Abstract
The replacement and repair of bone lost due to trauma, cancer, or congenital defects is a major clinical challenge. Skeletal tissue engineering is a potentially powerful strategy in modern regenerative medicine, and research in this field has increased greatly in recent years. Tissue engineering strategies seek to translate research findings in the fields of materials science, stem cell biology, and biomineralization into clinical applications, demanding the use of appropriate in vivo models to investigate bone regeneration of the long bone. However, identification of the optimal in vivo segmental bone defect model from the literature is difficult due to the use of different animal species (large and small mammals), different bones (weight-bearing and nonweight bearing), and multiple protocols, including the use of various scaffolds, cells, and bioactives. The aim of this review is to summarize the available animal models for evaluating long bone regeneration in vivo. We highlight the differences not only in species and sites but also in defect size, means of defect creation, duration of study, and fixation method. A critical evaluation of the most clinically relevant models is addressed to guide the researcher in his/her choice of the most appropriate model to use in future hypothesis-driven investigations.
Introduction
Therapy for Large Bone Defects
The current concept for treatment of a nonhealing bone defect is to fill the defect gap with a bone graft or a natural or synthetic material that may or may not contain stem/progenitor cells and/or growth factors to aid new bone growth. The first successful report of this technique may have been as early as 1889, 5 when decalcified chips of bovine bone were inserted into bone defects in the cranium of dogs. To date, due to its osteoconductive and osteoinductive nature, the most successful method used clinically is that of autogenous bone grafting, usually involving removal of bone from the patient's own donor site (e.g., the iliac crest) and transplantation into the bone defect. The advantage of using autogenous bone grafts is their capacity to provide the necessary elements for bone regeneration (e.g., osteogenic bone-forming cells, osteoinductive growth factors, and the natural organic/inorganic bone matrix as the osteoconductive scaffolds) and the lack of tissue rejection. However, this procedure is limited by the amount of healthy bone that can be taken from the patient at any one time (especially in the case of young patients, where there may be limited bone available at the donor site for grafting into large defects), additional surgery, and the pain and high morbidity associated with the donor site (10–30%).6–9 The other options to this are to use either allogenic bone (i.e., from other humans) or xenogenic bone (from nonhuman species), which introduces the risk of rejection,10,11 is reported to be less osteoinductive, 12 and does not revascularize as quickly as autograft bone. 13 Xenogenic bone grafts also carry the risk of zoonotic disease transmission as well as possible immune responses. 14 These limitations have lead scientists and clinicians to focus their efforts on the design and manufacture of a suitable artificial scaffold to support bone regeneration. 15
In addition to avoiding/minimizing possible complications, a scaffold needs to be strong enough to withstand the biomechanical forces traversing the defect. This may sometimes be overcome by the use of a rigid fixator, designed to withstand daily biomechanical loads. However, this in itself can cause further complications as healthy bone requires loading at different time points during fracture healing or bone regeneration (these periods vary between different locations and/or different species), and shielding healthy bone from daily biomechanical forces can cause further bone loss. 16 Moreover, complications can arise due to a lack of vascular invasion, leading to problems of gaseous exchange, waste elimination, and nutrient delivery within the center of the regenerating defect. In addition, if the nonhealing defect is a secondary symptom of disease or medication, it is possible that the reparative capacity of autogenic bone marrow stem cells (BMSCs) will be compromised. It has recently been shown that systemic factors are critical to the activation of tissue-specific progenitor cells. 17 However, it is still not known whether systemic factors may also affect implanted stem cells.
There have been numerous clinical cases of successful restoration and/or regeneration of bone using either autogenic bone,18–21 allogenic bone,22–24 or synthesized scaffolds,25,26 with or without cells and/or bioactives. However, although these cases show some positive results for the treatment of large bone defects, results are not consistent and optimal methods for regeneration of bone are still sought. Therefore, efforts to bioengineer bone either in vitro, using bioreactors, 27 or in vivo (either in situ 28 or within another location in the body 20 ) are ongoing.
Currently, tissue engineering is still in its infancy, with much research applied to optimizing protocols to gain the greatest bone regeneration in the quickest time. 29 To assess any bioengineered tissue, however achieved, there is a need to use a clinically relevant animal model to test the quality of the regenerated bone and its therapeutic bio-efficacy in vivo.
Provision of an Environment Suitable for Bone Regeneration/Restoration
The in vivo animal model provides a suitable (both biological and pathological) microenvironment for bone tissue engineering that needs the presence of not only bone-forming cells (osteoblasts and osteoprogenitors) but also osteoclasts and suitable conditions for remodeling (angiogenesis, supportive scaffolds, local/systematic growth factors, and mechanical stimuli). The segmental bone defect animal model can allow the manipulation of these factors for testing and improving fracture healing and bone regeneration. There does, however, appear to be species variation in osteogenic capacity of stem cells, 30 highlighting limitations of animal model usage. 31
Most researchers believe that the optimal cells for implantation into a bone defect come from patients themselves. However, this is not always necessary. Studies have shown the successful implantation of allogeneic stem cells within a rat 32 and canine 33 with/without (respectively) short-term immunosuppression and with no substantial adverse reaction. The anatomical source of the tissue from which implanted cells are derived also requires careful consideration as there are site-specific variations in tissue osteogenic capacity.34,35 For example, some workers36–38 have found that cells from the periosteum have a greater osteogenic potential than those derived from bone marrow. Soft tissues around the defect may also be an important source of osteoprogenitor cells. 39 When testing the reparative response of implanted cells, it may also be useful to flush the marrow cavity of the injured bone to remove all the host bone marrow stromal cells in situ 40 ; however, this may also remove essential growth factors and cytokines, thus impeding repair.
Both within the clinical and preclinical research fields, there is often only one cell population implanted, though it has been shown that at least two cell populations are required to work in concert for successful bone regeneration. 36 As bone is undergoing continual fine remodeling of the architecture, there is a need for osteoclasts39,41 as well as bone-forming cells. This is generally neglected, with only a few researchers commenting on the presence of these cells (an indicator of bone remodeling) for bone repair and regeneration.
Angiogenesis within the repairing area is essential for nutrient supply and gaseous exchange especially for the center of a large bone defect. If either is depleted the area will become necrotic.1,39 Further, circulating stem cells within the vasculature cannot migrate into the defect via chemotaxis 42 if blood vessels are not present within the vicinity. Neovascularization is one of the fundamental challenges to bone regeneration in large bone defects, which is only truly tested by the use of large animals and bone defects. It has been shown that there is a far greater reparative capacity using fresh autogenous periosteum with an intact vascular supply than an avascular periosteum. 43 Similar results have also been shown when comparing an intact vascular autograft of bone to a nonvascularized autograft. 13 Recently, Takeda et al. 44 and Elefteriou et al. 45 stated that nerves (innervation) are critical for osteoblast function. Indeed, the whole skeletal architecture and homeostasis has been reported to be under the control of the sympathetic nervous system. However, the mechanism behind this is still unclear. At present, this area of bone regeneration is overlooked.
Segmental bone defect models
The optimized animal model should have a pathology or injury that is similar to that seen in humans. The use of animal models allows researchers to investigate disease states in ways that would be inaccessible or unethical in a human patient. However, the use of segmental bone defect models in animals must be carefully designed to cause as little suffering to the animal as possible. In most countries, it is necessary to gain ethics approval before commencement. In all cases the investigator must abide by the three R's wherever possible by replacing animal work (using preliminary in vitro data), by reducing the number of animals to a minimum and refining their plan, surgical and experimental techniques, and proper assessment of the outcome.
For in vivo testing of any bioengineered bone, it is essential to design and choose the appropriate defect model according to the desired clinical applications. In general, the bone defect model must allow the engineered bone to be tested under similar conditions to that required in the clinic. For the testing of tissue destined for long bones, the use of nonhealing segmental bone defect models is well established and includes fracture and osteotomy models, based upon those described by Bonnarens and Einhorn 46 and Einhorn et al. 47 in 1984. The model ultimately chosen will partly depend upon the end goal of the research (such as nonweight-bearing bone defect model, e.g., calvarial and rib, or weight-bearing bone defect model, e.g., femur and tibia). Further, researchers must also consider if they want to heal cortical and/or cancellous bone. 48
This review will concentrate on the bone defect models within the long bones (e.g., femur, tibia, ulna, radius, and humerus). The varying defect locations add a complication when interpreting results from different studies as not all bone heals with the same efficiency. This has partly been shown to be due to site-specific differences in the periosteum; tibial periosteum has a greater capability to repair bone fractures than calvarial periosteum.34,35
The ideal animal model for bone repair and regeneration should (1) mimic clinical conditions of bone injury or defect, (2) utilize fixation of the defect, as within the clinic, (3) allow the animal to apply mechanical load through the defect, (4) permit angiogenesis, (5) provide all types of cells needed for bone repair in situ, and (6) minimize the suffering of the animal. When considering which animal model to be used, the type of animal, the model itself (location, size of defect, and fixing method), and how bone regeneration will be manipulated (scaffold, implanted cells, and bioactives) must be taken into account.
The species of animal chosen will depend upon the intended outcome of the study and whether human cells are to be used. For xenogenic stem cells, there is a requirement to use either immune-deficient, athymic (nude) rodents,49,50 or immune-response suppressants within the recipient. 32 However, conflicting studies suggest that the reparative response in athymic rodents is either similar to normal rodents after fracture 51 or is delayed, 50 although the latter study compared the response in athymic rats implanted with human BMSCs with that from a separate study using normal rats implanted with rat BMSCs. 52 For studies using autogenic stem cells or biomaterials alone, a greater choice of animal model is available, each with its own advantages and disadvantages, as previously discussed. 53
Creation of the defect
There are two main methods of creating a critical defect in the long bone using either an osteotomy approach or a traumatic approach (Table 1). Osteotomy utilizes a drill or saw to surgically remove the required length of bone from a predetermined site, producing a consistent defect in all subjects. The edges of the defect are usually cut straight (not jagged) with less trauma. Therefore, an osteotomy is similar to that seen after invasive surgery for tumors. To reflect the conditions after traumatic injury, the defect can be created via trauma, which will produce a jagged cut edge and traumatize both the bone and surrounding soft tissue. To achieve this, a three-point bending device, based on the Einhorn fracture model, 46 has been described for use on the rat. 54 Street et al. 55 used a dental burr (for mouse) or saw (for rabbit) to create bone defects in femurs after dissection of the periosteum. These studies clearly showed a nonunion fracture, with creation of atrophic tissue and highlighted the importance of the periosteum in fracture healing. The disadvantage of this method is the potential for larger variation in defect size between subjects.
Gigli saw, reciprocating saw, fretsaw, sterile saw blade, and/or threaded saw.
The numbers indicate references.
In addition, most animal models are not representative of the chronic nature of most bone defects, such as those often seen within the clinic associated with an open fracture with/without infection and/or fibril tissue (scar) formation or with secondary factors (e.g., diabetes). Hence, the internal milieu (e.g., growth factors and inflammatory cytokines) within an animal model will also fail to reflect the clinically relevant situation. Heckman et al. and Cook et al. designed models to take this into account by creating the defect 12–16 weeks 56 and 8 weeks 57 before implantation of scaffold and bioactives in dogs and rabbits, respectively.
Size of the defect
One of the fundamental factors for studying a nonhealing bone defect model is whether the size of the defect created remains critical. 58 Table 2 summarizes the range of sizes used in the various species at different locations. A problem with comparing different models is that there is often a lack of clear evidence that the model used was actually a critical defect. In 1934, Key 59 hypothesized that to create a critical size defect in canine ulnae, the defect must be over 1.5 times the diameter of the bone. Toombs et al., 60 testing this hypothesis within cats, suggested that this may be an overestimation but that studies should cite the length related to the bone diameter. Although some authors do cite this ratio,50,61–65 it is generally omitted from reports, with often only the length 56 rather than the diameter of the bone being provided. Additionally, Key suggested that the size of a nonunion defect needed to be larger in younger dogs, 59 highlighting age-related changes in capacity for repair. 66 It has also been repeatedly shown that the periosteum is very important for bone regeneration and that unless this is removed,54,67 the defect will often spontaneously heal.
ns, not stated.
The site of the defect within each bone must also be considered as fracture nonunions are most often seen clinically in (1) the proximal third of the tibia, 68 (2) the femoral neck (10–30%), 69 (3) the scaphoid (10%), 70 and (4) clavicle (0.1–15%). 70 Therefore, the exact anatomical location of the defect to be created needs to be considered carefully when designing the experiment.
Fixation of experimental defects
The method of fixation is another important variable to be studied in bone fracture/defect models. Fixation should not only reflect what occurs in the clinic but also provide sufficient rigidity to support bone healing. However, too rigid a fixator will also prevent healing.1,16 The method of fixation is more important for defects within weight-bearing bones than the nonweight-bearing bones or for bones that have natural internal fixation such as the radius and ulna. 71 At present, most long bone defect models use either bone plates or intramedullary rods to fix the defect (Table 3), thus reflecting the situation within the clinic. However, as most of these studies use fixators not originally designed for animal usage, it is not known whether the correct loads are permitted during the study, possibly leading to a detrimental effect on the response.
There is at present little research examining the effects of fixation removal on further bone healing, regeneration, and remodeling. Zhu et al. attempted to examine this by taking interlocking nails out of the femurs of goats 6 months postsurgery, 72 thereafter allowing the application of daily mechanical loading through the defect for two additional months before sacrifice. The authors reported that further bone remodeling occurred after fixation removal, highlighting the importance of this end-step for fracture repair.
Other considerations, including species-specific growth
To further manipulate defect repair, much effort has been devoted to the use of osteoinductive bioactives, including their optimum concentrations and methods of delivery. In addition to the use of recombinant bioactives, recent reports have described the use of gene therapy (e.g., implanted transfected BMSCs32,73,74) or in vivo viral transfection,75,76 although this technology has raised safety concerns. Evaluation of other nonunion therapies, for example, low intensity ultrasound 64 or pulsed electromagnetic field, 77 to manipulate bone regeneration has also been used with animal defect models. In some cases, the models have to be modified to accommodate analysis of bone healing. For example, metal fixation may not suitable when using microcomputed tomography and/or magnetic resonance imaging for assessment.
Finally, the time scale of an in vivo study will vary (Table 4) depending on the choice of model and species as well as the parameters to be monitored. The rodent, being a fast healer, has a distinct advantage over larger species (e.g., the study period will be significant reduced for fast screening)78,79 although the histology of the bone at different locations of the rodent and other species need to be possibly comparable to that of the homologous location in humans. Also, the postural difference between animal models and humans is an important consideration for bone tissue engineering research.121,122
Conclusion
The choice of animal model for bone tissue engineering must be carefully considered according to the proposed hypothesis to be tested. For example, for testing the feasibility of using cells and/or scaffold materials to repair bone defects, small animal models could be more advantageous. However, if the aim is to test the mechanical strength of the scaffold, a larger animal may be more appropriate. Ideally, a model should sufficiently match the clinical setting that it is to reflect, both in terms of creation of the defect and the cells, scaffolds, or bioactives that will be used for repairing it. Further, it is essential that an appropriate microenvironment (physiological, anatomical, pathological, and mechanical) reflecting the clinical situation will be created. Considerations should include the species and the type of defect (location, size, method of creation and repair, and fixation) as well as the postoperation care and monitoring. Some research, especially at the molecular and cellular level, can be achieved using in vitro analysis. However, this is limited as data gained in vitro have been shown to differ from that in vivo, particularly when investigating scaffolds. 80 In addition, the coordinated responses and cascade of cellular events that occur in vivo during tissue regeneration cannot be recreated within a test tube, nor can the biomechanical properties of the scaffold and the bio-efficacy of the technique be tested. Thus, a critical segmental bone defect animal model is essential for investigating the optimal conditions for bioengineering and bone regeneration for the treatment of a bone defect. However, it must be borne in mind that an animal model may have limitations, particularly if the animal is small. A protocol that works within an animal may not always translate successfully or directly to the clinic; a combination of different animal models for different strategies may be more realistic.
Disclosure Statement
No competing financial interests exist.