Introduction
Delayed or complete failure of fracture healing remains a problematic complication during fracture healing, with general incidences ranging between 5 and 10% [
1,
2] and up to date, the pathophysiologic mechanisms for delayed fracture healing are not completely elucidated [
3].
During the last decade(s), in vivo research in rodents has resulted in a wide range of different animal models for fracture healing and compromised healing resulting in delayed union and nonunion development [
4,
5]. These models have to be standardized and need to mimic the human clinical situation as close as possible. Previous studies have investigated closed induction [
6‐
8] of the fracture and open surgical procedures [
5,
9], with differences in osteotomy size (critical sized segmental defects vs non-critically sized), and different fixation techniques as bridging plates [
10-
12], intramedullary nails [
7,
13,
14] and external fixators [
15,
16].
Several factors have been shown to influence bony healing such as the biomechanical environment (interfragmentary instability), inadequate blood supply [
17] as well as the defect size [
18,
19]. Availability of knockout mice and senescence altered mice allows a broad spectrum of molecular biology-based investigations [
20] into developmental biological issues such as bone and cartilage formation [
9,
12,
17] in combination with these different healing models.
Another key player in adequate fracture healing is the periosteum and integrity of the periosteum must be retained to achieve a successful fracture healing [
21]. The periosteum consists of a thin, well-vascularized and innervated layer along the cortex of the bone and is primarily composed of osteogenic and fibroblastic cells [
22]. Especially during the soft callus formation, the periosteum has a major influence on fracture repair as the periosteal progenitor cells will differentiate into osteoblasts and, mainly, chondrocytes [
23,
24]. Consequently, in the present study, we hypothesize that periosteal cauterization would induce a significant and substantial delay in the bone healing process in mice. The aim of the current project is to describe and characterize the delayed healing process so that this developed novel model can be used for future biomechanical and molecular research to investigate the delayed bone healing process or its treatment.
Discussion
The aim of the current project was to develop an in vivo murine model for delayed union development, as an intermediary between normal fracture healing and the development of nonunions, for future possibilities in biochemical and molecular research to investigate enhanced and deficient bone healing processes. Results demonstrated that the fracture gap obtained after a standardized osteotomy reduced with semi-rigid internal plate-screw osteosynthesis and combined with periosteal injury prolonged the healing period for 7–14 days, with callus formation volumes after 42 days of fracture healing which were comparable with callus after 21–28 days in the control group. In contrast, in the control group without periosteal cauterization resorption of the fracture callus via remodelling processes was well advanced with restoration of the femur diameter and reconstruction of the medullary canal. Therefore, this model of delayed fracture healing provides an ideal intermediate between normal fracture healing and nonunion development, whereas larger sized osteotomies would result in critical segmental defects resulting in nonunion development without the ability to assess the enhanced healing capabilities of future bone-healing strategies.
Quantitative results from the µCT analysis showed that as a consequence of the periosteal injury, the typical healing response was inhibited with the amount of woven bone in and mostly around the osteotomy was significantly reduced. Bridging of the proximal and distal fragments with mineralized lamellar bone was delayed accordingly. Radiographic analysis showed similar patterns in fracture repair with a 1–2 week delay in the periosteal injury group. Immunohistochemical evaluation on formation, maturation, and hypertrophy of chondrocytes using Col II and Col X markers also demonstrated a shift in the fracture repair response as shown in Figs.
7 and
8. In the periosteal injury group, the normal healing cascade was delayed and prolonged with fibrous connective tissue and cartilage still present in the gap region, chondrocytes which just started to hypertrophy, limited presence of woven bone and no complete bridging of the cortices evident. As a result, postponed healing delayed functionality as bending stiffness increased over time for both the control group as the mice with periosteal injury. Stiffness at the end of the experimental period was significantly higher in control animals when compared with periosteal injured mice due to a larger callus supporting the osteotomy and a higher degree of bone mineralization. As other isoforms, i.e., collagen I are mainly found in mature bone, these were not investigated in the present study. Collagen III, which is found in scar tissue and connective tissue, next to the blood vessel walls, has been reported to regulate osteoblastogenesis [
27,
28]. However, the most pronounced delayed union and nonunions in our model are observed between day 28 and 42 whereas collagen III is mainly found between the 5th and 20th postoperative day and additionally does not significantly affect the callus volume in the early stages of fracture repair [
27], therefor making collagen III a less reliable marker in our current investigation.
The electro cauterization procedure performed in this study destroyed the integrity of the periosteum on the proximal and distal side of the osteotomy gap. Disruption of the periosteum leads to a markedly impaired blood supply [
22,
29-
31] and subsequent to a reduced release and proliferation of various cell types and to a reduced capacity to form bone and cartilage [
17,
32]. The critical role for the periosteum explains the obtained results in this study that in the periosteal injured group of mice during the first 2 weeks of fracture healing neither chondrocytes nor osteoblast-specific cells were migrating to the osteotomy gap and only fibrous tissue did develop.
Extensive reviews have been published on in vivo models of fracture healing and delayed union and nonunion development in rodents [
33‐
35]. A wide range of different models have been created to study biomechanical and biomolecular processes during fracture repair and compromised fracture healing.
Standardized closed fracture models have been developed which induce fractures by three of four-point bending [
8] or using a blunt guillotine combined with a dropping weight [
6]. In these models, the fracture will represent a more realistic situation as is seen clinically with a better containment of the fracture hematoma. As compared to our newly developed model, a disadvantage is, as this is not a model of compromised fracture healing, that a relatively low number of delayed unions/nonunions which will occur decreasing the usability for studying the biomolecular and biomechanical processes during delayed fracture repair. Also, since there is relatively thin soft tissue coverage of the tibia, its influence on fracture healing and possible interplays between different tissues is difficult to assess in this model [
34].
A range of different intramedullary fixation methods are presented in literature used in closed fracture models [
7] and in open [
5,
9] surgical procedures. Minimally invasive methods used are accompanied by a lack of rotational and axial stability and as a result have a high risk of dislocation [
7], making them not useful in standardized delayed union research. More adequate models using intramedullary pins are accompanied by locking nails [
8,
13] or compression screws [
14] making it possible to use segmental defects for studying compromised fracture healing. However, all intramedullary fixation techniques severely damage the medullary canal, making it impossible to study the different endosteal processes during healing of the bone [
34].
Until now, delayed union studies in mice and rats have been conducted using external fixators [
15,
16], intramedullary pins [
5,
9,
36] or no fixation at all [
17]. The use of unilateral or circular external fixation devices ensures minimal disturbances of the fracture/osteotomy location during healing but also in subsequent analysis. However, the relatively high weight of the fixators and possible excessive micromovement when using unilateral fixators will increase the unpredictability of the obtained results [
34,
35].
Plate-screw osteosynthesis with locking plates and screws [
11] as used in the current study is designed for minimal periosteal contact and can as such be used to investigate influence of periosteal modification on fracture healing and keeping the advantages of an intact medullary canal when compared with the intramedullary fixation methods. Reproducible results have been obtained in the current study and previously [
10,
12].
Although mice are not an exact model for human fracture healing, since rodents lack a Haversian system but use comparable resorption cavities for bone remodeling [
4,
37], a major advantage of murine models is the reduced time (and costs) necessary for experiments since the healing process under normal circumstances takes around 3 weeks until there is no detectable motion between the fracture parts [
33,
38]. In the current investigation, we had better controlled biomedical conditions as compared to other fixation techniques [
39‐
41], and advantages over models which use tibial fracture healing as the straight longitudinal axis of the femur makes standardized fracture stabilization and accuracy of biomechanical testing easier and more reproducible. Recently, titanium-covered PEEK (polyether ether ketone) was developed and used which mimic the titanium surface of human osteosynthesis materials [
10]. From an ethical point of view, every animal can then be monitored multiple times and during longer periods and without the need for euthanasia prior to collecting data, which is in compliance with the principles of reduction, replacement and refinement in lab animal experiments.
Mice also have a broad range of possibilities for usage of gene-targeted (knockout/knockin) animals, which enables molecular mechanistic studies on bone repair [
42] and different existing models are present e.g. in research aimed at osteoporosis based fracture healing in senescence accelerated mice [
43]. The periosteal injury model discussed in this current study has been used in the recently published study on the influence of nitric oxide (NO) deficiency on delayed bone healing and nonunion development [
12]. In short, in this study, knockout mice deficient for nitric oxide synthase (a key enzyme necessary for NO production) showed nonunion development when compared with normal wild type control animals, after a femur osteotomy combined with periosteal cauterization, as used in the current study. At the end of the experimental period after 42 days of fracture healing, the deficient animals showed no presence of callus formation and bone volumes which were between two- and fivefold lower when compared with mice in the control situation.
When interpreting the obtained results, some limitations need to be considered. In the periosteal injury group, some longer time points for the follow-up period would be needed to assess if the healing process continues and subsequently results in remodelling of the callus as is shown in the control group. With these extra time points, the final delay in healing could be assessed. Next to this, we only investigate one factor leading to the delay in fracture healing and control other confounding factors such as the biomechanical environment. In this model for delayed fracture healing this is a strength resulting in reproducible data; however, since bone healing in general is a multifactorial process, further research is needed into other influential factors. A final minor point of attention is the fair interobserver agreement which was reached in the radiographic analysis; however, this limited value underscores the micro-computed tomography results which show comparable and significant quantified results of bone and callus formation.
In conclusion, a moderate fracture gap produced by osteotomy and fixated by flexible plate-screw osteosynthesis in combination with additional periosteal injury induced by electro cauterization leads to a delayed union development in a murine in vivo model. The periosteal injury induced a delay of healing time of 1–2 weeks compared to control samples, visible as callus formation and gap bridging and the presence of collagen expression within the gap region. The observed delay is considered to be clinically relevant since normalized by averaged healing time in mice (4 weeks) [
41] and humans (16–20 weeks), it can be extrapolated that a delay of about 1–2 weeks in mice would correspond to delayed healing in humans by around 4–6 weeks. In the future, this mouse model with periosteal injury can be used to evaluate basic research questions regarding involvement of certain pathways or genes or to develop diagnostic tools and treatment options, in a model that provides a continuum between normal fracture healing and the development of nonunions.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.