Collagen Membrane for Guided Cortical Bone Regeneration: Characterization, Preclinical and Proof of Concept Clinical Studies

Background: Treatment of cortical bone defects is a clinical challenge. Guided bone regeneration (GBR), commonly used in oral in maxillofacial dental surgery, may show promise for orthopedic application in repair of cortical defects. However, a limitation in the use of GBR for cortical bone defects is the lack of an ideal scaffold that provides sufficient mechanical support to bridge the cortical bone with minimal interference in the repair process. We have developed a new collagen membrane, CelGro™, for use in GBR. We report the material characterisation of CelGro™, and evaluate the performance of CelGro™ in translational preclinical and clinical studies. Methods: Scanning electron microscopy (SEM), micro computed tomography (micro-CT) and transmission electron microscopy (TEM) were used to examine the structural morphology of CelGro™. Purity and biochemical composition of CelGro™ was evaluated by Western-blot, immunohistochemistry and confocal microscopy. Physical and chemical properties of CelGro™ were examined and compared with another commercially available collagen membrane. The pre-clinical evaluation was conducted using a cortical bone defect model in the New Zealand white rabbit. Cortical bone regeneration in defects of the femoral diaphysis were evaluated at 30 days and 60 days after intervention, by micro-CT and histology. A clinical study to evaluate the performance of CelGro™ in GBR for treatment of bone augmentation surrounding dental implants was also performed. The clinical outcomes were evaluated by semi quantitative tissue condition assessments and cone-beam computed tomography (CBCT) scan. Results: CelGro™ has a bilayer structure of different fibre alignment and is composed almost exclusively of type I collagen. CelGro™ was found to be completely acellular and a clinically significant xenoantigen, α -gal, was not detected. CelGro™ displayed less deformity and better mechanical strength as compared to Bio-Gide ® . In the preclinical study, CelGro™ demonstrated enhanced bone-modelling activity and cortical bone healing. Micro-CT evaluation showed early bony bridging over the defect area 30 days distance from the implant shoulder to first bone/implant contact (DIB) and increased horizontal thickness of facial bone wall (HT). Conclusion: The findings of our study demonstrate that CelGro™ is an ideal membrane for GBR not only in oral maxillofacial reconstructive surgery but also in orthopedic applications.

distance from the implant shoulder to first bone/implant contact (DIB) and increased horizontal thickness of facial bone wall (HT).

Conclusion:
The findings of our study demonstrate that CelGro™ is an ideal membrane for GBR not only in oral maxillofacial reconstructive surgery but also in orthopedic applications.

KEYWORDS
Collagen membrane, guided bone regeneration, animal study, clinical trial, dental implant, bone defect

Background
Critical bone defects impair biomechanics and structural stability of the skeleton, leading to poor mobility, ongoing functional deficits and an overall decrease in quality of life. Bone defects may be primary, as a result of trauma, or secondary to diseases including tumour, infection, rheumatoid arthritis and renal failure (1, 2). Defects in the diaphysis of the long bone is one of the common defects in orthopedic clinics (3). Due to a relatively poor blood supply compared to the epiphysis and metaphysis, diaphyseal injuries are at higher risk of nonunion as a result of inadequate vascularization and gapping at the defect site. Clinically, treatment of cortical bone defects is always a challenge. The current standard of care generally employs bone grafting techniques which are associated with high infection risk, sub-optimal osteointegration and the need for extensive surgical intervention (4).

Guided bone regeneration (GBR) is an established technique in oral maxillofacial
reconstructive surgery to regenerate the cancellous bone of the maxilla and mandible (5,6). In GBR, bone substitute is used to fill the defect, which is then overlaid by a barrier membrane.
The latter creates a favorable microenvironment for repopulation of osteoprogenitor cells into bone substitute to guide bone repair and prevent non-osseous tissue ingrowth from the gingiva (7). Ideal membranes for GBR are osteoconductive, in that they are able to directly stimulate the growth of osseous cells (8), promote vascularization of the healing tissue to fulfil tissue diffusion requirements and promote osseous regeneration (9). While several synthetic materials have been used for GBR, bio-resorbable scaffolds are favored as it negates the need for postoperative retrieval. Theoretically, GBR may enhance bone regeneration in orthopedic applications, but the use of GBR for this purpose has been limited by the lack of an ideal barrier membrane that able to support bridging of cortical bone with minimal interference in the repair process.
Bioactive membranes derived from collagen are a popular choice in tissue engineering applications. Collagens are the most abundant extracellular matrix (ECM) proteins (10), with low immunogenicity, and inherent bioactivity and biodegradability. The porous structure of a collagen-derived barrier membrane may also be ideal to facilitate tissue ingrowth and hence efficient bone formation (11). Despite many attempts, poor mechanical properties, as well as local toxicity from remnants of crosslinking chemicals such as glutaraldehyde, (12) has hindered the application of collagen-based GBR membranes in orthopedics. Previously, we used a new collagen membrane, CelGro™, in the repair of a cortical bone defect in a rat model.
CelGro™ was shown to facilitate bridging the gap of a cortical bone defect (13) and improve vascularization (14). The induction of cortical bone regeneration was evaluated using a combination of CelGro™ and recombinant human bone morphogenic protein-2 (BMP-2), which was found to significantly improve repair of both cancellous and cortical metaphyseal defects (13). Since CelGro™ has been shown to induce cellular recruitment, upregulate proosteogenic factors at the implant site(15) (16) and promote vascularization (17), we hypothesized that CelGro™ alone, without growth factors such as BMP-2, is able to directly participate in repair of cortical defects through osseointegration property.
In this study, we evaluated the physical characteristics of CelGro™, and determined its efficacy as a bioactive barrier membrane for GBR in a rabbit model of cortical bone regeneration and in a human clinical study for the treatment of periodontal bone defects around dental implants.

Biomaterials
CelGro™ is a porcine-derived collagen membrane developed at the University of Western Australia (18,19) and manufactured in Australia by Orthocell Ltd. Porcine connective tissue was defatted, followed by denaturing of non-collagenous proteins using 1% (v/v) sodium dodecyl sulphate and 0.5% (v/v) LiCl at 4 o C overnight. This was followed by further processing of the tissue in 0.5% (v/v) HCl to denature the collagen, then neutralizing with 0.5% (v/v) NaOH solution. The tissue was subjected to mechanical stretching to reach the desired size and thickness before further denaturation in a solution of 1% (v/v) HCl solution for 24 hours.
Bio-Gide ® (Geistlich Pharma) is a porcine-derived collagen membrane used previously for GBR and thus was used as for comparison. It has been shown that Bio-Gide ® is composed of Type I and III collagen and has a bilayer structure. The collagen fibrils are not crosslinked or modified from their native structure (20).

Characterization
Initially, structural morphology of CelGro™ was characterized by scanning electron microscopy (SEM; Supra 55, Zeiss), micro-computed tomography (micro-CT; Xradia 520, Zeiss) and transmission electron microscopy (TEM; JEM 2100, Jeol) at the Centre for Microscopy, Characterisation and Analysis, University of Western Australia. Protein composition of CelGro ™ was determined by Western blot and immunohistochemistry. The presence of residual cells and galactose-alpha-1,3-galactose (α-gal) on the membranes was investigated by confocal laser scanning microscopy (CLSM) (Nikon A1, Nikon).
Physical characteristics of CelGro ™ and Bio-Gide ® were compared, assessing structural morphology, physico-chemical properties and mechanical strength. To assess structural morphology, membranes were fixed in paraformaldehyde (4%) and embedded in paraffin.
Sections (4-6µm) were cut and transferred onto glass slides, dewaxed, and stained with Goldner's Trichrome and examined by light microscopy, or evaluated by SEM (Leica Cambridge S260) with measurements of pore size and collagen bundle diameter on the different sides. Physicochemical testing included pH, thickness, water retention and mechanical strength of each membrane. Membranes (10x10 mm) were immersed in 2 ml 0.9% NaCl at room temperature for 20 minutes. The pH of the immersion fluid was measured using a Horiba LAQUA twin compact hand-held pH meter after equilibration. Thickness and water retention capacity was performed using the method described by Pallela et al (21). Dry thickness of each article was measured using a calibrated THO1 electronic thickness gauge.
Water retention of the test articles at equilibrium (ER) was calculated by weighing the materials while dry (W dry ) and immediately after immersion in 0.9%NaCl solution (W wet ) and centrifugation to remove unbound moisture. The following equation was used to calculate water retention capacity: Mechanical strength was measured by evaluating the force required to tear the membrane.
Each article was cut into three 3cmx0.5cm samples, immersed in 0.9% NaCl and equilibrated at room temperature for 90 minutes. After draining excess moisture, the thickness of the membrane was measured at the top, middle, and bottom of the sample using an electronic thickness gauge. Testing grips were attached to the opposing 0.5cm sides of each test article strip and loaded into the Instron mechanical testing system (Model5566, Instron). The test article strips were pre-loaded with 0.1N and force loading was increased at a speed of 2mm/sec until failure at the centre of the membrane was reached. Ultimate tensile strength was calculated as failure force divided by the area of the testing material (~30mm x 5mm).

Animal study
To examine the performance of collagen membranes in cortical bone regeneration, CelGro™ were harvested from the right ilium, which was exposed via incision at the crista iliaca, using haemostatic forceps. Bone autograft was harvested for AG, AG+B and AG+C groups. The cortical bone in the diaphysis region of right distal femur was then exposed through a longitudinal medial parapatellar incision. A defect of 5mm in diameter, 5mm in depth, and 2 mm from the linea epiphysialis, was created in the cortical bone diaphysis of the right femur (in groups D, AG, AG+B, AG+C). Harvested trabecular bone graft was implanted into the bone defect site (in groups AG, AG+B, AG+C).
For treatment groups receiving a collagen membrane (AG+B and AG+C), a pre-sterilized circular piece of membrane (CelGro™ or Bio-Gide ® ), measuring 5mm in diameter, was saturated in PBS and placed over the defect site with the rough side facing the bone interface.
Gentle pressure was applied to the membrane to smooth the edge over the bone defect site.
For all surgical groups, the incision over the defect site was then sutured closed.
All animals received antibiotic prophylaxis (Cefazolin, 10mg/kg) administered intramuscularly once daily for three days post-operatively. Sacrifice via pentobarbital overdose occurred 30 days and 60 days post-operatively ( Fig S1), except for animals in the no treatment group, which were sacrificed at 60 days only.
To visualize and evaluate cortical bone defect healing in different groups, the area around the bone defect and normal cortical bone were scanned by micro-CT (Skyscan1176, Bruker) at 80 keV, 313 μA, isotropic resolution 8.89 μm, followed by reconstruction (NRecon, Bruker). The volume of interest (VOI) was precisely defined as 6mm x 5mm at the defect site (lateral) and on the normal cortical bone (medial) by CTANalyser (Bruker). The volume of bone tissue (BV) and porosity volume (Po.V) were expressed as a fraction of the selected total volume (TV).
Normalized cortical thickness (Δ Ct.Th) was calculated by the absolute value between the cortical thickness of the defect and normal bone tissue.
To determine the correlation between radiologic findings and histologic changes of the bone tissue at the cortical bone defect, histological examination was conducted on samples harvested HistoResin mounting medium and trimmed prior to sectioning. Sections (5 µm) were collected on silane-coated slides, covered with Kisoil Foil, pressed using a section mounting press and left at 37°C for 2-3 days. The resin was removed from the section prior to staining. Sections were stained with Goldner's Trichrome according to standard protocols (24) and analyzed microscopically (Aperio Scanscope, Leica).

Proof of concept human study
GBR is a standard procedure used routinely in clinical practice to preserve and restore bone A two-stage dental implant procedure was used for this study. The surgical procedures associated with GBR have been described elsewhere (25). In brief, following standard oral premedication and anesthesia, the implant site was prepared, removing all soft tissue remnants from the bone surface. An appropriately-sized screw-type titanium implant was placed into the alveolar bone of the mandible or maxilla using standard procedures. Void-filling material (Bio-Oss ® , Geistlich Pharma) was prepared and packed into the bone defect. CelGro™ was cut to size and placed in situ with the rough side facing the void-filling material and held in place with moderate pressure until adhesion to the site was achieved. The gingiva was sutured closed over the implant and patients received instructions for post-operative care (antiseptic mouthwash), antibiotics and analgesics as required according to standard protocols (26).
Re-entry surgery to anchor the implant abutment conducted 4-6 months post-operatively to Potential complications and adverse events related to the use of CelGro™ were also collected, including site specific reactions such as pain (unidirectional numeric rating scale -NRS), tissue condition (uneventful, inflammation, swelling, fibrin, plaque, necrosis) and soft tissue dehiscence, as well as systemic reactions.

Statistical methods
Data is presented as mean ± standard deviation unless otherwise stated. Due to the small sample size, micro-CT data from the rabbit femur cortical bone defect model was compared between two groups using a t-test. Statistical analysis, consisting of one-way analysis of variance (ANOVA), was performed on the clinical study CBCT data at baseline and re-entry surgery. All statistical analyses were carried out using SPSS software (v 24.0; IBM Corp, USA). P < 0.05 was considered as significantly different.

Characterization of CelGro™ collagen membrane
We first investigated topographical characteristics of CelGro™ to understand the structural configuration of the membrane. As observed by the SEM and micro-CT images (Fig.1A-E CelGro™ has a bilayer structure consisting of a smooth side (Fig. 1B, D) of densely aligned collagen bundles, and a rough side of randomly distributed collagen bundles that form a porous matrix (Fig. 1C, E). TEM results show the periodicity of CelGro™ collagen fibers range from 55 to 57nm, and the diameters range between 80 to 100 nm (Fig.1F, G).Immunohistochemistry showed that type I collagen staining was evident across the entire membrane (Fig 1H), but no type III collagen staining was detected (Fig S4). Western blot analysis showed high levels of collagen I proteins, including type γ, type β and type α, in CelGro™ membrane lysate (Fig. 1I).
These results indicate that CelGro™ is composed almost exclusively of Type I collagen.
co-contaminating non-collagen molecules and residual impurities of chemical processing may cause adverse biological reactions. For example, α-gal, which is commonly found in porcine tissue-derived biomaterials, may cause severe anaphylactic reactions in susceptible humans (29). To further examine the purity of CelGro™, we performed immunohistochemistry to evaluate the level of α-gal. CLSM images (Fig.1J) confirm that CelGro™ contains no cellular components and has no detectable levels of α-gal, evidenced by no fluorescence detected on both DAPI-stained and isolectin/GS-IB4-stained images respectively. Alpha-gal is labelled with Isolectin GS-IB4 and the cell nucleus is labelled with DAPI.
We then compared the physical characteristics of CelGro™ with a commercially available collagen membrane, Bio-Gide ® (30). Histology and SEM results (Figure 2 A-F) showed that pore size on the smooth side of each membrane, measured by SEM, ranged from 4-8µm for Bio-Gide ® and 3-7µm for CelGro™. As the average human fibroblast measures approximately 50-100µm in diameter (31, 32), the pore size of both CelGro™ and Bio-Gide ® is sufficient to prevent migration of epithelial cells across the scaffolds in vivo. The collagen bundles on the rough side were 5-30µm in diameter for Bio-Gide ® , and 5-20µm in diameter for CelGro™.
The gap between the collagen bundles, which is used to estimate the pore size on the rough side ranged from 44-171µm for Bio-Gide ® , and 75-142µm for CelGro™ (Table 1). CelGro™ exhibited greater uniformity in arrangement of collagen bundles, and the surface of the bundles was smoother compared to those observed for Bio-Gide ® (Figure 2 C-D).
values of the saline extract for both materials (Table 1) are close to the pH value of commercial saline (around pH 5.5), which is safe in vivo due to the buffering effect of body fluid (34). The presence of CelGro™ had little effect on the pH value of the saline solution, while Bio-Gide ® slightly alkalized the solution.
Moderate wettability promotes high levels of cell adhesion to scaffolds (35). We measured the change to thickness and mass of CelGro™ and Bio-Gide ® after wetting. CelGro™ had less variation in shape and thickness in both dry and wet conditions compared to Bio-Gide ® .
CelGro™ showed less deformation than Bio-Gide ® after wetting due to straighter alignment of collagen bundles in CelGro™, as seen in SEM, compared to Bio-Gide ® . However, there was no significant difference between two scaffolds in hydrophilicity, as water retention rates were very similar (Table 1).
Mechanical strength (ultimate tensile strength and failure force) of the two scaffolds was also tested. CelGro™ had a slightly higher failure force and, due to its thinner cross-sectional diameter, a slightly higher ultimate tensile strength. (Table 1)

Regeneration of cortical bone defects in a rabbit model
While the in vitro characterisation of CelGro™ showed that CelGro™ contains relatively pure type I collagen bundles and display less deformity and better mechanical strength, we next conducted a preclinical study to determine the efficacy of CelGro ™ for GBR in the repair of diaphysis cortical bones defects in rabbits. Previously, it has been shown that healing of a standard-sized defect in rabbit femoral bone treated with bone autograft occurs between 30 to 60 days after surgery (22) (23). To evaluate the efficacy of collagen membranes in conjunction At 30 days after surgery, the effect of a collagen membrane could be clearly observed by micro-CT (Fig. 3A). In both GBR groups (AG+B/AG+C), the regenerated bone had formed a continuous cortex over the defect area; the defects were completely sealed by the neo-cortex in At 60 days after surgery, lower Po.V/TV and increased BV/TV (Fig. 3C) were observed in the AG+B and AG+C treatment groups compared to the AG group and D group (P<0.05), and the regenerated bone cortex formed in the treatment groups with a collagen membrane (AG+B/AG+C) also showed significantly lower Δ Ct.Th (P<0.05) (Fig 3C). Together, the micro-CT data indicates that, compared to the treatment groups without a collagen membrane, use of a collagen membrane resulted in increased bone volume and a more mature cortex.
Dense, firm mineral precipitation with a similar shape and density to adjacent healthy cortical bone was observed by micro-CT (Fig. 3B) in the AG+B and AG+C treatment groups. Nearnormal corticalization of cortical bone defects in the CelGro ™ group (AG+C) was seen at 60 days, while a slight irregularity remained in the Bio-Gide ® group (AG+B) in sagittal and crosssectional imaging (Fig 3B). The micro-CT data was congruent with histological appearance in all groups. The cortical defect area was filled by linear neo-cortex surrounded with visible osteoid tissue, indicating highly active bone formation. The newly-formed cortical bone complete bridged the defect, and partially absorbed trabecular bone graft was found in the medullary compartment. CelGro™ showed improved cortical alignment and reduced porosity at the defect interface as compared to Bio-Gide ® (Fig 3D).

Proof of concept clinical study of CelGro™ in GBR
The promising pre-clinical data of cortical bone restoration in diaphyseal cortical bones defects of rabbits has promoted us to translate the study into the proof of concept clinical study in human. Owning the fact that selection of orthopedic patients with cortical bone defects is complex and since GBR is an established, commonly-used technique in dental implant procedures , we selected dental GBR for provide evidence for the clinical translation of CelGro™ in humans.
No device-related adverse outcomes, such as abnormal post-operative pain ( Figure S5) or wound infection were observed. Tissue conditions such as swelling and inflammation were generally mild and transient in nature, and did not affect treatment outcomes. One incidence of wound dehiscence was noted but the participant did not require intervention and no effect on bone regeneration was noted.

Discussion
The use of GBR to restore bone volume began in maxillofacial reconstruction. The advantage of barrier membranes in GBR is to prevent soft tissue ingrowth during bone regeneration.
Little evidence exists as to whether a barrier membrane can be used to guide cortical bone formation in large defects that do not heal adequately. We have developed CelGro™, a noncross-linked type I collagen-based membrane that can be used to simulate the ECM of bone for In the preclinical rabbit model, micro-CT analysis further confirmed that CelGro™ acts in an osteoconductive manner to facilitate cortical bone regeneration. In combination with the autogenous bone graft, use of CelGro™ leads to significantly increased bone volume and formation of a continuous bony bridge over the defect 30 days post-operatively. By 60 days post-operatively dense and firm mineral precipitation of similar shape and density to healthy cortical bone was observed. In general, autograft without GBR requires approximately 3 months (90 days) to achieve satisfactory bone healing in defects ranging from 4.5mm to 6mm in diameter on distal femur in rabbits (25) including peri-implantitis in 34% of cases (44) and flap dehiscence in 19.1% (45), our preliminary result of the clinical study is encouraging. We have seen no complications and successful restoration of bone tissue in GBR by CelGro™ in 16 implants among 10 participants

Conclusion
Type I collagen is major protein component of the ECM in bone. It has been shown that undenatured type I collagen has a dynamic, bioactive structure with the ability to regulate cellular proliferation, differentiation and repair (43,44). We showed that CelGro™ is a predominantly Type I collagen bilayer membrane with two distinct sides of differently-aligned collagen fibers.
We established in a preclinical model that the use of CelGro ™ together with autogenous bone grafting significantly improved cortical bone regeneration compared to autogenous grafting alone. Micro-CT imaging indicated CelGro ™ was superior to a similar comparator membrane, improved cortical alignment and decreased porosity at the defect site. The clinical study showed that CelGro™ is effective for bone regeneration in maxillofacial procedures with no adverse outcomes observed. The results of our study suggested that CelGro™ is ideal membrane for GBR not only in oral maxillofacial reconstructive surgery but also in orthopedic applications.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Ethics approval and consent to participate
The clinical study was registered at ANZCTR (trial ID ACTRN12615000027516). Animal experimental protocols were approved by the Ethics Committees of Kunming Medical University (Approval No. SCXK (DIAN) K2015-0004).

Consent for publication
All participants enrolled in this study provided written, informed consent and the study protocol were approved by the St John of God Health Care Human Research Ethics Committee.

Competing interests
One or more of the authors has declared the following potential conflict of interest: MH-Z is a scientific consultant to Orthocell Ltd, and holds a patent for CelGro™ collagen membrane; C-L is a Clinical Research Manager at Orthocell Ltd.