Keywords

Introduction

Bone fracture is a common injury. Severe bone fractures are usually comminuted, which may cause bone defects. Nonunion would occur if the size of bone defect is beyond the self-healing ability of bone, which will cause serious pain and medical burden to patients. Furthermore, the repair of bone defects is difficult in clinical treatment due to limited autogenous bone supply.

The development of osteoinductive biomaterials and bone tissue engineering technology brings hope to the treatment of bone defects caused by fracture. For bone defects caused by tumors and osteomyelitis, biomaterials with only osteoinductive capability have limited treatment potential. Thus, this chapter focusses on defects caused by fracture. After development of bioactive bone substitutes, including osteoinductive biomaterials and bone tissue engineering products, evaluation of whether the materials meet the requirements of clinical treatment is necessary. Considering the complexity of the human body, results from in vitro tests cannot properly reflect the performance of biomaterials in vivo. Animal experiments are therefore essential for biomaterial evaluations by simulating the clinical human application.

However, the establishment of a universal animal model applicable to all studies for bone tissue engineering and osteoinductive biomaterial is impossible. A wide variety of bone defect models has been created for testing various materials. The repair of bone defect is affected by many factors, such as stability of the fractured ends of bone, blood supply, infection, and treatment methods, all of which affect the evaluation results of biomaterials. An incorrect selection of model or failure of detail control may seriously affect the results and even mislead the researchers. Therefore, proper animal models should simulate the real conditions of fracture and bone defect well and should be repeatable and standardized to obtain reliable results.

Here, we review the current bone defect models for the researches of osteoinductive biomaterials and bone tissue engineering and provide prospects for the development of defect models, which may serve as a reference for relevant researchers. This review includes three parts. The characteristics and analyses of the each bone defect model are introduced in the first part. The model selection criteria are described and suggested in the second part. Finally, the need for improvement and development of new bone defect models are proposed in the third part.

Animal Bone Defect Models

An ideal animal bone defect model used in osteoinductive biomaterials and bone tissue engineering research needs to meet the following criteria.

Clinical similarity: The animal model should have a similar in vivo environment simulating a specific human disease. For the bone fracture defect model, it should simulate bone defect of human when fracture occurs. Meanwhile, the animal model should be consistent with the disease severity. In terms of bone defect models, the size of bone defects should exceed the self-repair ability of the body where nonunion occurs without immediate treatment. The smallest size causing nonunion is defined as the critical size defect (CSD) [1]. The CSD values vary with animal model and animal age, weight, defect location, and other pathological factors [2].

Repeatability: Standardized processes and operational methods should be used to produce animal bone defect model, which allows the model to be repeated under the same conditions, rendering the results from the bone defect model reliable and comparable.

There are many types of bone defect models used in the researches of osteoinductive biomaterials and bone tissue engineering. Based on the defect location, the bone defect models can be divided into calvarial bone defect model, long bone segmental defect model, metaphyseal defect model, vertebral body defect model, and long bone multiple defect model, etc. Moreover, each model can be created in different animal species. Within so many kinds of defect model, selecting a suitable model for research is a problem every researcher encounters. Before making choices, an in-depth understanding of these models is necessary.

Calvarial Bone Defect Models

Calvarial bone defects are usually created at the center of parietal bone by using a trephine with a specific diameter after cutting the skin and subcutaneous tissue [7]. This model is easy to create and the procedure is standardized to reach a high repeatability.

CSD in Calvarial Bone Defect Models

The CSD of a mouse skull is 4 mm in diameter for round defects [8]. In addition, bilateral skull defects with 3.5 mm in diameter [7] can be established in one mouse to obtain more reliable results. Table 1 summarizes CSD values for a number of different models.

Table 1 Calvarial bone defect models for biomaterial evaluations and relevant references
Full size table

Calvarial bone defect models are usually created in rats. However, the CSD of cranial bone defects is reported to be approximately 4–8 mm [3, 9,10,11,12,13] (Fig. 1a). Scholars initially recognized that the CSD was 8 mm in diameter [9], which is generally chosen in the level of the dura and sigmoid sinus [9]. However, these large-size cranial defects may damage the sagittal sinus, resulting in increasing the probability of bleeding. Therefore, scholars studied small-size skull defects. Sakata et al. created a 7-mm diameter cranial defect [13]. At 14 days post grafting, no calcified tissue was visible in the defect, and more calcification was seen after 35 days, but this did not fill the calvarial defects [13]. Subsequently, experiments of rat skull defects with diameters of 6, 5, and even 4 mm were reported [12,13,14]. These small-size bone defect models could make bilateral defects so that it might be used to compare different materials in one single animal, thus improving the reliability of the experiment. The most widely used model is the 5-mm defect model, which is considered to be CSD with bilateral [15, 16] or single defect [3, 11, 17]. With regard to the skull defect with 4-mm diameter, Glowacki et al. [12] found no obvious signs of bone healing even after 6 months of observation. It should be noted that the observation time period for the determination of CSD in the rat skull should be >8 weeks because skull repair is active within the first 4 weeks and essentially stops at the eighth week [15].

Fig. 1
figure 1

Calvarial bone defect models : (a) a rat calvarial CSD of 5 mm in diameter [3]; (b) the composite hydroxyapatite/cell sheet transplanted into the calvarial defect of rabbit (15 mm in diameter) [4]; (c) four 8 mm-diameter defects were drilled into the rabbit calvarial bone with a trephine [5]; (d) canine calvarial defect model: (I) bilateral full-thickness bone defects with a size of 20 mm × 20 mm were created; (II) the experimental side (left) was repaired with a cell-coral construct and the control side (right) was filled with a coral alone; (III) coronal CT scan image of the same animal was taken immediately after surgery [6]

Full size image

For the rabbit calvarial defect model, the CSD is generally considered as a full-thickness defect with a diameter of 15 mm [4] (Fig. 1b). Bilateral bone defects with smaller size were also established on rabbit skull. Lin et al. drilled an 8 mm-diameter bilateral full-thickness defect, which was used to compare the repair of bone defect with autologous mesenchymal stem cells (MSCs) [18]. Liu et al. created four 8 mm-diameter defects in the calvarium of rabbits using a trephine attached to a low-speed hand-piece [5] (Fig. 1c). Furthermore, Tovar et al. also drilled two bilateral 8.0 mm-diameter defects without dural injury in the proximal to the coronal suture of the parietal bone [19]. In this way, a variety of bone substitutes have be evaluated by the rabbit calvarial defect model (Table 1).

The CSD in canine skull defect model is usually 20 mm in diameter [20]. With regard to large animal models, making bilateral skull defects is convenient. Sato et al. manufactured bilateral 14-mm trephine defects in dog skull to evaluate the performance of β-sheep bone morphogenetic protein (β-BMP) and a carrier consisting of matrix y-carboxyglutamic acid rich protein [21]. Cui et al. created bilateral full-thickness bone defects with a size of 20 × 20 mm at the parietal bone [6] (Fig. 1d). In addition, a large skull defect area, measuring 35 × 35 mm, was also created in another study [22].

Comments on the Models

Some studies performed periosteal removal in the operative region [12, 23], which might have caused the failure of the skull repair. For example, the stripping of periosteum could even lead to the failure of 2-mm diameter defect healing in rats [24, 25]. Thus, the periosteal layer should be sutured to avoid the obstruction of blood supply when creating animal skull defects.

The operation procedure of the calvarial defect model seems simple, but many things needs special attention to create a successful, repeatable model. The skull is located under the skin; thus, direct skull exposure without intraoperative bleeding is simple. However, prevention of secondary damage is difficult. The surgeon should be careful not to damage the dura mater and the surrounding vessels intraoperatively, or it would cause uncontrollable bleeding. Also, the sagittal sinus may be injured during when creating large-size defects, resulting in obstruction of blood supply. In addition, bone marrow progenitor cells exist in the connective tissue near the sagittal suture, which might be involved in bone remodeling after injury. These factors may critically impact the repair of bone defects.

Segmental Defect Models of Weight-Bearing Long Bone

Weight-bearing bones refer to the bones that bear the main weight of the body in the extremities. The femur and tibia are the most common weight-bearing bones in animal models. The femur size is large, which is convenient for the establishment of bone defect model. And the infection resisting ability of femur model is believed to be strong. However, much muscle tissue around the femur can complicate bone exposure during operation. Because part of tibia is directly under the skin, tibial exposure is relatively simple, but infection is easy to occur. The selection of model should be based on the purpose of the study and the degree of familiarity of the anatomy.

Fixation Used in the Segmental Defect Model

After the long bone segmental defect forms, the body weight would cause the movement and displacement of the broken ends of the bone, which could affect bone healing and test results. Therefore, internal or external fixation must be used to ensure the stability of defect site.

Rats are often used to create the segmental defect models of weight-bearing long bone. It is however challenging to perform surgery on the slender bones. Kirschner wire is often used for fixing the tibia [30] and femur [31] of rats. However, Kirschner wire in bone marrow cavity occupies much space, causing inconvenience for implanting testing materials. In addition, Kirschner wire has no anti-rotation ability and disturbs the blood supply of defect ends. These shortcomings limit the application of Kirschner wire in the small animal model. Besides the Kirschner wire, external fixator can be used to secure the femoral defect [32, 33], which reveals good anti-rotation ability and preserves the blood supply of defect ends as well as adequate space for implantation. Moreover, the external fixator usually allows animals move freely without the joint being fixed. In recent years, plate fixation [26] (Fig. 2a) and a four-hole high-density polyethylene with stainless-steel screws [34] have been developed for rat femur models. Also, small steel plates and screws have been used to immobilize the femur [35] and tibia [27] (Fig. 2b) of rabbits (Table 2).

Fig. 2
figure 2

Weight-bearing long bone segmental defect models with fixations : (a) Photo and radiographs of rat femoral defects with fixation (I) prior to implantation, (II) immediately post-implantation, (III) treated with HA/TCP loaded human MSCs at 12 weeks post-implantation, and (IV) treated with HA/TCP alone at 12 weeks post-implantation [26]. (b) A 5-mm bone defect was created in the middle of rabbit tibia and stabilized with a plate and screws. The defect was implanted with a complex of β-TCP granules and collagen, either with 200 mg of FGF-2 (I) or without FGF-2 (II). For control group, the bone defect was left empty (III) [27]. (c) The 40 mm defect of goat shank with fixation before (I) and after implantation of biomaterial (II) [28]. (d) Goat tibia defects model with external fixation. (I) A periosteal segmental defect of 26 mm length was created at the right tibia; (II) the defect was filled with BMSCs/β-TCP construct; (III) radiographs of goat tibial defects taken at different time points post-operation [29]

Full size image
Table 2 Weight-bearing long bone segmental defect models for biomaterial evaluations and relevant references 
Full size table

With regard to large animals, the weight of the goat and sheep is closer to that of the human body and the size of the bone is also similar to that of the human bones. So more fixation options are available for these animal models. The standard dynamic compression plate (DCP, ten-hole) [36, 37], external fixator [29, 38] (Fig. 2d), screws [28] (Fig. 2c), and internal fixation rods [39, 40] can be used as fixation devices for tibial defects. For femoral defects, internal fixation rods or interlocking nails can be used as fixation devices [39].

CSD in the Segmental Defect Models of Weight-Bearing Long Bone

The CSD of long bone is considered as 1.5–2.5 times the circumference of the diaphysis, or >1/10 of the length [1]. Different animals and bones have different CSDs.

The CSD of the tibia and femur in the rat model are 4 mm [30] and 5 mm [31,32,33,34], respectively. For the rabbit model, the CSD of the tibial defect model is determined as 5 mm [27]. However, there are different opinions on the femoral CSD in rabbit models. Fialkov et al. confirmed 12 mm as CSD for femoral diaphyseal defect model [41], but Fan [42] and Duan et al. [35] used a 15-mm defect as CSD.

With regard to large animals, goats and sheep have relatively large limbs that are convenience for modeling surgery using internal fixators. Both the femur and tibia of goat may be used in bone defect models. The segmental defects in the femur is located in the middle of the shaft, and the length of the osteotomy defect could reach up to 25 mm [39]. The sizes of the tibial defect in goats are reported to be different (e.g., 26 mm [38] and 40 mm [28]). For sheep models, the tibia is the most widely used bone and the lengths of defect is either 3.0 cm [36, 37] or 3.5 cm [43], or even 5 cm [40].

Comments on the Models

Segmental defect models of long bone model requires osteotomy to make bone defects and needs to fix the defect ends with fixator, which makes this type of model complicated. However, in real clinical practice, fractures and bone defects usually occur on the weight-bearing bones of the extremities and internal and external fixators are frequently used in treatment. Thus, the segmental defect models are close to clinical scenarios. Besides, if the design purpose of a novel bone substitute material is to repair a long bone defect of the limbs, the segmental defect model of weight-bearing bone is more suitable. In addition, real bone fractures are often accompanied by peripheral soft tissue injury and inadequate blood supply. The middle and lower segments of the tibia are in contact with subcutaneous tissue without muscle in between, rendering the tibia defect model similar to the clinical factures.

Segmental Defect Models of Non-weight Bearing Long Bone

The non-weight-bearing bones include ulnar, radius, and the fibula. When these bones are truncated, the stability of the broken ends may not be destroyed and the defect ends usually need no fixation, which obviously simplifies the preparation process of the model. To prevent the displacement of the testing material, cerclage wires can be used to prevent dislocation during modeling [46, 47, 49].

CSD in Segmental Defect Models of Non-weight Bearing Long Bone

For mice, the shaft of the radius is most commonly used and the lengths of defects are usually 1.5 and 2.5 mm [50, 51].

For rats, 5-mm osteotomy is usually created on the bilateral radial shaft [45, 52] (Fig. 3a). Although the segmental fibular defect model has two different defect sizes (2 mm and 4 mm) [53], 4 mm is generally considered to be the CSD for fibula [1, 54]. The segmental fibular defect model can also be performed in bilateral fibular bone [54] which can reduce the required number of animals. Besides, two different materials can be compared on one individual rat, which could reduce the experimental error and make the results more consistent.

Fig. 3
figure 3

Non-weight-bearing long bone segmental defect models : (a) Postoperative radiographs of chitosan (CS)-gelatin (Gel) scaffolds in rat radial bone defect (5 mm) at 2, 5, and 8 weeks [45]. (b) Radius of rabbit exposed by dissection of surrounding muscles and an osteoperiosteal segmental defect was created on the middle portion of each radius at least twice that of the diameter of the diaphysis [46]. (c) (I) Xenogenic growth plate grafting in radius defect of rabbit; (II–III) Radiographs of forelimb on 14th day post opertation (II-autograft, III-growth plate xenograft) [47]. (d) Radiographs of the postoperative changes in canine segmental ulnar defects filled with BMP and cells [48]

Full size image

In rabbit model radius is the most commonly used bone to generate a CSD. It is generally believed that the CSD value in long bone should be at least two times the diameter of the diaphysis [46, 47, 49] (Fig. 3b, c). The radius diameter of adult New Zealand white rabbit is approximately 5–6 mm. Therefore, some scholars have determined 10 mm-defect as the CSD [16, 55,56,57,58], although 15 mm-defect has also been used [59, 60]. Moreover, the osteotomy site of the defect is suggested to be selected at 2.5 cm above the radio-carpal joint. Besides the radius, the rabbit ulnar shaft can also be used. Two sizes, 15 mm [61] and 20 mm [62], are commonly adopted as the CSD values of the ulnar. Only a few scholars adopted nonstandardized ulnar stem defects. For example, evaluate recombinant human bone morphogenetic protein-2 in promoting bone healing, Bouxsein et al. built a blade-width (0.5–1 mm) defect in the ulna [63].

Ulna defects are often made in dog models. The ulna is exposed through craniolateral approach [64], and the osteotomy site is located at the proximal 4 cm of the ulnar styloid process [65]. The length of the osteotomy has various sizes in literature: 20 mm [65], 22.5 mm [64], and 25 mm [48, 66, 67] (Fig. 3d). The radius of the dog can also be used to build segmental bone defects with a size of 10 mm [68].

Comments on the Models

Compared with the weight-bearing long bone defect models, the non-weight bearing long bone segmental defect models are relatively simple to create. However, the forearm defect model cannot avoid the interactions between radius and ulna, and the pathological and biomechanical changes in one could be compensated by another. Other shortcomings of these models include the great variations in the bone diameter and resultant sizes of defects created (Table 3).

Table 3 Non-weight bearing long bone segmental defect models for biomaterial evalutions and relevant references
Full size table

Metaphyseal Defect Model

When preparing a metaphyseal defect model, a small drill is used to make a lateral cylindrical defect which is vertical to the axis of the femoral shaft [71,72,73,74], and then the bone substitute materials with proper diameters are implanted into the cylindrical defect.

Femur or tibia is commonly selected for this model due to their large sizes and good accessibility to their condyles. The metaphyseal bone is located subcutaneously and usually no extra fixation is needed after defect creation, which makes the model relatively simple to create with a high success rate [75].

CSD in Metaphyseal Defect Models

In mice metaphyseal defect model , defects with 1 mm in diameter were usually created at the proximal tibia [76] and distal femur [77, 78]. However, these small defects are difficult to evaluate bone substitutes. These defects are mainly used to test drug or investigate the mechanism of fracture healing.

For rats, bone defects with 3.5 mm in diameter at the proximal tibial diaphysis are often used for evaluating biomaterials [79]. Bone defects with different sizes, 5 mm in diameter and 5 mm in depth [80], 6 mm in diameter and 10 mm in depth [69, 73] (Fig. 4a), and 6.1 mm in diameter [70] (Fig. 4b) were fabricated on the metaphysis of rabbit femur. The size of defect in canine femoral condyles could reach 20 mm in depth and 8 mm in diameter [81] or 20 mm in depth and 10 mm in diameter [82].

Fig. 4
figure 4

Metaphyseal defect models : (a) Rabbit femoral condyle bone defects surgical procedure; (I) CSD (6 mm in diameter and 10 mm in length) was transversally created in the femoral condyles of rabbits; (II) the defect was filled with nHA/CS composite scaffolds [69]. (b) Micro-CT sagittal images of the rabbits’ distal femurs from a top view of the CO3Ap blocks (I–IV) and HAp blocks (V–VII) immediately after implantation (day 0) (I, V) and at 4 (II, VI), 12 (III, VII), and 24 (IV, VII) weeks after implantation in rabbits’ distal femurs [70]

Full size image

Comment on the Models

The metaphyseal defect model only simulates a bone defect environment, not a long bone fracture or defect in clinical scenarios. Only a few categories of materials are suitable for evaluating by this type of models, including bone filler or substitutes for localized bone defects, granular or colloidal materials without structural support capabilities, and bone cement, etc. Besides simple modeling without the requirement for fixation, another advantage of the condyle defect model is that it can avoid the displacement of testing material (Table 4).

Table 4 Metaphyseal defect models for biomaterial evaluation and relevant references
Full size table

Vertebral Body Defect Models

In some studies, the injectable biomaterials were evaluated on long bone defect models [79]. However, the different parts of the bone have varied mechanical environment, which may lead to varied material degradation behavior and bone tissue response in vivo [83, 84]. To simulate the actual environment, materials specifically for vertebral repair or augmentation are strongly recommended to be evaluated by vertebral defect models (refer to [85] for illustrative images).

Small Animal Vertebral Body Defect Models

In small animals, rats are often used to fabricate vertebral body defect models. The CSD of rat vertebra was explored by Liang et al. [85]. Two different sizes of defect, 2 mm × 3 mm × 1.5 mm and 2 mm × 3 mm × 3 mm, were made in anterior part of L4, L5, and L6 vertebra, and the former bone defect was found to be the CSD. Besides, the bone defect model can also be constructed in the caudal vertebra of rats. For example, a cylindrical defect (1.8 mm in diameter and 2 mm in depth) in the center of the third caudal vertebra was created through a left posterolateral approach [86]. The advantages of caudal vertebra model include simple anatomy approach and free of the risk to important organs and vessels. Meanwhile, there is even simpler method of modeling, such as PMMA directly injected into the L1–L6 vertebral body of New Zealand rabbits [87]. However, the clinical significance of such modeling methods for biomaterial evalution is questionable.

Large Animal Vertebral Body Defect Models

Sheep vertebral body defect model is most frequently used. The similarity of body weight and bone size of sheep to those of humans allows testing biomaterials designed for clinical study. The L2-L5 vertebrae are usually used to create the model. Usually, a cylindrical hole with a certain diameter and depth is drilled in the center of vertebral body to simulate the space created by balloon in kyphoplasty (KP) and then the defect is filled with injectable biomaterials. The diameters of the hole include 4 mm [88], 6 mm [88, 89], 8 mm [90, 91], and 10 mm [92], while the depths of the hole are 9 mm [88], 10 mm [89], 15 mm [90,91,92], and even 20 mm [93].

In addition to these common cylindrical defects, the rectangular defect with 8 mm high, 10 mm deep, and 20 mm long, was used to test injectable calcium phosphate cement [94]. Verron et al. [94] built an osteoporosis vertebral bone defect model, which was in line with the fact that most of the patients receiving the treatment of bone filling material suffer from osteoporosis. However, the establishment of this osteoporotic model needs a long modeling time (4–6 months) and high costs.

Canines are less used in vertebral defect model than sheep. In the modeling process, the lateral side of the lumbar spine is shown through the dorsal to the transverse process. The central vertebral defect with the size of 18 mm × 5 mm × 22 mm was produced, and the researcher injected the biomaterial into the defect for testing [95].

The bone defect in swine vertebral body can be created by percutaneous puncture under fluoroscopic guidance [96]. The lumbar vertebrae of mini-swine were also used for preparing defect model (15 mm in depth and 4 mm in diameter) [97]. Compared with other species, swine models are noisier and more aggressive, and their daily management is more difficult [98] (Table 5).

Table 5 Vertebral body defect models for biomaterials evalutions and relevant references
Full size table

Other Bone Defect Models

Femoral Wedge Bone Defect Model

Volker et al. produced femoral wedge bone defect model in rat to resemble clinically relevant situations of osteoporotic bone [100]. In Volker’s research, a 5-mm wedge-shaped osteotomy as critical size defect was conducted at the distal metaphyseal area of rat femur and fixed using a T-shaped plate.

Multiple-Defect Model in Canine Femur and Tibia

Bone defects could be established at various sites on canine bones. The middle section of the femur can produce two bone defects with 4-mm in diameter for testing biomaterials [101]. The canine femoral multiple defect (CFMD) is in the lateral cortex of the upper segment of the femur. During modeling, a specific device is fixed to the lateral cortex of the proximal femur by two screws: one upper and one lower. The four consecutive fabricated bone defects are cylindrical with 10 mm in diameter and 15 mm in length [102,103,104]. In addition, each bone defect is at least 1.5 cm [102] or 1.2 cm [105] apart from the normal bone and bone marrow. This model is suitable for evaluating different materials in the same animal at the same time. Thus, accurate data comparison could be reached, and the number of animals could also be reduced. It is also an effective comparison tool for the evalution of cell delivery [103] and growth factor delivery [102, 105].

The canine tibia could also be used to prepare tibil a multiple-defect model (CTMD) [106]. In a previous study, the tibia was exposed via a medial approach, and then an electrical motor was used to create consecutive cylindrical bone defects with 4 mm in diameter, where different materials could be implanted at the same time [106].

Nonunion Models

The short stumpy limbs of swine might restrict the creation of standard CSD models. Thus, some special models, such as nonunion model, have emerged. Schubert et al. [107] created a critical bone defect with a size 1.5 times that of the diameter of the femoral shaft in miniature swine (aged not older than 6 months and weighing ≤50 kg) and used two 4.5-mm locking compression plates for stabilization. Canines can also be used to produce nonunion model. Saifzadeh et al. produced a 2-mm transverse defect in the right medial radial diaphysis [108], where the defect ends were not fixed to create a nonunion model for testing autogenous greater omentum (Table 6).

Table 6 Other bone defect models
Full size table

Selection of Bone Defect Model

A suitable bone defect model for biomaterials and tissue engineering study is critical. The selection of bone defect model involves the considerations of characteristics of animal models, research purposes, and biological characteristics and design specifications of bone substitute materials.

Age and Sex of Animal

The bone repair ability is very high in young animals. For rats, the speed of bone repair in immature bones was more than that of the mature ones [109, 110]. For example, the healing time of femoral fractures is only 4 weeks for 6-week-old rats, whereas the healing time is 26 weeks for 52 week-old rats [111]. The similar results have been found in rabbit models [112], which might be related to the varied secretion of bone morphogenetic protein (BMP) [113]. The ability of fast bone repair could affect the test results of bone substitute material. Therefore, adult rats were usually selected in the establishment of bone defect models [1, 109, 114, 115]. Similarly, rabbit bone growth stops within 19–32 weeks [116] and therefore 6–9 month-old male rabbits are usually selected for study [72, 73].

Estrogen cycle also has a significant effect on bone repair and turnover. In ovariectomized rats, particularly in elderly rats, the delayed union of the femoral fractures and reduction of bone mineral density (BMD) would occur [117]. Therefore, the ovariectomized rats were often used to establish osteoporotic fracture models. Studies on non-osteoporotic bones should select male animals to avoid this interference. But it is worthy mentioning that male animals usually have a strong sense of territory and the use of individual cages may be required to prevent fights, injuries, or even deaths among male animals. Thus, the age and sex of the animal should be well controlled for study and reported in details.

Comparison between Different Animal Species

In bone tissue engineering and osteoinductive biomaterial researches, the animal species used for model is also critical for reliable and consistent results. The animals of each species have their own unique skeletal characteristics including bone morphology, bone microstructure, bone turnover, and bone remodeling. The ideal animal for bone defect model should have bone characteristics highly similar to that of human, in order to guarantee that the results of animal experiments could be safely extrapolated to humans. However, the availability, costs, and ethical consideration will also be involved in the selection of animal species. A general summary of various commonly used species is provided below.

Rodents

There is a great discrepancy between the bone microstructures of rodents and human. Different from the osteon structure in adult humans, the long bone cortex of rats consisted mainly of primary and cancellous bone, which lacks a Haversian canal system [118].

Intramembrane and endochondral ossifications are two typical bone formation mechanisms during fracture healing [119]. In the fracture healing of rodents, endochondral ossification is more predominant although both kinds of fracture healing mechanisms are present [120, 121]. The bone healing capability of rodents is greater than that of human [1]. Unlike human bone that ceases to grow after sexual maturity, the bones of the sexually mature rodent would continue to grow for extended time [122].

Mice are inexpensive and widely available, easy to manage, and have pure strain and strong infection-resistance. The mouse genome is also thoroughly researched, and it means that bone substitute material can be studied at the molecular level. When the inbred line mice are used in the model, the differences between individuals are small, which makes the results more reliable. Furthermore, nude mice are well-developed animal model that has no immunological rejection for bone tissue engineering substitutes that incorporated with human cells. However, the mouse skeleton is too small to establish a large defect model. Moreover, for some bone implants, shrinking to the appropriate size of the defect to evaluated the clinical outcome is difficult [123]. In addition, compared with mouse model, the rat model has better practical value in biomechanical tests.

Rabbit

Both the macro- and microstructure of rabbit’s bones are distinctly different from those of human [124]. Unlike the mature human bones with secondary hierarchical structure, rabbit skeletal bone has a primary vascular longitudinal tissue structure. It forms a vascular access to the Haversian canal that is wrapped around the medullary cavity and distributed on the periosteal surface [125]. The bones between these lamellar structures are composed of dense Haversian bone [125]. However, the bone structure of rabbit is still more representative for human than that of mouse [126].

Compared with primates and rodents, rabbits have faster bone remodeling, particularly in Haversian remodeling of cortical bones [127,128,129]. Thus, it is not advised to extrapolate from the rabbit’s findings to possible human clinical responses [130].

Rabbits are easy to manage, and the cost is less than larger animals. Compared with rodents, the biggest advantage of rabbit is the large size of bones that benefits for bone defect creation. Also, the ear margin vein of rabbit is convenient for obtaining blood samples, which is beneficial for blood analysis.

Canine

Canine bone components are highly similar to those of humans, including hydroxyproline, extractable proteins, IGF-1 content [131], as well as water fraction, organic fraction, volatile inorganic fraction, and ash fraction [132]. Microscopically, canine bone has secondary osteonal structures, known as plexiform bone or laminar bone in the adjacent endosteum and periosteum [124]. The plexiform bone is found in the rapid growth of children, or in rapidly growing large animals [133]. Therefore, canine bone is considered as one of the best representatives of human bone structure [131]. Thus, dogs are suitable for animal models in biomaterials research and bone tissue engineering. However, canine bone defect models have not been widely used due to the high costs and difficult breeding as well as ethical limitations in some countries.

When using canine as animal model, it should be noted that bone remodeling process of dogs is quite different from that in humans. The mean bone turnover rate of young adult female beagles is much higher than that of humans [134, 135], which could partially explain the high fusion rates in canine lumbar fusion models and the low nonunion rates [98]. The rate of bone remodeling in different bones varies significantly for dogs. Considering adult male beagles, the bone turnover rate is close to 200% and 12% annually in the lumbar vertebra and talus, respectively. Although, the mean total turnover of trabecular bone is approximately 100% annually [134].

Goat and Sheep

Sheep and goats have the similar weight as humans, and their limbs can provide sufficient testing space for clinically sized implants [128, 136, 137].

The structure of sheep bone is remarkably different from that of human histologically. Compared with human bone that mainly consists of secondary bone structure [138], the sheep bone has mostly primary bone structure where osteons is <100 μm in diameter and contains at least two central blood vessels. Bone structure changes in sheep are related to age. Before the age of 3–4 years, the sheep’s plexiform bone structure consists of a combination of woven and lamellar bone with sandwiched vascular plexuses [128]. When the sheep is 7–9 years old, Haversian remodeling occurs [128] and becomes more common as age increases [139]. Furthermore, the location of bone remodeling seemed to vary with bone types and the earliest indications of remodeling include the distal femur, radial shaft, and humerus [128]. Similar to sheep, the Haversian systems of goats are unevenly distributed in the whole skeleton, which mainly located in skull and the medial part of tibial shaft [140]. After comparing sheep with menopausal women in terms of bone mass, bone volume, and mineral deposition rate, Turner et al. concluded that old sheep were suitable as models for osteoporosis in the elderly [141].

Although the microscopic structure of sheep bone is different from humans, many studies have confirmed that bone turnover and remodeling activities in the sheep model are similar to those of humans [142], which makes the sheep as a valuable model animal for testing bone substitutes [136, 143]. Although the amount of bone ingrowth in sheep was greater than that in humans, Willie et al. found that sheep and humans had similar patterns in terms of bone ingrowth into porous implants [144]. Lamerigts et al. found similar sequences of events occurring in humans and goats for bone grafts [145]. Furthermore, Dai et al. found that the bone healing capacity and the blood supply of goat tibia were both similar to those of humans [38].

There are some differences between goats and sheep, including microscopic structure, bone metabolism, and remodeling. However, these two models were very similar in general. Therefore, choosing goats or sheep as animal models depends on availability and other personal preferences.

Swine

Compared with sheep model, swine model are closer to humans in bone morphology, bone metabolism, bone healing, and remodeling [98, 146, 147]. The cross-sectional diameter and area of femur are very similar to those of human [148]. In mature bone, the swine have well-developed Haversian systems [149] and a layered bone structure similar to human [150]. Furthermore, its bone density and bone mineral content are almost equal to those of human bones [131], although the trabecular meshwork of swine is denser than that of humans [151]. The bone growth rate in swine is 1.2–1.5 mm daily, which is comparable to that of human beings (1.0–1.5 mm daily). The processes of bone remodeling in swine, including trabecular and intracortical basic multicellular remodeling units (BMU) based remodeling, are all similar to those in humans [150, 151]. Moreover, mineralization rate of cortical bone was the same as that in humans [152]. Thus, swines are considered suitable models for bone-related studies [147].

Nevertheless, pigs weighing over 150 kg are considered not suitable to produce animal models for biomaterials and bone tissue engineering study. Even for miniature swine, the daily management is difficult, particularly because they are noisy and aggressive. Thus, in many cases sheep and dogs are preferred over swine as large animal models [128, 153].

Small animals, such as mice, rat, and rabbits account for >70% [154, 155] of all the animal models. In general, the advantages of small animals include good affordability, convenient management, and small individual difference. However, some problems exist when using small animal model to test the potential clinical products. Considering cell-seeded scaffolds for example, the viability and osteogenic potential of cells in small scaffolds are easy to maintain to repair bone defects in small animals but it becomes difficult in large scaffolds used for large bone defect repair. Besides, small animals have strong bone healing ability and weak immune response, which is different from the actual bone repair of the human body.

For large animal models, such as sheep, dogs, and swine, its large bones would allow the evaluation of implants with the size close to those used in humans. Moreover, compared with small animals, large animal bones have higher similarity to human bones. Although with many advantages, large animal models were not usually used due to the difficulties of accommodation, management, availability, and ethics.

Research Related Criteria

For experiments that are expected to get results in a short time, the small animal model (rodents and rabbits) is recommended due to their fast bone healing process. When long-term observation is required, large animal models are recommended, since the bone structure, bone repair mechanism, and immune response in large animals are closer to those in human body.

The study purpose of osteoinductive biomaterials and bone tissue engineering is also an important factor when choosing the model. If the experiment was to test the osteoinductive capability of the materials, skull defect models or long bone metaphyseal defect model could be used to simplify the experimental procedure.

However, more complicated models are needed for studies with specific research goals. For example, long bone segmental defect model should be utilized to test the substitute materials designed to repair long bone defects, in which fixation devices would be utilized to fix the ends of defects to simulate clinical practice. When bone substitute materials are used as bone cement for vertebroplasty (VP) and kyphoplasty (KP) surgery, the vertebral defect models are recommended.

The characteristics of materials also affect the selection of models. If the bone substitute materials were colloidal, shapeless granule, or powder, the long bone metaphyseal defect model or the multiple defect model might be a better choice, since it can avoid the dislocation and loss of biomaterials or scaffolds.

For materials containing human tissue cells, nude rodent model may be suitable models to prevent the immune system from interfering with cells. Nude rodent models are capable of ignoring the source of cells and the reaction between scaffold and the host; hence, it could be used to study the efficacy of engineered tissue products directly [156]. However, because nude rodents are immune-deficient animals and cannot reflect the normal immune functions, the data obtained from nude rodent models may need to be further confirmed by other animal models.

Bone Defect Models Simulating Clinical Scenarios

Actual Situation of Bone Defect in Clinical Settings

It should be recognized that most fractures were simple two-part fractures or minor comminuted fractures without bone defects. Bone defect is usually caused by comminuted broken ends of the fracture. In fact, serious violent injuries not only cause fracture end comminution, but also force the fractured fragment to move, rendering it impossible to maintain the position of the broken ends.

When the bone defect exceeds the self-repairing ability, nonunion would occur without external intervention therapy. However, the bone defect size is not the only factor that affects bone healing. The real bone defect is accompanied by severe damage of blood supply at the broken ends of the fracture, which would seriously hamper the fracture healing. Therefore, the occurrence of bone defect and fracture healing are not simple processes in clinical practice. Unfortunately, most of the current bone defect models did not well imitate the actual state of clinical bone defects.

Fracture Model and Nonunion Model

There are two models that simulate the actual state of a bone fracture: closed fracture model and nonunion model. The closed fracture model is the most widely used animal model for the study of fracture healing, which is caused by exerting strength damage without incision to completely simulate the clinical practice. To make consistent fracture models, Jackson made fracture of rat femur on a force-controllable blunt guillotine device after inserting intramedullary needle into marrow cavity [157]. Bonnarens and Einhorn [158] improved the device to make the whole modeling tool portable and easy to operate. Thus, the tools and concepts of closed fracture model making have been adopted by many scholars. Marturano et al. [159] made further improvements to the fracture device, making it suitable for preparing models on the femur of mice. The tibia of mouse may also be used to make a fracture model using similar methods [160,161,162,163]. Compared with femur, the location of tibia has less muscle attachment. However, the tibia is not an ideal model for fracture-related research due to the curved long axis, complicating the biomechanical test and the insufficient soft tissue around the bone [164, 165]. Furthermore, the fibula is accidentally fractured at a rate up to 30% during the modeling process, which can change the healing rate of the tibia [166].

For creating nonunion model, closed [167,168,169] or open [170] fractures are created first. Subsequently, the 2-mm periosteum of the two fracture ends are cauterized and corroded, which destroy the blood supply at the fracture ends, resulting in atrophic nonunion [167,168,169,170]. These nonunion models are obviously different from the nonunion caused by severe trauma in clinical practice, which causes not only the crushing of fracture ends, but also large segmental bone defects, and destroys the blood supply around the broken ends.

Fracture-Bone Defect Models

Closed fracture models can simulate the fractures seen in clinical practice. However, the closed fracture models are suitable for fracture healing research without space for implanting biomaterial or bone tissue engineering scaffolds. More importantly, these fracture models can self-heal without external intervention. Although current nonunion models cannot self-heal, they are not consistent with the clinical practice. Fracture defect models mimicking the clinical scenarios are still lacking.

For producing fracture-defect model, fracture is first produced. Blunt guillotine and three-point bending are classic methods of creating fractures, and the tools are simple and easy to operate. However, unlike aforementioned fracture models that have transverse fracture, fracture-defect models aim to create comminuted fracture.

After creating the fracture, open operation can be used to expose the fracture ends and remove the broken pieces of bone to create a bone defect. The incision at the broken ends can be regarded as an open fracture, which is consistent with the open fractures in severe traumatic injury.

Fracture-defect model can use tibia to prepare. This is because tibia has less muscle encircling, convenient positioning and is directly located in the subcutaneous tissue, which makes the removal of broken bones much easier. In clinical practice, the open comminuted fracture also occurs mostly in the tibia.

Challenges and Future Prospects

There is no standard answer for how much bone should be removed to reach the state of nonunion. In bone defect model, there is a concept of CSD and CSD values have been established in many animal models. While in fracture defect models, the factors affecting bone healing include not only the size of the defect, but also the degree of soft tissue injury at the fracture ends. Unfortunately, quantification of soft tissue damage in the process of modeling is difficult.

The present fracture creation tools are designed for small animal models. When produce similar models in large animals, redesign and development of new tools becomes necessary. However, it is not known whether the previous classical weight fall and the three-point bending principle are applicable to or effective on large animal models. In addition, small animals, such as mice and rats, used for fracture model have less individual difference and high repeatability. When fracture defect model is made using large animals, the individual difference between animals enlarges and the repeatability of the model decreases. 

Summary

Compared with other animals, canine and swine are closer to humans in bone morphology, bone metabolism, fracture healing, and remodeling. However, the rat model is the most widely used bone defect model in the study of biomaterials and bone tissue engineering. It can be seen that in the actual selection of animal models, the factors that need to be considered are multifaceted. In the selection of animal species, researchers should consider the ultimate purpose of the biomaterial evalution, the material characteristics, their own facility capacity, costs and other resources. In addition, it should be recognized that there is a significant difference between the animal bone defect models and clinical. In the future, the development of fracture defect models that are close to clinical reality may be come possible, making the evaluation results more reliable for the studies of biomaterials and bone tissue engineering.