60 mm stirring blade (a, c–e).
than three times higher phosphate content was detected in the cell culture medium (without cells) released by struvite samples compared to HA (Table 2). Simultaneously no Mg2+ ions were released by HA, but more Ca2+ ions were detracted from the medium by this material.
Table 1 pH value of a (3.5−x) M (NH4)2HPO4/x M NH4H2PO4 solution as a function of the NH4H2PO4 concentration (X). NH4H2PO4 concentration X [M]
pH value
0.00 0.50 1.50 1.75 2.00
8.19 7.24 6.31 6.09 5.87
4. Discussion 4.1. Fabrication of granules 4.1.1. Common materials and methods Ceramic granules are commonly used for bone augmentation applications in maxillofacial surgery. Commercially available synthetic granules consist of high-temperature phases such as β-tricalcium phosphate or hydroxyapatite [18], which are synthesized by either sintering pre-granulated powder or by crushing and sieving sintered monoliths [3]. The present study introduces the granulation of struvite ceramics as a low-temperature phase with improved solubility properties. Struvite is naturally found as pathological calcification in kidney stones and is chemically degradable under physiological conditions similar to brushite [11]. A key benefit of the produced granules is their spherical morphology. Irregularly sharp-edged formed granules are supposed to promote irritation and inflammation of surrounding tissues [16,21],
134
T. Christel et al. / Materials Science and Engineering C 42 (2014) 130–136
Fig. 3. (a) Struvite granule sizes prepared by dispersing a cement paste of 10 g Ca0.75Mg2.25(PO4)2 and 5 mL (3.5−x) M (NH4)2HPO4/x M NH4H2PO4 solution in Mygliol 812 with 3.0% Tween 80 using different concentrations of NH4H2PO4. Further parameters were a stirring level of 4 and the 70 mm stirring blade. (b) compressive strength of samples with appropriate compositions after 24 h in the water bath at 37 °C and conversion degree of the granules using 20 wt.% internal standard CaF2. (c) compressive stress–strain curves of exemplary samples with appropriate compositions after 24 h in the water bath at 37 °C (d) X-ray diffraction pattern of Ca0.75Mg2.25(PO4)2 (iii) and of granules that were made of the raw material and a 2.0 M (NH4)2HPO4/1.5 M NH4H2PO4 solution with (i) and without (ii) post-curing in a 3.5 M (NH4)2HPO4 solution for 24 h. The peaks correspond to raw material farringtonite (* PDF-No. 33-0876) and stanfieldite (# PDF-No. 11-0231) or to struvite (° PDF-No. 15-0762). The strong peak at 28.3° belongs to the internal standard CaF2.
even if the shape of granules seemed to have no negative influence on the function of osteoblasts in cell culture [22]. In the case of roundshaped granules, Misiek et al. [21] proved a faster healing in vivo. Another benefit related to their morphology is their improved injectability as demonstrated by Oliveira et al. [23]. 4.1.2. Parameterization of the emulsification process Granulation of cement pastes by using cement-oil suspensions has been recently described by our group for a brushite forming cement
system [16]. However, the adaption of the process to the struvite forming cements of the current study was difficult, since it was practically impossible to dissolve the used surfactants in the highly concentrated ammonium phosphate cement liquid. This problem could be solved by either adding a highly concentrated surfactant solution to the cement paste after mixing, or more efficiently by adding the surfactant directly to the oil phase. Although this worked well in producing spherical struvite granules, it was difficult to obtain granules within a clinically relevant size range of 200 to 1000 μm. By varying different
Fig. 4. (a) Cell number and (b) activity per human fetal osteoblast cell on different surfaces for 4, 6 and 10 days. Struvite samples were made of Ca0.75Mg2.25(PO4)2 and 2.0 M (NH4)2HPO4/1.5 M NH4H2PO4 with a powder to liquid ratio of 2.0 g/mL. Polystyrene and HA surfaces were used as controls.
T. Christel et al. / Materials Science and Engineering C 42 (2014) 130–136 Table 2 Released ions from different samples placed in DMEM + 10% FCS for 10 days. Struvite samples were made of Ca0.75Mg2.25(PO4)2 and 3.5 M (NH4)2HPO4 solution with a powder to liquid ratio of 3.0 g/mL.
DMEM Hydroxyapatite Struvite a
Phosphate as PO3− 4
Mg2+
Ca2+
35 ± 2 ppm 70 ± 2 ppm 157 ± 22 ppma
15 ± 3 ppm 12 ± 1 ppm 108 ± 17 ppma
79 ± 11 ppm 25 ± 2 ppm 39 ± 9 ppm
p b 0.01 compared to hydroxyapatite.
process parameters it was tried to optimize the emulsification procedure in order to preferably produce smaller granules. The application of Mygliol 812 instead of safflower oil showed a shift of the granules' size to larger diameters, because the decrease in viscosity promotes coalescence of the non-set cement droplets. This result is in agreement both with a study of Moseke et al. [16] about the granulation of a brushite forming cement system and with an investigation of Perez et al. [24] about HA granules. In contrast, an increase in application of energy (i.e. by using a bigger stirring blade or higher stirring rate) led to a slight reduction of the granules' diameter. However, caution has to be exercised to avoid turbulences in the oil, which will ultimately lead to big granules with deviating shape according to a study of Perez et al. [24]. This process can by described by the so-called Reynolds number R [24]: 2 R ¼ ρ N D =μ where ρ is the oil density, N the number of rotations, D the diameter of the vessel and μ the oil viscosity. The flow is laminar when R b 2300, otherwise it is turbulent [24]. A high number of rotation and a low oil viscosity results in high Reynolds numbers and promote turbulent flow. The analysis of different surfactants revealed that Tween 80 matches most with the applied cement system, although Tween 80 has an hydrophilic lipophilic balance (HLB) value of 15 [25]. Therefore, it should actually be more suitable as a surfactant for an oil-in-water emulsion [26]. The HLB value describes the ratio of water soluble and oil soluble compartments and indicates the type of resulting emulsion. However, the surfactants were added to the oil phase of the emulsion. Hence, the predominant hydrophilic molecules most likely interacted more with the hydrophilic cement paste to stabilize cement droplets in oil rather than oil droplets in the cement phase. 4.1.3. Fabrication of possibly small granules So far, the presented attempts to fabricate granules smaller than 1000 μm had only little success. Although initially smaller cement paste granules were formed when the paste was dropped into the oil, these small droplets showed a coalescence effect to form much larger granules due to the relatively long setting time of the cement. To avoid this problem, the setting time of the struvite cement had to be decreased. This was recently demonstrated by Moseke et al. [16] who showed that a brushite forming CPC with shorter setting time leads to the fabrication of smaller granules. In contrast to brushite forming cements, where simply the absence of setting retarders leads to setting times of a few seconds [4], struvite cements are generally slow-setting cement systems. A successful reduction of the setting time was achieved by partially replacing the secondary ammonium phosphate (NH4)2HPO4 with primary phosphate NH4H2PO4. By this modification of the liquid cement phase, a significant reduction of the granules' size was realized and most of the fabricated granules exhibited sizes in the range between 200 and 1000 μm. The impact of this compositional change of the cement on other material characteristics is also fundamental with respect to its application. The compressive strength and degree of conversion to struvite can have an important influence on the stability and solubility of the implanted granules. The combination of Ca0.75Mg2.25(PO4)2 with a 2.0 M (NH4)2HPO4/1.5 M
135
NH4H2PO4 solution not only led to the fabrication of granules within the desired size range, but also to a material with improved mechanical properties without affecting the struvite conversion. Besides, the latter can be improved by a post-curing process. 4.2. Evaluation of the biological performance On the analyzed struvite surfaces, the activity of single human fetal osteoblasts (hFOB) was, considering the standard deviations, similar to osteoblasts on polystyrene and partially better than on HA surfaces. The lower cell number and activity per cell on the HA surfaces at day 6 may be a result of a decreased calcium ion concentration in the medium, as it is known that calcium deficient HA will remove these ions from solution and has therefore a detrimental effect on the cell viability grown on these surfaces. It is a well-known fact that both Mg2 + and Ca2+ promote proliferation and differentiation of osteoblasts and play an important role concerning bone deposition and mineralization [7]. Witte et al. [27] showed that implants consisting of Mg stimulated the formation of new bone in vivo. In contrast to HA cement, a high release rate of Mg2 + ions by struvite samples was observed in our study. In addition, less Ca2+ ions were detracted from the cell culture medium and probably adsorbed by struvite. The removal of Ca2+ ions from the surrounding medium is also described in literature for the case of HA and is linked to its recrystallization into stoichiometric HA [5]. In spite of the high activity on struvite surfaces, only few cells were counted on this material. The low cell number can be explained by the high solubility of the material as shown by mass spectrometric analysis. Gradually, the struvite surfaces consist more and more of loose cement particles. If a cell adheres on such a particle that separates from the bulk, the cell will be removed during change of medium and washing procedures. In vivo, this effect is of minor importance, because the cells are in their native environment and are steadily recruited to the defect site. Thus, the bone healing can progress simultaneously to the resorption of the implant by passive and active mechanisms. Furthermore, the biocompatibility of struvite based mineral systems was not only proved in many cases by means of MG36 cell line [6,7,11,28], but also verified in vivo [12]. Klammert et al. [12] already showed that an implant on the basis of Ca0.75Mg2.25(PO4)2 and 3.5 M (NH4)2HPO4 solution stays in the animal model for at least 15 months without the appearance of infection, inflammation or rejection. 5. Conclusions The presented emulsion technique was proven to be an effective method to produce spherical granules on the basis of a hydraulic struvite forming cement paste. Among the many parameters investigated, the setting time of the cements was found to be the key factor controlling the final size of the cement granules. Granules with a clinically relevant size range could be fabricated by adjusting the setting time. Acknowledgments The authors would like to acknowledge financial support from the Federal Ministry of Education and Research (BMBF) under the grant number 13EZ1208B (NAKKRO). References [1] P.V. Giannoudis, H. Dinopoulos, E. Tsiridis, Injury 36 (2005) S20–S27. [2] C.J. Damien, J.R. Parsons, J. Appl. Biomater. 2 (1991) 187–208. [3] F.M. Tamimi, J. Torres, I. Tresguerres, C. Clemente, E. López-Cabarcos, L.J. Blanco, J. Clin. Periodontol. 33 (2006) 922–928. [4] M. Bohner, Injury 31 (Supplement 4) (2000) D37–D47. [5] C. Großardt, A. Ewald, L.M. Grover, J.E. Barralet, U. Gbureck, Tissue Eng. A 16 (2010) 3687–3695. [6] J. Wei, J. Jia, F. Wu, S. Wei, H. Zhou, H. Zhang, J.-W. Shin, C. Liu, Biomaterials 31 (2010) 1260–1269. [7] F. Wu, J. Wei, H. Guo, F. Chen, H. Hong, C. Liu, Acta Biomater. 4 (2008) 1873–1884.
136 [8] [9] [10] [11] [12] [13] [14] [15] [16] [17]
T. Christel et al. / Materials Science and Engineering C 42 (2014) 130–136 G. Mestres, M.-P. Ginebra, Acta Biomater. 7 (2011) 1853–1861. K. Kowalczyk, G. Palavit, J. Mater. Sci. Lett. 12 (1993) 1276–1278. D.V. Ribeiro, M.R. Morelli, Mater. Res. 12 (2009) 51–56. E. Vorndran, A. Ewald, F. Müller, K. Zorn, A. Kufner, U. Gbureck, J. Mater. Sci. Mater. Med. 22 (2011) 429–436. U. Klammert, A. Ignatius, U. Wolfram, T. Reuther, U. Gbureck, Acta Biomater. 7 (2011) 3469–3475. H. Zhou, T.F. Luchini, S. Bhaduri, J. Mater. Sci. Mater. Med. 23 (2012) 2831– 2837. K.J. Lilley, U. Gbureck, J.C. Knowles, D.F. Farrar, J.E. Barralet, J. Mater. Sci. Mater. Med. 16 (2005) 455–460. J. Lu, J. Wei, Y. Yan, H. Li, J. Jia, S. Wei, H. Guo, T. Xiao, C. Liu, J. Mater. Sci. Mater. Med. 22 (2011) 607–615. C. Moseke, C. Bayer, E. Vorndran, J. Barralet, J. Groll, U. Gbureck, J. Mater. Sci. Mater. Med. 23 (2012) 2631–2637. W. Paul, C.P. Sharma, J. Mater. Sci. Mater. Med. 10 (1999) 383–388.
[18] M. Bohner, S. Tadier, N. van Garderen, A. de Gasparo, N. Döbelin, G. Baroud, Biomatter 3 (2013) e25103. [19] S. Singh, V. Singh, S. Aggarwal, U. Mandal, Chem. Pap. 64 (2010) 491–498. [20] L. Alexander, H.P. Klug, Anal. Chem. 20 (1948) 886–889. [21] D.J. Misiek, J.N. Kent, R.F. Carr, J. Oral Maxillofac. Surg. 42 (1984) 150–160. [22] D.P. Pioletti, H. Takei, T. Lin, P. Van Landuyt, Q. Jun Ma, S. Yong Kwon, K.L. Paul Sung, Biomaterials 21 (2000) 1103–1114. [23] S.M. Oliveira, C.C. Barrias, I.F. Almeida, P.C. Costa, M.R.P. Ferreira, M.F. Bahia, M.A. Barbosa, J. Biomed. Mater. Res. B Appl. Biomater. 87B (2008) 49–58. [24] R.A. Perez, S. Del Valle, G. Altankov, M.-P. Ginebra, J. Biomed. Mater. Res. B Appl. Biomater. 97B (2011) 156–166. [25] G. Ben-Et, D. Tatarsky, J. Am. Oil Chem. Soc. 49 (1972) 499–500. [26] W.C. Griffin, J. Soc. Cosmet. Chem. 1 (1949) 311–326. [27] F. Witte, H. Ulrich, C. Palm, E. Willbold, J. Biomed. Mater. Res. A 81A (2007) 757–765. [28] A. Ewald, K. Helmschrott, G. Knebl, N. Mehrban, L.M. Grover, U. Gbureck, J. Biomed. Mater. Res. B Appl. Biomater. 96B (2011) 326–332.
