Optimal Internal Fixation of Anatomically Shaped Synthetic Bone Grafts for Massive Segmental Defects of Long Bones

Abstract

Background

Large segmental bone defects following tumor resection, high-energy civilian trauma, and military blast injuries do not heal well and present significant clinical challenges. Tissue engineering strategies which use scaffolds are being considered as a treatment of this clinical problem but there has been little research into optimal fixation of such scaffolds.

Methods

Twelve fresh-frozen paired cadaveric legs were utilized to simulate a critical sized intercalary defect in the tibia. Poly-ε-caprolactone and hydroxyapatite composite scaffolds 5 cm in length with a geometry representative of the mid-diaphysis of an adult human tibia were fabricated, inserted into a tibial mid-diaphyseal intercalary defect and fixed with a 14-hole large fragment plate. Optimal screw fixation was tested by comparing non-locking and locking screws. Specimens were tested in axial, bending, and torsion in a non-destructive manner. A cyclic torsional test to failure under torque control was then performed.

Findings

Biomechanical testing showed no significant difference for bending or axial stiffness with non-locking vs locking fixation. Torsional stiffness was significantly higher (p=0.002) with the scaffold present for both non-locking and locking compared to the scaffold absent. In testing to failure angular rotation was greater for the non-locking compared to locking constructs at each torque level up to 40 N-m (P < 0.05). The locking constructs survived a significantly higher number of loading cycles before reaching clinical failure at 30 degrees of angular rotation (P < 0.02).

Interpretation

The presence of the scaffold increased the torsional stiffness of the construct. Locking fixation resulted in a stronger construct with increased cycles to failure compared to non-locking fixation.

1.0 Introduction

Massive skeletal defects of the tibia following high-energy civilian and military trauma or extensive resection of tumors remain a therapeutic challenge. Large bony defects occur in a small minority (0.4%) of fractures overall, but are common (11.4%) in open fractures.¹ In the tibia defects greater than 2 cm are unlikely to heal without additional bone grafting and defects greater than 5 to 6 cm are difficult to manage with autologous bone graft alone.[1-3] A number of surgical techniques exist to address these larger defects but require multiple procedures and significant morbidity with a high rate of complications.

Segmental allografts have been advocated to reconstruct defects after tumor resection and trauma but have a non-union rate as high as 17.3% and fracture rate of 17.7%.[4, 5] Distraction osteogenesis has been shown to be effective but also requires multiple procedures, an extended period of protected weight bearing with a bulky external fixator, and carries a re-fracture rate of 5% and amputation rate of 2.9%.[6] It also loses effectiveness for defects larger than 6 to 8 cm.[6, 7] Vascularized free fibula grafting is a viable option as well but requires specialized surgical skill as well as significant donor site morbidity.[8, 9] More recently a staged procedure with placement of a poly(methyl methacrylate) spacer to induce an osteogenic membrane followed by autogenous bone grafting has been advocated and shown to be effective for defects up to 20 cm.[2, 10, 11]

While effective, these techniques all require multiple procedures and carry significant donor site morbidity for bone grafting.[9] Tissue engineering has emerged as a promising approach for addressing large bony defects. The triad of tissue engineering has been proposed as a scaffold to provide structural stability as well as a platform for delivery of osteogenic growth factors and/or cells.[12] Scaffolds are now able to be designed with the correct micro and macro architecture for bony in growth and vascularization with promising results in animal studies.[13-17] The load-bearing capacity of the bone scaffold-fixation construct in human patients is an important factor in order to guide post-operative rehabilitation. However, most of previous studies focused on bone regeneration rather than the weight-bearing capacity of the bone scaffold-fixation construct.

Plate and screw fixation is commonly used for forearm bone fractures. For femur and tibia, intramedullary nailing is a common choice of internal fixation. For the humerus, both plate and screws and intramedullary nailing are used. However, for skeletal defects following resection of malignant tumors, intramedullary nailing is not conceptually preferred due to a theoretical concern of spreading tumors into the proximal and distal host bone and soft tissues during nail introduction and reaming.

In the present study, we examined the load-bearing capacity and optimal internal fixation of a bone/poly-ε-caprolactone-hydroxyapatite (PCL-HA) composite scaffold/plate construct that was anatomically fabricated in order to match the 5 cm segmental defect in the human tibia. We compared the non-locking versus the locking plate and screw constructs because one of the future indications for biocompatible and biomechanically competent scaffolds will be massive defects following tumor surgeries. Numerous biomechanical studies have demonstrated the mechanical superiority of locking fixation in a fracture model but none have focused on fixation of the bone scaffolds.[18-20] We hypothesized that the locking plate fixation provides more stable internal fixation under cyclic axial, bending, and torsional loading. The study is clinically relevant in that it will guide treating surgeons with respect to internal fixation and guide post-operative regimens for the patients with massive intercalary skeletal defects treated with bone scaffolds.

2.0 Materials and Methods

2.1 Fabrication of Intercalary Tibia Scaffolds

The anatomical morphology of a native cadaver adult human tibia was acquired from computed tomography scans on a clinical machine (SIMENS/Biograph 40, Malvern PA, USA) and manipulated using computer-aided design software for 3D modeling (Mimics, Materialise Co., USA). A 5 cm mid-shaft anatomically correct model of the tibia was designed. A composite polymer scaffold was fabricated using layer-by-layer deposition with a 3D printing system (Bioplotter, EnvisionTec, Berlin, Germany). The composite consisted of 90 wt% poly-ε-carpolactone (PCL) and 10 wt% hydroxyapatite (HA) (Sigma, St. Louis, MO, USA). PCL-HA composite was molten in the chamber at 120°C and dispensed through an 18 Ga needle to create interlaid strands and interconnected micro-channels (diameter 400 μm). Gaps between strands were 0.5mm and the porosity was 35%. Each layer was dispensed with a 0.9 mm height. We selected this PCL-HA composite in accordance with previous findings of cell adhesion and osteochondral histogenesis using the same material.[14]

2.2 Implantation of Structural Scaffolds into the Intercalary Defect of the Tibia

Twelve fresh-frozen cadaveric limbs from six individuals (6 pairs, average age 75, range 72 – 78, 4 males, 2 females) were utilized in this study. An antero-lateral incision was made and the lateral aspect of the tibia was exposed sharply. The surrounding soft tissue and fibula were maintained and a 5 cm section of tibia removed from the diaphysis. A 14-hole 4.5 mm broad, 260 mm large fragment locking compression plate from Synthes (Paoli, PA) was placed provisionally with bone clamps. The scaffold was provisionally placed with suture augment around the plate. The plate was fixed with five 5.0 mm diameter locking screws superior and inferior to the scaffold. Two 4.5 mm diameter cortical screws were placed into the scaffold to secure it. All screws bridged both cortices. The plate was contoured slightly at the distal end to conform to the distal tibia surface. All soft tissue was removed for biomechanical testing.

2.3 Biomechanical Testing

The tibial plateau was cut to create a square end that was placed within a 10.2 cm square fiberglass tube. Orthogonal Steadman pins were passed through the tube and bone and the entire construct filled with poly(methyl methacrylate). A metal plate with a spherical depression and a steel ball was placed on top of the fiberglass tube when axial compression was applied and the tibia was loaded via a flat platen. When torsion loading was applied, the plate and ball were removed and square clamp was lowered around the fiberglass tube that allowed for axial motion while preventing any slippage in rotation. The distal end of the tibia was potted in the same manner as the proximal end except a 5.1 cm square polycarbonate tube was used and placed within a square clamp rigidly connected to the biaxial load cell. Torsional loading was applied at the proximal end of the tibia. (Fig. 1)

Figure 1.

Mechanical testing of human cadaveric tibia with scaffold in segmental defect, (a) axial loading via ball bearing, (b) torsional loading via square clamp allowing axial displacement, (c) three-point bending of tibia-plate-scaffold construct, bending load is applied to the middle of the locking plate.

Loading was applied with a MTS 858 Bionix (MTS Eden Prairie, MN, USA) testing system. Axial loading was applied via a series of increasing sinusoidal loads from a baseline of 50 N to a peak load of 400 N, 500 N, 600 N and 700 N, respectively. At each step, the load was applied in a sinusoidal fashion for 50 cycles at a rate of 2 Hz. Data were collected at a rate of 10 Hz. Three-point bending loads were applied to the middle of the plate also via a series of increasing sinusoidal loads from a baseline50 N to a peak load 200 N, 250 N, 300 N and 350 N in the same manner as was done for the axial loading. Bending supports were placed 27.9 cm apart for all specimens. Mid-span displacements were recorded with a MTS Model 632.03 extensometer. Finally, a sequence of increasing sinusoidal torsional moments of +/-3, +/-6 and +/-9 N-m was applied for 20 sinusoidal cycles at a rate of 0.5 Hz. This testing protocol was adapted from well-established, clinically relevant, previously validated testing models, [21-24] The loading sequence was randomized for each pair, but was the same for each leg of a pair. Following sub-failure testing with the scaffold in place, the scaffold was removed and the same series of tests repeated without the scaffold. The order of scaffold / no scaffold was not randomized because the main intent of the study was to determine the stiffness of the construct with the scaffold in place. Specimens were then loaded to failure in torsion with the scaffold absent. A sequence of increasing sinusoidal torsional moments of +/-10, +/-15, +/-20, +/-25, +/-30, +/-35, +/-40, and +/-45 N-m were applied at 0.5 Hz for 25 cycles at each level of torsional loading and allowed to continue indefinitely at the +/-45 N-m level until failure.

2.4 Statistical Analysis

Separate statistical analyses were performed for the non-destructive and failure testing. Two-way repeated measures ANOVAs (SAS, Cary, NC) were performed for non-destructive testing. Factors were fixation type (locking, non-locking screws) and scaffold condition (scaffold present, scaffold absent). Axial stiffness, bending stiffness and torsional stiffness were the measured parameters. Two-way repeated measures were also performed for failure testing with the factors of fixation type (locking, non-locking screws) and torsional load (+/-10, +/-15, +/-20, +/-25, +/-30, +/-35, +/-40, and +/-45 N-m). Student-Newman Keuls multiple comparisons tests were used to discern differences between loading levels. Statistical significance was taken as P < 0.05.

3.0 Results

3.1 Non-Destructive Testing

There was no significant difference in axial stiffness, bending stiffness for locking vs. nonlocking constructs both with and without the scaffold (Figs 2 and 3). For torsional stiffness there was no difference for locking vs. non-locking constructs but constructs with the scaffold present were significantly stiffer (P < 0.002, Fig 3).

Figure 2.

Axial, bending, and torsional stiffness for locking vs non-locking constructs. Error bars denote 1 standard deviation.

Figure 3.

Axial, bending, and torsional stiffness for with and without scaffold. Error bars denote 1 standard deviation.

3.2 Failure Testing

Specimens tended to gradually loosen as indicated by increasing angular rotation at each increasing torsional moment. The non-locking constructs demonstrated increased angular displacement at each step torque (P < 0.05, Fig 4) compared to the locking constructs. It was appreciated that by 30 degrees of angular displacement specimens were grossly loose though still intact. We defined this amount of angular displacement as “clinical failure”, i.e. the point at which constructs had irreversibly started loosening. Comparing the number of cycles to clinical failure for 30 degrees of angular displacement, for all locking and non-locking constructs showed that the locking constructs lasted an average of 203 cycles vs 66 cycles for the non-locking constructs (P < 0.02, Fig 5). At physical failure, the non-locking constructs tended to fail by screw loosening and gradual leveraging open of the bone whereas the locking constructs tended to fail with little widening of the crack along the screw line until the bone suddenly failed catastrophically (Fig 6).

Figure 4.

Maximum angular displacements as a function of torque for non-locking and locking constructs. Error bars denote 1 standard deviation.

Figure 5.

Cycles to clinical failure for non-locking and locking constructs. Error bars denote 1 standard deviation.

Figure 6.

Depiction of locking screws and non-locking screws showing mechanism of loosening and crack formation. Note the screw loosening, toggling and back-out for non-locking screw as compared to the locking screw that results in progressive widening of the crack. Plots are of representative failure tests showing tibial rotation as a function of time and applied torque for a paired specimen. Note the larger rotation at higher applied torques for the non-locking screw.

4.0 Discussion

Massive segmental bony defects of the tibia remain very challenging to treat with a high percentage of patients experiencing complications of infection, non-union, fracture, and amputation. Tissue engineering strategies may provide improved implants for providing temporary stability as well as delivery of osteogenic growth factors and cells to promote bony healing.

A number of studies have looked at many different materials for bony scaffolds with a wide range of material properties.[25, 26] Some are very soft such as alginate gels or collagen sponges, intermediate such as PGA/PLA polymers and others are hard but brittle such as bioactive glasses and calcium triphosphate. PCL-HA demonstrates advantageous material properties, biocompatibility, and the ability to bind and deliver growth factors and cells as shown by previous work.[13, 14] Two types of internal fixation methods are commonly used for simple fractures. There are no well-established methods for reconstruction of segmental defects using tissue-engineered intercalary synthetic bone substitutes. Our study indicates that both conventional and locking constructs exhibit similar stabilization under cyclic loading. This study showed that in either a locking or non-locking construct the presence of a PCL-HA scaffold has the ability to provide extra rigidity in torsion. Studies looking at bony incorporation of metal implants in arthroplasty have shown that too much motion at the interface can lead to fibrous tissue formation where more rigid constructs result in bony ingrowth. It may be the same in a bone-scaffold interface situation in fracture repair as well.

Our failure tests demonstrate significantly greater cycles to clinical failure for locking plate fixation compared to non-locking. No matter what strategy is used, large segmental bony defects are very slow to heal putting tremendous stress on fixation constructs. Use of locking fixation with a scaffold proved to be a more robust construct with increased resistance to failure than non-locking in this biomechanical study. Current treatments used in the treatment of bony defects are now most commonly an induced membrane followed by bone grafting or bone transport with distraction osteogenesis. With the induced membrane technique, regardless of whether the tibia is fixed with an IM nail or external fixator, an extended period of non-weight bearing is required, as the bone graft has no real mechanical stability when first implanted. Distraction osteogenesis requires an extended use of bulky external fixators as well as requiring that the patient remain non-weight bearing. Use of a PCL-HA scaffold and locking plates produce a rigid construct, which may allow earlier mobilization and partial weight bearing while treating large bone defects. A locking plate also has the theoretical advantage of less disruption of the fracture healing biology because the plate is not tightly compressed against the bone as with a non-locking construct.

There are several limitations to this study. This is an in-vitro lab biomechanics study and therefore only able to assess the stability of the construct at time zero. In vivo, the biomechanics of healing fractures and fixation constructs change dynamically as the bone heals, forms callous, and remodels. This model also represents a best-case scenario where the bone defect was placed in the middle of the diaphysis with clean cuts. It was also assumed when performing the statistical analyses that the left and right legs of an individual represented a repeated measures condition, i.e. it was assumed that there were no differences between left and right legs. The average age of 75 of the specimens is also much older than that of actual patients, though it is thought that this factor would not have a significant impact on the findings. Another limitation, as mentioned in the Methods Section, is that the order of scaffold / no scaffold was not randomized because the main intent was to determine the stiffness of the construct with the scaffold in place. However, doing this may have introduced a systematic bias in the results that favor the construct with the scaffold since screw loosening in the first series of tests may affect the results of the second series of tests. Significant work remains to elucidate what is the ideal scaffold to use for bone tissue engineering applications and which growth factors or cellular augmentation should be employed. This study only sought to examine what fixation strategy would be best, independent of those issues. Future studies are needed in large animal models testing the feasibility and in vivo stability of scaffold-based repair of large bony defects.

5.0 Conclusions

Tissue engineering with anatomically correct scaffolds presents an exciting new strategy for the treatment of critical size defects of the tibia. The presence of the scaffold was found to increase the torsional stiffness of the construct. Locking fixation resulted in a stronger construct with increased cycles to failure compared to non-locking fixation. It is envisioned that currently available locking plate-screw scaffold constructs may provide rigid enough fixation to allow early mobilization and weight bearing of patients with critical sized intercalary bone defects. Based on our findings, we recommend the use locking pate and screw constructs be considered for emerging tissue-engineering based skeletal reconstruction of segmental defects.

Highlights.

• The effects of scaffolds for segmental defects in tibia were studied.

• Locking and non-locking constructs were compared in cadaveric human tibiae.

• The presence of scaffolds increased torsional stiffness.

• Locking fixation increased the number of cycles to failure.