Cementless Titanium Mesh Fixation of Osteoporotic Burst Fractures of the Lumbar Spine Leads to Bony Healing: Results of an Experimental Sheep Model

Introduction. Current treatment strategies for osteoporotic vertebral compression fractures (VCFs) focus on cement-associated solutions. Complications associated with cement application are leakage, embolism, adjacent fractures, and compromise in bony healing. This study comprises a validated VCF model in osteoporotic sheep in order to (1) evaluate a new cementless fracture fixation technique using titanium mesh implants (TMIs) and (2) demonstrate the healing capabilities in osteoporotic VCFs. Methods. Twelve 5-year-old Merino sheep received ovariectomy, corticosteroid injections, and a calcium/phosphorus/vitamin D-deficient diet for osteoporosis induction. Standardized VCFs (type AO A3.1) were created, reduced, and fixed using intravertebral TMIs. Randomly additional autologous spongiosa grafting (G1) or no augmentation was performed (G2, n = 6 each). Two months postoperatively, macroscopic, micro-CT and biomechanical evaluation assessed bony consolidation. Results. Fracture reduction succeeded in all cases without intraoperative complications. Bony consolidation was proven for all cases with increased amounts of callus development for G2 (58.3%). Micro-CT revealed cage integration. Neither group showed improved results with biomechanical testing. Conclusions. Fracture reduction/fixation using TMIs without cement in osteoporotic sheep lumbar VCF resulted in bony fracture healing. Intravertebral application of autologous spongiosa showed no beneficial effects. The technique is now available for clinical use; thus, it offers an opportunity to abandon cement-associated complications.


Introduction
In osteoporosis rarefication and thinning of intrinsic bone structures leads to reduced bone quality with inferior biomechanical properties [1,2]. Vertebral compression fractures (VCFs) after low-velocity trauma currently occur with increasing numbers due to demographic age pattern changes [1,3]. VCFs predominantly affect the thoracolumbar levels, with their high level of degrees of motion making implant fixation strength of particular interest; this is because internal fixation devices are likely to break out of or displace in osteoporotic bone [4]. Therefore cement-based treatment procedures, such as kyphoplasty or vertebroplasty techniques used alone or in combination with internal fixation, 2 BioMed Research International play a major role in this age group regardless of the fracture severity [5][6][7][8][9][10][11]. Using kyphoplasty or vertebroplasty an instant pain relief is achieved by transpedicular polymethylmethacrylate cement (PMMA) application [8][9][10][11][12]. Disadvantageous of intravertebral applied PMMA cement is the change of vertebral elasticity parameters, which results in a change in biomechanical properties that increases the likelihood of adjacent fractures (reported incidence: 8-26%) [3,7,9,[13][14][15]. In addition, complications such as pulmonary embolism (up to 20%), cement leakage into the epidural space, or injury of adjunct neurovascular structures can occur due to exothermic reactions during the polymerization process (4-13%) [5,7,10,11,15]. Apart from those complications, fracture treatment using cement depends upon bone-cement compound-fixation while disabling bony healing capacity of the fracture.
Astonishingly, VCF treatment modalities promoting fracture healing currently have no value in clinical practice. This finding is contrary to the customary clinical practice that provides osteoporotic fracture healing capabilities for other sites such as the proximal femur or distal radius [16,17]. The question arises: what would be the ideal treatment modality to avoid the aforementioned problems? Theoretically, in order to promote bony healing, this could be a minimally invasive placed device (e.g., cage) that could provide intravertebral reduction while avoiding cement use. Recently developed intravertebral reduction systems enable this intravertebral reduction with reduced amounts of PMMA; however, they are not designated for cementless use [14,18,19]. Their use has proven reduction of cement-associated complications in initial clinical studies; anyhow their annihilation did not succeed [18,20].
In order to enable experimental improvements and new treatment techniques for VCF, a workable spinal fracture model for creation of VCFs in osteoporotic Merino sheep was developed in a prior study [21]. Weekly corticosteroid therapy and a calcium/phosphorus/vitamin D-deficient diet was administered for 5.5 months following ovariectomy and severe osteoporosis of the lumbar spine was demonstrated by pQCT and micro-CT analyses. Creation of a VCF type A3.1 succeeded via an ALIF (Anterior Lumbar Interbody Fusion) approach and standardized compression method. Using this fracture model, it was hypothesized that (1) lumbar VCFs type A3.1 in osteoporotic sheep can be adequately restored using titanium mesh cages without cement application; (2) osteoporotic VCFs have bony healing capabilities that can be proven via micro-CT analyses; and (3) additional application of autologous spongiosa graft accelerates fracture healing with superior biomechanical properties.

Animal Preparation, Surgical
Procedure, and Randomization. For osteoporosis induction twelve 5-year-old female Merino sheep (mean body weight: 66.3 ± 2.0 kg; range: 57-79) were ovariectomized bilaterally (OP1), housed in a shed protecting them from sunlight, and fed a calcium/ phosphorus/vitamin D-deficient diet and were injected weekly with dexamethasone (1.3 mg/kg body weight plus 0.028 mL/kg body weight dexamethasone-sodiumphosphate) for 5.5 ± 0.1 months (range: 5-6). Micro-CT and pQCT analyses documented significant lowering of BMD (bone mineral density) values for the second lumbar vertebrae for this model [21].
Animal preparation was performed under general anesthesia induced by 2.0 mg/kg body weight ketamine, 0.5 mg/kg body weight midazolam, and 0.010 mg/kg body weight fentanyl intravenously (IV) and maintained with 1.0-1.5% isoflurane and a FiO 2 of 0.4 after 24 hours of fasting. Preoperatively, 0.1 mg/kg body weight xylazine was injected intramuscularly (IM) for sedation. Intra-and postoperatively, extreme effort was made to minimize suffering via the administration of perioperative analgesia with 2.5 g metamizole IV for the second half of the procedure and 0.005-0.010 mg/kg body weight fentanyl IV if the duration of the procedure exceeded 1 hour. The postoperative analgesia regimen included a combination of 25-50 mg/kg body weight metamizole IV/IM q6 h (later administered orally) and 0.005-0.010 mg/kg body weight fentanyl IV on demand. An IV prophylactic antibiotic (cefuroxime) was administered intraoperatively and postoperatively; 1 g menbutone IM was administered to promote absorption.
A standard minilumbotomy of 4-5 cm and ALIF approach to the second lumbar vertebra was performed in the left lateral position. Creation of a type A3.1 VCF of the second lumbar vertebral body was achieved in a standardized manner in all cases and resulted in a significant decrease in body angle and vertebral height as previously reported [21]. Immediate fracture care (OP2) was performed by extrapedicular placement of one intervertebral mesh cage (OsseoFix®, Alphatec Spine, Carlsbad, California, USA) for fracture reduction and maintenance of reduction. The mesh cage (4.5 mm) was deployed according to the manufacturer's recommendations, using the specific implant insertion devices while referencing intraoperative fluoroscopic imaging in anterior-posterior and lateral views. The implant was deployed until the sagittal index (SI) and vertebral body kyphotic angle (KA) as measures of segmental kyphosis at the level of the spine segment were adequately reduced ( Figure 1). PMMA cement was not applied. Randomly, additional intravertebral autologous iliac crest spongiosa graft application (G2) or no additional augmentation (G1) was performed ( = 6 each). For G2, approximately 2 g autologous iliac crest spongiosa graft was acquired from a 1.5 cm incision on the right iliac crest and applied in the cage via the manufacturer's implant inserter cannula. After removal of instruments the surgical site was closed and immediate mobilization in an all-fours position was made.
The calcium/phosphorus/vitamin D-deficient diet was continued until euthanization was planned (2.2 ± 0.1 months; range: 2-2 after OP2) using IV propofol/T61 and the lumbar spine harvested immediately after death.

Macroscopic Analysis.
The former fracture line and amount of callus development was assessed after death. The circumference area 0.5 cm below the upper second lumbar vertebrae endplate was compared to the third lumbar vertebrae circumference via analogue localization.

High-Resolution Three-Dimensional (3D) Micro-CT and
pQCT Analysis. For quantitative and 3D micro-CT analysis a Skyscan 1076 micro-CT scanner and software package for reconstruction and analysis (CTAn, V1.1; NRecon, Version 1.6.6.0; Bruker Corporation, Billerica, MA, USA) was used according to manufacturer's recommendations (Method note: Phil Salmon, Bruker Inc.). The former fracture region of the second lumbar was analyzed for callus formation. A group of 7 healthy sheep and 4 osteoporotic sheep was used as control in order to compare nonosteoporotic to osteoporotic bone structural parameter. After 12 hours in 0.9% saline, each vertebra was wrapped in parafilm and placed in a polypropylene tube to prevent drying and associated motion artefacts during scanning. A foam tube allowed horizontal mounting in the sample chamber for CT imaging. With this approach, all samples were oriented with the longitudinal axis parallel to the optical axis of the instrument. To determine BMD specimens with known mineral density and adequate size (bone phantoms) were assessed under the same conditions.
Projection images were analyzed at 49 kV with 200 A Xray intensity using a 0.5 mm aluminum beam-flattening filter. Spatial resolution was 18 m and the rotation step was 0.4 ∘ . Image reconstruction was performed with beam hardening correction of 30%, defect pixel masking of 20%, ring artifact reduction of 6, and an individual alignment correction calculated by the software. The reconstructed volume was oriented along the long axis of the vertebra and the region of interest covering trabecular bone tissue was drawn using a standardized circular region of interest for L3 and manually drawn region of interest for L2; this was done to diminish artifacts caused by the implanted cage. From the upper vertebral endplate, the first slice containing only cancellous bone was selected and then 101 lower slices were selected. Regions with artifacts or cortical bone were excluded. Three-dimensional reconstruction of the series of images visualized the degree of osseous integration of the cages and the proportion of the segmented bone volume relative to the total volume of interest (BV/TV), the average number of trabeculae per unit length (Tb.N), the mean thickness of trabeculae (Tb.Th), the mean distance between the trabeculae (Tb.Sp), and the structure model index (SMI). In order to estimate the precision of the analysis, the CT scans were analyzed independently three times by the same investigator.
Osteoporosis induction and retention were evaluated by changes in BMD using peripheral quantitative computed tomography (pQCT) of the right distal radius according to World Health Organization (WHO) osteoporosis guidelines [22,23]. PQCT of the right distal radius was conducted for fracture creation/treatment as well as postmortem using a pQCT scanner (Stratec Medizintechnik, Pforzheim, Germany). An untreated group of sheep (7 slaughtered 5-yearold Merino sheep) served as control since in vivo pQCT-BMD analysis of the lumbar spine is not possible. In order to increase precision, three scans were performed and analyzed independently by the same investigator.

Biomechanical Analysis.
A servohydraulic material testing machine type 322.21 (MTS, Eden Prairie, Minnesota, USA, 100 kN, MultiPurpose Testware, v 5.6A4585) was used in compression testing of lumbar vertebra L2 and 3 of each cadaver. Average failure load ( max ) and stiffness ( wk ) were calculated for each of the three groups for L2 and compared to the values of L3. Both the vertebrae's upper and lower endplates were embedded in a special two-component adhesive (Demotec 95; Nidderau, Germany) in order to ensure parallel application of force (Figure 2). In a displacementcontrolled mode at a rate of 2 mm/min axial compression forces were applied between two parallel plates until macroscopic failure of the vertebra occurred or 16 kN was reached (at a data rate of 5 Hz). The load displacement diagram was documented and failure load was manually defined at the first significant decrease of slope; vertebral stiffness was defined from the first approximately linear range and the slope of the corresponding regression line. For a better comparability, data were calculated to a structural modulus ( wk ) and limit stresses ( wk ) according to their cross section area ( wk ) and height (ℎ wk ) ( wk = wk / wk ⋅ ℎ wk ; wk = max / wk ).

Statistical Analysis.
Results are presented as mean ± SEM (Standard error of mean; range). After satisfying the assumption of normality (Kolmogorov-Smirnov test), the paired t-test analyses or the Mann-Whitney U Test (nonnormal distribution) was performed to analyze the differences in both groups. Significance was defined at < 0.05. Statistical analysis was performed using SPSS Statistics version 20.0 software (IBM, Armonk, New York, USA).  (Figure 3(a)). The degree of fracture healing was assessed by evaluation of the fracture line and amount of callus development. The former fracture line showed bony consolidation in all cases. Four vertebrae (33.3%) showed an increased callus development with more than 1.5 cm increase in vertebral circumference (G1 = 2; G2 = 2). This callus development resulted in a 58.3% increase in mean vertebral circumference difference for the group with additional intravertebral autologous spongiosa grafting (G2 1.1 ± 0.4 cm, range: 1-2) when compared to G1 (0.6 ± 0.3 cm, range: 0-1).
Biomechanical testing revealed similar vertebral failure load and vertebral stiffness when compared to unfractured

Discussion
Fracture reduction and fixation using titanium mesh implants in a validated sheep model of osteoporotic lumbar spine incomplete burst fractures (A3.1) result in bony fracture healing without cement application. As hypothesized, bony 6 BioMed Research International healing was proven with macroscopic, micro-CT, and biomechanical evaluation. Additional intravertebral application of autologous spongiosa leads to increased callus formation; however, without superior biomechanical properties. Reports of fracture models in animals, which are able to experimentally reproduce a specific fracture type that frequently occurs in humans, are rare; to date, only studies of bone defect models have been published [24,25]. In addition, studies of this nature have not been conducted on healthy vertebrae [26][27][28][29][30]. Sheep spines have been demonstrated to be biomechanically similar to human spines with similar loads; however, higher axial compression stresses concomitant with slightly higher bone densities are due to quadrupedal locomotion [31]. Multiple studies have demonstrated the sheep's usefulness for osteoporosis research and bone healing due to their genetical similarity to humans with their estrus cycle, hormone profiles, and Harversian bone remodeling [32][33][34][35][36]. A current literature review of Sheng et al. [37] found the sheep spine as the most useful experimental in vivo model of the lumbar spine when compared to human anatomical parameters involving cost and availability thoughts. Furthermore, the comparison of micro-CT structural changes of the vertebral cancellous bone in between our results and human data indicates that a similar osteoporosis degree to that in the clinical situation was induced in our sheep. Due to these similarities to the human bone remodeling and human spine, this study comprised a previously published validated osteoporotic VCF sheep model.
For VCF surgery, it is well known that osteoporosis due to decreased BMD affects the implant-bone interface, leading to inferior biomechanical properties [27]. Thus, cementassociated solutions such as vertebroplasty or kyphoplasty techniques are the gold standard for clinical practice; however, they are associated with the specific aforementioned complications [5-15, 19, 27]. In order to diminish these cement-related complications, this study compromised a specific titanium mesh implant designated for cementless application (currently, an off-label use) in 12 osteoporotic sheep with unstable burst VCFs. Vertebral body reduction succeeded in all cases. These reduction capabilities concur with a few previously published biomechanical studies of the titanium mesh implant: Upasani et al. [38] compared the implants to standard balloon kyphoplasty and reported a significantly greater height maintenance as well as decreased cement amounts. Ghofrani et al. [14] reported similar results in human spine cadavers that spared additional cement application. The first subsequent clinical trial revealed improvements in pain (Visual analogue scale 7.7 to 1.4), Oswestry-Disability Index (71% to 30%), and adequate reduction (Cobb angle 11.7 ∘ to 10.4 ∘ ) in a 12-month follow-up of 32 osteoporotic VCFs stabilized with the titanium mesh cage; however, PMMA cement was still used [20]. Admittedly, reduction properties in our study do not fully match the situation in humans since surgical resolution of osteoporotic spine fractures without neurologic impairments is rarely performed immediately after the injury. This denoted a study limitation because clinical experience indicates a more difficult fracture reduction with delayed surgery. Nevertheless, one first published clinical trial presented preliminary results in four patients with off-label cementless use of the titanium mesh cage and revealed similarly adequate fracture reduction (KA 14.5 ∘ to 10.7 ∘ ; Cobb 10.1 ∘ to 8.3 ∘ ) [18].
In addition to the adequate reduction capabilities of the implant, the cementless procedure demonstrated bony healing of the osteoporotic VCFs two months postoperatively. Although it is well documented that patients suffering from osteoporosis have an increased fracture risk, and prevention methods are well established, the properties of osteoporotic fracture healing have received little attention [17,39,40]. In contrast, current clinical practice with kyphoplasty/vertebroplasty techniques leading the way disables bony healing capacity via intravertebral cement application. Recent animal studies in rats and sheep confirm fracture healing capabilities under low BMD conditions, and interestingly both decreased [41][42][43] as well as increased callus formation [44] was observed. In addition, clinical trials in humans confirm healing capabilities in osteoporotic bone, although they reveal an abnormality of bone remodeling [16,17].

7
Osteoporosis fracture healing capabilities can be explained by the different pathways of bone remodeling resembling different aspects of embryological skeletal progresses [45].
In addition to the aforementioned fracture healing impairments, low BMD-associated problems such as implant loosening or breakage have been reported, which might be reduced by an enhancement of fracture repair. This was aimed at with additional intravertebral autologous iliac crest spongiosa graft application for one group. Anyhow, despite the distinct increased amount of callus, we observed no superior biomechanical properties. This concurs with other studies [46]; for example, Peter et al. [47] described an alendronate-induced increase in callus size in dogs that had undergone a radius fracture, without an accompanying improvement in flexural rigidity or failure-load. This might be due to the fact that the rapidly induced increase in callus size leads to an early gain of strength (strength, resp., stiffness is related to the third, resp., fourth power of the radius), while bone remodeling in osteoporosis is delayed; consequently bone formation rates and mineral apposition are reduced [48]. Thus, mechanical integrity of the callus in this early stage might be impaired and a time-dependent coefficient may be implied as other studies have revealed improved mechanical strength after 6 months in similar models [24]. Therefore, our data do not reveal any advantage to additional spongiosa graft application. Moreover, a lack of biomechanical data after the course of 2 months is a limitation of the study, coming with the study design.
Further study limitations in terms of bony healing include a certain inherent subjectivity bias that occurs with the macroscopic evaluation of the fracture gap. In clinical practice healing is defined as combined clinical and radiological findings, whereas radiological evaluation is difficult to assess [49]. Eventually, the biomechanically demonstrated stability proves bony healing.

Conclusions
Our study results indicate that fracture fixation using the titanium mesh implant provides a sound mechanical basis for fracture reduction and healing even in osteoporotic bone. Thus, cement-associated complications can be avoided. Prospective clinical trials are needed to determine its value for clinical use.