An overview of de novo bone generation in animal models

Abstract

Some of the earliest success in de novo tissue generation was in bone tissue, and advances, facilitated by the use of endogenous and exogenous progenitor cells, continue unabated. The concept of one health promotes shared discoveries among medical disciplines to overcome health challenges that afflict numerous species. Carefully selected animal models are vital to development and translation of targeted therapies that improve the health and well‐being of humans and animals alike. While inherent differences among species limit direct translation of scientific knowledge between them, rapid progress in ex vivo and in vivo de novo tissue generation is propelling revolutionary innovation to reality among all musculoskeletal specialties. This review contains a comparison of bone deposition among species and descriptions of animal models of bone restoration designed to replicate a multitude of bone injuries and pathology, including impaired osteogenic capacity.

1. INTRODUCTION

The goal and focus of innumerable scientific efforts throughout recorded history was to decipher and harness the power and unlimited potential of the cell. Discovery, isolation, and culture of cells that can assume characteristics of numerous lineages, including those from distinct embryonic layers, ignited a virtual explosion of discovery in the vast arena of cell therapies over the last two to three decades. A natural trajectory of the therapeutic momentum is to replace musculoskeletal tissue compromised by trauma, disease, or malformation with healthy tissue via de novo tissue generation. Broad approaches include in vitro generation of viable, implantable tissue, and application of exogenous cells and materials to recruit and direct endogenous cells. Carriers for cell delivery are composed of materials that facilitate tissue formation by progenitor cells, and they are routinely customized at the macro‐, micro‐, and ultra‐structural levels to replicate tissue matrix, including organic and inorganic components. Tremendous advances in de novo tissue generation provide unlimited opportunities to restore musculoskeletal tissue and impact the health and wellbeing of global community members at any stage of life.

The process of moving innovative de novo musculoskeletal tissue generation from concept to clinical reality is incremental and iterative. Key elements of successful translation from bench to bedside are reproducible animal models that recapitulate targeted musculoskeletal pathology. Models vary widely among joints and limbs and between traumatic, degenerative, and developmental conditions. Many are induced by surgical or chemical means, and test therapies are applied immediately or after a period of time following the initial injury. Numerous considerations are associated with selection of an animal model. There are specific factors related to the scientific questions or techniques to be tested and practical considerations like animal cost, availability, and regulation‐compliant surgical and housing facilities. Published substantiation of, and investigator experience with a model in addition to validated outcome assessment assays, including proteomic and genomic panels, also guide selection. Customized genetic makeup and immunodeficiency are found primarily in rodents. Findings from such highly tailored models require testing in larger mammal models before clinical translation and implementation.

Orthotopic evaluation of bone healing in a large animal model is frequently part of the final preclinical testing stages. In addition to anatomy and magnitude of load bearing, ¹ bone formation and microstructure are critical assessments of an animal model (Table 1). It is also important to remember that, while bone composition is relatively highly conserved, it is not identical among species ² ; canine and porcine are relatively close in composition and density to human, while rat has few similarities. Additionally, bone regeneration declines and morphology ³ changes differently with age among distinct life spans. ⁴ , ⁵ This is especially relevant to defining critical size defect (CSD) ⁶ sizes at various maturity levels in animals (Table 2). The following sections provide an overview of animal bone regeneration models beginning with a general comparison of bone turnover rates.

Table 1.

Animal model long bone characteristics

	Small mammal	Large mammal	NHP and human
Sexual maturity age	Murine: 6–8 weeks 7	Canine: 7–21 months 8	Human: ~17 years 9
	Rat: 6 weeks 10	Ovine: 7–8 months 11	NHP: 4–6 years 12
	Lapin: 10–12 weeks 13	Porcine: 5–6 months 11
		Equine: 7–14 months 10
Skeletal maturity age	Murine: 16–24 weeks	Canine: 10–11 months	Human: ~25 years
(Growth plate closure age/life expectancy age x 100) 14	(13.9–27.8) 7	(4.3–6.9) 8	(16.7–25) 15
	Rat: 24–32 weeks	Ovine: ~40 months	NHP: 7.2‐10 years
	(22–35) 10	(9.4) 16	(11.2–17.5) 17 , 18
	Lapin: 28–30 weeks	Porcine: 18–22 months 11 , 19 , 20
	(5.5–8.1) 13	Bovine: 12‐37 months
		(6.7–20.1) 14
		Equine: ~3 years
		(5.8–6.3) 21
Fractional area of secondary bone (FASB)	Rat: minimal 22	Ovine: 2%–91% 23	Human: ≈48% 24
	Lapin: minimal 25	Bovine: ≈11% 24	NHP: 61%–74% 26
		Equine: 5%–75% 23 , 27
Bone remodeling period	Murine: ~2 weeks 7	Canine: ~2 months 28 , 29	Human: 6–9 months 7
	Rat: ~6 days 30	Ovine: ~80 days 31	NHP: 8–24 months 32
	Lapin: 70 days 33	Porcine: 1–5 months 32
Bone formation rate/bone volume (BFR/BV) at skeletal maturity (bone type)	Murinea: ≈1900% (cancellous) 34 , 35 , 36	Canine: 0.5%–6.4% (cortical) 28 , 37 , 38	Human: 3%–4% (cortical) 39 , 40 , 41
	Rat: ≈19% (cortical) 42	20%–50% (cancellous) 38	≈26.3% (cancellous) 43
	≈1158% (cancellous) 44	Ovine: 55%–72% (cancellous) 45	NHP: 13%–38% (cancellous) 46 , 47
	Lapin: ≈20.7% (cortical) 33	Porcine: ≈53% (cancellous) 32
		Equine: ≈10% (cortical) 48
Pelvic limb axial force	Lapin: 201% BW 49	Caprine: ≈100% BW 50 , 51	Human: 470% BW 52
		Ovine: 48% BW 53

Abbreviations: BW, body weight; NHP, nonhuman primate.

^akeletally immature.

Table 2.

Critical defect size and fixation among bones and species

Bone	Species	Defect size (mm)	Fixation	Potential advantages
Calvarium	Murine	>Ø 2 54 , 55
	Rat	>Ø 5 56 , 57
	Guinea Pig	10 58
	Lapin	>Ø 6 59 , 60
	Canine	20 61 , 62
	Ovine	>30 63 , 64 , 65
	Porcine	>Ø 10 66 , 67		Bone composition similar to human 2
Rib	Canine	>50 68 , 69		Thoracic wall kinetics similar to human 70 , 71
	Ovine	40 72	Plate
	Porcine	100 73
Ilium	Lapin	>Ø 5 74 , 75
	Caprine	>Ø 8 76 , 77 , 78
Humerus	Lapin	>7 79 , 80 , 81	Plate, intramedullary rod
	Canine	>Ø 5 82 , 83 , 84 , 85 , 86
	Ovine	>Ø 6 87 , 88 , 89
Radius	Rat	>5 90 , 91
	Lapin	>14 92 , 93 , 94		Segmental defect without fixation possible
				Established radiographic and histologic scoring system 95
		>10 96 , 97 , 98	Plate
Femur	Rat	>4 99 , 100 , 101 , 102 , 103	Plate, external fixator, Intramedullary rod	Highly standardized fixation systems
				Macrostructurally similar to human 1
	Canine	>21 104 , 105 , 106 , 107	Plate, Intramedullary Rod	Macrostructurally similar to human 1
				Bone composition similar to human 2
	Caprine	Ø 8 76
Tibia	Lapin	15 108	Plate
	Ovine	>30 4 , 109 , 110 , 111 , 112	Plate, external fixator	Defect strain similar to human
Vertebrae	Rat	>Ø 3 113 , 114 , 115
	Caprine	Ø 5 76
	Ovine	>Ø 6 116 , 117 , 118 , 119 , 120 , 121 , 122
Mandible	Rat	>Ø 3 123 , 124 , 125 , 126 , 127 , 128
	Porcine	>17 129 , 130 , 131		Macrostructure and microstructure and masticatory force similar to human 132 , 133

Note: Ø, cortical defect diameter.

2. BONE FORMATION DURING NORMAL HOMEOSTASIS

2.1. Bone remodeling

For comparisons among species, the rate of natural bone formation during normal homeostasis should be considered (Table 1, Figure 1). Two measures of bone activity are the extent of remodeling and rate of remodeling. Both vary with species, age, bone, and bone region. The fractional area of secondary bone (FASB), area comprised of secondary osteons in cortical bone, represents the amount of remodeling present. It is defined as the percentage of the total area of secondary bone relative to the total area of interstitial bone and secondary bone together. ¹³⁴ The higher the FASB, the greater the extent of bone remodeling. In general, the amount of remodeled bone increases with age. Additionally, bone regions under compressive stress have the highest extent of remodeling and therefore secondary osteons while those under tensile stress retain more primary bone. ²³ , ¹³⁵ , ¹³⁶ , ¹³⁷ Due to anatomical differences among species, there are regional differences within bones. As an example, the human femur experiences largely compressive stresses, ¹³⁸ , ¹³⁹ while most quadruped femurs are subjected to both tensile and compressive stress; this leads to important species‐specific characteristics in regional bone remodeling. ¹⁴⁰

Figure 1.

Animal and human remodeled cortical bone (fractional area of secondary bone) and bone formation rate (bone formation rate per unit of bone volume) with wedge area representing relative amounts and rates [Color figure can be viewed at wileyonlinelibrary.com]

Bone maturity is an important consideration in animal models. Broadly speaking, humans and nonhuman primates have highly remodeled cortical bone at maturity followed by large mammals; small mammals like lapin, rat, and murine have minimal remodeled bone as adults. ²⁴ , ²⁵ , ²⁶ , ¹⁴¹ Among large mammals, the canine and equine FASB are closest to that of human. ²³ , ²⁷ , ¹⁴² , ¹⁴³ Another distinct difference between animal and human bone is the prominent proportion of plexiform bone, a form of primary bone present during bone growth in rapidly growing mammals, that can result in a relatively low FASB (Figure 2). ²³ , ²⁴ , ¹⁴⁴ Large mammals typically develop secondary bone near the endosteum while plexiform bone remains adjacent to the periosteum. ²² Additionally, the relative size of the osteonal resorption and Haversian canal areas within secondary osteons is positively correlated with body mass; the higher the body mass, the greater the area of each. Animals close in size to human counterparts may have similar secondary osteon structures. ¹⁴⁵

Figure 2.

Photomicrograph of an undecalcified section of ovine endosteal cortical bone from the radius. Plexiform cortical bone is on the left. Active remodeling is indicated by the presence of secondary osteons (white arrow heads) on the right. Toluidine blue stain. Scale bar = 100 µm. (Photo courtesy of Dr. Clifford Les) [Color figure can be viewed at wileyonlinelibrary.com]

2.2. Bone formation rate

Distinct from the amount of remodeled bone is the rate of trabecular or cortical bone turnover, often measured as the bone formation rate per unit of bone volume (BFR/BV). ¹⁴⁶ In humans, the BRF/BV is a well established measure ¹⁴⁷ , ¹⁴⁸ that is affected by age, ¹⁴⁹ use, ⁴² , ¹⁵⁰ and comorbidities. ³⁹ , ¹⁵¹ , ¹⁵² In animal models, it is used to assess both the rate of bone remodeling and bone healing. ¹⁵³ In general, BFR/BV is higher in cancellous bone than cortical bone, and tends to be higher in small versus large mammals and lowest in human cortical and cancellous bone. ²⁸ , ³² , ³³ , ³⁴ , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² , ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ , ⁴⁷ , ⁴⁸ Age has a large impact on BFR/BV in animal models. In rats, the BFR/BV of the proximal tibial metaphysis varies from 290.9% and 335.2% at 1 and 3 months of age, to 61.9% and 80.1% at 6 and 14 months of age, respectively ¹⁵⁴ ; in dogs, the BFR/BV of the femoral mid‐diaphysis is 72% in immature and 1%–6.4% in mature animals. ²⁸ , ³⁷ The process of bone remodeling during normal homeostasis is somewhat demonstrative of, but not identical to, bone healing capacity. ¹⁵⁵ Additionally, the FASB and BFR/BV permit some relative comparisons among species, but they are only two representative measures of normal bone remodeling (Figure 1). Any number of measures may be used or combined to monitor inherent bone forming capability, ¹⁴⁶ an important consideration when utilizing animal models to test bone regeneration strategies.

3. MODELS OF BONE REGENERATION

3.1. Flat Bone

Common flat bone models include the calvarium, costae, and ilium. These non‐load bearing bones permit use of multiple CSDs without fixation, and intramembranous ossification is highly conserved among species. ¹⁵⁶ , ¹⁵⁷ Among the three, full‐thickness (bicortical), round defects in the rodent and lapin calvarium are the most popular for initial in vivo, orthotopic, and non‐orthotopic testing (Figure 3). ⁵⁶ , ⁵⁷ , ⁵⁹ , ⁶⁰ Notably, the dura mater is reported to be a source of bone morphogenetic protein 2 (BMP‐2) in young animals that seems to diminish with age. ⁵⁸ , ¹⁵⁸ The surgical procedure for calvarial defect creation is relatively simple, and the thin murine calvarium permits in vivo cell imaging with multi‐photon microscopy to investigate spatiotemporal coordination of cells that contribute to bone healing. ¹⁵⁹ As an example, two‐photon microscopic imaging was used to confirm that exogenous bone marrow derived multipotent stromal cells on collagen/hydroxyapatite scaffolds were primarily responsible for new bone formation in murine calvarial defects while host cells participated most in periosteum regeneration. ¹⁶⁰ Both circular defects and craniectomies are reported in large mammals like canine, ovine, and porcine in which ostectomies recapitulate craniectomies for congenital malformation, trauma, and neoplasia. ⁶¹ , ⁶² , ⁶³ , ⁶⁴ , ¹⁶¹ Instrumented transport osteogenesis models are reported as well as self‐retaining materials that facilitate detailed magnetic resonance imaging. ⁶¹ , ⁶² , ⁶³ , ⁶⁴ Collectively, calvarial defect models in both small and large mammals are valuable models for proof‐of‐concept testing in non‐load bearing bone.

Figure 3.

Schematic representation of common animal bone defect models [Color figure can be viewed at wileyonlinelibrary.com]

Costal bone is a common harvest site for autologous bone and costochondral grafts, and thoracic surgery or trauma can necessitate rib resection. ¹⁶² , ¹⁶³ , ¹⁶⁴ Rib ostectomy models to test rib regeneration options are designed to address pain, instability, and cosmesis associated with large defects. ⁷² , ⁷³ , ¹⁶⁵ , ¹⁶⁶ In part, due similarity in size to human, ovine models of rib resection are common. ⁷² , ¹⁶⁷ , ¹⁶⁸ Thoracic wall reconstruction models are typically in species that share human thoracic cavity dynamics like canine and lapin. ⁶⁸ , ⁶⁹ , ⁷⁰ , ⁷¹ , ¹⁶⁹ A concavoconvex costovertebral joint ¹⁷⁰ in cursory mammals like humans facilitates thoracic cavity expansion by intercostal and diaphragmatic musculature. ¹⁷¹ , ¹⁷² Non‐cursory mammals like caprine and ovine species have a flat costovertebral joint that relies on diaphragmatic musculature for thoracic expansion. ¹⁷⁰

The iliac crest is another non‐load bearing bone used for materials testing with the important distinction of healing by endochondral ossification. Caprine and ovine models of circular unicortical or bicortical defects along the iliac crest are popular because microstructural cancellous bone volume and connectivity are similar to human ¹⁷³ , ¹⁷⁴ and ovine models of osteoporosis are well established. ¹⁷⁵ , ¹⁷⁶ The ilium is one of the most common sites of autologous cancellous and corticocancellous bone graft harvest. ¹⁶³ , ¹⁷⁷ Vascularized iliac bone block resections for treatment of avascular bone lesions or multiple corticocancellous bone harvests for staged surgical reconstructions drive efforts to enhance iliac bone regeneration. ¹⁷⁸ , ¹⁷⁹ , ¹⁸⁰ For large defects and iliac bone blocks, the ovine is particularly advantageous due to anatomical properties that are close to that of human. ¹⁸¹ The ovine ilium has only a slightly longer iliac shaft and smaller wing than the human female.

3.2. Long Bone

Many animal models of long bone generation correspond to the most prevalent long bone fractures in humans. ⁷⁶ , ⁹² , ¹⁸² , ¹⁸³ , ¹⁸⁴ Multiple, round, unicortical or bicortical, CSDs and non CSDs are used for orthotopic testing in virtually every bone of large and small animals. ⁸⁷ , ¹⁸⁵ Defect creation typically requires minimal soft tissue trauma and internal fixation is not required. Models that incorporate ostectomy or osteotomy require internal or external stabilization. ⁵⁸ , ¹⁸⁶ , ¹⁸⁷ , ¹⁸⁸ , ¹⁸⁹ , ¹⁹⁰ Large animal models have advantages of large defects and use of standard surgical tools and devices that are not possible in small animals. ¹⁹¹ As a general rule, long bone diaphyseal CSDs correspond to approximately 2–2.5 times the diaphyseal diameter, about 3–5 cm in ovine ¹⁰⁹ , ¹⁹² , ¹⁹³ , ¹⁹⁴ or 3 cm in porcine ¹⁹⁵ adult tibiae. Ostectomies are typically used to represent comminuted or unstable fractures while osteotomies represent minimally displaced fractures with limited comminution. ¹⁹⁶

As indicated above, load bearing varies between quadrupeds and bipeds, especially in the forelimb equivalent of human arms. Based on bone mineral density changes in astronauts, bone deposition in the human pelvic limb is especially responsive to frequency and magnitude of loading, while the thoracic limb is less so. ¹³⁵ , ¹⁹⁷ , ¹⁹⁸ , ¹⁹⁹ , ²⁰⁰ , ²⁰¹ The relative physiologic load is comparably higher in the forelimbs and lower in the hind limbs of most animal models. Large mammals bear about 60% of body weight on the forelimbs while rodents and lapin bear approximately 55%. ²⁰² , ²⁰³ In terms of tibial loading at a walk, the lapin ⁴⁹ is reportedly closest to human, ⁵² 201% and 470% of body weight, respectively, while caprine is about 100% of body weight. ⁵⁰ These important distinctions must be carefully considered when selecting animal models and comparing results among species. Specific information about popular long bone models is provided below (Figure 3).

3.2.1. Thoracic Limb

There are a number of animal models to assess proximal humeral epi‐ and metaphyseal bone regeneration. ⁸² , ⁸³ , ⁸⁴ , ⁸⁷ , ⁸⁸ , ⁸⁹ , ²⁰⁴ In part due to anatomical congruity between the human and canine humerus, canine cylindrical defect models are commonly employed in young to aged adult dogs. ⁸² , ⁸³ , ⁸⁴ , ⁸⁵ , ²⁰⁴ Additionally, proximal humeral osteosarcoma occurs naturally in adolescent to young adult dogs, 18–24 months, somewhat analogous to human adolescents. ²⁰⁵ These points support the value of canine models to optimize osteogenesis in the proximal humerus. A typical critical size cylindrical defect in the canine proximal humerus is about 5 mm wide and 4 mm deep in middle‐size dogs (25–35 kg). ²⁰⁶ A valuable stage to assess treatment effects in the model is reportedly during the fibrous to lamellar bone transition between 4 and 6 weeks postoperatively. ⁸² , ⁸⁴ , ⁸⁵ , ⁸⁶ In addition to histologic and histomorphometric analyses, electron probe microanalysis can be used to determine regenerated bone maturity based on chemical composition, typically the calcium/phosphorus ratio. ⁸² , ⁸³ , ⁸⁴ , ⁸⁵ , ²⁰⁴ , ²⁰⁷ Using the outcomes above and a proximal metaphyseal cylindrical defect model in young (1–2 years) and senior (10–12 years) adult dogs, transforming growth factor‐β2 on titanium cylinders increased bone volume to tissue volume by three‐fold compared with implant alone, though regenerated trabeculae were thinner and unmineralized osteoid higher in senior animals. ⁸⁴

A popular segmental defect model in the humerus is in the lapin mid‐diaphysis. ⁷⁹ , ⁸⁰ , ⁸¹ CSDs in skeletally mature rabbits are around 7 mm long and stabilized with an intramedullary rod or bicortical plates. ⁷⁹ , ⁸⁰ , ⁸¹ In vivo monitoring is typically via radiography and nuclear scintigraphy, and postmortem histology is standard. ⁷⁹ , ⁸⁰ , ⁸¹ Determination of torsional strength and stiffness via mechanical testing is well established and consistent with predominant physiologic stresses. ⁸¹ , ²⁰⁸ , ²⁰⁹ Complete healing of segmental defects can be achieved as early as 6 weeks, but there is a high rate of non‐union up to 8 months after injury, 43%–100%, reportedly a result of poor healing capacity. ⁷⁹ , ⁸⁰ , ⁸¹ This makes the model appealing for developing treatments to overcome similar complications in human humeral fractures. ²¹⁰ , ²¹¹ , ²¹² In one report, titanium mesh implants with BMP‐2 in polymer gel had 100% complete bony bridging of 15 mm humeral defects 6 weeks after implantation, while none of the defects without implants achieved bridging. ⁸⁰ This and other reports help establish that the lapin humeral segmental defect model is amenable to testing therapies for suboptimal healing capacity. ⁷⁹ , ⁸¹

Most lapin and rat species have a radio‐ulnar synostosis. ⁹⁰ , ⁹¹ , ⁹² , ⁹³ , ⁹⁴ , ⁹⁶ , ⁹⁷ Though load bearing is shared between the bones, radial ostectomies are stable and do not require internal fixation. ⁹⁰ , ⁹¹ , ⁹³ , ⁹⁴ At lapin skeletal maturity, segmental radial defects range from 10 to 14 mm, though 14 mm is recommended for a CSD; the segmental radial CSD in a skeletally mature rat is greater than 5 mm. ⁹⁰ , ⁹¹ , ⁹³ , ⁹⁴ , ⁹⁶ , ⁹⁷ Radiography and microcomputed tomography (µ‐CT) as well as histology outcome measures are standard, ⁹⁰ , ⁹¹ , ⁹³ , ⁹⁴ and the Lane–Sandhu scoring system for both radiograph and histologic quantification of bone healing ⁹⁵ facilitates comparisons among studies. ⁹⁰ , ⁹¹ , ⁹⁴ Serum biomarkers are also possible outcome measures; prolonged healing in aged rats is associated with significantly lower levels of bone biomarkers like osteocalcin and alkaline phosphatase. ²¹³ Evidence of rat and lapin radiographic bony bridging typically coincides with full recovery of mechanical strength in compression and bending. ⁹⁰ , ⁹¹ , ⁹⁶ Nanoindentation of thin sections (~ 100 μm) to measure modulus and hardness of new bone has also been reported in the lapin model. ⁹⁸ Although less common, segmental radial ostectomies are reported in Yucatan miniature swine which also have a radio‐ulnar synosthosis. ²¹⁴ Previously, 25–30 mm long defects filled with polymeric membrane in one‐year‐old animals were bridged radiographically by 8 weeks. ²¹⁴ The miniature swine model has unique advantages of a large size without the need for internal or external fixation.

3.2.2. Pelvic Limb

The most common femoral segmental defect models are rat and murine. Anatomically, the rat femur resembles that of human, and femoral neck and greater trochanter ossification centers do not coalesce in either species. Closure of the rat and murine femoral and tibial physes relative to lifespan are comparable to humans and later than other mammals. ¹⁴ Immunocompromised rodent strains permit testing of xenogeneic cells and biomaterials. Recently, human adipose stromal vascular fraction cells on ceramic scaffolds that enhanced bone formation in immunocompromised rats in preclinical testing also promoted proximal femoral fracture healing in a clinical trial. ⁹⁹ Detection of human cells in immunocompromised animals can be accomplished by standard methods including identification of human genetic sequences and antigens by in situ hybridization and immunolabeling, respectively. ⁹⁹ Commercially available fixation systems for rodent femoral stabilization range from radiolucent plates ⁹⁹ and interlocking nails ¹⁰⁰ to external fixators. ⁹⁹ , ¹⁰¹ Segmental defects greater than 4 mm in adult rats require about 8 weeks for complete bridging, though study end points typically range from 4 to 12 weeks, and bone formation is monitored similarly to the rodent forelimb. ⁹⁹ , ¹⁰⁰ , ¹⁰¹ , ¹⁰² Evaluation of bone mineral density in regenerated rat bone with scanning electron microscope‐based quantitative backscattered electron imaging has been reported. ¹⁰¹ , ²¹⁵ Mechanical tests are frequently designed to assess torsional properties, though mechanical testing varies widely. ⁹⁹ , ¹⁰³ A potential disadvantage of the rodent femoral defect model is the well‐recognized robust healing capacity with and without fixation that can necessitate outcome validation in larger mammals. ²¹⁶

Similarities between human and canine femoral anatomy contribute to the value of the canine femoral segmental defect model despite thinner cortices in the canine bone. ¹ , ²¹⁷ , ²¹⁸ Intramedullary rods or cortical plates are used to stabilize CSDs of at least 21 mm in skeletally mature, middle‐ to large‐size dogs (12–55 kg). ¹⁰⁴ , ¹⁰⁵ , ¹⁰⁶ , ¹⁰⁷ Outcome assessments are similar to other species, bony bridging typically occurs around 12 weeks, and remodeling has been monitored for extended periods, 24 weeks or more, postoperatively. ¹⁰⁴ , ¹⁰⁵ , ¹⁰⁶ , ¹⁰⁷ Unlike small mammals, however, recovery of mechanical strength does not always coincide with radiographic healing; this is likely a consequence of extensive remodeling associated with canine bone healing, similar to human bone. ¹⁰⁷ , ²¹⁹ Use of gait kinetics to quantify limb use are fairly common in canine studies. ²²⁰ , ²²¹ Ground reaction forces measured with a force platform are positively correlated with bone healing and have a strong association with callus mineralization and defect stiffness. ²²² A recent study showed that addition of human osteogenic protein‐1 to cortical allograft strips in canine femoral defects improved limb use over allograft alone 10 weeks after surgery. ¹⁰⁴ Wide use of canine gait kinetic measures permits comparisons among a multitude of orthopedic studies, including those with a focus on accelerated bone formation. ²²⁰ , ²²¹

Among long bones, tibial ostectomies are frequently used to model traumatic bone loss. ²²³ As mentioned above, lapin tibial loading is closer to human than other small mammals, so the lapin tibial mid‐diaphyseal segmental defect model, typically stabilized with a bone plate, may have the strongest translational value. ⁴⁹ , ⁵² , ⁵³ , ¹⁰⁸ In a large animal model, ovine tibial defects of 30 mm or more in the mid‐diaphysis are often treated with plates or external fixators and monitored by standard means up to 3 to 12 months followed by histology and mechanical testing. ⁴ , ¹⁰⁹ , ¹¹⁰ , ¹¹¹ , ¹¹² , ²²⁴ The ovine tibial diaphysis has a relatively simple cylindrical macrostructure and loading mechanism compared to more structurally complex bones. ²²⁵ , ²²⁶ , ²²⁷ This, in addition to similarity in weight to adult humans, lends itself to testing of three‐dimensional printed grafts with varied microstructure and composition. ¹⁰⁹ , ¹¹⁰ , ¹¹¹ , ¹¹²

3.3. Vertebrae

Animal models of de novo vertebral bone synthesis are largely divided into two types, vertebral body defects and spinal fusion. In small mammals, vertebral body defects are typically spherical ¹¹³ , ¹¹⁴ , ¹¹⁵ , ²²⁸ , ²²⁹ as in an osteoporotic rat model that showed increased bone formation and improved stiffness of new bone with platelet‐rich plasma combined with a gelatin/β‐tricalcium phosphate (TCP) sponge. ²³⁰ Both kyphoplasty materials and novel implants are frequently tested in ovine. ¹¹⁶ , ¹¹⁷ , ¹¹⁸ , ¹¹⁹ , ²³¹ , ²³² Defect models often replicate highly prevalent lumbar (L) 2–5 vertebral body compression fractures. Midbody defects up to 6 mm in diameter and burst fractures created by manual compression are reported in skeletally mature animals with 12–36 weeks postoperative follow up. ¹¹⁶ , ¹¹⁷ , ¹¹⁸ , ¹¹⁹ , ¹²⁰ , ¹²¹ , ¹²² , ²³³ As with all bones, there are important anatomical differences between human and animal vertebrae, however, the immature domestic porcine (55–65 kg) vertebral macrostructure resembles the human in pedicle dimensions, ²³⁴ vertebral body height, and end‐plate and spinal canal shape. ²³⁵ Polymethylmethacrylate cement containing magnets injected into porcine thoracic vertebrae to mimic kyphoplasty attracted systemically administered magnetic nanoparticles. ²³⁶ Lumber posterolateral spinal fusion is modeled in large and small animals, among which the rodent and lapin are commonly used to evaluate de novo bone formation and remodeling in a non‐instrumented model (Figure 4). ²³⁷ , ²³⁸ , ²³⁹ , ²⁴⁰ Bilateral decortication of the L4–L5 or L5–L6 transverse processes, with or without decortication of the spinous processes and lamina, produces stable fracture beds to which materials are applied topically. As illustrated by a rat spinal fusion study that showed syngeneic adipose tissue‐derived multipotent stromal cells (ASCs) on β‐TCP/collagen type I matrix enhanced bone formation over matrix with allogeneic ASCs or matrix alone, ²⁴⁰ bone formation is readily assessed with radiographs, μ‐CT and routine histology. ²³⁷ , ²³⁸ , ²³⁹

Figure 4.

Demineralized bone matrix (A, white circles) and corticocancellous bone (B, white circles) during implantation in a rat model of lumbar spinal fusion. The lumbar spine is evident between the circles in each image [Color figure can be viewed at wileyonlinelibrary.com]

3.4. Facial Bone

The mandible, orbital, zygoma, maxilla, and frontal bones are frequently sites of congenital malformation and trauma. ²⁴¹ , ²⁴² Facial bone regeneration is somewhat distinct in that minimal soft tissue coverage and esthetics require close replication of the original structure. In the rat model, CSDs in the mandible beneath the pterygomasseteric sling are used to test materials. ¹²³ , ¹²⁴ , ¹²⁵ , ¹²⁶ , ¹²⁷ , ¹²⁸ A distinct feature of the rat mandible is absence of a bony symphysis between hemimandibles that reduces load transfer and allows asynchronous motion between them. ²⁴³ , ²⁴⁴ , ²⁴⁵ Asymmetric masticatory function and dominant and nondominant hemimandibles are described in rats. ²⁴⁶ As such random assignment of treatment among hemimandibles or bilateral defects may be more important in rats than other models. Segmental defects in large animal models are frequently reported in the mandible, ¹²⁹ , ¹³⁰ , ²⁴⁷ , ²⁴⁸ , ²⁴⁹ , ²⁵⁰ , ²⁵¹ hard palate, ²⁵² , ²⁵³ , ²⁵⁴ and zygoma. ¹³¹ Unicortical and bicortical defects consisting of partial and full thickness bone resections are reported throughout the mandible in multiple large animal species including canine, ovine and porcine. Unicortical, partial bone thickness alveolar bone “saddle back” defects are frequently used to test bone regeneration in the unique bone‐tooth interface. ²⁴⁹ , ²⁵⁵ Stabilization varies with defect size and configuration as does time required for bone healing with 12 weeks typical for canine and ovine and porcine requiring slightly longer. ¹²⁹ , ¹³⁰ , ¹³¹ , ²⁴⁷ , ²⁴⁸ , ²⁴⁹ , ²⁵⁰ , ²⁵¹ As in long bones, the best models are in those bones with proportionately similar loading and comparable anatomy to human. In terms of bone volume, trabecular thickness, and trabecular spacing, ovine and porcine mandible are among the closest to human. ¹³² Additionally, the porcine temporomandibular micro‐ and macrostructure resembles that of the human, and the joints in both species experience similar masticatory forces. ¹³³ This makes the porcine mandibular condylectomy model well aligned with condyle and ramus regeneration studies. ¹³⁰

4. FUTURE DIRECTIONS

Numerous novel interventions tested in animal models are the foundation on which current standard bone regeneration therapies are based. The future is bright as humanized animal models make it possible to more closely align outcomes between human and nonhuman species. Further advances may include larger species with bone size, shape and stresses that are similar to human. Continued efforts to identify shared conditions that occur naturally in animals and humans may increase parallel clinical trials, especially for age‐related tissue changes. Three‐dimensional printing with organic and inorganic materials has limitless possibilities for treatment customization, not only for optimal bone size and shape, but composition and therapeutics. Cellular therapies will be enhanced by mechanisms to control cell migration and measure cell longevity in vivo. These are among the innumerable other ambitious goals that are the basis for discovery efforts that will change the future of health care options.

5. CONCLUSIONS

The information above provides a limited glimpse of the burgeoning scientific efforts focused on bone restoration. It also highlights that shared goals to improve treatment options benefit all members of our global community. The importance of carefully selected animal models contributes to advances in de novo bone formation daily. This drives development and translation of targeted therapies that improve the health and well‐being of humans and animals alike. The concept of one health has gained renewed attention recently. In a nutshell, the message supports the benefits of sharing discoveries to address medical challenges afflicting numerous species among medical disciplines that attend to them. Naturally, it is vital to both recognize and respect inherent differences among species that limit direct translation of scientific knowledge. Nonetheless, the rapid progress of ex vivo and in vivo de novo bone generation is clearing propelling a wealth of revolutionary innovation to reality among scientific and clinical specialists.

AUTHOR CONTRIBUTIONS

Mandi J. Lopez conceived the original idea. Takashi Taguchi and Mandi J. Lopez performed the literature search and wrote the manuscript. All authors read and approved the final submitted manuscript.

Taguchi T, Lopez MJ. An overview of de novo bone generation in animal models. J Orthop Res. 2021;39:7–21. 10.1002/jor.24852