Ten percent of all fractures require further surgical procedures because of impaired healing [1]. Coles [2] reported up to 17 percent of nonunions after treatment of closed tibial shaft fractures. Impairment of fracture healing is linked to demographic changes in society e.g. the growing proportion of elderly people and causes not only individual but also economic damage [3]. Hence, by means of in vitro and in vivo experiments, considerable efforts have been made in fracture research to develop therapeutic approaches. Before promising concepts can be used in humans, it is necessary to verify their efficacy and safety in a variety of animal models (for critical review and guidelines see [4]).

Models simulating impaired fracture healing in animals are not easy to conduct. Animal models of bony nonunion mainly utilize techniques with large segmental defects, thermic treatment of the defect region, instable fixation or combinations of these procedures [5–8].

Descriptions or comparative studies of the different methods used in fracture research are often limited in detail. However, understanding of the biological course of healing is essential for a therapeutic approach and the choice of an adequate animal model can be crucial for the experimental results. Furthermore, with respect to laboratory animal stress and avoiding recurrent failure, we believe that critical reports on an in vivo approach are essential.

The bone healing process is a special form of wound healing. According to Cruess and Simmons [9, 10] the regular course of fracture healing can be broken down into different phases: i the reactive phase, including fracture, haematoma and inflammation whereby the initial inflammation is regarded as an activator of fracture healing [11]; ii the reparative phase, characterized by callus formation and lamellar bone deposition, and iii the remodeling phase, creating the original bone contour. During this process two types of bone formation are possible: desmal with direct bone formation from mesenchymal progenitor cells and chondral with bone formation from cartilage intermediate.

Impairment of regular osseous healing can result in delayed union or nonunion (also called pseudarthroses). Classification of these cases in humans depends on whether nonconsolidation of bone fragments occurs within 4 or 6 months [12, 13]. Nonunions can be classified as hypertrophic and are termed as vital on the basis of radiological evidence of proliferative external callus formation on the fragment side; if callus formation does not occur, they are termed atrophic [14, 15]. Common reasons for the impaired healing are poor end-to-end contact of the bone fragments, excessive interfragmentary movement and poor blood supply due to insufficient vascularization at the site of the defect or the interaction of all factors [16, 17]. Infection is mentioned as another frequent cause of nonunion [18].

Rats are the most widely used animal species in bone healing research and osteotomy and closed fracture of long bones are frequently used model techniques [19]. Here we describe the course of delayed healing in an open rat tibial osteotomy model in comparison to a closed fracture model with regular healing in order to find a possible explanation for the impairment of healing.

In order to evaluate an applicable in vivo model for local therapeutical targeting, we investigated the impaired healing course of a rat tibial osteotomy with intramedullary stabilization compared to a previously described closed fracture model with regular healing.

Surgical procedure of rat tibiae osteotomy as performed in this study or applied in other studies [27, 28] generates a unique phenotype of bone defect according to the AO-classification for fractures (42A3). Hence, it provides consistent preconditions for further investigations. In contrast, the use of a fracture device could lead to inconsistent fracture types [22].

Radiological, biomechanical and histological screening during the observation showed a clear course of impaired bone healing. Even after 84 d the osteotomized tibiae were far from the stage of remodeling the original bone structure. At best, defect bridging occurred periosteally but not interfragmentary as seen in the fracture model [24]. Primary periosteal bridging is an integral part of the approximately 5 week bone healing process in rats and has been described as species-specific [29]. On the other hand, a contributory factor may be the use of the intramedullary stabilisation technique.

The development of delayed healing after tibial osteotomy in direct comparison to the closed fracture technique was previously documented in a rabbit model by Park et al. [30] using an external fixator for stabilization of the bone defect. These authors showed respective differences two weeks post bone defect, but these had disappeared at later observation time points. In a study with rats Kokubu et al. [31] developed a femoral model with 100% nonunion up to 56 d post surgery. However, as an effect of cauterization on each side of the fracture, the radiographical appearance of nonunions was atrophic.

Compared to our previously reported study in rats which underwent closed tibial fracture with the same intramedullary stabilization, clear differences in the biomechanical properties of the osteotomized tibiae could be observed in contrast to the fractured tibia. The osteotomized tibiae showed significant lower values for maximum torque and torsional stiffness after 42 and also 84 days as compared to the fractured tibiae. Even though sample preparation for biomechanical testing at an earlier time point (28 d) was not possible for all osteotomized bone samples, mainly due to the very instable callus, the results were in accordance with those in a study from Shefelbine et al. [28] demonstrating also very low biomechanical data 21 d after rat tibial osteotomy.

Although our results can only be interpreted cautiously because of the small sample size investigated, the comparative examination of osteotomized and fractured bones in the present study gave evidence, that the course of delayed healing in osteomized animals is determined already before day 28 after surgery.

Histological investigations comparing the two different techniques even in earlier healing phases, at 10 d post bone defect, also revealed differences: The histomorphometric analyses showed that the amounts of accumulated desmal formed reparative bone tissue and of differentiated cartilaginous tissue were retarded in the region of the osteotomy defect compared to the fracture.

Less deposition of trabecular bone could be found on the anterior tibial side for both groups. One compounding parameter may be the weakly developed soft tissue envelope on the anterior tibial side. Moreover, comparison of the regional accumulation of differentiated cartilage revealed that osteotomized animals accumulate lower amounts of differentiated cartilaginous cells on the anterior side where the surgical approach was located. We therefore conclude that the choice of surgical approach can influence the local formation of reparative cartilaginous tissue.

Furthermore, in the callus of osteotomized rat tibiae we found numerous amounts of inflammatory cells. Additionally the number of vessels in callus was significantly higher than in the callus of fractured tibiae, where the amount decreased between 5 and 10 d post intervention. Comparable results i.e. an initial relatively constant number of vessels in the callus of tibiae with induced impaired healing in contrast to a decreased number in the regular healing group were shown in osteotomized sheep [32]. We would interpret these observations in the present study as a prolonged and pronounced inflammatory reaction occurring in all osteotomized rats. In fact, microbiological screening revealed persistent subclinical infection in a subset of the osteotomized animals. This can hardly be avoided given the keeping conditions and activities of the laboratory animals. For a rabbit model, Melcher and colleagues [33] showed even higher infection rates. Therefore, adjuvant systemic antimicrobial prophylaxis should be considered in future animal studies. Thus, tibial osteotomy has a higher chance of infection than closed fracture, something that also occurs in surgical treated closed tibial fractures in humans [2].

What could be a reasonable explanation for the delayed bone healing in all osteotomized animals in this study? Radiologically, we could rule out poor end-to-end contact of osteotomy gap. Another frequently quoted cause could be a high degree of instability in the osteotomized bone, concerning the rotation movement [34]. But all animals were stabilized in the same manner as described for the healing fracture model and breed, weight and sex were analogous. Although this was not examined in any comparative measurements, differences in instability were not obvious in either group. However, variations within minimal interfragmentary movements cannot be ruled out. In fractured animals we observed the deposition of more defect-associated cartilage, pointing to a possible higher degree of instability capable of modulating the healing process [35, 36]. The comparative analysis of tibia and fibula in osteomized animals may support this hypothesis as more cartilaginous tissue formation was detected around the fractured and unstabilized fibula than around the stabilized tibial gap in histological bone slices from the same individual. However, the fibula was broken manually and the tibia by osteotomy. Both of these parameters can influence the outcome.

Poor vascularization would be an implausible reason for non-union. The calluses appeared hypertrophical radiologically and well vascularised histologically. Bone necroses at the fragment sides were not observed and severe injuries of the bone's nutrient vessels were not conspicuous during surgery.

Furthermore, proliferating mesenchymal stem cells from bone marrow are though to play a key role in bone repair [37, 38]. Primarily used for cooling purposes, open techniques require irrigation and it cannot be ruled out that progenitor cells were washed out. However, haematoma removal within the first hour was shown to be of no relevance for the healing progress [39].

The most plausible explanation for nonunions in this study may be an extended spatio-temporal inflammation and granulation phase, identifiable by dimensions, prolonged inflammatory cell composition and initial vascularization. Local infection as detected in a subset of animals may be a contributory factor but aseptic inflammation leading to nonunion [40] could be assumed to play a main role in our model.

As known from in vitro cell culture systems, environmental changes have the potential to influence the differentiation and functioning of cells. Regardless of the importance of inflammation for initializing the healing process, severe soft tissue trauma and the linked excessive release of inflammatory mediators can be discussed as factors to have a negative impact on bone healing [41].

Irrespective of the initial trigger, consideration should be given to whether persistent early inflammation could be a basic principle for nonunions.

The open tibial osteotomy technique combined with intramedullary stabilization in rats leads to pronounced impairment of bone healing. Prolonged local initial inflammation is suggested as a causative mechanism.

We conclude that the reported osteotomy model is not suitable to provide insight into normal bone healing. Nevertheless, it could be used as a delayed healing model. As known from previous studies, PDLLA coating of the titanium implant has no negative effect on bone healing (21,24). Thus, further investigations with therapeutic application of pharmaceutical agents (e.g. growth factors) via the PDLLA-coated implant should allow evaluation of the potency of locoregionally administered drugs for nonunion therapy.