Bone Regeneration in Load-Bearing Segmental Defects, Guided by Biomorphic, Hierarchically Structured Apatitic Scaffold

Elizaveta Kon; Francesca Salamanna; Giuseppe Filardo; Berardo Di Matteo; Nogah Shabshin; Jonathan Shani; Milena Fini; Francesco Perdisa; Annapaola Parrilli; Simone Sprio; Andrea Ruffini; Maurilio Marcacci; Anna Tampieri

doi:10.3389/fbioe.2021.734486

. 2021 Sep 27;9:734486. doi: 10.3389/fbioe.2021.734486

Bone Regeneration in Load-Bearing Segmental Defects, Guided by Biomorphic, Hierarchically Structured Apatitic Scaffold

Elizaveta Kon ^1,², Francesca Salamanna ³, Giuseppe Filardo ^4,^5,⁶, Berardo Di Matteo ^1,^5,^*, Nogah Shabshin ^6,⁷, Jonathan Shani ⁸, Milena Fini ³, Francesco Perdisa ⁹, Annapaola Parrilli ¹⁰, Simone Sprio ¹¹, Andrea Ruffini ¹¹, Maurilio Marcacci ^1,², Anna Tampieri ¹¹

PMCID: PMC8503888 PMID: 34646817

Abstract

The regeneration of load-bearing segmental bone defects remains a significant clinical problem in orthopedics, mainly due to the lack of scaffolds with composition and 3D porous structure effective in guiding and sustaining new bone formation and vascularization in large bone defects. In the present study, biomorphic calcium phosphate bone scaffolds (GreenBone™) featuring osteon-mimicking, hierarchically organized, 3D porous structure and lamellar nano-architecture were implanted in a critical cortical defect in sheep and compared with allograft. Two different types of scaffolds were tested: one made of ion-doped hydroxyapatite/β-tricalcium-phosphate (GB-1) and other made of undoped hydroxyapatite only (GB-2). X-ray diffraction patterns of GB-1 and GB-2 confirmed that both scaffolds were made of hydroxyapatite, with a minor amount of β-TCP in GB-1. The chemical composition analysis, obtained by ICP-OES spectrometer, highlighted the carbonation extent and the presence of small amounts of Mg and Sr as doping ions in GB-1. SEM micrographs showed the channel-like wide open porosity of the biomorphic scaffolds and the typical architecture of internal channel walls, characterized by a cell structure mimicking the natural parenchyma of the rattan wood used as a template for the scaffold fabrication. Both GB-1 and GB-2 scaffolds show very similar porosity extent and 3D organization, as also revealed by mercury intrusion porosimetry. Comparing the two scaffolds, GB-1 showed slightly higher fracture strength, as well as improved stability at the stress plateau. In comparison to allograft, at the follow-up time of 6 months, both GB-1 and GB-2 scaffolds showed higher new bone formation and quality of regenerated bone (trabecular thickness, number, and separation). In addition, higher osteoid surface (OS/BS), osteoid thickness (OS.Th), osteoblast surface (Ob.S/BS), vessels/microvessels numbers, as well as substantial osteoclast-mediated implant resorption were observed. The highest values in OS.Th and Ob. S/BS parameters were found in GB-1 scaffold. Finally, Bone Mineralization Index of new bone within scaffolds, as determined by micro-indentation, showed a significantly higher microhardness for GB-1 scaffold in comparison to GB-2. These findings suggested that the biomorphic calcium phosphate scaffolds were able to promote regeneration of load-bearing segmental bone defects in a clinically relevant scenario, which still represents one of the greatest challenges in orthopedics nowadays.

Keywords: hydroxyapatite, tricalcium phosphate, vascularization, segmental bone defect repair, biomimetism, ion doping, biomorphic scaffold

Introduction

The regeneration of load-bearing segmental bone defects is today considered among the greatest challenges in orthopedics, due to the lack of biomaterials effective for long bone substitution and to the frequent occurrence of various complications such as nonunions (Giannoudis and Atkins, 2007).

In the last 20 years, there have been numerous attempts in the development of new materials for the treatment of long bone defects, particularly based on hydroxyapatite (HA) and other calcium phosphates (CaPs), which most closely resemble the mineral composition of natural bone (Habraken et al., 2016). However, despite preclinical promising results, the clinical application of these materials in various forms and with different physico-chemical features is still restricted for long bone defects consequent to excision of tumors, infection, major trauma, and nonunion (Ebrahimi et al., 2017; Roffi et al., 2017). Indeed, currently available CaP-based devices for large bone defects often fail in recreating the anatomical and functional features of the lost tissue, due to the complexity of bone in terms of composition, structural, and mechanical properties (Hutmacher et al., 2004). This is a critical issue when it comes to regenerate load-bearing segmental bones where multiaxial biomechanical stresses are particularly demanding, so that insufficient structural organization of the newly formed bone tissue can result in impaired functionality. In this respect, the typical osteon structure of the long bone has a multi-scale hierarchical architecture which is highly functional in ensuring: 1) the propagation of mechanical forces from the macro- to micro-scale and 2) the activation of mechano-transduction processes responsible for the bone self-repair ability.

Hence, the main obstacle to the regeneration of critical segmental bone defects is the lack of technologies enabling the fabrication of 3D porous scaffolds capable to recreate in vivo the complex ensemble of cell-instructive physico-chemical, topological, and structural signals inducing the formation and remodeling of mechanically competent bone tissue in large, load-bearing defects. To meet this goal, such a scaffold should have several different properties, difficult to achieve at the same time: 1) bioactive chemical composition, 2) bio-resorbability, 3) interconnected, cell-conducive porous architecture to enable ossification and vascularization in the whole scaffold volume, and 4) bone-mimicking mechanical performance promoting mechano-transduction phenomena and bone remodeling from woven to mature, mechanically competent tissue (Huang and Ogawa, 2002; LeGeros, 2002; Hannink and Arts, 2011; Stegen et al., 2015). Particularly, when dealing with inorganic materials such as CaPs, the need of a sintering process for scaffold consolidation yields to phase degradation and crystal growth which impair bioactivity and bio-resorbability. In addition, sintered CaPs are usually characterized by brittle fracture behavior, which is a concern for their use in load-bearing models due to the risk of sudden rupture (Ginebra et al., 2018). Previous studies attempted to solve this problem by developing ceramic-polymer composites, but in spite of the advantage of limiting the occurrence of fragile fracturing, the regenerative ability of such devices is limited by the lack of compositional and mechanical affinity with natural bone, and by the delivery of harmful by-products through nonenzymatic acidic dissolution, which easily induce inflammatory reactions, thus damaging bone cells at the implant site and reducing the bone tissue regeneration process, particularly in the presence of large bone defects (Liu et al., 2006).

In the attempt to overcome the limitations inherent to the classical approach for bio-ceramics fabrication, recent studies reported innovative nanotechnological processes based on the biomorphic transformation of natural wood structures into 3D porous CaP ceramics showing hierarchical architecture (Tampieri et al., 2009; Tampieri et al., 2018; Filardo et al., 2020). In particular, rattan wood was used as a template for such a process based on its bone-like hierarchical structure mimicking the osteon structure typical of the long bone (Tampieri et al., 2009; Tampieri et al., 2018; Filardo et al., 2020). Previous studies in 3D bioreactor showed that, in comparison with sintered hydroxyapatite scaffolds featuring similar porosity extent, biomorphic scaffolds made of hydroxyapatite/tricalcium phosphate doped with Mg and Sr ions are capable of triggering dramatic overexpression of genes involved in osteogenesis and to promote cross-talk between osteoblasts and fibroblasts, thus potentially stimulating angiogenesis (Tampieri et al., 2018; Sprio et al., 2019). Furthermore, the scaffolds were found to have mechanical performance unusual for a pure ceramic material, resulting in high anisotropic compressive and tensile strength, high toughness and bone-like, damage-tolerant behavior, making them very promising for application as bone substitutes in load-bearing bone defects (Tomlinson et al., 2001; Bigoni et al., 2019).

In the present work, the performance of a scaffold (GreenBone™), produced via biomorphic transformation of rattan wood structures, was tested in a sheep model of segmental bone loss and compared to that of an allograft implant. The purpose of the work was to assess the regenerative ability of the scaffold in a load-bearing site and to show whether the enhanced biological efficiency and mechanical ability previously demonstrated in vitro can be confirmed in a clinically relevant animal model.

Materials and Methods

Scaffolds Preparation

The scaffolds tested in this work were obtained by a biomorphic transformation process as previously reported (Tampieri et al., 2018), using cylindrical rattan wood pieces (Calamus manna) as 3D templates guiding the process. Briefly, the wood template was pyrolyzed at 1,000°C under nitrogen gas flow to obtain a biomorphic (i.e., reproducing the original wood structure) carbon template, then placed in a vessel containing metallic Ca granules (Sigma Aldrich, United States), and subjected to heating up to 1,250°C to allow sublimation of calcium in an oxygen-free environment, and reaction with the carbon template to obtain a 3D biomorphic calcium carbide (CaC₂). The resulting CaC₂ was converted into calcium oxide (CaO) by heating at 900°C under air atmosphere. Conversion of the resulting CaO into a calcium carbonate (CaCO₃) template was carried out by thermal treatment under a flux of carbon dioxide under non-isothermal and isobaric conditions (p = 100 atm) under dry or wet conditions from room temperature to 800°C. To activate CO₃ ↔ PO₄ ion exchange in the as-obtained CaCO₃ template, it was placed in a closed reactor containing a 2.0M solution of (NH₄)₂HPO₄ (Sigma Aldrich, United States ) and heated at 220°C under water vapor pressure (∼20 bar) to obtain a biomorphic CaP scaffold. Then, a conclusive treatment was carried out by soaking the obtained scaffold in an aqueous solution containing 0.5M Sr(NO₃)₂ and 0.5M Mg(NO₃)₂ ions for 24 h at 50°C, thus obtaining a Mg, Sr-doped biomorphic scaffold (hereinafter coded as GB-1). The undoped biomorphic scaffold is hereinafter coded as GB-2.

Scaffolds Characterization

The phase composition of the scaffolds was obtained by X-ray diffraction (XRD), using a D8 Advance diffractometer (Bruker Karlsruhe, Germany) equipped with a Lynx-eye position sensitive device (CuKα radiation: λ = 1.54178 Å). XRD spectra were recorded in the 2θ range of 20–60° with a counting time of 0.5 s and a step size of 0.02°. XRD spectra were subjected to full profile analysis (TOPAS software v.5, Bruker, Karlsruhe, Germany) to evaluate cell parameters and coherent domain sizes.

Chemical analysis was performed on dried samples using an ICP-OES spectrometer (Agilent 5,100). About 20 mg of the sampling material was dissolved in 2 ml of nitric acid (Sigma Aldrich; 65 vol%) and then diluted with Milli-Q water to obtain 100 ml of solution. The solution was then analyzed using a standard prepared from primary standards (1,000 ppm, Fluka). The carbonate content was evaluated by thermogravimetric analysis (Stanton STA 1500, Stanton, London, United Kingdom), measuring the weight loss in the temperature range 600–1,100°C.

Mercury porosimetry was used to evaluate pore size distribution (<50 μm) by two different equipment (Carlo-Erba Porosimeter 2000 and Macropores Unit 120), working on separate pore size ranges. The open and total porosities of the studied ceramics were measured by Archimede’s method and geometrical weight volume evaluation, respectively.

The compressive strength of the scaffolds was measured on cylindrical specimens sized 20 × 14 mm (height × diameter) using a Zwick/Roell instrument, model Z050 (Ulm, Germany). Microscopic evaluation of the scaffold was made by field emission-scanning electron microscopy (FEG-SEM) (Sigma NTS GmbH, Carl Zeiss, Oberkochen, Germany) in order to evaluate the porous architecture of the scaffolds at the multi-scale.

Scaffold Liquid Absorption

The scaffold’s liquid absorption ability was evaluated by soaking into body fluid. Five dried scaffolds (three GB1 and two GB2) were weighted (0.1 mg accuracy) (SI-64, Denver Instrument GmbH) and then immersed in 10 ml of sheep blood for 1 min at room temperature. After this procedure, the scaffolds were placed for a 10 s on a qualitative filter paper disk to remove the blood in excess, and the weight was measured again. The liquid absorption was calculated with the following equation:

L i q A = \frac{m_{2} - m_{1}}{m_{1}} \times 100

where LiqA is the blood absorption (%), m ₁ is the dry mass, m ₂ is the wet mass.

Mechanical Stability of the Scaffold

The scaffold’s mechanical stability was examined by the drill test. Ten scaffolds were drilled by using the orthopedic power tool Colibri II (DePuy Syntheses GmbH; Oberdorf—Switzerland), using a drill bit of 2.5-mm diameter, to place a 3.5-mm diameter cortex self-tapping screw. During this procedure, digital photos of each scaffold were taken, and the images were elaborated with Adobe Photoshop software v.6.0.1.

In vivo Study in Sheep Model

The study was approved by the institutional Committee for Ethical Conduct and Care and Use of Laboratory Animals of Israeli Ministry of Health (approval 09/2016). Study procedures were carried in accordance with the requirements set out in the Animal Welfare Requirements Part 2, Animal Welfare Law, the Guide for Care and Use of Laboratory Animals by the Institute of laboratory Animal Resource (ILAR), the guidelines by the National Institute of Health (NIH), and by the Association For Assessment And Accreditation Of Laboratory Animal Care (AAALAC), and in accordance with Parts 1, 2, and 6 of regulations for Biological evaluation of medical devices. The study has herein been reported following the ARRIVE guidelines.

Study Design

The present study was performed as a joint research between the Assaf-Harofeh Medical Center (Israel) and the IRCCS Istituto Ortopedico Rizzoli (Bologna, Italy) experimental facilities. The study evaluated the safety and performance of both GB-1 and GB-2 against allograft transplantation in the repair of a critical bone defect in 24 healthy female Assaf-Awassi sheep, 59–90 kg in weight. After selection, animals were housed in a dedicated enclosure at least 5 days prior to the surgical operation, where they could roam freely. Feeding was withheld the day before surgery.

The sheep were randomized into three study groups to receive either the scaffold GB-1, GB-2 or the allograft (AG, control group). Implants for the AG group were obtained from the bone segments removed from both GB-1 and GB-2 animal groups. Bone segments removed from the animals were washed with saline and antibiotics and conserved in a refrigerator for less than 48 h.

After implantation, the animals were closely monitored: follow-up procedures consisted of daily inspection of the animals, a weekly health assessment by a certified veterinarian, a monthly radiographic evaluation, and CT scans at 3 and 6 months. Animals were sacrificed at the end of the 6 months follow-up period.

Surgical Procedure

Surgeries were performed in aseptic conditions. Sheep were placed on the right side, and the right hind limb was shaved and disinfected. The metatarsus shaft was exposed through a medial approach directly above the bone to reach the medial side. A 3.5 mm broad dynamic compression plate with eight holes was contoured to the shaft, and holes were drilled with a high-speed perforator: three distally and three proximally to the defect. A standardized 2 cm defect was created with an oscillating saw between the fourth and fifth screw hole, under constant irrigation, while preservation of soft tissues was obtained by using two retractors. The 2-cm bone segment with periosteum was then removed, and after implant insertion, the plate was fixed to the bone with 3.5 mm screws, and the soft tissues were closed in layers. The correct positioning of the implant and plate was ensured using intraoperative fluoroscopy.

The lesion gap was treated according to the following groups:

Group 1— biomorphic scaffold GB-1
Group 2— biomorphic scaffold GB-2
Group 3— allograft (AG).

Removed bone segments from the animal’s Group 1 and 2 were conserved in a refrigerator (−20°C) for 24–48 h and used as allograft implants in Group 3.

After wound dressing, the animals were anesthetized and anterior–posterior and medial-lateral X-rays were performed by means of an X-ray generator (6.4 mA; 55 kV). Then, a full cast was put on the operated leg. The sole of the claws was closed for 4 weeks and then left open to ensure weight-bearing while preventing torsional or shear forces on the fracture site through the cast immobilization of the metatarsus shaft. The cast was changed at 4 weeks or upon need, and at 8 weeks, the sheep were let out to graze.

Anesthesia and Postoperative Medication

Prior to the surgery, the animals were subjected to sedation via an intramuscular injection of ketamine (19 mg/kg) and xylazine 2% (1 mg/kg). Subsequent anesthesia was induced by intravenous injection of midazolam (0.1–0.5 mg/kg or to effect). The animals then underwent endotracheal intubation using an AG-EET. Anesthesia was maintained using isoflurane (1–3% with 100% O₂) and aided by mechanical ventilation. Postoperative medication included antibiotics for at least 2 weeks following implantation (cefazolin 1 gr bid). Analgesic administration included dipyrone (1 gr bid on operation day) and flunixin (2.2 mg/kg/day, 1M) for the first 2 weeks following surgery.

Euthanasia

Euthanasia was carried out under general anesthesia and performed via intravenous injection of potassium chloride, in accordance with facility procedures. After euthanasia, a segment from the operated leg (which included the scaffold and an additional 3 cm of bone above and below it) was removed for subsequent analyses.

Radiography and CT Scan

Implant performance was assessed monthly by means of blind radiographic assessment in the antero-posterior (AP) and lateral planes (6.4 mA; 55 kV) at months 1, 2, 4, 5, 6 plus by CT scans including coronal and sagittal reformats (Siemens, Polymobil III, Philips ICT256) at months 3 and 6 by an independent radiologist with 17 years of experience. The implant performance and the healing process were quantified following the modified RUST scoring system, which focuses on bridging callus formation and fracture appearance, obtained as a cumulative score of single scores. In detail, the following parameters were evaluated: 1) callus formation, 2) new bone formation, and 3) implant resorption. The scoring system for callus formation ranged from 1 to 3: 0, no callus formation; 1, non-bridging; 2, partial bridging callus; and 3, complete bridging callus. New bone formation referred to the medullary cavity and was evaluated on CT as a high-density material in the medullary canal. It was scored as 0, none; 1, some; and 2, massive new bone formation. Finally, implant resorption was evaluated on both radiographs and CT, with the scores indicating 0, none; 1, partial; 2, complete resorption. The highest possible score was given if the boneimplant was barely seen.

Macroscopic Assessments

On the same day in which the animals were euthanized, regional lymph nodes of each animal were harvested and macroscopically examined to detect any alteration of the normal structure. Furthermore, popliteal lymph nodes were fixed in 4% formaldehyde within 1 h following euthanasia, dehydrated in a graded series of alcohols, and finally, embedded in paraffin. Histopathological evaluation examined: subcapsular sinus, follicles, germinal centers, high endothelial venules of cortex and medullary cords, and the sinus of the paracortex medulla.

Tissue reaction was assessed in terms of nature and extent of any of the following: hematoma, edema, encapsulation, and/or additional gross findings; presence, shape, and location of implant including possible remnants of degradable material; lymph node presentation.

Micro-CT

Architectural characteristics and measurements were assessed in a non-destructive manner by micro-CT at postmortem dissection and plate removal. Explanted metatarsus bone segments containing implanted scaffolds or AG were scanned with a high-resolution microtomography system (Skyscan 1,176, Bruker Micro CT, Belgium) at 80 kV and 300 μA using a copper and aluminum filter. The nominal resolution (pixel size) was set for all samples at 17.50 μm. The acquired images were then reconstructed using the NRecon software (Bruker MicroCT, Belgium) and were corrected for alignment (depending on the acquisition), beam hardening, and ring artifact reduction. The resulting images were saved in jpg 8-bit format. Quantitative 3D analyses were carried out with CTAn software (Bruker Micro CT, Belgium, v.1.16.4.1) on all samples. Volumes of interest (VOIs) of 20 mm or 40 mm in height and 35 mm × 35 mm on the axial plane were defined in each sample, using the implanted material as reference. Morphological parameters evaluated were:

- nBV (in mm³), newly formed bone volume;
- Cl.V (in mm³), callus volume at the defect level;
- Ct. V in %, cortical volume,
- Ct. Th in mm, cortical thickness;
- Tb.Th (in mm), trabecular thickness of the newly formed bone;
- Tb.N (in mm⁻¹), trabecular number of the newly formed bone;
- Tb. Sp (in mm), trabecular separation of the newly formed bone.

Histology and Histomorphometry

To evaluate the local histopathological response, metatarsi were dehydrated in ethyl alcohol solutions of increasing concentration (one passage at 70%, two passages at 95%, and two passages at 100%) in intervals of 24–48 h per solution. The samples were then soaked in methyl methacrylate. After polymerization, the blocks containing the samples were sectioned along a plane parallel to the long axis of the implant (EXAKT Cutting Systems, GmbH and Co., Norderstedt, Germany). From each sample, 5–0 and 100–200 µm sections were obtained. Sections were evaluated by an optical microscope (Olympus BX51), and two of them were selected for further histological evaluations. The two selected sections were attacked on a special microscope slide (Microscope Slides, 50 × 100 × 1.5 mm, EXAKT); the precise adherence of the sections to the acrylic glass was performed with a cyanoacrylate glue which allows a complete adhesion in about 20–30 min. After 1 day, the sections were thinned with a grinding system (550, ATM GmbH, Mammelzen, Germany) using abrasive papers of different granulation (Struers), from 600 to 4,000 grit, up to a thickness of 40 ± 10 µm. Before staining, sections were treated with a polycrystalline diamond spray and polished automatically with a polishing system (Saphir 550 Grinding/Polishing System). The two sections were stained with toluidine blue/fast green and mounted with a cover glass using a mounting medium (contains acrylic resin and xylenes).

Histological and quantitative histo-morphometric analyses were carried out with digital scanner with a resolution of 0.5 μm (CS System, Aperio Technologies, Vista, CA—United States) and with an optic microscope (Olympus BX51). Regions of interest (ROIs) of 20 and 5 mm in height were defined in each sample, by using the implanted material as reference. Parameters assessed included changes in tissue morphology, quality of bone ingrowth, material fragmentation, presence of debris, shape and position of the implanted materials, extent of fibrous capsule (Fb.Ar), and inflammation–as compatible with methyl methacrylate embeddingpresence/extent of necrosis, presence of calcified tissue (osteoid surface, OS/BS), thickness of non-calcified tissue (osteoid thickness, Os.Th), bone surface covered by osteoblasts (Ob.S/BS).

To further assess inflammation (polymorphonuclear cells, lymphocyte plasma cells, macrophages) and neovascularization (number of vessels/microvessels) of the implanted materials, a region of interest (ROI) of 1 cm × 1 cm, on the lateral side of the metatarsus (opposed to the osteosynthesis plate), was obtained from two unstained sections. The obtained sections were attacked on a microscope slide (as previously described) and thinned down to a thickness of 15 ± 3 µm. These sections were stained with Stevenel’s blue/picro-fuchsin and mounted with a cover glass using a mounting medium. Histological and quantitative histo-morphometric analyses were carried out as previously described with a digital scanner and with an optic microscope. The histological and histo-morphometric parameters assessed were inflammation (i.e., presence of polymorphonuclear cells, lymphocytes plasma cells, macrophages, giant cells) and neovascularization (vessels/microvessels numbers).

Microhardness

Bone maturation, mineralization, and mechanical strength of newly formed bone were evaluated with the Vickers indentation test in the polymerized blocks containing the scaffolds (sectioned along a plane parallel to the long axis of the implant, as described above), and in 6 non-implanted scaffolds embedded in methyl methacrylate. This test allows evaluation of resistance of bone to indentation which can reflect biochemical proprieties such as mineral content. The microhardness evaluations were performed through a Vickers indentation test using a microdurometer (DuraScan 70; EMCO-TEST Prüfmaschinen, Kuchl, Austria) connected to an optical microscope, which also allows to identify and differentiate the newly formed bone, formed by thin and dense trabeculae with new osteonic systems, from the preexisting host bone, cortical bone formed by dense and compact bone. Microhardness measurements were performed perpendicularly to the surface with a Vickers indenter (quadrangular pyramid with an angle of apical 136° 15′) applying a load of 0.05 Kg and for a time of 5 s.

The index of bone microhardness (HV) was calculated by dividing the indentation strength for the imprint surface on the bone and observed under the microscope (Table 1). In the non-implanted scaffolds, the microhardness average value was calculated on a mean of 10 measurements. The microhardness average value for each metatarsus was calculated on a mean of 10 measurements for each examined area: 1) pre-existing host bone, 2) bone at the scaffold and AG interface, 3) scaffold/AG, and 4) newly formed bone. Finally, the bone mineralization index (BMI) was calculated as a percentage by dividing the HV of newly formed bone by the HV of preexisting host bone, as reported in Table 1.

TABLE 1.

Microhardness parameters.

Parameter	Description
The index of bone microhardness (HV)	Calculated by dividing the indentation strength for the imprint surface on the bone and observed under the microscope
Bone mineralization index (BMI %)	$BMI = \frac{HV of new bone}{HV of host bone} \times 100$

Open in a new tab

HV, Vickers Hardness Index; BMI, Bone Mineralization Index.

Statistical Analyses

Statistical analysis was performed using R v.3.3.3 software (R Team, 2002). The significant level considered was α = 0.05. For dependent variables, the normal distribution (ShapiroWilk test) and homogeneity of variance (Levene test) of data were evaluated. Data are reported as box-plot graphs or histograms representing mean ± SD, at a significant level of p < 0.05.

Data from imaging (X-ray and CT) were analyzed with one-way ANOVA. Data on vessels and microvessels number were analyzed with nonparametric KruskalWallis test followed by MannWhitney U test. The other ex-vivo and microhardness data of scaffolds were analyzed with a one-way ANOVA by considering “scaffold” factor—three levels (GB-1, GB-2, and AG)—as fixed effect. When ex-vivo data violated the assumption of homogeneity of variances, Welch’s ANOVA was used. Two-way ANOVA was used to analyze bone microhardness results by considering factors “scaffold” and “measurement site”—three levels (host bone; newly formed bone within scaffold; and cortical bone at the interface with scaffold)—as fixed effects. Then, the post hoc pair-wise comparison Tukey HSD test was used to compare data among groups. Finally, the Bonferroni correction of p-value was applied rejecting the null hypothesis for each p ≤ α/m, where α is the desired overall alpha level and m the number of hypotheses, in order to control the family-wise error rate.

Results

Scaffolds Preparation and Characterization

Figure 1 shows a macroscopic view of the GB scaffold, highlighting the channel-like porosity running throughout the whole scaffold length, as well as the central hole, created to mimic the medullary cavity of long bones and to improve new bone penetration and vascularization in vivo. Figure 2 shows the X-ray diffraction patterns of GB-1 (Figure 2A) and GB-2 (Figure 2B), showing that both scaffolds are made of hydroxyapatite, with a minor amount of β-TCP in GB-1. In this respect, Table 2 shows the semiquantitative phase composition of the scaffolds, as well as the chemical composition obtained by ICP-OES spectrometer, highlighting the carbonation extent and the presence of small amounts of Mg and Sr as doping ions in GB-1.

XRD spectra of GB1 **(A)** and GB-2 **(B)**. The black signs point to major XRD reflections belonging to the TCP phase.

TABLE 2.

Composition and porosity extent of GB-1 and GB-2 scaffolds.

Scaffold type	GB-1	GB-2
	HA + β-TCP	HA only
	—	—
Phase composition	—	—
HA	85 ± 10 vol%	100%
β-TCP	15 ± 10 vol%	0%
	—	—
Chemical composition	—	—
Ca/P molar ratio	1.6 ± 0.1	1.6 ± 0.1
Mg	< 1.5 wt%	—
Sr	< 1.5 wt%	—
CO₃	< 8 wt%	—
	—	—
Physical properties	—	—
Shape	cylinder	cylinder
Total porosity	>45%	>45%
Micropore size <10 μm	>30%	>30%
Macropore size >10 μm	>30%	>30%

Open in a new tab

SEM micrographs in Figure 3 show the typical hierarchical structure and pore morphology of biomorphic scaffolds at different size scales. In particular, Figures 3A,B show the channel-like wide open porosity of the biomorphic scaffolds, running throughout the whole scaffold, whereas Figure 3C shows the typical architecture of internal channel walls, characterized by a cell structure mimicking the natural parenchyma of the rattan wood used as a template for the scaffold fabrication [17]. Figure 3D shows a detail of the nano-size lamellar structure characterizing both the biomorphic scaffolds. Both GB-1 and GB-2 scaffolds show very similar porosity extent and 3D organization, as also obtained by mercury intrusion porosimetry (Table 2).

SEM micrographs showing: **(A)** the upper scaffold surface showing large channel-like pores, better evidenced in **(B)** where a different tilt angle shows the external channel walls; **(C)** a longitudinal section of the scaffold showing the internal channel wall, revealing a cell-like pore structure interconnecting the different large channels; **(D)** a detail of the scaffold structure, showing its constituting building blocks, made of nano-sized hydroxyapatite lamellae.

The mechanical behavior of the biomorphic scaffolds, tested under compression along the direction of the main channels (Figure 4), shows, after a linear elastic stage, a stress plateau in the range 11–15 MPa where the scaffold starts to fracture but maintaining good structural integrity up to a deformation of ∼30%. Such a damage-tolerant behavior is quite unusual for pure ceramic materials and can be ascribed to the hierarchical porous architecture of the biomorphic scaffolds that facilitates dissipation of the mechanical forces, thus preventing sudden rupture, similarly as occurs in bone (Gibson et al., 1999). Comparing the two scaffolds, GB-1 shows slightly higher fracture strength: particularly we measured compressive strength values of 14.8 ± 1.2 MPa and 11.1 ± 1.4 MPa for GB-1 and GB-2 scaffold, respectively, as well as improved stability at the stress plateau for GB-1 than for GB-2.

Compressive strength analysis on GB-1 **(A)** and GB-2 **(B)** scaffolds.