Biocompatibility, resorption and biofunctionality of a new synthetic biodegradable membrane for guided bone regeneration

Medical grade PLGA copolymer (85/15 molar ratio, inherent viscosity midpoint of 3.1 dl g⁻¹) was purchased from Corbion Purac (ref. Purasorb PLG 8531, Corbion Purac BV, Gorinchem, The Netherlands). The PLGA dental membrane (Tisseos^®, Biomedical Tissues SAS, Nantes, France) was manufactured using a proprietary process in an ISO 6 clean room in respect of the ISO 13485 certification for medical devices. Briefly, the polymer was dissolved in chloroform and a dense layer was prepared by film casting. A micro-fibrous layer was then prepared by jet-spraying with compressed air. After drying overnight, the bi-layered membrane was cut into 15  ×  25 mm or 20  ×  30 mm pieces, packaged in double sealed pouches and sterilized with gamma irradiation at 25 kGy (Ionisos SA, Dagneux, France).

A commercially available PLA dental membrane (PLA; Epiguide^®, Kensey Nash Corp., Exton, USA) was purchased for comparison of the biocompatibility and resorption rates in the subcutis of rats. A cross-linked porcine collagen membrane (COLL; CovaMax^®, Biotech Dental, Salon-de-Provence, France) and β-TCP granules (TCP; 0.5–1 mm, Crossbone^®, Biotech Dental) were used in bio-functionality models of bone regeneration of calvaria defects in rats and mandible defects in rabbits.

The microstructure of the membrane was investigated by using a scanning electron microscope (SEM; Hitachi TM3000 Tabletop Microscope, Tokyo, Japan). Thickness was measured with a non-contact laser device (Keyence LK-G87; Keyence, Itasca, IL, USA) at several locations (n  =  10). The mechanical properties of the PLGA membrane were measured using a mechanical test bench (FAVIMAT Fiber Test) equipped with holders of 7.5 mm in width at a distance of 10 mm and a load cell of 3200 cN. A pre-load of 0.5 cN and a crosshead speed of 50 mm min⁻¹ were applied. The tensile strength, elasticity and maximum tear resistance values were determined (n  =  5). The resistance to suturing was measured according to the NF 94-801 2007 standard. A 4/0 suture (Prolene, Ethicon) was passed through the membrane, 5 mm away from the side. The membrane was fixed with a 15 mm wide holder while the suture was fixed on the load cell at a distance of 20 mm. A crosshead speed of 100 mm min⁻¹ was applied. The maximum load at rupture was recorded (n  =  5).

The degradation of the PLGA membrane was first studied in vitro. After weighing, membranes measuring 20  ×  30 mm and 400 µm thick were immersed in closed glass vials containing 17 ml of phosphate buffer saline (PBS; Lonza) in a thermoregulated oven at 37 °C. After 2, 4, 12, 26, 38 and 48 weeks, the pH was measured and the solution was passed through cellulose filters. After drying, a digital calliper and laser device were used to measure the dimensions and thickness of the membranes. The remaining membranes were weighed to determine the weight lost during hydrolysis and dissolved in pure chloroform at a concentration of 0.1 g dl⁻¹. The intrinsic viscosity was measured using an Ubbelohde tube in a thermoregulated water bath at 25  ±  1 °C. For each time point n  =  4 membranes were used.

The surgery protocols were conducted in accordance with European regulations on animal welfare and complied with the guidelines of the Animal Care and Use Committee of the University of Nantes, France. The animals were housed in cages with pelleted food and water in a temperature-controlled room with a 12 h artificial day/night cycle at the experimental therapeutic unit of the Faculty of Medicine. They were acclimatised for at least 2 weeks prior to surgery.

The biocompatibility and resorption of the membranes were assessed in the subcutis of rats in accordance with the ISO 10993-6 standard [25]. A pilot experiment of subcutis implantation of the porcine derived collagen (COLL) membrane indicated its complete resorption within 8 weeks. For these reasons, the study was conducted with another commercially available dental membrane, PLA, which has a comparable degradation rate than the experimental PLGA membrane. Forty albino rats of the Wistar strain (male, 7 weeks old, body weight 125–150 g) were purchased from a certified stockbreeder (Janvier Labs, Le Genest Saint Isle, France). The rats were placed under general anaesthesia by inhalation of 1.5% isoflurane (Forane^®, Baxter Healthcare Corp. USA). Pre- and post-operative analgesia was provided by intramuscular injection of buprenorphine (30 µl kg⁻¹, 2 times d⁻¹ for 3 d; Buprécare, Axience, Pantin, France). The animals were identified with numbered ear tags. An area approximately 8 cm in length and 4 cm in width was depilated on the back of each rat using an electric shaver. After disinfection with povidone iodine 10% solution (Betadine^®), a sterile field was placed on the back of the animal. A skin incision was made para median, along the vertebral column, followed by separation of two unconnected subcutaneous pouches by means of blunt dissection. Each animal received a 10 mm diameter PLGA (Tisseos) and PLA (Epi-Guide) disc in two separate subcutis pouches. The PLGA disc was implanted on the right side with the micro-fibrous layer facing the muscle, and the PLA disc was placed on the left. The skin was sutured using non-absorbable 4/0 sutures (Peters Surgicals, Bobigny, France). After 4, 8, 16, 26 and 52 weeks of implantation, the animals were euthanized by prolonged inhalation of carbon dioxide gas. Immediately after euthanasia, the back of the animal was shaved and a wide skin incision was made to expose the subcutis-implanted membranes. The sites were examined and photographed with a graduated ruler in order to determine the diameter of the membrane remaining. After dissection, samples isolating each membrane were fixed in 10 volumes of neutral 4% formaldehyde (MicromMicrotech, France) and stored at 4 °C for 2 d.

The capacity of the membranes to guide the regeneration of a critical-sized bone defect was assessed on the calvaria of rats. Three experimental groups (defect left EMPTY, PLGA, and COLL membranes) and two time points (4 and 8 weeks) were designed. For each experimental group, n  =  6 samples were used giving a total of 36 animals. The rats (Wistar, male, 7 weeks old, body weight 125–150 g, Janvier Labs) were operated under anaesthesia by inhalation of 1.5% isoflurane. Analgesic (Buprenorphine, 30 µl kg⁻¹, 2 times d⁻¹ for 3 d) was injected before and after surgery, as described in the previous section. The head was shaved, disinfected with iodine solution and covered with a sterile fenestrated sheet. Local anaesthesia was performed by injection of adrenaline articaine hydrochloride (0.5 ml, Alphacaine SP, Dentsply). A 1.5 cm skin incision was made and the skull was exposed by blunt dissection. The periosteum was detached from the calvaria bone. A critical-sized defect of 5 mm in diameter was carefully created on the calvaria bone using a trephine mounted on a dental micro-motor (Nouvag NM3000, Switzerland). Constant saline irrigation was maintained during drilling. The disc of calvaria bone was carefully removed. The PLGA membrane was cut into an 8 mm diameter disc and placed on the calvaria defect with the micro-fibrous layer facing the bone, while the dense layer was against the subcutis tissue. The skin incision was closed with non-resorbable 4/0 sutures. The surgical procedure was repeated with a commercially available collagen membrane (COLL) used as the control. Calvaria defects that were left empty served as negative controls. Following surgery, the animals were observed daily, and body weight was determined every week. After 4 and 8 weeks, the rats were euthanized by inhalation of an overdose of carbon dioxide gas. The calvarias were dissected and cut using a circular diamond saw mounted on a dental hand piece (Nouvag NM3000). The explants were observed for signs of tissue necrosis, inflammation or infection. The samples were fixed in buffered 4% formaldehyde solution for 2 d at 4 °C prior to histology. Radiographs were taken using a digital x-ray microradiography apparatus (Faxitron MX-20). Semi quantitative scoring [26] was used to assess bone regeneration in accordance with a previously published method. Briefly, a score of 0 corresponded to no bone formation within the defect, while a score of 4 was for bone that bridged the entire span of the defect.

This study was conducted to evaluate the GBR of a PLGA membrane covering a mandibular bone defect [27]. The defect was left empty or filled with TCP granules. The bone regeneration was compared with a dental membrane made of porcine collagen (COLL). A total of four groups were tested with two possible sites of implantation in the rabbit mandible with a healing time of 8 weeks. Given a sample size of n  =  6/group and bilateral implantation, 12 New Zealand White rabbits were used in this study (adult females, average body weight of 3.2–3.5 kg). The animals were purchased from a certified breeder (Hypharm, Roussay, France) and acclimatized in individual cages at the Experimental Unit. For analgesia, 1 h before and for 2 d after the surgery, the rabbits received buprenorphine (0.05 mg kg⁻¹, s.c. 12 hourly). General anaesthesia was obtained by injection of a mixture of xylazine/ketamine (intramuscular, Imalgene 0.35 ml kg⁻¹; Rompun 2% 0.25 ml kg⁻¹). Prophylactic antibiotherapy was administered by injection of Marbofloxacine (i.m. 2 mg kg⁻¹ for 3 d, Marbocyl 10% Vétoquinol). The mandible was shaved and disinfected with povidone iodine solution. The sites were covered with a sterile adhesive fenestrated sheet. Local anaesthesia was administered by injection of adrenaline hypochloride (0.5 ml, Alphacaine SP, Dentsply). A 2 cm cutaneous incision was made on the ramus, and the mandibular bone was exposed by smooth dissection. A bone defect with the dimensions 10  ×  5  ×  5 mm was created under saline irrigation using an oscillating saw mounted on a dental micro-motor (Nouvag NM3000) on both side of the mandible. The defects were filled with TCP granules or not, and covered with the bi-layered PLGA membrane, with the micro-fibrous layer facing the bone defect and the dense film facing out. The PLGA membrane was placed on the right side of the mandible. The skin tissue was closed with absorbable 4/0 braided sutures (4/0 Vicryl^®, Ethicon, USA). The collagen membrane was implanted on the left side with or without TCP granules. The animals were euthanized 8 weeks after the surgery under general anaesthesia by intra-cardiac injection of an overdose of sodium pentobarbital (1 ml Dolethal^®, Vetoquinol France). A broad dissection of the mandible was carried out and both upper rami were cut with a diamond saw. The specimens were immediately fixed in 10 volumes of buffered 4% formaldehyde solution and stored at 4 °C.

The mandibles were scanned by micro-computed tomography (microCT; Skyscan 1076, Belgium). The x-ray source was operated at a voltage of 50 kV and a current of 200 µA. The detector was rotated to 180° in 0.6° steps with an exposure time of 1.7 s, giving a spatial resolution of 18 µm per voxel. The 3D reconstructions were made using CTVOX software.

Once fixed, the rats' subcutis and calvaria samples were processed for decalcified paraffin histology. The samples were placed in cassettes and decalcified in 4.13% EDTA/0.2% PFA in PBS pH 7.4 for 96 h at 50 °C using an automated microwave decalcifying apparatus (KOS Histostation, Milestone Med. Corp. Kalamazoo, MI, USA). The samples were then rinsed with water and dehydrated in ascending series of 80, 95 and 100% ethanol baths, and finally in butanol for 30 min (Automated dehydration station, Microm Microtech, France). The samples were then immersed in liquid paraffin at 56 °C (Histowax) and embedded at  −16 °C. Blocks were cut by using a standard microtome (Leica RM2250). Thin histology sections (3–5 µm thick) were made perpendicular to the plane of the skin for the subcutis implants, and in the middle of the calvaria defects. The sections were mounted on microscopy glass slides and stained with hematoxylin/eosin or Masson's trichrome using an automated coloration station (Microm Microtech).

Non-decalcified histology with poly-methylmethacrylate (PMMA) resin embedding was used for the mandible defects in rabbits. The samples were dehydrated in ascending concentrations of ethanol and in acetone for 24 h each. They were then impregnated with methyl methacrylate for 48 h at  −20 °C. Finally, they were embedded in PMMA resin in the presence of an initiator and propagator for radical polymerization at 4 °C for 4 d. The samples were cut transversally into two parts using a circular diamond saw (Leica SP1600, Solms, Germany). One part was used to produce 30 µm stained sections and descriptive analyses of the bone defect. The section was mounted on a plexiglass microscope slide with UV light epoxy curing resin, and polished with SiC papers using grinding/polishing equipment (Meatserv 250, Büehler, Dusseldorf, Germany). After 1 min etching with hydrochloric acid ethanol (2 ml HCl 37 wt.%/98 ml absolute ethanol), the sections were stained by soaking in 1% toluidine blue for 90 s, rinsing with demi-H₂O and in 0.3% basic fushin for 60 s. Contiguous SEM images (Hitachi TM3000) of the block were taken at 50  ×  magnifications and assembled. The resulting image allowed the quantitative measurement of the newly formed mineralized bone and the quantity of biomaterial based on the grey intensity in the defect that defined the region of interest (ROI). Image analysis was carried out using the freeware ImageJ (National Institutes of Health, Bethesda, MD, USA). The newly formed bone surface was expressed as a percentage of the ROI.

The stained slices were scanned (NanoZoomer 2.0RS, Hamamatsu Corp. Japan) and observed with the virtual microscope (NDP view software, Hamamatsu Corp). Tissues and cells in the different membranes were identified and scored by M F Heymann, a medical doctor specializing in anatomic pathology. Tissue necrosis, neovascularization, fibrosis and fat tissue infiltration were scored according to the 10993-6 standard [26]. The number of granulocytes, lymphocytes, plasmocytes, macrophages and giant cells were numbered in the membranes at a magnification of  ×400. A score of 0 was given in the absence of cells and score 4 corresponded to a high number of cells gathered in wide-ranging areas

Quantitative data were presented as the mean  ±  standard deviation. Analyses of variances followed by least significant difference post hoc assessments were applied to compare the groups using GraphPad Prism 5.0 software (GraphPad Software Inc., La Jolla, USA). Differences were considered significant if their p-values were less than 5% (^*p  <  0.05).

The synthetic PLGA membrane is shown in figure 1. The membrane exhibited two layers, a micro-fibrous mat side and a dense glossy side (figure 1(A)). The membrane was designed with a dense layer to prevent epithelial tissue ingrowth, and a micro-fibrous layer to guide bone regeneration. These two layers are clearly visible on the cross-section observed by SEM (figure 1(B)). The thin and dense film appeared in contact with a thick and porous layer made of micro-fibres. The overall thickness of the membrane was around 400 µm while the thickness of the dense layer was approximately 25 µm. A flat, smooth, non-porous surface with no evidence of surface irregularity was observed on the dense side (figure 1(C)). In contrast, the micro-fibrous layer exhibited a highly porous structure with interlaced non-woven fibres. The diameters of the fibres ranged from 0.2 to 2 µm (figure 1(D)). The membranes had a weight of 50  ±  5 mg on average for a 20  ×  30 mm size, corresponding to a basis weight of approximately 80 g m⁻². This basis weight was comparable to standard cellulose paper.

Zoom In Zoom Out Reset image size

The mechanical properties of the PLGA membrane were measured to ensure both their resistance to surgical handling, and their compatibility with suturing. The ultimate stress and strain were determined as 6.7  ±  2.6 N and 7.6  ±  4.0%, respectively. The maximum resistance to initial tear was 47  ±  5 cN. The maximum resistance to suture traction was 2.83  ±  1.09 N.

The hydrolysis of the PLGA membrane was studied in an aqueous physiological environment (figure 2). After 2 weeks in PBS, the PLGA membrane had a similar shape with a bi-layered structure. After 4 weeks, the PLGA membranes had bent and swelled, but maintained their physical integrity with a bi-layered structure. After 12 weeks of hydrolysis, the membrane remained in one piece, but exhibited significant bending and swelling. No membrane could be found after 26 weeks of hydrolysis in PBS. As shown in figure 2, there was a good correlation between the weight loss of the PLGA membranes and the pH variations in the PBS solution. The weight loss was not significant in the first 4 weeks but began to increase at 12 weeks, reaching a plateau of around 100% at 26 weeks. Similarly, the pH remained stable, at around 7.4, for the first 12 weeks of hydrolysis in PBS. The pH then dropped to 3.2 at 26 weeks and remained stable at this value until 48 weeks. Intrinsic viscosity was also measured, as it is a good indication of hydrolysis of the polymer chains. The viscosity of the PLGA polymer decreased rapidly as a function of soaking time in PBS. After only 4 and 8 weeks, the viscosity had reduced two- and five-fold respectively, indicating a significant breakdown in the polymer chains. Overall, these results indicated that the PLGA membrane had completely degraded in approximately 26 weeks in vitro.

Zoom In Zoom Out Reset image size

All rats recovered with no complications and macroscopic examination showed a good wound closure without signs of inflammation or infection. After 4 and 8 weeks of subcutis implantation, both the PLA and PLGA membranes were clearly visible and encapsulated within a vascularized fibrous tissue. Both types of disc had already reduced in diameter. After 16 weeks, both the PLA and PLGA membranes appeared to be fragmented. After 52 weeks, the PLGA membrane could not be found while the PLA membrane discs had greatly reduced in diameter.

Histology images of the PLA and PLGA membranes are shown in figure 3. After 4 weeks, fibrosis was observed around both types of implant. The fibrous capsule was narrow, with a maximum of 1–4 cell layers around both types of membrane. No fat tissue infiltration was observed in either case. Neovascularization was focalized with a maximum of 4–7 capillaries. In the PLA membrane, there was little cell colonization, particularly after a short implantation time. In contrast, many cells had infiltrated the polymer micro-fibres of the PLGA membrane at as early as 4 weeks. The dense film (white) was visible up to 8 weeks but appeared highly distorted and fragmented. The polymer micro-fibres of the PLGA membranes appeared in white bundles and loose strands. At 8 weeks, many macrophage-like cells and multi-nucleated giant cells have infiltrated both types of polymer membrane (figure 4). Collagen (green) was also present between the PLGA micro-fibres. After 16 weeks, the diameters of the PLGA polymer micro-fibres increased sharply because of hydrolysis and swelling. The PLA membrane appeared fragmented at 16 weeks with the presence of collagen between the fragments. At 26 weeks, the PLGA membrane was almost completely degraded with only a few polymer micro-fibres visible. As shown in figure 4, these remaining PLGA fibres were surrounded by macrophages and multi-nucleated giant cells, embedded in collagen fibrous tissue. At 26 weeks, the PLA membrane appeared highly fragmented and similarly invaded by macrophages and multi-nucleated cells, as well as by collagenous fibrous tissue.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The reduction in the diameter of both membranes was plotted as a function of the implantation time (figure 5(A)). Both types of membrane considerably reduced in diameter after 4 weeks. For the PLA membrane, the diameter decreased gradually over 52 weeks. For the PLGA membrane, the reduction in diameter was more rapid, and only four out of six implanted PLGA membranes could be retrieved at 26 weeks. A good correlation was therefore found between in vitro hydrolysis and in vivo degradation of the PLGA membranes.

Zoom In Zoom Out Reset image size

Counting of the macrophages and multi-nucleated giant cells indicated less of these cells in contact with the PLGA membranes than with the PLA membranes (figures 5(B) and (C)). These cellular elements were characteristic of local inflammation due to the presence of a foreign body reaction. Counts of the other cells, such as granulocytes, plasmocytes and lymphocytes, as well as tissue necrosis, neovascularization and fibrosis always scored less 2 for both membranes (data not shown). Both types of membrane can therefore be considered to be biocompatible.

All the animals recovered from surgery without complications and gained weight normally. As shown in the radiographs in figure 6, the 5 mm diameter calvaria defect had a critical size, as only marginal bone healing was observed at 4 and 8 weeks. When a collagen membrane (COLL) was applied to the defect, bone healing was observed on the edge of the circular defect. Centripetal bone growth and, in some cases, partial bone bridging of defects were observed in the case of the synthetic PLGA membrane. Semi-quantitative scoring corroborated these observations with an increase of GBR using the synthetic PLGA membrane in comparison to the defects left empty (table 1). However, the difference was not statistically significant.

Zoom In Zoom Out Reset image size

Histology sections of the calvaria defects for the three groups at 4 and 8 weeks are shown in figure 7. After 4 weeks, the defects left empty appeared weakly healed. The collagen membrane was visible but did not completely cover the bone defect. In contrast, the PLGA membrane was clearly recognizable with its bi-layer structure, consisting of a dense film and a layer of micro-fibres. The PLGA membrane filled the defect area and guided the centripetal bone healing. After 8 weeks, the same histology results were observed. Partial bone healing and a few bone spicules were observed in the critical-sized defects left empty. As previously observed, the porcine collagen membrane only partially covered the defect, but still allowed GBR. The PLGA membrane was clearly visible across the bone defect. The dense film had functioned as a barrier, while the micro-fibrous layer guided the bone healing. As expected, the new PLGA membrane had the properties necessary to guide bone regeneration.

Zoom In Zoom Out Reset image size

All the animals recovered from surgery without complications and gained weight normally. After 8 weeks, micro-tomographies of the rabbit mandibles were taken (figure 8(A)). The sub-mandibular bone defects covered by the collagen or the PLGA dental membrane were still visible. No signs of osteolysis were observed around the defects covered with either the COLL membrane or the synthetic PLGA membrane. As shown in figure 8(B), bone growth was comparable in the defects covered with either type of membrane. When the defects were filled with TCP granules and covered with the membranes, more bone growth was observed than without the synthetic bone filler (figure 8(C)). In both cases, newly formed bone was observed in direct contact with the TCP granules. These qualitative observations were corroborated by histomorphometry analysis of the bone growth. The amount of bone in the mandibular defects was 24.6  ±  3.6% and 30.0  ±  8.7% for the collagen and PLGA membranes, respectively. When TCP granules were used in combination with membranes, newly formed bone was 39.1  ±  3.4% and 39.4  ±  2.7% for the COLL and PLGA membranes, respectively. These histomorphometry results were, however, not statistically significant. In this model, both types of membranes formed a physical barrier to maintain the TCP granules, favoured bone regeneration and prevented epithelial tissue ingrowth in the bone defects.

Zoom In Zoom Out Reset image size