Introduction
Because of the limited availability of autologous cartilage grafts and the associated morbidity with their harvest, allogenic implants have been increasingly used for tracheal and nasal reconstruction. Unfortunately, extrusion and inflammatory reactions remain an infrequent but real problem associated with their use. Because autologous tissue is still considered the implant of choice, investigators have pursued tissue engineering as a potential cartilage source for head and neck reconstructive and plastic surgery.
Tissue engineering involves the creation of a solid material, bone, skin, cartilage, etc, out of cells and other biodegradable matrices. Although engineered skin and articular chondrocytes are currently available as commercial products, cartilage formation is still being investigated at the basic science level. Several distinct problems with current cartilage engineering models are of note. When cartilage is formed in vitro, it exhibits poor biomechanical properties, with a modulus of elasticity 10% to 20% of normal cartilage, even after implantation into a living host.1,2 Authors have postulated that the in vivo environment during chondrogenesis may be too complex to recreate in a laboratory setting.
lmplantation of these cultured cartilage samples helps to mature the cartilage to a variable degree, but significant mechanical deformation and volume loss occurs, making results unpredictable. Extended cell culture causes chondrocyte de-differentiation, resulting in unpredictable tissue results: fibrocartilage, articular type cartilage, or a mixture. One may reasonably postulate that in vivo formation would lead to improved cartilage strength and volume retention if tissue engineering were performed in a natural milieu.
Previous studies using immunosuppressed animals have yielded pertinent information about feasibility and techniques.1-4 However, a single experiment by Park et al. used an immune competent animal for in vivo chondrogenesis. An intramuscular injection of alginate and autologous auricular chondrocytes was performed.5 Park et al. demonstrated an 80% success rate, as well as near complete volume retention.
Cartilage architecture was normal, but significant bone formation occurred. Stains for elastic fibrils were not performed. Although highly encouraging, the translation of this model to a clinical setting is difficult because intramuscular retrieval of cartilage would carry unacceptably high patient morbidity in most cases.
Therefore, we have designed a model for in vivo cartilage engineering that could be directly translated to clinical use. After evaluating the existing literature, we decided to use fibrin glue (Tisseel) because of its availability, biocompatibility, and rapid polymerization.6-9 Autologous cells in a viscous matrix (fibrin glue) are delivered into the subcutaneous tissues with a simple injection. This site was chosen for the ease of implantation, monitoring, and harvest. In addition, we used only currently available commercial materials and technologies in constructing our design.
The goal of this study is to validate the feasibility of our model. Using this technique, we would ultimately like to define alternative strategies for improved cartilage yield and quality using a single application of growth factors (GFS) at the time of injection. One specific combination of GFs at a single concentration was evaluated. This choice was based on prior studies at our institution using these materials in the same species of rabbit.10 It is our belief that cartilage of high quality can be created in an easily accessible site and that this model can be extrapolated, ultimately, for clinical use before reconstructive or esthetic surgery within the head and neck.
Materials and Methods
Eight New Zealand white rabbits were anesthetized with ketamine and xylazine for general anesthesia. Cartilage was removed from the right ear by incising the dorsal surface and creating a perichondrial flap. A 3><3 cm segment (approximately 0.45 mL) was removed in standard surgical fashion. Cells were prepared using a standard protocol. After digestion using 0.2% collagenase (Worthington) at 37°C for 15 hours, cells were filtered using a 120 p.m pore mesh (Tetko). Cells were centrifuged and suspended three times using Ham’s F-12 medium
Overall cell yield and injection concentrations were computed using these calculated values (Table I).
Fibrin glue (Tisseel) was prepared in standard fashion according to package insert instructions. Because of the rapid polymerization of the glue if undiluted and combined in a single syringe, the Duplijet system was used for all injections. This system incorporates two separate syringes and channels and only allows materials to mix in the injection needle itself. The chondrocytes were suspended in 0.5 mL of the thrombin solution component, which was not diluted to ensure rapid polymerization in vivo. GFs, when used, were added into 0.5 mL of fibrinolysis inhibitor solution. Combined volume after injection was, therefore, 1 mL.
For samples containing GFs, both 200 ng/mL recombinant human insulin-like growth factor (rhIGF)~1 and 500 ng/mL recombinant human basic fibroblast growth factor (rhb-FGF, Research Diagnostics, Inc.) were used together. The decision to use both was based on several factors. Although b-FGF at 250 ng/mL demonstrated improved cartilage performance using intact ear cartilage grafts, prior unpublished pilot studies in our laboratory in the injectable model showed no discernible benefit from this concentration of b-FGF.10 A higher concentration was chosen for this investigation.
Because historic in vitro data demonstrated synergistic effects when combined, we decided to use both b-FGF and IGF-1 for this study.
One milliliter total of solution was injected subcutaneously into the donor rabbit’s dorsum. Implants were palpated for size, integration, and polymerization. Implants were checked weekly for 3 months and then harvested after animal sacrifice. Hematoxylin-eosin (H&E), alcian blue (AB), and elastin Verhoff-van Geison (EVG) staining was performed. Samples were assessed for a variety of conditions: the presence of cartilage, the presence of inflammatory reactions, necrosis, vascular ingrowth, active ground matrix deposition, and the presence of elastic fibers.
Historical controls from 6-month-old rabbits were used for comparison purposes.10 Chondrocyte dropout (CD) counts were performed by a single observer (RW) on viable samples using two ><40 high-power fields per sample stained with H&E. Counts for each sample were then averaged and compared with historic controls.
Glycosoaminoglycoside (GAG) concentration was evaluated using AB stains, which were scored on a scale of 1 to 10 by two separate observers. A score of 5 was set for “normal” GAG concentrations, as seen in historic controls. Scores were then averaged and compared using standard statistical analysis.
Results
Fourteen of 16 implants were found at the time of harvest. Seven were from non-growth-factor containing (GFC) and seven were from GFC samples. Most implants were seeded into the subcutaneous tissues with minimal mobility, whereas others had migrated into the dorsal fat pad (Fig. 1B). Most samples were easily removed from the surrounding fascia without extensive dissection. Some had become incorporated into the dorsal fat pad of the rabbit but were easily extracted. There was limited capsule formation around all of the samples. Several were noted to have subdermal blood vessels coursing into and around the implant itself at the time of harvest. They were mobile but adherent to the underside of the dermis and easily seen as a contained mass after reflection of the skin.
All samples were palpated with Adson forceps and assessed for overall appearance and volume (Table ll). Several had the look and feel of natural cartilage, with excellent rigidity (Fig. 1). Other samples had areas of cartilage formation at the periphery but did not exhibit significant firmness or resiliency.
All samples were histologically assessed for overall appearance both at ><1O and ><40. Samples that formed cartilage had normal lacunar structure with ground substance deposition. There were a paucity of empty lacunae as well as higher density of lacunae and cells at the periphery. Samples with cartilage throughout the implant demonstrated extensive vascular channels with ingrowth to the center of the implant. Samples with failed central areas had a minimal blood supply to the center of the implant and avascular necrosis with inflammatory reactions of these portions. (Fig, 2A) Cartilage formation at the periphery was normal in appearance. Other samples demonstrated evolving chondrogenesis throughout the implant with the periphery being most mature. Replacement of fibrous tissue appeared to be occurring in the central portions and was accompanied by moderate vascular infiltration to those areas. There was very limited or no osteoid formation seen in successful implants. All cartilage-producing samples had formed a perichondrial-like layer. Just adjacent to this perichondrium, a high density of cells was found, indicating active chondrogenesis and new chondrocyte production. Incorporation of the implant into surrounding fat was also seen in most samples. AB (staining for sulfonated glycosoaminoglycogans) was assessed subjectively from 1 to 10 by two separate observers (Fig. 2, B and C) This was performed after examining all areas of cartilage formation, and scores were based on the average staining pattern for each sample. There was variability within samples representing evolving ground-substance deposition, but overall staining was generally consistent with or more than normal cartilage controls in most samples. Elastic phenotype was noted by the presence or absence of elastic fibrils on EVG stains. These fibers were seen in all cartilage forming samples. Again, variability was seen within samples. A lower density of fibrils, when compared with native elastic cartilage, was noted in all specimens. There were no observable differences in CD counts between GFC and non-GFC samples. CD counts were significantly lower that control values (Table III).
Discussion
Since the advent of modern reconstructive surgery in the head and neck, a search for the optimal alloplastic material has achieved significant advances in terms of stability of form, biocompatibility, and host tissue integration. Despite these accomplishments, most reconstructive surgeons agree that autografts are the material of choice if small defects are to be addressed. However, when revision surgeries or larger deficits are encountered, a paucity of material exists that can be harvested with minimal patient morbidity. For these clinical scenarios, tissue engineered autologous cartilage would be an invaluable tool and would, in many ways, revolutionize how surgeons approach challenging problems.
Prior studies of tissue engineering with immunosuppressed animals used both xenogenic and allogenic materials. They demonstrated that cartilage could be formed in vitro or in vivo using a variety of matrices: alginate, fibrin glue, collagen copolymer, Vicryl scaffolds, and hydrogels. When these experimental designs were conducted in immunocompetent animals, rejection uniformly occurred.1,6-8,11,12 In general, immunologic reaction to foreign chondrocytes and inflammatory reactions to Vicryl or polyglycolic acid scaffolds was thought to be the cause.
Recent work by Park et al.5 using immunocompetent animals showed that autologous chondrocytes with alginate could be injected intramuscularly to form cartilage in approximately 80% of samples. Volume retention was excellent, and overall histology was normal, with the exception of bone formation in all samples. Muscle, with its rich blood supply, would intuitively seem to be an optimal recipient bed for implantation. However, intramuscular chondrogenesis lacks clinical applicability because harvest would cause significant morbidity.
Our study demonstrates that cartilage can be formed in 85% of samples injected into a relatively avascular site. This in vivo engineered cartilage reproduces the histologic and gross look of natural cartilage and forms elastic cartilage. It is likely that generating the cartilage in vivo, as opposed to an artificial in vitro environment, aided in maintaining the elastic chondrocyte and cartilage phenotype.
The use of GFs within in vitro chondrocyte cultures is a widely accepted practice. However, their application in living systems is still under investigation. b-FGF, IGF-1 and 2, and transformation growth factor (TGF)-beta have been extensively studied in vitro. b-FGF promotes cell survival and has mitogenic effects, whereas IGF-1 and 2 promote differentiation of immature chondrocytes and increased ground substance matrix deposition. b-FGF and IGF-1, when used together, appear to have additive effects, TGF-beta appears to promote fibrocartilage and fibroblastic-like chondrocyte morphology.13-16 Fibrin glue has also been shown to stabilize GFs and other proteins, preventing natural enzymatic degradative processes.
Our current study not only failed to show a benefit from the GFs but demonstrated a negative effect on chondrogenesis at the concentrations given (P = .015) (Table IV), However, the ultimate utility of these powerful reagents remains unknown because both dose-dependent and state—dependent factors influence how GFs effect cells. The optimal in vitro concentrations for FGF and IGF-1 are well known, but no investigator has examined their levels in evolving tissues. Additional studies at varying concentrations need to be performed to elucidate whether an optimal concentration exists for in vivo chondrogenesis. However, the complexity of living tissues is a severely confounding factor.
Stains for AB (Fig. 2) and elastin show that cells retain their physiologic and morphologic characteristics, They are not only capable of laying down ground substance to replace a fibrin glue matrix but also recreated elastic cartilage in 100% of successful samples. No methods to date have demonstrated the ability to produce elastic cartilage, even when auricular chondrocytes are implanted. The low CD counts imply viability of the newly formed cartilage with respect to long-term survival.
An extremely interesting finding of the current study was the formation of a perichondrium-like layer (Fig. 3C, white arrow). Perichondrium was removed from all samples before digestion. Therefore, one must conclude that cells de-differentiated in vivo to more pluripotential cells to form this layer. This finding raises several salient questions.
Do chondrocytes de-differentiate in vivo after implantation to form this perichondrium? If so, did b-FGF or IGF alone or in combination prevent this crucial step by promoting terminal differentiation? Could this explain the decreased success rates (28% with GFs vs. 85% without GFS) associated with their use? These questions will have to be explored further in future studies.
Some problems with the current study are of note. Most importantly, historical controls have demonstrated that the in vitro seeding concentration for optimal cartilage formation ranges from 10 to 40 >< 106 cells/mL. Our samples averaged 4.8 >< 106 cells/mL. Although 85% of non-GF-treated samples formed cartilage, a higher concentration may have helped with central portions, where regional implant failure was most commonly encountered. Although the chondrocytes were self derived, other components of the matrix were not rabbit specific. Fibrin glue (Tisseel) is made up of both human and bovine proteins. Prior experimental evidence suggests biocompatibility across species. The GFs used were rhGFs, and the lack of positive findings associated with their use may represent incompatibility with the rabbit’s growth factor receptors. Kaufman et al.10 used New Zealand white rabbits and recombinant human b-FGF and fibrin glue without noting biocompatibility problems. Additional concerns regarding GF degradation in the fibrinolysis inhibitor solution are present. However, prior studies in our laboratory using similar preparations noted GF-associated effects, showing retained function. Because of the relative avascular implantation site, we hypothesized that including one or more GFs into the original injection material might lead to greater cell survival, higher rates of cartilage formation, and improved cartilage quality. Unfortunately, we were not able to demonstrate this and will have to reexamine this issue with additional studies. Several more concentrations may have to be tried to truly characterize the role of both FGF and IGF. Angiogenic factors may need to be incorporated as well because we believe that avascular necrosis of the implant’s center was a significant contributing factor to sample failure.
Conclusion
The aim of these and future studies is to provide information that ultimately leads to clinical application in human beings. We believe that our design, using immunocompetent animals and an implantation site with little ultimate harvest morbidity, should be investigated further in future studies. Cartilage was formed in 85% of non-GFC samples, and the elastic phenotype was maintained in all samples. In addition, all components of our design draw from existing technologies: readily available injection materials (Tisseel) and GFS, as well as simple laboratory protocols. Within the orthopedic community, private companies are already expanding chondrocytes from biopsy specimens to be used for autologous subperiosteal articular implantation.
Our experimental findings provide an alternative perspective in the way we approach tissue engineering models. Three to 6 months before planned surgical procedures, a patient’s own cells could be obtained with a small biopsy. Although commercially performed in vitro cell expansion would be needed to have the necessary number of chondrocytes for implantation, these cells can be re-differentiated in culture before harvest and subcutaneous implantation. This material could then be monitored pre-operatively and harvested easily with a simple skin incision.
We believe that immunocompetent animals and autologous cells should be used for future tissue engineering protocols. It remains to be determined which method(s) produce optimal cartilage results, and further investigation is warranted. However, our results validate the subcutaneous model and fibrin glue (Tisseel) as a feasible model for cartilage engineering studies.
Bibliography
- Chang S, Rowley J, Tobias G, et al. Injection molding of chondrocyte/alginate constructs in the shape of facial implants. JBiomed Mater Res 2001;55:5103—5111.
- Ting V, Sims C, Brecht L, et al. In vitro prefabrication of human cartilage shapes using fibrin glue and human chondrocytes. Ann Plast Surg 1998;40:413-421.
- Britt J, Park S. Autogenous tissue-engineered cartilage. Arch Otol HNS 1998;124:671—677.
- Kamil S, Kojima K, Vacanti M, et al. In vitro engineering to generate a human sized auricle and nasal tip. Laryngoscope 2003;113:90—94.
- Park D, Bong J, Park S, Hong K. Cartilage generation using alginate encapsulated autogenous chondrocytes in rabbits. Ann Otol Rhino] Laryngol 2000;109:1157—1161.
- Silverman R, Passaretti D, Huang W, et al. Injectable tissue engineered cartilage using a fibrin glue polymer. Plast Reconstr Surg 1999;103:1809—1818.
- Sims C, Butler P, Cao Y, et al. Tissue engineered neocartilage using plasma derived polymer substrates and chondrocytes. Plast Reconstr Surg 1998;101:1580—1585.
- Saim A, Coa Y, Weng Y, et al. Engineering autogenous cartilage in the shape of a helix using and injectable hydrogel scaffold. Laryngoscope 2000;110:1694—1697.
- Yang W, Chen S, Mao T, et al. A study of injectable tissue-engineered autologous cartilage. Chin J Dent Res 2000;4:10—15.
- Kaufman M, Westreich R, Ammar S, et al. Autologous cartilage grafts enhanced by a novel transplant medium using fibrin sealant and fibroblast growth factor. Arch Facial Plast Surg 2004,6:94—100.
- Vacanti C, Langer R, Schloo B, Vacanti J. synthetic polymers seeded with chondrocytes provides a template for new cartilage formation. Plast Reconstr Surg 1991;88:754—759.
- Cao Y, Vacanti J, Paige K, et al. Transplantation of chondrocytes using a polymer-cell construct to produce tissue engineered cartilage in the shape of a human ear. Plast Recvnstr Surg 1995;96:1390—1398.
- Bos P, van Osch G, Frenz D, et al. Growth factor expression in cartilage wound healing; temporal and spatial immuno-localization in the rabbit auricular cartilage wound model. Osteoarthritis Cartilage 2001;9:382—389.
- Jakob M, Demarteau D, Hintermann B, et al. Specific growth factors during the expansion and redifferentiation of adult human articular chondrocytes enhance chondrogenesis and cartilaginous tissue formation in vitro. J Cell Biochem 2001;81:368—377.
- Arevalo-Silva C, Cao Y, Vacanti M, et al. Influence of growth factors on tissue engineered pediatric elastic cartilage. Arch Otol HNS 2000;126:1234—1238.
- Van Osch G, Van Der Veen S, Verwoerd-Verhoefl In vitro redifferentiation of culture expanded rabbit and human auricular chondrocytes for cartilage reconstruction. Plast Reconstr Surg 2001;107:433—440.
- Nakanashi R, Hashimoto M, Yasumoto K. Improved airway healing using basic fibroblast growth factor in the canine tracheal autotransplant model. Ann Surg 1998;227:446—454.
Additional readings:
Fujimoto E, Ochi M, Kato Y, et al. Beneficial effect of basic fibroblast growth factor on the repair of full thickness defects in rabbit articular cartilage. Arch Orthop Trauma Surg 199%},119:139—145.
Van Susante J, Buma P, Van Beuningen H, et al. Responsiveness of bovine chondrocytes to growth factors in medium with different serum concentrations. J Orthop Res 2000;18:68—77.