Controlling calcium and phosphate ion release of 3D printed bioactive ceramic scaffolds: An in vitro study

This paper characterizes in an in vitro setting the release of calcium (Ca) and phosphate (PO4) of 3D printed bioactive ceramic scaffold prepared from extrudable paste containing hydroxyapatite and β-tricalcium phosphate (β-TCP). Hydroxyapatite and β-TCP were calcined at 800 °C for 11 h, fabricated into four experimental groups (100% HA, 100% β-TCP, 15%/85% HA/β-TCP, and 15%/85% HA/β-TCP (design)), sintered to 1100 °C for 4 h. Calcium and phosphorus concentrations were evaluated using ICP spectroscopy, and the release of Ca and PO4 ions during dissolution of the CaP-based scaffolds was measured by submerging in 0.05 mol/L Tris(hydroxymethyl)aminomethane-HCl and maintaining a temperature of 37 °C. The Ca and PO4 concentrations of the solutions were measured with the utilization of a calcium assay kit and a phosphate assay kit and read in a UV–visible spectrophotometer. The 100% HA scaffold group showed the greatest concentration of Ca ions (~1.9 mg/dL), but ultimately released at a lower amount as time increased; the 100% HA scaffold also showed the lowest total amount of calcium ions released over the course of evaluation. The results for the 100% β-TCP were on the opposite of the HA with the highest amount of calcium ion release over the study. While the PO4 ion release showed a similar trend as those observed with Ca ions with an apparent difference in the 100% HA scaffold group. There was nearly 0 mg/dL of the phosphate ions released in the first 24 h, in comparison to the amount of Ca ions released during the same time frame. Since various formulations can lead to different properties of these bioactive ceramic scaffolds, it is important to understand how the tailoring of this important biphasic material can impact the long-term outcome of an ever-important in vivo clinical trial in the future.


Introduction
Bone wound healing is a complicated area of research, which has doctors, engineers, and scientists continuously searching for the most appropriate solutions. Currently there are no "ideal" options on the market to repair large, critical sized defects or load bearing areas. The future of bone grafting is leading towards the combination of materials and fabrication  processes, which will allow for fabrication of "tailor-able" devices.
The repair of critical sized long bone defects is of special interest in the orthopedic field due to its reconstructive complexity. Much attention has been given to the defect site's blood supply, with focus between bone and immune cells and mechanical reestablishment. Most commonly, the defects are caused by birth defect, trauma, or tumor resection. These issues may leave the patient with a critical sized defect, which must be controlled with suitable dead space management followed by bone reconstruction utilizing a grafting material or device, which is osteoconductive and degradable, while simultaneously minimizing complications and shortening the duration of recovery.
Currently, the viable bone grafting material solution is calcium phosphate (CaP)-based, which possesses two (osteoconduction and osseointegration) of the following three properties: osteoconduction-the capability to support the growth of bone over its surface; osteogenesis-the process of extra cellular matrix deposition and subsequent mineralization by osteoblasts; and osseointegration-the ability to intimately interact with bone without an extra layer of soft tissue. However, different host-to-grafting material responses have been reported and should be taken into consideration while determining treatment options.
The stability of calcium phosphate-based ceramics is known to be compositionally and structurally dependent [1], depending on the materials hydroxyapatite (HA), beta tri-calcium phosphate (β-TCP), or biphasic calcium phosphate (BCP). When these CaP-composed devices are implanted, the ceramic will undergo dissolution at different rates (HA < β-TCP < α-TCP) [2,3]. Additionally, when placed in vivo, the CaP ceramics induce a precipitation of a phase which is comparable to the HA-like mineral found in bone [4]. While in some instances it has been seen that a bone apatite-like material was formed in nonosseous sites [3,4]. Therefore, an important factor with regards to bioactivity (bone tissue formation rate and bone tissue bonding) is the release of ions from the dissolution of CaP ceramics, which in turn may affect cellular activity, organic matrix deposition, or mineral deposition [5,6].
There was an analysis and assessment of the dissolution of 3D printed lattice-based bioactive ceramic scaffolds (LBBCS) [7]. In an in vitro setting, it is of value to qualitatively and quantitatively analyze and observe any possible phase changes, which can possibly occur during the fabrication process or dissolution process [8,9]. In the past, there has been work done to try to interpret as well as predict the long-term effects of the dissolution of CaP ceramics. This study is set out to determine the relative dissolution behavior of four experimental groups (Table  1) specifically the mean concentrations (mg/dL) of Ca and PO 4 ions which were released from the 3D printed grafting devices in a Tris(hydroxymethyl) aminomethane-HCl buffer environment [4].

1 Ink preparation
Prior to 3D printing, the formulated colloidal gel was subjected to a series of powder processing steps at the beginning (e.g., calcining, attrition milling, and drying of the individual base materials), which is described in detail by Witek et al. [7]. The aforementioned processed ceramic powders were utilized for the colloidal ink suspension, with a solid volume fraction ( solid  ) of approximately 46% ceramics, (i.e., 100% HA (0.7 μm in particle size), 100% β-TCP (0.4 μm in particle size), or pre-determined ratio of HA:β-TCP) [10]. Upon calculating the appropriate amounts of ceramic powders, H 2 O, and dispersant, Darvan ® 821A (R.T. Vanderbilt Company, Inc., Norwalk, CT, USA), ~15 mg/g of ceramic, the materials were added to a mixing container in which the bottom was charged with zirconia milling media (5 mm; Union Process, Akron, OH, USA). The mixture was spun in a conditioning mixer (THINKY AR-250; Thinky, Tokyo, Japan) for 180 s at 2000 rpm. The next step was the addition of a thickening agent, Methylcellulose F4M (Dow Chemical Company, Midland, MI, USA), approximately 8 mg/mL of ceramic, followed by spinning for ~120 s at 2000 rpm. The final step was the addition of a gelling mechanism (polyethlyenimine (PEI); Sigma-Aldrich St. Louis, MO, USA). About 200 mg of polyethyleneimine per 30 mL of ink was added followed by spinning to achieve the desired consistency.
After the colloidal gel was prepared, it was 3D printed into circular lattice scaffolds (Fig. 1).

2 Sample fabrication
The 3D direct-writing printer used a gantry robot with affixed syringe pumps to extrude the colloidal inks through a fine nozzle to build the 3D scaffold [11,12]. The 3D disc-shaped scaffolds (10 mm in diameter, ~2 mm in thickness, 330 μm in struts, and 500 μm in pore size) were designed on a computer-aided design (CAD) software (RoboCAD 4.3, 3D Inks, LLC Tulsa, OK, USA) (Fig. 1). The 3D disc-shaped scaffolds were printed layer-by-layer at 10 mm/s deposition rate onto a flat ceramic substrate (ADS-96R, Coors Tek, Grand Junction, CO, USA) in low-viscosity paraffin oil (Ultra-Pure lamp oil, Lamplight Farms Inc., Menomonee Falls, WI, USA) to prevent drying of the structure during assembly. After the scaffolds were fabricated, they were removed from the oil bath and allowed to dry in air for ~12 h prior to the sintering process. Heat treatment of the scaffolds was performed utilizing an electric furnace (Model LHT 02/17, Nabertherm GmbH, Lilienthal, Germany) to enhance the mechanical strength for handling during the experiments (e.g., in vitro dissolution study) [11]. During the sintering processing, the green body ceramics were densified and organic components were burned off. The firing protocol began by heating at 2 ℃/min until 400 ℃, dwelling at 400 ℃ for 2 h, then heating until 900 ℃ (~8 ℃/min), dwelling for 2 h at 900 ℃, heating at 3.3 ℃/min to ultimate temperature of 1100 ℃, dwelling at 1100 ℃ for 4 h, and finally cooling at ~8.0 ℃/min until the samples were cooled to room temperature (~30 ℃). The binder burnout during the sintering process resulted in a predictable volumetric shrinkage (~20%) and defects such as cracking and warping [10]. The sintering shrinkage of ceramics is an unavoidable step, and all dimensions of the designed geometric structure must be enlarged to account for this phenomena and to ultimately attain the final desired dimensions.

3 Chemical characterization
Calcium (Ca) and phosphorus (P) concentrations were evaluated using inductively coupled plasma (ICP) spectroscopy (Thermo Jarell Ash, Trace Advantage, Franklin, MA, USA). For this objective, 10 mg of powders from each experimental group were dissolved in 35% HCl solution and mixed 100 mL of double distilled water in a volumetric flask. Suitable standard solutions for Ca (0, 20, and 40 ppm) and P (0, 10, and 40 ppm) were prepared from 100 ppm standard solutions of the respective elements (Fisher Scientific, Pittsburgh, PA, USA). Each specimen in solution and standard solution was pumped through argon plasma, which was excited by a 2 kW, 27.12 MHz radio frequency generator [13]. Additionally scaffold groups were subjected to energy-dispersive X-ray spectroscopy (Hitachi S-4800, Schaumburg, IL, USA) for further verification of their respective Ca:P ratios.
X-ray diffraction (XRD; Bruker AXS D8 Discover, Madison, WI, USA) was used to determine the crystalline phases present within the raw materials and experimental groups. Three different spectra from each testing group were obtained (n = 3 per group). The diffractometer was operated by using a curved crystal monochromator, operating at 45 mA and 45 kV, and scanned in 2θ range from 20° to 65° [7], with a step size of 0.02° at 3 s/step.
Rietveld refinement analysis used the data collected from X-ray diffraction, and subsequently converted data to ".xy" format. The raw data was then input into Material Analysis Using Diffraction (MAUD) software for quantitative analysis. Rietveld analysis utilized the samples from previous XRD spectra (i.e., peak, height, width, and position). This refinement method was used to quantify the percentage of each phase (HA, β-TCP, and other commonly observed phases in calcium-and phosphate-based materials such as calcium oxide, CaO) present in different calcination temperatures and sintering groups. A least-squares fit approach was utilized to measured scans until replicating a theoretical scan (based on the structure in Inorganic Crystal Structure Data Base (ICSD)) [14,15]. The background, cell parameters, preferred orientation, peak asymmetry, atomic positions, site occupancy factors, and global vibrational parameters were refined.
The in vitro observation for the release of Ca and PO 4 ions during dissolution of the CaP-based scaffolds, was measured by submerging LBBCS in a 24 well plate (Corning ® Costar ® , Sigma-Aldrich, St. Louis, MO, USA) (n = 1 scaffold/well) containing 0.05 mol/L Tris(hydroxymethyl)aminomethane-HCl (bioWORLD, Dublin, OH, USA) solution with a pH of ~7.3 and maintaining a temperature of 37 ℃ (in an incubator). The weight of scaffold to volume of solution ratio was ~1 mg/mL as recommended by Ducheyne et al. [4]. The time periods to be examined for the release of ions was from 1 h ranging to 2 weeks, with samples measured in triplicate

4 Statistical analysis
For statistical evaluation, analysis of variance (ANOVA) at a 95% confidence interval by using IBM SPSS v23 (IBM Corporation, Armonk, New York, USA) was utilized considering HA and β-TCP ratios as independent variables, thus Ca:P ratio and Ca and PO 4 ion concentrations were considered as dependent variables.

Results
The ICP results were used to verify the Ca:P ratio of the two individual materials, in each of the 3 experimental states: (i) raw (as received), (ii) after calcination (800 ℃), and (iii) post sintering (1100 ℃) (Fig. 2(a)). The trend in the β-TCP group shows an initial decrease from raw state to after calcination state, but then an increase in post sintering state. The Ca:P ratio slightly increases while going from raw, to calcined, and ultimately sintered state in the HA group. The one-way ANOVA statistical analysis results in p = 0.0204 and p = 0.022, for HA and β-TCP respectively, indicating significant effect due to the state of the material. Energy-dispersive X-ray spectroscopy (EDS) was completed to evaluate the Ca/P ratio for experimental scaffolds of 100% HA, 100% β-TCP, and 15%/85% HA/β-TCP. The Ca/P ratios observed for the different groups are similar to what is expected aside from 100% HA, which seems to be slightly on the lower end of the atomic value of ~1.67 ( Fig. 2(b)). The EDS was done as a supplemental quantitative verification measurement on the primary sample groups. These values are taken as an estimate of final average of the groups.
XRD spectra were collected of the individual raw materials for crystalline verification. Their spectra were matched to the theoretical HA and β-TCP. After confirmation of the materials was complete, and the experimental colloidal gels were fabricated and subjected to sintering at 1100 ℃, XRD spectra were obtained from the groups for crystalline phase determination and verification ( Fig. 3(a)). The β-TCP phase is indicated by three high intensity peaks located Fig. 2 (a) Mean ± 95% confidence intervals for Ca/P ratio depicting the two primary groups (HA and β-TCP) and their state. (b) Mean ± 95% confidence intervals for Ca/P ratio of the experimental groups. at 27.9°, 31.2°, and 34.4°, while the highest peaks for HA are 31.8° and 32.9° (Fig. 3(b)). High crystallinity is indicated by the narrow diffraction peaks described as sharp peaks. This result is in good agreement with other researchers' works that found that biphasic mixing on a crystallite level is feasible [16][17][18].
The calcium (Ca) and phosphate (PO 4 ) concentrations (mg/dL) of scaffold dissolution in Tris-HCl solution are summarized in Table 2. Figure 4 indicates that the data, which begin at time point of 0 h to indicate the initial, are used for analysis of both calcium and phosphate. The 100% HA scaffold group shows the greatest concentration of Ca ions (~1.9 mg/dL), but ultimately releases at a lower amount, which can be seen at hour 336 (day 14). The 100% HA scaffold has the lowest amount of calcium ions released.
On the other end of the materials, the 100% β-TCP has the highest concentration released with approximately a measurement of 7.6 mg/dL. The 15%/85% HA/β-TCP scaffold group falls in between the 100% HA and 100% β-TCP groups, and at day 14 has a Ca ion release of approximately 5.95 mg/dL.
While the PO 4 ion release shows a similar trend as that in Ca with one apparent difference in the 100% HA scaffold group. There is nearly 0 mg/dL of the phosphate ion released in the first 24 h, in comparison to the amount of Ca ions released during the same time frame. The 100% HA group eventually ends up having a release of approximately 5.1 mg/dL at day 14, but it is short of what the other three groups (100% β-TCP, 15%/85% HA/β-TCP, and 15%/85% HA/β-TCP (design)) release in the same time period. 100% β-TCP,  15%/85% HA/β-TCP, and 15%/85% HA/β-TCP (design) follow the same trend, and at day 14, they all have an average PO 4 ion concentration of ~8.61 mg/dL, with no significant difference between these three experimental scaffolds' compositions. Additionally, the experimental scaffolds' compositions were analyzed after the time in vitro, to verify no changes in phases. As seen in Figs. 5(a)-5(d), when evaluating the relevant peaks, no differences are observed from time 0 thru 28 days, indicating that the material remains unaffected over time.

Discussion and conclusions
Being that a wide range of physical and chemical properties can be achieved with this particular bioactive ceramic by modifying specific parameters, such as the HA/β-TCP ratio, calcining and/or sintering temperature [19], these materials have been used to fabricate pre-formed implants and have shown considerably greater mechanical strength in comparison to other bioceramics used for similar bone grafting purposes [19]. The present study aimed to chemically characterize bioactive ceramic LBBCS prepared as 100% HA, 100% β-TCP, 15%/85% HA/β-TCP, and a model 15%/85% HA/β-TCP group with raw materials calcined at 800 ℃, and final LBBCS subjected to a sintering regimen of 1100 ℃. The advantage of this material, which is fabricated to an aqueous paste, allows for fabrication of scaffolds utilizing solid free form (SFF) equipment also known as robocasting. This 3D printing technique allows for the production of variety of complex structures utilizing CAD/CAM software in addition to utilizing computed  tomography scans exported to a specific file type which can in turn be fabricated to a specific defect, capable of reproducing the same structure consistently [20].
XRD of the as-received materials prior to calcination primarily showed HA and β-TCP crystalline phases for the individual HA and β-TCP powders. However, phase quantification through Rietveld refinement showed trivial fractions of β-TCP, calcium pyrophosphate (Ca 2 P 2 O 7 ), and calcium oxide (CaO) for the HA as-received material, and minor fractions of HA, Ca 2 P 2 O 7 , and CaO for the β-TCP starting material. Given that the aqueous pastes were fabricated without considering the presence of minor quantities of other CaP-based phases, the intended composition of the 15%/85% HA/β-TCP group was slightly shifted from what was originally intended.
Phosphorus (P) is widely distributed in the body. More than 80% of total phosphate (PO 4 ) is present in the bone and teeth in the form of apatite. The remaining phosphate is mostly present in the skeletal muscle, with trace amounts in the extracellular fluids (< 0.1%) [21][22][23][24]. Intracellular phosphate ions are essential for oxidative phosphorylation and approximately 20% of cellular phosphate is present in the mitochondria. Roughly 30% of total cellular phosphate is stored in the endoplasmic reticulum and is used in phosphorylation of various proteins. The remaining cellular phosphate is present in the nucleus, Golgi complex, and lysosomes. Transporting phosphate in and out of the cell according to the need of the body is a complex process and the exact molecular mechanisms of such sensitive transport are not entirely clear. It is important to note that cells use PO 4 to transport cellular energy through the formation of ATP by oxidative phosphorylation.
With the aforementioned, phosphorus is an essential nutrient for the body and is routinely consumed through food. After consumption, phosphorus binds with oxygen and exists as PO 4 in the body. Both organic and inorganic forms of PO 4 are present in regularly consumed foods. The amount of total PO 4 ingestion can be significantly influenced by consumption of processed food and beverage, as there are phosphate metabolites additives in these items. Following a meal, inorganic phosphate can be rapidly absorbed across the small intestine and enter the blood stream causing an elevation in blood phosphate levels. The net efficiency of intestinal phosphate absorption is double that of Ca [25]. Taking into consideration just the amount of phosphate consumed and adding a calcium phosphate (CaP)-based scaffold can lead to a potential toxic situation. Phosphate toxicity due to excessive retention of PO 4 in the body can cause a variety of cellular and tissue damage. An example is seen with the higher occurrence of vascular calcification, encountered in patients with chronic kidney disease, is related to the increased retention of phosphate in the body [26,27]. The normal range of PO 4 should be in the order of 2.4-4.1 mg/dL, so this is of importance as when scaffold is placed in the system, the levels in the local region may be slightly higher than normal, but should not increase the overall levels to prevent any complications such as cell death, which in turn would lead to killing new bone cells.
Chen et al. [28] and Sweet et al. [29] have studied the in vitro analysis of ion release in the SBF solution, with regards to cytotoxicity. The cytotoxicity assay results indicated that the 100% β-TCP material, which is most commonly used, did not have a negative effect on altering the Schwann cells in vitro.
The data from the in vitro portion of this study is in agreement with literature. Using the 100% HA and 100% β-TCP as controls and the two 15%/85% HA/β-TCP groups as the experimental groups, the results are similar to what was expected. Literature presents the bulk or pressed powder properties of these two control group materials, but does not present the biphasic mixture release for Ca or PO 4 in an in vitro setting. The results indicate the trend for the two biphasic groups, in which the rates of calcium and phosphate ion release should fall above the 100% HA but below 100% β-TCP. This was the result, which was observed when doing a qualitative analysis of the results and further identified with the extrapolation of data from previous bulk property results.
Overall, since various chemical formulations can lead to different physiochemical properties of these extruded materials processed, it is important to understand how the tailoring of this important biphasic material can impact the long-term outcome of an ever-important in vivo clinical trial in the future.
As the scaffolds can be fabricated from mixing various ratios of the bioactive ceramic, HA and β-TCP, both of which are biocompatible, resorbable, and can facilitate in the regeneration and remodeling of bone at defect site, the results are important on how to tailor composition. Custom ratios of the materials in scaffolds for use in load and non-load bearing models are warranted to explore the rate of osseointegration, as the release of these ions has the potential to dictate which material would be of use in the final scaffold design.