Animal experiments using the induced membrane procedure for bone tissue engineering purposes have provided evidence that the membrane has structural characteristics and biologic properties that may be used for bone tissue engineering purposes. Clinically relevant animal models have demonstrated that standardized particulate bone constructs can be used to repair large bone defects using the procedure and that the osteogenic ability of these constructs partially approaches that of bone autografts.
Clinical management of extensive segmental long-bone defects occurring after high-energy trauma or debridement for infected nonunion poses a substantial orthopedic challenge. Current solutions include the use of cancellous autografts when defects are 4 to 5 cm or smaller, whereas other procedures such as transfer of vascularized bone or intercalary bone transport methods are required to treat longer defects.
Masquelet and colleagues reported successful repair of wide diaphyseal defects (≤25 cm) with concurrent severe soft tissue loss in human patients using fresh autologous cancellous bone grafts in an original two-step surgical procedure. In the first step, a polymethylmethacrylate (PMMA) cement spacer was inserted into the defect inducing formation of an encapsulation membrane. After 8 weeks, the spacer was removed and the cavity filled with autologous cancellous graft harvested from the iliac crest. Bone union was usually achieved within 8.5 months, with patients recovering normal gait and motion. Interestingly, the procedure applied to all sorts of defects in contrast to the use of a massive allograft. Clinical trials in which the procedure has been performed have highlighted the advantages of using the induced membrane: (1) prevention of protrusion of adjacent soft tissues in the bone defect; (2) restraint of the graft in place; and, according to Masquelet, (3) maintenance of graft volume over time through either protection of the graft against resorption or local production of osteoinductive substances. Interestingly, animal studies have demonstrated that the membrane indeed has histologic characteristics and biologic properties that might facilitate bone healing.
Large quantities of autograft and sometimes repeated surgery to achieve bone union are nevertheless needed for successful outcome with this procedure. The development of tissue-engineered bone constructs (TEBC) combining osteoconductive scaffolds with autologous mesenchymal stem cells (MSC), the availability of bone morphogenetic proteins, have opened new avenues for designing efficient bone substitutes. The possibility of placing either skeletal stem cells or bone morphogenetic proteins (BMPs) onto scaffolds in the cavity delineated by the membrane is a very attractive prospect.
The purpose of this article was to review data from animal studies that have allowed better understanding of the structure and properties of the membrane as well as the preliminary and preclinical evaluations of new therapeutic strategies using the induced membrane procedure.
Animal models using the induced membrane procedure
The structure and biologic properties of the induced membrane have been studied in small (rabbit) and large (sheep) animal species in which implantation of PMMA cylinders in either the dorsal paraspinal subcutaneous tissue (ectopic implantation sites) or segmental bone defects (orthotopic implantation sites) have induced the formation of a membrane. These animal models have been further used to assess the osteogenic potential of different TEBC.
Subcutaneous implantation of PMMA cylinders (four 10-mm-diameter, 5-mm-long cylinders in rabbits, eighteen 15-mm-diameter, 28-mm-long cylinders in sheep), have made possible the analysis of the structure and biologic characteristics of the induced membrane at different time points after cement implantation.
Briefly, induced membranes were sampled at different time points postoperatively (2, 4, 6, and 8 weeks in rabbit experiments; 6 and 14 weeks in sheep experiments) and processed for histology and immunohistochemical analysis to assess structure and inflammatory reaction. Samples from membranes induced in sheep were immunostained to assess the presence of Col-1 and vascular endothelial growth factor (VEGF), as well as CD14 (for macrophages) and CBFA-1 (for osteoblast precursor cells) positive cells. Samples from membranes induced in rabbits were processed for protein extraction and dosage of growth factors (VEGF, transforming growth factor [TGF]-β1, BMP-2). Proliferation and differentiation of human bone marrow stromal cells grown into a culture medium containing protein extracts from either the induced membrane or subcutaneous tissue (controls) were also evaluated.
Implantation of PMMA cylinders in either 25-mm-long metatarsal or 30-mm-long femoral segmental bone critical size bone defects (CSD) stabilized with a bone plate in sheep has made possible the validation of the procedure described by Masquelet and colleagues using morselized corticocancellous bone autograft in conditions similar to the one encountered in clinical situations (eg, replacement of a CSD in a load-bearing bone, in an animal with bone healing and remodeling characteristics close to humans’). Briefly, a mid-diaphyseal segmental bone resection was created and filled with a cement spacer for 4 or 6 weeks. At these time points, the cement was carefully removed through an incision made along the formed encapsulation membrane surrounding the cement, and the resultant cavities were either left empty or filled with a morselized autologous bone graft sampled from the iliac crest. Bone healing was obtained in both femoral and metatarsal defects within 4 to 6 months, respectively. Membrane induced at the metatarsal site was sampled and processed for histology and immunohistochemistry at time of cement removal and bone grafting, 6 weeks postoperatively, and at the time the animals were humanely killed 6 months later, making possible the characterization of its structure in a bone site.
Characteristics and biologic properties of the induced membrane: data from animal experiments
Both subcutaneous and orthotopic implantation of PMMA cylinders in rabbit and sheep experiments induced the formation of a fibrous membrane. Implantation in ectopic and orthotopic sites in sheep have shown that at time of cement removal, 6 weeks after cement implantation, the membrane was 1 to 2 mm thick, not adherent to underlying cement thus allowing easy removal of the later and bled when incised. It was also mechanically competent in both locations and incised edges could be sutured without tension thus delineating a cavity in which either bone autograft or TEBC could be contained.
As shown in sheep in which PMMA cylinders were implanted into segmental metatarsal bone defects, the encapsulation membrane prevented adjacent soft tissue protrusion in the defect. In these experiments, the membrane was indeed found adherent to the resected bone edges and did not collapse after cement removal, thus delineating a cavity corresponding to the volume of the retrieved cement spacer. This was an important outcome because it has been reported that the use of pliable and nonrigid membranes (ie, silicone sheeting) across a defect may result in nonunions because of collapse of the membrane and subsequent interposition of tissue in the defect.
Interestingly, the membrane defined the shape of the regenerating bone as replacement bone retained the original size and contour along the defect throughout the study. The induced membrane indeed constrained the autograft, which was a critical issue, as in human patients, a major problem encountered with use of morselized graft for segmental bone losses greater than 4 cm long is constraint of bone chips to the site and prevention of soft tissue protrusion into the defect.
At last, use of an enclosed space to constrain autologous bone chips within the defect prevented ectopic bone formation. This is of particular relevance in tissue engineering strategies where maintenance of biologic agents within defect areas is critical.
Histologic and Immunohistochemical Characteristics
Histologic and immunohistochemical analysis of membrane specimens sampled 6 weeks after subcutaneous or orthotopical implantation of PMMA cylinders in both rabbit and sheep models have shown that the induced membrane was made of a collagenous matrix in which numerous elongated fibroblasticlike cells were found embedded ( Fig. 1 ). Collagen fibers were orientated parallel to the PMMA surface.
The membrane was highly vascularized in both ectopic and orthotopic sites ( Fig. 2 ). A mild inflammatory reaction with the presence of multinucleated giant cells was observed within induced membranes in subcutaneous and bone sites. Subcutaneous implantation of PMMA cylinders in rabbits showed that it decreased from 2 to 8 weeks postoperatively. These observations were further confirmed in sheep in which inflammatory cells were identified within encapsulation membranes in subcutaneous membranes sampled at time of cement removal (6 weeks) and constructs explantation (14 weeks postoperatively). Immunoreactivity for CD14 was not observed in 6-month membrane explants from sheep having undergone bone grafting using the induced membrane procedure in metatarsal location suggesting that at that time point, the inflammatory reaction had ceased.
The mild inflammatory reaction in the membranes induced experimentally is an interesting observation in tissue engineering. It differs from the histologic response observed when silicone spacers were used as an interposition material for reconstruction of traumatic bone loss in humans. The later indeed included T cell and macrophage reactions, with foreign-body giant cells embedded in a vascularized pseudosynovium. Silicone debris was frequently observed under polarized light in the connective tissue and within giant multinucleated cells. A similar foreign-body reaction has been reported after use of liquid PMMA (instead of the dough PMMA cement we used) implanted around cemented hip implants, and for kyphoplasty.
Although foreign body reaction within PMMA-induced membrane remained mild in animal experiments, further investigation is yet needed to test the hypothesis that the membrane can protect bone graft or resorbable osteoconductive materials from premature resorption as some authors have suggested.
Animal experiments have pointed out biologic characteristics that may be of clinical interest for tissue-engineering purposes.
Secretion of angiogenic factors
Protein extraction from membranes induced after subcutaneous PMMA implantation in rabbits has shown a higher concentration of VEGF compared with control subcutaneous samples. Furthermore, it was found that membranes induced in subcutaneous and bone sites were highly vascularized in both rabbit and sheep. Interestingly, vessels were found oriented perpendicular to the long axis of the cement implanted in the metatarsal site, toward the bone defect ( Fig. 2 ).
These findings may be of clinical relevance in bone tissue engineering, as angiogenesis is an essential step in osteogenesis. However, the hypothesis that the encapsulation membrane has a substantial role in graft or constructs vascularization remains to be validated.
Osteoinductive and/or osteogenic properties
Osteoinductive or osteogenic properties have been attributed to the induced membrane.
Protein extraction from membranes induced after subcutaneous PMMA implantation in rabbits have shown that they contained a higher concentration of TGF-β1 at different time points (2, 4, 6, and 8 weeks postoperatively) and of BMP-2 (4 and 6 weeks postoperatively) compared with control subcutaneous samples. It has further been shown that protein extraction from these membranes promoted human MSC proliferation and differentiation into bone-forming cells. BMP-2 production was found a maximum of 4 weeks postoperatively, suggesting that there might be an optimal time for bone grafting.
In one experiment in sheep in which cement spacers were implanted in segmental metatarsal defects, fibroblastlike cells positive for CBFA1, a critical transcription factor for osteoblast differentiation, were identified throughout the encapsulation membranes collected from two sheep at time of PMMA removal. These cells were not found in membranes induced subcutaneously. Yet, neither woven nor mature bone was found in the core of the encapsulation membranes in subcutaneous and bone locations in both rabbit and sheep experiments. Some bone formation did occur occasionally from the bone defect edges on the internal side of, but not within, the membranes.
Interestingly, sheep femoral bone defects left empty after removal of the cement spacer 4 weeks postoperatively, showed higher bone formation, 4 months postoperatively, when the induced membrane was preserved compared with defects in which the membrane had been resected at time of cement removal. Similar trends have been observed in metatarsal defects. Yet, in the latter, different observation time periods between defects with (6 months) and without (4 months) preservation of the induced membrane have prevented objective comparisons. In all cases, bone formation though remained limited and confined to the internal aspect of the membrane, in the close vicinity of resected cortices suggesting that the membrane per se had limited bone-regeneration capacity. Further investigation is therefore needed to test the hypothesis that the encapsulation membrane does have a substantial osteogenic role.
Maintenance of graft volume
As observed in clinical studies in humans, morselized bone autografts used to fill the cavity delineated by the membrane in femoral and metatarsal segmental defects in sheep retained their original shape and size throughout the studies (4 and 6 months, respectively), grossly matching the ones of the resected bone ( Fig. 3 ).