Vascularized Periosteal Flaps Accelerate Osteointegration and Revascularization of Allografts in Rats

Abstract

Background

Surgical reconstruction of large bone defects with structural bone allografts can restore bone stock but is associated with complications such as nonunion, fracture, and infection. Vascularized reconstructive techniques may provide an alternative in the repair of critical bone defects; however, no studies specifically addressing the role of vascularized periosteal flaps in stimulating bone allograft revascularization and osseointegration have been reported.

Questions/purposes

(1) Does a vascularized periosteal flap increase the likelihood of union at the allograft-host junction in a critical-size defect femoral model in rats? (2) Does a vascularized periosteal flap promote revascularization of a critical-size defect structural bone allograft in a rat model? (3) What type of ossification occurs in connection with a vascularized periosteal flap?

Methods

Sixty-four rats were assigned to two equal groups. In both the control and experimental groups, a 5-cm critical size femoral defect was created in the left femur and then reconstructed with a cryopreserved structural bone allograft and intramedullary nail. In the experimental group, a vascularized periosteal flap from the medial femoral condyle, with a pedicle based on the descending genicular vessels, was associated with the allograft. The 32 rats of each group were divided into subgroups of 4-week (eight rats), 6-week (eight rats), and 10-week (16 rats) followup. At the end of their assigned followup periods, the animals were euthanized and their femurs were harvested for semiquantitative and quantitative analysis using micro-CT (all followup groups), quantitative biomechanical evaluation (eight rats from each 10-week followup group), qualitative confocal microscopic, backscattered electron microscopic, and histology analysis (4-week and 6-week groups and eight rats from each 10-week followup group). When making their analyses, all the examiners were blinded to the treatment groups from which the samples came.

Results

There was an improvement in allograft-host bone union in the 10-week experimental group (odds ratio [OR], 19.29 [3.63–184.50], p < 0.05). In contrast to control specimens, greater bone neoformation in the allograft segment was observed in the experimental group (OR [4-week] 63.3 [39.6–87.0], p < 0.05; OR [6-week] 43.4 [20.5–66.3], p < 0.05; OR [10-week] 62.9 [40.1–85.7], p < 0.05). In our biomechanical testing, control samples were not evaluable as a result of premature breakage during the embedding and assembly processes. Therefore, experimental samples were compared with untreated contralateral femurs. No difference in torsion resistance pattern was observed between both groups. Both backscattered electron microscopy and histology showed newly formed bone tissue and osteoclast lacunae, indicating a regulated process of bone regeneration of the initial allograft in evaluated samples from the experimental group. They also showed intramembranous ossification produced by the vascularized periosteal flap in evaluated samples from the experimental group, whereas samples from the control group showed an attempted endochondral ossification in the allograft-host bone junctions.

Conclusions

A vascularized periosteal flap promotes and accelerates allograft-host bone union and revascularization of cryopreserved structural bone allografts through intramembranous ossification in a preclinical rat model.

Clinical Relevance

If large-animal models substantiate the findings made here, this approach might be used in allograft reconstructions for critical defects using fibular or tibial periosteal flaps as previously described.

Introduction

Limb salvage has become the standard of treatment for patients with malignant bone tumors. Structural bone allografts are one of the most commonly used reconstructive options in large bone defects. As a result of its anatomic and biologic properties, over time there is slow ongrowth of the host bone over the allograft at the junctions [6]. However, allografts are associated with high risk of nonunion (12%-57%), fracture (7%-30%), or infection (5%-21%) [5, 18, 21]. These complications have been associated with the avascular nature of the allografts, which persists over time [6]. Techniques that combine structural bone allografts with vascularized grafts have recently been described. In this setting, investigators believe that the structural allograft provides mechanical support, whereas the vascularized graft supplies osteogenic properties, potentially accelerating union of the allograft-host junction [3]. Capanna et al. [2] described a technique that utilized a vascularized fibular graft associated with a structural bone allograft. More recently, Soldado et al. described vascularized tibial and fibular periosteal flaps in the treatment of recalcitrant bone nonunion and the prevention of allograft-host bone junction nonunion, reporting excellent results [29, 30]. Vascularized periosteal flaps are versatile and adaptable to different anatomic defects, and some authors have suggested that because they provide periosteal apposition, they offer more osteogenic surface at the allograft-host bone junction than vascularized fibular grafts, although this suggestion has not been demonstrated [28, 29].

The osteogenic properties of cells located in the cambium layer of the periosteum have been widely studied. Small clinical series have reported higher allograft-host bone union proportions [29, 30] and revascularization of osteonecrosis [27]. However, to our knowledge, despite intriguing results in clinical practice, no studies have been reported that might specifically explain the biologic mechanism of a periosteal flap when associated with an allograft. We sought to determine whether vascularized periosteal flaps stimulate union, revascularization, and remodeling of structural bone allografts in segmental critical-size defect reconstructions in a preclinical rat model and the biologic mechanism by which it is carried out.

We therefore asked: (1) Does a vascularized periosteal flap increase the likelihood of union at the allograft-host junction in a critical-size defect femoral model in rats? (2) Does a vascularized periosteal flap promote revascularization of a critical-size defect structural bone allograft in a rat model? (3) What type of ossification occurs in connection with a vascularized periosteal flap?

Materials and Methods

Our experimental study was approved (CEEA 11/14) in compliance with Spanish national, local, and EU legislation (Real Decreto 53/2013; Decret 214/97; 2010/63/EU). Experimental procedures were performed at the Vall d’Hebron Research Institute (Universitat Autonoma Barcelona, Barcelona, Spain).

Sixty-four adult male Sprague-Dawley rats (average weight 300-400 g) were obtained from Janvier Labs (Roubaix, France). All animals complied with the usual requirements and were fed, watered, and kept under standard conditions for 2 weeks before surgery.

The animals were randomly assigned to two groups of 32 rats each. In each of the rats in the control and experimental groups, we created a 5-mm critical-size segmental defect [27] in the left femur, which we then reconstructed using a cryopreserved structural femoral allograft from a rat of the same strain. We created the same segmental defect in each of the rats in the experimental group. We repaired these defects using the same allograft reconstruction as in the control group, but in the experimental group, we added a medial femoral condyle vascularized periosteal flap that covered the allograft [22]. Both groups were then subdivided into three subgroups each based on the followup they were scheduled to receive: eight rats in a 4-week followup group, eight rats in a 6-week followup group, and 16 in a 10-week followup group (Fig. 1).

Fig. 1.

Flowchart of the experimental study is shown. Sixty-four Sprague-Dawley rats were randomly assigned to two equal groups. In all rats, a critical-size femoral defect was created and then reconstructed with a bone structural allograft (control group) or a bone allograft with an associated vascularized periosteal flap (experimental group). Both groups were subdivided into 4-, 6-, and 10-week followup groups. Fluorochrome labeling and analyses performed on each group are detailed in the graph. In the 4-week followup control and experimental groups, oxytetracycline (yellow arrow) was administered at the time of surgery and again before euthanasia at 4 weeks postsurgery. In the 6- and 10-week followup groups, oxytetracycline (yellow arrow) was administered at the time of surgery and calcein green (green arrow) at 4 weeks postsurgery. In the 10-week followup groups, a third labeling was made at 6 weeks using xylenol orange (orange arrow).

After an overnight fast, the rats were anesthetized using an intraperitoneal injection of ketamine hydrochloride (75 mg/kg) and medetomidine (0.5 mg/kg). Before incision, we injected 20,000 IU/kg of penicillin G intramuscularly as antibiotic prophylaxis. The rat was placed in a lateral position and the left hindquarter was shaved, cleaned, and sterilized with povidone-iodine. A longitudinal incision was made lateral to the femur, and the intermuscular septum was divided to expose the femur. After removing and excising the periosteum, a 5-mm critical-size defect (measured using calipers) was designed and created in the middle third of the femoral shaft [24]. The removed bone segment was processed as an allograft for a different rat. Next, a fully threaded intramedullary 1.6-mm Kirschner wire (Stryker Inc, Kalamazoo, MI, USA) was drilled in a retrograde direction from the defect through the proximal femur until it appeared at the greater trochanter. After this step, a 5-mm cryopreserved structural bone allograft was placed in the defect. The Kirschner wire was then advanced (antegrade) through the bone graft, into the distal femur, and finally to the trochlear groove of the knee. Correct intramedullary Kirschner wire placement was checked with fluoroscopy. This reconstructive procedure was carried out on all members of both the control and experimental groups. The muscular layer was sutured (4-0 Novosyn®; Braun Surgical, Tuttlingen, Germany), and the wound was closed with interrupted nylon suture (4-0 Dafilon®; Braun Surgical). In the experimental group, an additional medial longitudinal incision was made over the femur. The procedure was performed with a x 10 magnification lens (Carl Zeiss OPMI 9-FC; Zeiss, Jena, Germany). The femoral vessels were identified and individualized. An incision in the fascia exposed the vastus medialis and biceps femoralis muscles, which were separated using blunt dissection to identify the descending genicular artery and vein, running along the femur and nourishing the periosteum at the medial condyle. The patella was laterally dislocated, exposing the trochlea and the medial femoral condyle. A 5 x 4-mm vascularized periosteal pedicle flap of the medial condyle, based on the genicular descending vessels, was then harvested [22]. To achieve greater coverage of the allograft, the flap was rotated and transferred deep to muscle from medial to lateral through the anterior side of the femur with the aid of suture threads. Finally, the flap was fixed with nylon sutures (8-0 Prolene®; Ethicon Inc, Somerville, NJ, USA). The muscular layer was sutured (4-0 Novosyn®; Braun Surgical), and the wound was closed with interrupted nylon sutures (4-0 Dafilon®; Braun Surgical). Meloxicam (1 mg/kg) was administered subcutaneously twice daily after the surgical procedure. At the end of each group’s designated followup period, the rats were surgically explored and euthanized, and the femurs were harvested for analysis.

Independent veterinarians (AR, MR) at the animal facility where the animals were kept clinically evaluated all animals daily during the first week and weekly thereafter until they were euthanized. No signs of pain, stress, or discomfort were reported.

Clinical Analysis

Rats were euthanized at 4, 6, or 10 weeks, depending on the group to which they had been assigned. After a rat was euthanized, we harvested the surgically treated femur and tested the discernable allograft-host bone junction displacement in both proximal and distal junctions when manual force was applied as a preliminary analysis to detect nonunion. Any evidence of local infection (wound inflammatory signs, fistulas, presence of purulent material) or other change was registered. This analysis was performed on all study animals.

There was a progressive increase in body weight in both control and experimental groups throughout the study. Survival rate was 100% in both groups. Two rats developed surgical wound dehiscence (in the 10-week control and 4-week experimental groups) and underwent reoperation for wound closure without any further complications.

Micro-CT Evaluation

To evaluate bone healing, we scanned all harvested rat femurs of both control and experimental groups with a desktop micro-CT imaging system (Quantum FX microCT Imaging System; PerkinElmer, Waltham, MA, USA). The femur was longitudinally scanned, including both allograft-host bone junctions, at 70-kV energy and 114-μA intensity, with 16-μm (2048 x 256) resolution. All images were analyzed by the same radiologist (MDA) who was blinded to the treatment groups from which the samples came. Each was analyzed for signs of allograft-host bone union and for evidence of allograft fracture or periimplant osteolysis. Mean radiographic scores were calculated for each group according to the modified scoring system proposed by Yasuda et al. [38] (Table 1). Quantitative evaluation of new bone formation was performed using A Medical Image Data Examine (AMIDE) software [17]. Cylindrical regions of interest of 5 mm were established in each sample including allograft and surrounding tissue. Newly formed bone volume (BV) and BV/total volume ratio were measured by manual segmentation followed by standardized thresholding at a gray scale corresponding to soft and bone tissue. Newly formed bone mineral density was measured using a phantom of known hydroxyapatite concentrations for gray scale conversion.

Table 1.

Micro-CT scoring system proposed by Yasuda et al. [38]

Biomechanical Testing

To evaluate the reconstruction durability, we performed biomechanical torsional testing on the fracture in eight rats from each 10-week followup group. Once the femurs were harvested, the head and distal ends each were embedded in bone cement (Cemento 3®; Surgival, Nijmegen, The Netherlands) and fixed in a metallic socket designed for mounting in the testing device (Bionix 358®; MTS Systems, Eden Prairie, MN, USA). A torsional load was applied to each specimen, subjecting it to dynamic mechanical analysis until fracture occurred; a constant load of 25 kN was used at a torsional speed of 1° per second. In this biomechanical testing, samples from the control group were not evaluable as a result of premature breakage during the embedding and assembly process. Experimental samples were instead compared with untreated contralateral femurs.

Confocal Microscopic Analysis

To assess whether new bone would be formed and undergo mineralization, we administered fluorescent dyes to the animals at set time points. If new bone formation occurred, these dyes would be incorporated and detectable by microscopy. Sequential administration of fluorescent dyes would allow us to appreciate both the direction and topographic localization of new bone formation. Confocal microscopic analysis was performed on all remaining rats of both control and experimental groups (those not used in the biomechanical testing, two sections per sample). We performed polychromatic fluorescence labeling. At the time of their initial surgeries, all of the animals received intramuscular injections of 25 mg/kg oxytetracycline dye (Engemicina®; MSD Animal Health, Merck Sharp & Dohme Corp, Inc, Kelinworth, NJ, USA). At 4 weeks postsurgery, those in the 4-week control and experimental groups received a second injection of oxytetracycline. Also at 4 weeks, rats in the 6-week and 10-week control and experimental groups received a second labeling injection using 20 mg/kg calcein green (Calcein®; C0875 Sigma; Sigma-Aldrich, Inc, St Louis, MO, USA). Finally, at 6 weeks postsurgery, rats in the 10-week groups received a third labeling using 90 mg/kg xylenol orange dye (Xylenol Orange tetrasodium salt®; 33825 Fluka; Sigma-Aldrich, Inc) (Fig. 1). At the end of each group’s followup period, the animals were euthanized; the harvested femurs underwent fixation in 10% buffered formalin and posterior sequential dehydration with ethanol. After being longitudinally cut, the femurs were set in methacrylate resin (Technovit 7200VLC®; Heraeus Kulzer, Hanau, Germany) and polished to a thickness of 50 μm for laser confocal microscopy (Leica TCS SP2; Leica Microsystems, Wetzlar, Germany). Wavelengths 365 nm to 490 nm (oxytetracycline), 436 nm to 495 nm (calcein green), and 375 nm to 570 nm (xylenol orange) were measured. Qualitative descriptive analysis was done to evaluate the new bone apposition pattern in each specimen. All observations were performed by the examiner (CM) who was blinded to the treatment groups from which the samples had come.

Backscattered Electron Microscopy Analysis

After confocal microscopic analysis, all samples were prepared for evaluation under backscattered electron microscopy (BS-SEM). Samples were polished and carbon-coated using sputtering equipment. The BS-SEM analysis makes it possible to clearly differentiate the structure of calcified cartilage from other calcified tissues (eg, chondroid tissue, woven bone, and lamellar bone). This allows differentiation of bone tissue ossification processes as well as determination of maturity, depending on the state of mineralization in the specimen [16, 19]. Newly formed bone was appreciated using a Focused Ion Beam Scanning Electron Microscope Neon40 with GEMINI column (Carl Zeiss, Jena, Germany). All samples were evaluated using a power of 15 kV and a working distance of 8 mm at x 75 magnification. Image stitching was performed on one sample of each group (SmartStitch; Carl Zeiss). Qualitative descriptive analyses of the ossification process in samples from all control and experimental groups were carried out. All observations were made by the same examiner (CM), who was unaware of the slides’ origins.

Histologic Analysis

After BS-SEM analysis, four representative samples from each 10-week followup group were selected. The samples were polished to remove the graphite carbon layer (Exakt 400CS; Exakt Technologies, Inc, Oklahoma City, OK, USA). Each sample was then affixed to a histologic plate and cut into 1-mm slices with a cold diamond bandsaw (Exakt 300; Exakt Technologies, Inc). Finally, the samples were repolished to 50 μm to 100 μm thickness and Trichrome-Masson-Goldner staining was performed. The prepared histologic sections were observed under a light microscope (Leica DMD108; Leica Microsystems, Wetzlar, Germany). Qualitative descriptive analyses were made, evaluating for the presence of acute or chronic inflammatory signs, osseous trabeculae through the allograft-host bone junction, and bone neoformation related to the vascularized periosteal flap. All observations were performed by the same examiner (CM), who was unaware of the slides’ origins.

Statistical Analysis

We calculated a sample of 64 rats, based on Yasuda et al. [38], as an analogous study. We also considered the minimum sample size required for pathologic anatomy and biomechanics because there are no clinical studies with statistically significant results that specify the revitalization of an allograft with periosteal flap and assumed a loss of 10% per group.

Statistical analysis was performed using a SPSS Statistics for Windows, Version 18.0 (SPSS Inc, Chicago, IL, USA). Data were expressed as a mean and SD if normally distributed or as a median and range. For all tests, p < 0.05 was considered significant. Results were analyzed with the Kruskal-Wallis variance analysis. Chi-square analysis was used with categorical variables, logistic regression analysis when comparing categorical variable with quantitative variable, and Student’s t-test to compare means.

Results

We found that allograft-host bone union occurred more frequently in the presence of a vascularized periosteal flap. There was a discernible proximal or distal allograft-host bone junction displacement in > 78% of all control specimens (100% [16 of 16] in the 4-week control group, 87.5% [14 of 16] in the 6-week control group, and 78.1% [25 of 32] in the 10-week control group), whereas there was discernible displacement in < 37.5% of the experimental groups (31.3% [five of 16] in the 4-week experimental group, 37.5% [six of 16] in the 6-week experimental group, and 18.8% [six of 32] in the 10-week experimental group). Differences were found in all followup groups (4-week followup groups: odds ratio [OR], 0 [0-0.13], p < 0.001; 6-week followup groups: OR, 0.04 [0.01-0.43], p = 0.008; 10-week followup groups: OR, 0.09 [0.02-0.33], p < 001). We detected septic pseudarthrosis in four cases: three in the control and one in the experimental group. More than 50% osseous bridging was observed in 69% (11 of 16) of the distal allograft-host bone junctions in the 10-week experimental group and in 56% (nine of 16) of proximal junctions within the same group. In the 10-week controls, bone union was observed in 13% (two of 16) of the proximal and 6% (one of 16) of the distal junctions (Fig. 2). When comparing control and experimental groups, we only found the distal junction union to be higher in the 10-week experimental group, acting as a protective factor (4-week followup groups: absolute risk reduction [ARR] -0.06 [-0.76 to 0.06], p = 0.083; 6-week followup groups: ARR -0.31 [-2.50 to 1.88], p = 0.53; 10-week followup groups: OR, 19.29 [3.63-184.50], p < 0.001). The BV of samples from all followup experimental groups (4-week experimental group: 156.78 ± 13.46 mm³; 6-week experimental group: 151.80 ± 26.11 mm³; 10-week experimental group: 177.20 ± 41.51 mm³) was higher than their respective control group (4-week experimental group: 93.35 ± 28.05 mm³; 6-week experimental group: 108.38 ± 15.37 mm³; 10-week experimental group: 114.30 ± 17.92 mm³; 4-week followup groups: OR, 63.3 [39.6–87.0], p < 0.001; 6-week followup groups: OR, 43.4 [20.5–66.3], p = 0.0012; 10-week followup groups: OR, 62.9 [40.1–85.7], p < 0.001) (Table 2). Newly formed tissue was more mineralized in the 4-week (3259.37 ± 209.95 mgHA/cm³ versus 2802.15 ± 216.14 mgHA/cm³) and 10-week followup groups (3812.12 ± 292.16 mgHA/cm³ versus 3524.95 ± 329.32 mgHA/cm³) when the periosteal vascularized flap was present without finding differences between 6-week control (3278.82 ± 304.47 mgHA/cm³) and experimental groups (3567.00 ± 217.22 mgHA/cm³; 4-week followup groups: OR, 1.02 [1.002–1.044], p < 0.001; 6-week followup groups: OR, 1.006 [0.99–1.01], p = 0.025; 10-week followup groups: OR, 1.003 [1.0004–1.006], p = 0.007) (Table 2). In our biomechanical testing, control samples were not evaluable as a result of premature breakage during the embedding and assembly process. Experimental samples were compared with untreated contralateral femurs. Biomechanical stability at week 10 in the experimental group was compared with the contralateral femur. Torsional resistance was higher in untreated femurs (559.78 ± 96.86 Nmm versus 374.87 ± 180.67 Nmm, p < 0.05). However, in both groups, the torsion test produced a similar pattern, ascending gradually to a peak corresponding with the moment of the fracture, then abruptly descending (Fig. 3). Confocal laser analysis in control group specimens showed unstructured bone apposition. In contrast, linear and structured new bone apposition patterns related to the vascularized periosteal flaps were observed in samples from the experimental group. In these samples, linear and organized synchronous perivascular bone apposition was also observed (Fig. 4). In the 10-week experimental group, we found the bone apposition rate to be higher within the first 4 weeks (2.32–6.21 μm/day) and decreasing between the fourth and sixth weeks (0.85–4 μm/day). BS-SEM analysis corroborated our observations using micro-CT. In the experimental group, primarily in the distal junctions of the 10-week followup specimens, calcified tissues were observed sealing the distal interface between fragments in contrast to what was observed in the control groups. In the control samples, we observed mineralized bone tissue corresponding to the bone allograft and immature mineralized tissue from proximal and distal ends of the host bone trying to incorporate the allograft. In the vascularized periosteal flap samples, the immature mineralized tissue also appeared on the surface of the allografts, related to the overlying vascularized periosteal flap (Fig. 5 A-D).

Fig. 2.

Micro-CT tomography representative images are shown from each group. Note that there is no new bone formation on the allograft surface (^*) in the control groups, which contrasts with the wide area of new bone formation (yellow arrow) in the experimental groups.

Table 2.

The table shows the mean (SD) of each variable evaluated in the micro-CT analysis according to the scoring system used^*

Fig. 3.

Biomechanical testing analysis is shown. The upper image is an untreated control femur with the graph of its torsional assay. The lower image is an example specimen from the experimental group. Both femurs exhibit similar fracture patterns: ascending gradually to a peak (red arrow) corresponding with the moment of the fracture and then abruptly descending.

Fig. 4 A-C.

Confocal microscopic analysis with images corresponding to the 10-week control (A) and experimental groups (B-C). Images obtained in the confocal laser analysis show areas labeled in blue (oxytetracycline), green (calcein green), and red (xylenol). (A) Unstructured bone apposition is observed in the sample of the control group. Note the linear and structured labeling associated with a vascularized periosteal flap (parallel fluorescence lines, yellow arrow), (B) which does not interfere with organized synchronous perivascular bone apposition (circular-osteonal bone labeling, white arrow).

Fig. 5 A-D.

Backscattered scanning electron microscopy images from the 10-week experimental (A-C) and control groups (D) are shown. (A) Complete union of the distal allograft-host bone interface is observed (red arrow), (B) whereas the proximal junction shows nonunion when endochondral ossification did not occur (white arrow). (C) Higher magnification of the allograft area showing allograft material (whiter, ^*) in the highly revascularized, enlarged cortical bone wall. (D) In contrast, a sample from the control group shows mature mineralized bone tissue corresponding to the bone allograft (^*) without revascularization, not remodeling signs, is observed in the control group sample. Immature mineralized tissue arising from host bone (★) trying to incorporate the allograft is observed.

Allograft Revascularization and Remodeling

We observed that bone allografts were revascularized and remodeled in the presence of vascularized periosteal flaps. In control group samples, both BS-SEM and histologic qualitative analysis showed a large amount of nonmineralized tissue around an unstructured allograft, fragmented and exhibiting no signs of incorporation or bone neoformation on its surface (Fig. 6 A-C). In contrast, in the experimental group samples, histologic analysis revealed newly formed bone tissue containing osteocytes and osteoblasts located within the margin of the bone trabeculae in a linear pattern. Osteoclast lacunae with osteoclasts were observed at the border with the nonmineralized tissue (Fig. 7 A-C). These findings indicate a regulated process of bone regeneration with osteoclastic resorption followed by osteoblastic apposition. Furthermore, the allograft had been incorporated with the host bone and a remodeling process had begun, consisting of allograft reabsorption and revascularization (Fig. 8 A-D). Because of this process, the initial allograft had adopted a similar structure and histologic characteristics as the host bone. BS-SEM analysis of experimental group specimens showed that the allograft had been revascularized and remodeled, making it indistinguishable from the host bone. In micro-CT analysis, rats studied showed more than 75% (six of eight) periimplant osteolysis in the distal segment and < 25% (two of eight) within the proximal segment in all followup groups of both control and experimental samples (Table 2) without finding differences between them (4-week followup groups: OR, 1 [0.02-4.93], p = 1.00; 6-week followup groups: OR, 0.77 [0.15-3.93], p = 0.72; 10-week followup groups: OR, 1 [0.32-3.10], p = 1.00). Allograft fracture was observed in 12.5% (one of eight) of the samples in the 4- and 6-week control groups and in 25% (four of 16) of the 10-week followup group. Among the experimental groups, the fracture proportion was 50% (four of eight) in the 4-week group, 25% (two of eight) in the 6-week group, and 12.5% (two of 16) in the 10-week group. No differences were found between groups (4-week followup groups: OR, 0.14 [0.003-2.42], p = 0.10; 6-week followup groups: OR, 0.43 [0.006-10.79], p = 0.52; 10-week followup groups: OR, 1.17 [0.07-18.32], p = 0.89).

Fig. 6 A-C.

Ten-week control femur histologic section with Trichrome-Masson-Goldner staining and x 4 and x 10 magnifications is shown. (A) Large amount of nonmineralized tissue around unstructured nonvascularized allograft (^*), fragmented, and exhibiting no signs of incorporation or revascularization. (B-C) A characteristic endochondral ossification structure is observed in one of the ends of the allograft (★) in an attempt of union to the host bone.

Fig. 7 A-F.

Ten-week experimental femur histologic section with Trichrome-Masson-Goldner staining at x 4, x 10, and x 20 magnifications is shown. (A-C) Newly mineralized chondroid bone (★) in contact with the metallic implant and continuous with the allograft segment (^*) has been formed. The allograft is being revascularized and remodeled, showing osteoclast lacunae with osteoclasts (yellow arrow), osteoblasts forming bone lining cells (blue arrow), and vascular structures (red arrow). (D-E) Complete union through intramembranous ossification is observed. Dense trabeculae formed by chondroid and woven bone (★) intermingled with vascular spaces with blood cells inside (F, red arrow) are visible in the interface area.

Fig. 8 A-D.

Ten-week experimental femur histologic section with Trichrome-Masson-Goldner staining at x 10 magnification is shown. Complete union of allograft-host bone junctions is observed (yellow arrow). (A-B) The interface repair tissue structure corresponds to an intramembranous ossification process: dense trabeculae constituted by chondroid bone surrounded by woven bone (★) and lamellar bone remodeling (^*) around the vascular structures (red arrow). Tissue corresponding to allograft has been revascularized and remodeled, showing cemental lines (white arrow) characteristic from a remodeling process around vascular structures with blood cells (purple arrow) inside (D, x 20). In the allograft images (C), linear structured periosteal fibers are observed (blue arrow).

Ossification Type

Intramembranous ossification was observed in areas in contact with the vascularized periosteal flap. In our BS-SEM qualitative analyses, allograft-host bone unions showed intramembranous ossification with chondroid tissue-based trabeculae (Fig. 5). In comparison, in the junctions not covered with a vascularized periosteal flap (control group and some proximal allograft-host bone junctions in the experimental group), we observed nonunions when endochondral ossification was unsuccessful (Fig. 6 A-C). The histologic study also confirmed complete unions with intramembranous ossification observed in distal junctions in the experimental group (Fig. 7 D-F), like in BS-SEM analysis. This type of ossification was present in all locations in contact with vascularized periosteal flap. No evidence of inflammatory signs such as presence of infiltrate cells in the samples was found.

Discussion

Structural bone allografting, despite its complications, is one of the most used options in the reconstruction of large segmental bone defects because of its mechanical properties and its potential for restoring bone stock [32]. The osteogenic properties of cells located in the cambium layer of the periosteum have been widely studied. Small clinical series have reported a higher incidence of allograft-host bone union [29, 30] and revascularization of necrotic bone with the use of vascularized periosteal flaps [27]. To our knowledge, however, no studies have been reported that specifically address the biologic mechanism of a periosteal flap and its effects when associated with an allograft. In our rat model, we have found that vascularized periosteal flaps increase allograft-host bone union, revascularization, and remodeling of the bone allograft and that this occurs through an intramembranous ossification.

Because it was performed in an animal model, the current study presents inherent limitations in the clinical translation of the information obtained. The critical-size defect [24] was reproduced and reconstructed in the same manner as would be used in treating a malignant bone tumor in a clinical scenario. However, the postoperative protocol was different from clinical practice: although a patient who has undergone a critical bone defect reconstruction may be restricted to protective nonweightbearing initially, our rats were allowed to bear weight immediately after surgery. A high rate of distal periimplant osteolysis was observed in all our study groups. We acknowledge that the fixation device chosen, although it has been the implant of choice in rat models [9], did not provide sufficient stability for our reconstructions.

The vascularized periosteal flap chosen, as described by Nau et al. [22], has a limited surface and pedicle length, so, although well covering the distal junction and middiaphyseal defect, the proximal allograft-host bone junction was not completely covered in all experimental samples, probably altering and reducing bone union at these junctions. However, this situation is similar to what we can find in clinical practice, where critical bone defects are barely fully covered by a vascularized periosteal flap.

Qualitative analysis was performed using confocal microscopic, BS-SEM, and histologic evaluations; thus, no numeric data were obtained, and no statistical analysis was performed in these studies. Furthermore, samples had to be embedded in plastic and fixed in metal for confocal microscopy and BS-SEM analysis before histologic examination under optical microscopy. This meant that sample thickness had to be greater than what would have been optimum for histologic examination; this resulted in lower histologic image resolution than could be expected if paraffin histology had been used. Finally, in biomechanical testing, control samples could not be included as a result of their premature breakage during the embedding and assembly process. Experimental samples therefore had to be compared with untreated contralateral femurs. This meant that an originally intended analysis–biomechanical comparison between control and experimental group specimens–could not be performed.

We found that host-allograft osseous union occurred more frequently in the presence of a vascularized periosteal flap. Some previous models had demonstrated low union rates when reconstructing critical bone defects with only bone allografts at 6, 8, and 12 weeks of followup [15, 38], in line with our present results. In micro-CT evaluation of our control samples, very few signs of bone union were observed in either proximal or distal bone junctions. Histologic and BS-SEM analyses corroborated these findings. Several authors have reported favorable results with the use of bone morphogenetic proteins, activin receptor-like kinase 2, prostaglandin E2 parathyroid hormone, and synthetic periosteum for stimulating osteogenic potential [12, 13, 15, 23, 31, 36, 38]. Bone neoformation can be increased by 50% when a vascularized periosteal flap is used with a biodegradable scaffold [22, 36]. The osteogenic properties of vascularized periosteal flaps have been demonstrated in the present study. Allograft-host bone union and bone neoformation at the allograft surface were higher in the experimental group. This capacity could potentially offer a great advantage over vascularized bone grafting, where such osteogenic potential only appears at the ends of the grafts. Some data have been reported regarding the role of vascularized periosteal flaps in the union of bony junctions, primarily citing clinical findings. Several authors have reported that when using a vascularized fibular graft associated with an additional vascularized periosteal flap extension at its ends, time to fibular graft-host bone union was reduced (compared with use of a vascularized fibular graft alone) [28, 34]. Soldado et al. [29, 30] published excellent results with vascularized periosteal flaps in the treatment of recalcitrant bone nonunion and its prevention in massive bone allografts, even achieving improvements in healing times over the Capanna technique. Similarly, flaps composed of cortex and periosteum from the medial femoral condyle, described by Doi and Hattori [4], are used in recalcitrant scaphoid pseudoarthrosis and other similar bone pathologies. At the 6-week followup examination, the degree of union of all proximal junctions in the experimental group was found to be lower than in control subjects. The flap we used has a limited surface and pedicle length, which meant that in some cases, the proximal bone junction was only partially covered. These observations were supported by histologic analysis, which showed some proximal nonunions in the experimental group, in which endochondral ossification was not achieved, as occurred in all control samples. Furthermore, linear periosteal fibers were observed at the distal bone junctions in the experimental group; they were disrupted in the proximal junctions in cases where union did not occur. When we correlated our BS-SEM images with those obtained in our confocal microscopy analysis, we corroborated the relationship between the periosteum and structured bone neoformation in the experimental group. We performed a single bone apposition rate measurement per sample, as a descriptive analysis, which was comparable to results reported previously [8, 26]. Although the bone apposition rate decreased, the osteogenic potential of the periosteum remained active during the followup period. Furthermore, such an approach did not inhibit organized synchronous perivascular bone apposition, which was not observed in the controls.

We found that bone allografts were revascularized and remodeled when there was an associated vascularized periosteal flap and that the biomechanical torsion resistance pattern of the allograft was not different from that of contralateral untreated femurs. Allograft revitalization is a process of revascularization, incorporation, and bone remodeling, converting the allograft to viable new bone with the biomechanical properties and healing capacity of host bone through a process known as creeping substitution [14]. Enneking and Campanacci [6] reported only 20% allograft incorporation in critical bone defects at 5-year followup; Wheeler and Enneking [37] observed a 50% loss of bone allograft strength after 10 years in vivo. Neovascularization took place predominantly at the surface and allograft-host bone junctions, whereas the remainder of the graft was avascular, increasing the risk for allograft fracture [6, 37]. In clinical scenarios, vascularized periosteal flaps have shown promise in the pediatric setting for the treatment of osteonecrosis [4, 25, 27]. However, because no preclinical studies have been performed, the biologic changes produced in a bone allograft in contact with a vascularized periosteal flap have not been evaluated, because it would be necessary to harvest the bone for confocal, BS-SEM, histology, or biomechanical analysis, which cannot be performed in clinical practice. In the present study, these clinical findings have been reproduced. Evidently, the allograft adopts the same structure and characteristics as the host bone, making the transition between the two indistinguishable. These findings, which were not observed in the controls, clearly favor the regenerative capacity of the vascularized periosteal flap. It is also interesting to note that 4- and 6-week control subjects had a lower incidence of fracture than the experimental group. Some authors have suggested that there may be a relationship between early reabsorption and revascularization in bone autografts with their lower initial resistances, supporting the remodeling process [7, 10]. Combining a vascularized periosteal flap with a bone allograft may produce the same changes as seen in autografts; however, further studies are necessary to confirm this hypothesis.

BS-SEM and histologic analysis showed that bone neoformation corresponded to immature bone from periosteum in contact with an allograft, which was remodeling and incorporating it. Both revascularization and allograft-host bone junction healing occurred through intramembranous ossification. Study reports reveal controversy over the type of periosteal ossification taking place. The type of matrix synthesized by osteochondral cells is determined by oxygen density, which is largely conditioned by biomechanical circumstances: a low oxygen concentration produces an endochondral ossification, whereas a high concentration yields intramembranous ossification [1]. Tiyapatanaputi et al. [33] observed a structural bone autograft with an intact periosteum elicited intramembranous ossification; they found that when the periosteum was removed, the ossification was endochondral. With intramembranous ossification, the vascularized periosteal flap protected against nonunion and allograft-host bone junction torsional displacement in the 10-week followup group. In a clinical setting, when long bones are formed and healed through an endochondral ossification process in the formation of callus osseous, progress is slower and depends on good vascularization and stability [20]. Similarly, intercalary bony defect reconstructions are healed through endochondral ossification within the allograft-host bone junction [6, 11], whereas the more efficient distraction procedures show evidence of intramembranous ossification [16, 35]. Stimulation of intramembranous ossification through the association of a vascularized periosteal flap demonstrates the extraordinary osteogenic and osteointegrative potential of such flaps.

Bone allografts with vascularized periosteal flaps were superior in terms of histologic and radiologic results when compared with isolated bone allografts in a preclinical rat model. If large animal models substantiate the findings made here, this might be an approach that could be explored in humans for the treatment of segmental bone defects. We believe that multiple properties of vascularized periosteal flaps make them helpful complements to bone allografting in reconstruction of critical bone defects, where biologically unfavorable factors are usually present.