Strength of Thoracic Spine Under Simulated Direct Vertebral Rotation: A Biomechanical Study

Spine Deformity 4 (2016) 85e93 www.spine-deformity.org

Biomechanics

Strength of Thoracic Spine Under Simulated Direct Vertebral Rotation: A Biomechanical Study Sean L. Borkowski, PhDa,b, Sophia N. Sangiorgio, PhDb, Richard E. Bowen, MDc, Anthony A. Scaduto, MDc, Bo He, MDb, Kathryn L. Bauer, MDb,c, Edward Ebramzadeh, PhDb,* a Lucideon, 2210 Technology Dr, Schenectady, NY 12308, USA The J. Vernon Luck, Sr., M.D. Orthopaedic Research Center, Orthopaedic Institute for Children/UCLA, 403 West Adams Blvd, Los Angeles, CA 90007, USA c Orthopaedic Institute for Children and the Department of Orthopaedic Surgery, University of California, 403 West Adams Blvd, Los Angeles, CA 90007, USA Received 22 May 2015; revised 12 August 2015; accepted 13 September 2015

b

Abstract Background: Direct vertebral rotation (DVR) has gained increasing popularity for deformity correction surgery. Despite large moments applied intraoperatively during deformity correction and failure reports including screw plow, aortic abutment, and pedicle fracture, to our knowledge, the strength of thoracic spines has been unknown. Moreover, the rotational response of thoracic spines under such large torques has been unknown. Purpose: Simulate DVR surgical conditions to measure torsion to failure on thoracic spines and assess surgical forces. Study Design: Biomechanical simulation using cadaver spines. Methods: Fresh-frozen thoracic spines (n 5 11) were evaluated using radiographs, magnetic resonance imaging (MRI) and dual-energy x-ray absorptiometry. An apparatus simulating DVR was attached to pedicle screws at T7eT10 and transmitted torsion to the spine. T11eT12 were potted and rigidly attached to the frame. Strain gages measured the simulated surgical forces to rotate spines. Torsional load was increased incrementally till failure at T10eT11. Torsion to failure at T10eT11 and corresponding forces were obtained. Results: The T10eT11 moment at failure was 33.3  12.1 Nm (range 5 13.7e54.7 Nm). The mean applied force to produce failure was 151.7  33.1 N (range 5 109.6e202.7 N), at a distance of approximately 22 cm where surgeons would typically apply direct vertebral rotation forces. Mean right rotation at T10eT11 was 11.6 5.6 . The failure moment was significantly correlated with bone mineral density (Pearson coefficient 0.61, p 5 .047). Failure moment also positively correlated with radiographic degeneration grade (Spearman rho O 0.662, p ! .04) and MRI degeneration grade (Spearman rho 5 0.742, p 5 .01). Conclusion: The present study indicated that with the advantage of lever arms provided with DVR techniques, relatively small surgical forces, !200 N, can produce large moments that cause irreversible injury. Although further studies are required to establish the safety of surgical deformity correction surgeries, the present study provides a first step in the quantification of thoracic spine strength. Ó 2016 Scoliosis Research Society. Keywords: Thoracic spine; Biomechanics; Strength; Torsion; Direct vertebral rotation

Introduction The present work was supported by the Orthopaedic Institute for Children Pediatric Fund. This investigation was performed at the J. Vernon Luck, Sr., M.D. Orthopaedic Research Center, Orthopaedic Institute for Children/UCLA. Author disclosures: SLB (none), SNS (none), REB (none), AAS (none), BH (none), KLB (none), EE (none). *Corresponding author. The J. Vernon Luck, Sr., M.D. Orthopaedic Research Center, Orthopaedic Institute for Children and UCLA Department of Orthopaedic Surgery, 403 W Adams Blvd, Los Angeles, CA 90007, USA. Tel.: (213) 742-378; fax: (213) 742-1365. E-mail address: [email protected] (E. Ebramzadeh). 2212-134X/$ - see front matter Ó 2016 Scoliosis Research Society. http://dx.doi.org/10.1016/j.jspd.2015.09.044

The increase in popularity of pedicle screwebased instrumentation systems for deformity correction has led to developments in correction maneuvers employed during surgery. With the high boneescrew interface strength provided by pedicle screws [1,2], substantial forces and torques can be applied to the spine intraoperatively to achieve necessary sagittal plane and rotational deformity corrections. For example, using direct vertebral rotation (DVR) [3,4], axial torsion is applied directly to the spine through a device with large lever arms attached bilaterally to pedicle

86

S.L. Borkowski et al. / Spine Deformity 4 (2016) 85e93

screws at multiple levels along the length of the deformity. With such techniques, torques purportedly in excess of 100 Nm have been applied to the thoracic spine intraoperatively [3], and satisfactory curve corrections have been reported. Notwithstanding studies that have reported satisfactory curvature corrections and low complication rates using DVR [4-8], others have reported intraoperative complications with DVR and other rotational correction maneuvers [9-12]. Wagner et al. [9] reported seven cases (2.6%) of lateral pedicle screw plow following DVR. This is a potentially debilitating complication, particularly in patients with right thoracic scoliosis whose aorta is positioned more laterally and posteriorly [13], where lateral screw plow could cause aortic abutment and injury. Unfortunately, the true frequency of vascular injury is unknown and likely underreported [14,15], especially since the majority of DVR reports in the literature have not obtained postoperative CT scans which are important in evaluating pedicle screw placement [6-8,16-27]. Although the large intraoperative torques afforded by these new maneuvers may indeed achieve curve correction effectively in most cases, they may approach or even cross the strength limits of the spine in some cases. Despite the potential risk involved with using DVR, the magnitude of torsional loads that can be safely applied to the thoracic spine has not been established. Boneescrew interface strengths under torsional loading in the thoracic spine have been reported at torques of less than 50 Nm [28,29]; however, unlike intraoperative conditions, for these measurements, the boneescrew interface was isolated, allowing no motion of the spine or transfer of load to the intervertebral discs. Previous biomechanical studies have not measured the torsional structural strength of the thoracic spine, that is, the strength of the intact thoracic spine under axial rotational torque. In the lumbar spine, torsional failures have been reported to be in the range of 50 Nm [30,31]. The thoracic spine, because of its smaller dimensions, is likely weaker. Facetectomies may reduce the torsional strength further still. Clearly, it is necessary to establish torsional strength of the spine as a first step toward reducing the risk of injury or failure during deformity correction surgery. The purpose of the present study was to measure and establish the axial torsional strength of the thoracic spine under simulated DVR loading using an in vitro cadaveric thoracic spine model.

Methods Specimen Preparation Eleven fresh-frozen human cadaveric thoracic spines (T1eT12) were obtained from Science Care (Science Care, Phoenix, AZ). Specimens were wrapped in saline-soaked gauze, and frozen at 20 until testing [32]. Before experimentation, the specimens were dissected to remove all skin, muscle, and fat tissue, while maintaining the

integrity of the discs, bony structures, and stabilizing ligaments. Additionally, the posterior 5 cm of the ribs were preserved, with the costovertebral joints intact. The dissection was performed according to methods established previously in the literature [33,34]. Each specimen’s T11 and T12 vertebrae were potted together in a low-temperature setting epoxy resin. Transverse screws were inserted partway into both T11 and T12 to provide additional fixation within the pot. The potted vertebrae were then placed in a custom-designed 15-cmdiameter aluminum ring for mounting onto the load frame. Two tri-planar laser levels (Stanley Crossline Level MaxCL2; Stanley Tools Product Group, CT) were used to align the pot within this ring such that, once mounted, the anatomical planes of the T10eT11 disc would align with the axes of the load frame [35]. Specimen Imaging and Health Evaluation The bone mineral density (BMD) of each thoracic spine specimen was determined by dual-energy x-ray absorptiometry (DEXA) using a Hologic 2000 bone densitometer (Hologic Inc., Waltham, MA). Although t- and z-scores could not be assessed for the thoracic spine, the raw BMD score was used as a measure of bone health. Localized BMD measurements were obtained for each vertebra along the length of the thoracic spine. Two other measures of bone health were included. First, standard high-resolution, 15% magnification anteriorposterior and lateral radiographs were taken using an HP Faxitron Series high-resolution radiography system (43805N, Hewlett Packard Company, Palo Alto, CA). The T10eT11 disc level of each specimen was then graded according to the grading systems established by Mimura et al. (1-4 grading system) [36] and Lane et al. (0-3 grading system) [37]. Finally, T1- and T2-weighted sagittal MRI images were taken before testing, and the T10eT11 disc levels were graded according to the degeneration grading system established by Pfirrmann et al. (I-V grading system) [38,39]. Instrumentation Bilateral facetectomies were performed from T7 to T10 to allow for pedicle screw insertion. Polyaxial pedicle screws (Medtronic, Memphis, TN) were then inserted by trained pediatric spine surgeons bilaterally from T7 to T10 using the free hand technique. Standard high-resolution, 15% magnification anterior-posterior and lateral radiographs were taken using an HP Faxitron Series radiography system (43805N, Hewlett Packard Company) to assess pedicle screw placement. Although the radiographs could not definitively rule out misplacement of the pedicle screws, they were used to ensure that the screws were placed parallel to the endplates and directed along the pedicle axes, and that the screw tips converged toward but did not cross the midline.

S.L. Borkowski et al. / Spine Deformity 4 (2016) 85e93

87

Surgical Release Before load-to-failure testing, a Ponte osteotomy was performed at T10eT11 to simulate an intraoperative surgical release for deformity correction. The Ponte osteotomy included a bilateral facetectomy, as well as resection of the inferior half of the T10 spinous process, the interspinous ligament, and the ligamentum flavum [40].

DVR Simulation Device and Strain Measurements A custom-made Vertebral Derotation Simulator (VDS) was fabricated to 1) to function as a typical surgical maneuvering device used intraoperatively for DVR [3] and 2) to mount on the upper actuator of an MTS load frame, allowing precise control and accurate measurements of the applied torsional moment and loads (Fig. 1). The VDS provided attachments that could be adjusted to mount on pedicle screws on different spines but could then be tightened to transmit the torque from the load frame (Fig. 2). First, bilateral polyaxial pedicle screws (5.540; Medtronic, Inc., Memphis, TN) were attached to each of the T7eT10 vertebrae. A short cylindrical spacer was placed in the rod slot of each polyaxial screw head to allow tightening and locking the head position. For each vertebra, the bilateral pedicle screw heads were connected to each other using a horizontal bar while also locking the polyaxial head (Fig. 1). Each bar was then attached to a single-level linkage rod. The VDS included four single-level linkage rods, one for each of the four vertebrae. Once attached, the four single-level linkage rods were then linked together, creating a rigid quadrangular structure (Fig. 1). One main linkage rod extended further posteriorly (Fig. 2). The load frame actuator connected to this main linkage rod to apply rotational forces to the spine via the VDS system. This connection was made through connective bars and a linear bearing which allowed free transverse (AP) sliding of the bar on the rod. The rotation was transmitted through the quadrangular-linked VDS, thereby sharing the applied load across four vertebral bodies, and eight pedicle boneescrew interfaces. The simulator was equipped with strain gages to measure the simulated surgical loads. One set of strain gages was mounted on the anterior end of the single main linkage rod, measuring the transverse bending on that rod, which was calibrated to obtain the total bending load (transverse force to the left) applied to the posterior end of the rod to create right rotation in the spine. In addition, strain gages were mounted on each one of the four single-level linkage rods to measure the bending load applied to each individual vertebra (Fig. 1). The strain gages were biaxial precision strain gauges (OMEGA Engineering, Inc., Stamford, CT), with a resistance of 350 ohms, and were applied near the anterior end of each linkage rod. Two biaxial strain gauges were applied at each location, one to measure the tension on the

Fig. 1. Custom Vertebral Derotation Simulator (VDS) attached to the spine through each of eight pedicle screws, creating a quadrangular linkage. For each vertebra, the bilateral pedicle screw heads were connected to each other using a horizontal bar (1). Each bar was then attached to a single-level linkage rod (2). There were four single-level linkage rods. Strain gages were attached on the anterior end of each linkage bar (3). The four single-level linkage rods were linked together (4), creating a rigid quadrangular structure.

Fig. 2. One main linkage rod extended posteriorly (arrow). The load frame’s torsional actuator was aligned with the axis of T10eT11 (top right). The actuator’s rotation pushed the main linkage bar in transversely, thereby transmitting the load through the quadrangular-linked VDS, and sharing the applied load across four vertebral bodies and eight pedicle boneescrew interfaces.

88

S.L. Borkowski et al. / Spine Deformity 4 (2016) 85e93

right side of the bar, and one to measure the compression on the left side of the bar. The pairs were wired in a single fullWheatstone bridge configuration. The bridge was connected to a signal conditioning unit, which in turn was connected to a computer dedicated to the collection of the strain signals. These strains were then used to calculate the applied bending loads simulating a surgeon’s maneuver. To calibrate and relate the measured strains to the amount of bending load and the resulting torsional load on the spine, each of the four bars was tested independently in bending before assembling the VDS. Specifically, an independent bending test was conducted to determine the linear response of the strain gauges as a function of applied load. This calibration value was then used in the calculation of moments that were applied to each bar, to rotate the spine. In the process of calibration, the strains were predicted also using calculations from beam theory, and then compared to the resulting experimental measurements from the strain gauges. The comparison showed very low magnitudes of errors. Specifically, bending moments that are 1 Nm and above showed errors of less than 5%. 5 Nm and above showed far below 3% error. In each specimen, the VDS was attached to the bilateral pedicle screws from T7eT10, effectively protecting the connective tissues at those levels of the spine. Additionally, the T11eT12 vertebrae were rigidly attached to the bottom gimbals of the load frame. Therefore, torsional load was effectively transmitted through the T10eT11 motion segment, and motion occurred predominantly at this level. Torsional loading to the spine was measured and controlled using a bi-axial load cell mounted below the spine.

was defined as a quick, drastic drop in the torsional moment-rotation curve, creating an inflection point at the maximum applied moment. Relative motions of the thoracic spine vertebrae were recorded during load to failure using an Optotrak 3020 3-dimensional motion tracking system (Northern Digital, Inc., Waterloo, Ontario, Canada). Custom motion flags, equipped with four noncollinear light-emitting diode (LED) markers, were attached to T10 and T11 to characterize motion of the T10eT11 disc level. Additionally, one flag was mounted on the MTS load frame post to establish a fixed coordinate axis. Planes of motion for T10 and T11 were established using a digitizing probe. Right axial rotational range of motion (ROM) was recorded as the motion of T10 with respect to T11. This method for measurement of motion is necessary to isolate an accurate measurement of the relative motion between T10 and T11, and specifically to exclude any motion of the other vertebrae relative to each other and any motion of the spine relative to the potting compound or the fixtures. The relative axial rotation of T10eT11 was used to produce the torsional moment-rotation curves from which failure strength was derived. The torsional moment applied to the vertebral bodies through the VDS system was measured and controlled using a two-degree-of-freedom axial-torsional load cell (MTS Systems, Eden Prairie, MN), mounted inferior to the potted vertebrae. The loads on the main linkage rod and each of the four single-level rods were calculated using the strain gage measurements and the length of each linkage bar.

Loading Protocol

Statistical Analysis

The VDS was mounted into an eight-degree-of-freedom MTS 858 mini-bionix servohydraulic load frame equipped with the Flextest system (MTS Systems Corporation, Minneapolis, MN) [35,41,42]. The load frame had gimbals and actuators providing the following eight degrees of freedom: on the top, flexion-extension, lateral bending, axial rotation, and axial loading (four); on the bottom, flexion-extension, lateral bending, anteroposterior translation (shear), and left-right translation (shear) (four). Each specimen was mounted in the VDS in an upright vertical position, with the inferiorly potted vertebrae attached to the bottom gimbals of the load frame. The MTS load frame was programmed to apply torsional moments through the VDS, simulating a correction maneuver (ie, DVR). Specifically, the MTS was programmed to apply a ramping, stepwise axial rotation right moment at a rate of 0.1 /s, in 4-Nm increments. When the first moment step of 4 Nm was reached, a 10 s hold was applied to allow for viscoelastic relaxation. Then, rotation was applied at a rate of 0.1 /s until the second moment step of 8 Nm was reached. This process was continued until gross failure occurred. Failure

The primary input variables were the applied torsional moment, specimen health (thoracic BMD), and specimen age. The primary outcome variables were strain in the VDS at failure, T10eT11 axial rotation right ROM at failure, and moment at the point of failure at the T10eT11 level. Descriptive statistics were used to describe the moment at failure of the thoracic spine specimens. Pearson correlation coefficients and Spearman rho coefficients were used to determine the correlation between failure moment, failure strains, and specimen health variables. Results Failure Results On average, under simulated DVR, the moment at failure of T10eT11 was 33.3  12.1 Nm (Table 1). The mean applied force to produce failure was 151.7  33.1 N. This force was calculated for the posterior end of the single-level rod linkages, at a distance where a surgeon would typically apply direct vertebral rotation forces. This distance was

S.L. Borkowski et al. / Spine Deformity 4 (2016) 85e93

89

Table 1 DVR-to-failure results and observations. Experiment no.

Torsional moment (Nm)

Max force (N)

Observations

1

43.2

190.6

2

21.7

109.6

Loosening(s) Fracture(s) Disc instability Loosening(s) Fracture(s)

3

30.7

135.2

4

13.7

N/A

5

36.9

160.1

6

35.8

154.6

7

47.4

172.8

8

54.7

202.7

9

23.3

N/A

10

25.5

114.4

11

33.2

125.5

Disc instability Loosening(s) Fracture(s) Disc instability Loosening(s) Fracture(s) Disc instability Loosening(s) Fracture(s) Disc instability Loosening(s) Fracture(s) Disc instability Loosening(s) Fracture(s) Disc instability Loosening(s) Fracture(s) Disc instability Loosening(s) Fracture(s) Disc instability Loosening(s) Fracture(s) Disc instability Loosening(s) Fracture(s) Disc instability

R-T7; L-T9; R-T10; L-T10 Fracture at left lamina of T10, just inferior to screw and superior to osteotomy site; fracture at T10 just posterior to the ALL Yes R-T10; L-T10 Fracture at left T10 inferior to screw insertion site None noted

L-T10 Osteophyte fracture Yes. Disc bulging when bending. L-T7; R-T9; L-T9; Rt-T10

R-T10; L-T10

L-T7; R-T10; L-T10

R-T8; L-T8; R-T10; L-T10

ALL, anterior longitudinal ligament; DVR, direct vertebral rotation; N/A, not available. The failure mode of the specimens, along with the maximum torsional moment at T10eT11 at failure, and the maximum transverse force.

Table 2 DVR-to-failure strength, force, and moment results.

Mean Standard Deviation Minimum Maximum

T10eT11 moment

Total applied force (surgical)

Force (T7)

Force (T8)

Force (T9)

Force (T10)

Nm

N

N

N

N

N

33.3 12.1 13.7 54.7

151.7 33.1 109.6 202.7

26.3 38.6 0.1 132.7

38.3 24.95 5.5 85.8

68.5 35.5 7.4 135.7

67.9 35.9 26.8 138.1

DVR, direct vertebral rotation.

approximately 22 cm. (Because of strain gage failure, force data were not available for two specimens; only torsion data is presented for these.) At failure, mean unidirectional right axial rotation at T10eT11 was 11.6 5.6 . The load was applied by the MTS actuator on the main linkage rod was at an average distance away from the spine of 351 mm.

On average, load was largest at the T10 vertebral linkage, where the average force on its single-level rod at failure was 67.9 N. Single-level rod loads decreased across the linkages, moving away from the T10eT11 disc level. In other words, load was highest at T10, decreasing to a minimum load at failure of 26.3 N at the T7 linkage (Table 2).

90

S.L. Borkowski et al. / Spine Deformity 4 (2016) 85e93

Table 3 Cadaveric specimen characteristics and BMD. Donor number

Age

Cause of death

T7 BMD (g/cm2)

T8 BMD (g/cm2)

T9 BMD (g/cm2)

T10 BMD (g/cm2)

Average BMD (g/cm2)

510A00632-001 S141592 S141094 S141548 S141474 S141197 S141107 S141251 S141230 S141253 S141570

50 78 89 49 85 59 66 59 72 95 65

Head trauma, motorcycle-related Esophageal cancer, stage III Uremia; end-stage kidney disease; HTN; CAD End-stage liver disease Thrombocytopenia, prostate cancer, DM, HTN Acute respiratory distress; pneumonia Cardiovascular disease; central sleep apnea Malignant neoplasm esophagus Cardiac arrest; end-stage liver disease Respiratory failure; natural causes Cardiac failure, CAD, CHF

0.885 0.657 0.616 0.713 0.708 0.726 0.992 0.888 0.793 0.952 0.737

0.950 0.625 0.648 0.727 0.751 0.763 0.991 0.971 0.805 0.943 0.791

1.035 0.661 0.643 0.746 0.937 0.814 1.032 0.963 0.813 0.926 0.780

1.071 0.659 0.652 0.801 0.885 0.834 1.151 0.908 0.833 1.061 0.804

0.985 0.651 0.640 0.747 0.820 0.784 1.042 0.933 0.811 0.971 0.778

BMD, bone mineral density; CAD, coronary artery disease; CHF, congestive heart failure; DM, diabetes mellitus; HTN, hypertension.

Table 4 Correlation with average BMD.

Pearson p value Spearman p value

T10eT11 moment

Applied force

.609 .047 .664 .026

.514 .129 .636 .048

BMD, bone mineral density.

Fig. 4. Correlation between the transverse force applied to the VDS main linkage rod and the average BMD across the instrumented T7-T10 vertebral bodies.

Fig. 3. Correlation between the moment applied at failure and the average BMD across the instrumented T7-T10 vertebral bodies.

Failure Measurements as a Function of Bone and Disc Health The failure moment was significantly correlated with average BMD across the quadrangular linkage, with a Pearson correlation coefficient of r 5 0.609 (p 5 .047) (Tables 3 and 4, Fig. 3).

Similarly, the applied force to produce failure was correlated with average BMD, with a Pearson correlation coefficient of r 5 0.514 (p 5 .129) (Fig. 4). BMD ranged from 0.616 to 0.992 g/cm2 at T7, 0.625 to 0.992 g/cm2 at T8, 0.643 to 1.035 g/cm2 at T9, and 0.652 to 1.151 g/cm2 at T10 (Table 3). Average BMD across the linked vertebrae, that is, T7eT10, ranged from 0.640 to 1.042 g/cm2 (Table 3, Fig. 5). There were also strong and significant correlations between failure moment and intervertebral disc health. Specifically, failure moment was positively correlated with radiographic degeneration grade (Spearman rho O 0.662, p ! .04) and MRI degeneration grade (Spearman rho 5 0.742, p 5 .014).

S.L. Borkowski et al. / Spine Deformity 4 (2016) 85e93

Fig. 5. BMD value ranges measured in the T7eT10 bodies.

Discussion Despite the evident large torque magnitudes applied intraoperatively, both from the purported torsional loads (eg, 100 Nm) [3] and the intraoperative failures (eg, screw plow, aortic abutment, pedicle fracture) [9-12], to our knowledge, the strength limits of the thoracic spine column under such torques have not been measured previously. Moreover, the rotational response of the thoracic spine under such large torques is also unknown. Therefore, with the simulated DVR model, we aimed to provide a first step toward a quantitative analysis and understanding of the thoracic spine biomechanical properties in relation to deformity correction. Unlike surgical studies, which have reported safe application of DVR moments in excess of 100 Nm [3], in the present study, thoracic spine torsional failure occurred at applied moments ranging from 13.7 to 54.7 Nm, with an average moment at failure of 33.3 Nm. The corresponding applied force was, on average, 151.7 N. These values are substantially lower than 100 Nm, as was also expected based on previous in vitro evaluations of the lumbar spine [30,31]. Four studies have produced relevant in vitro torsional load simulations using cadaveric spines. For the thoracic spine, only two of these studies evaluated the safety of applying derotational-type torques by bending through pedicle screws in the thoracic spine [28,29]; however, neither study evaluated structural strength of the spinal column. Rather, these two studies constrained all of the vertebral bodies using cement, to allow testing of the screw(s) within the lateral wall pedicle complex. Therefore, all of the intervertebral discs and connective ligaments were protected from torsional loading. According to these two studies, single screwebone interface failure occurred at average values ranging from 4 to 12 Nm [28,29]. However,

91

using a quadrangular linkage similar to the one used in the present study, Cheng et al. [29] showed an average torque at failure of 42.5  16.5 Nm. The present study showed an average failure torque 33.0  12.1 Nm, which is similar to that of Cheng et al., with standard deviations overlapping. However, in the present study, motion was allowed at T10eT11, leaving the intervertebral discs and ligaments at this level subject to torsional loading. Therefore, in the present study, failure at T10eT11 was allowed to occur not only at the boneescrew interfaces but also as a result of failure of the disc or ligaments; in some tests, the latter indeed occurred. Although the previous studies by Parent et al. [28] and Cheng et al. [29] provided valuable information on thoracic pedicle screwebone interface strength, their simulations were limited to only one type of failure. Before the present work, to our knowledge, no previous thoracic spine studies evaluated the strength of the entire thoracic spine column under torsional loading and axial rotation motion, with or without pedicle screw fixation. Both types of simulations provide information relevant to the safety of derotation maneuvers conducted during deformity correction surgery. In contrast to the thoracic spine, for the lumbar spine, two other studies compared torsional loads on unconstrained lumbar vertebrae to measure boneescrew interface strength. Miller et al. [31] and Bisschop et al. [30] both evaluated the torsional strength of lumbar spine FSUs. In Miller et al., lumbar spine torsional strength was less than 59 Nm. Similarly, in Bisschop et al., the strength was on average 58.9 Nm. These results are generally consistent with the results of the present study because the strength of the lumbar spine is expected to be somewhat larger than that of the thoracic spine because of larger dimensions. Intact lumbar strengths were less than 59 Nm and on average 58.9 Nm, respectively [30,31]; as expected, the strength of the lumbar spine was larger than that of the thoracic in our study. Bisschop did show lower strength in specimens with lower BMD, as in the present study. Moreover, as in Bisschop et al. [30], thoracic spine BMD in the present study had a significant effect on the resulting failure moments. Specifically, thoracic spine BMD in our study was significantly correlated with both the applied force at failure (r 5 0.932, p ! .001) and the applied moment at failure (r 5 0.609, p 5 .047). This correlation should be taken into account in future in vitro studies for adolescent AIS and other pediatric and adolescent simulations, as cadaveric specimen selection may be of utmost importance. The majority of in vitro studies are performed using elderly cadaveric spines, as adolescent specimens are virtually unattainable. Consequently, BMD should be considered in specimen selection. As in any in vitro study, the present study had some limitations. First, the cadaveric spines used in the study were from older patients, and none of them had untreated deformities. Adolescent cadaveric spines are very difficult to obtain for research in the United States, and untreated

92

S.L. Borkowski et al. / Spine Deformity 4 (2016) 85e93

scoliotic spines are nearly impossible to find. Despite the differences between the elderly cadaveric specimens we used and the spine of an AIS patient, the study provides approximations of torque versus rotation curves that may prove useful as a first step in establishing safety guidelines for deformity correction using DVR. In assessing the results, the differences between elderly spines with some degeneration and adolescent scoliotic spines must be taken into account. Specifically, differences in bone quality, ligament stiffness, and intervertebral disk geometry and stiffness may affect the results. The only other approach to verify the applied forces obtained from the present study would be to measure this type of force and moment intraoperatively. However, notwithstanding the technical difficulties in conducting in vivo measurements, measuring the strength of the thoracic spine would not be possible intraoperatively. To address this limitation, we contrasted the BMD values of the specimens in the present study to those of AIS patients reported in the literature. The mean BMD value in our specimens was 0.833 g/cm2 (SD 5 0.134). There are very few studies that have reported BMD in AIS patients. None of these studies have directly reported the BMD for thoracic regions, because this is not routinely measured in clinical DEXA studies. For example, one representative study reported the lumbar spine BMD for AIS patients as a mean of 0.66 (SD 5 0.10) in those younger than 13, and a mean BMD of 0.80 (SD 5 0.13) in those 13 and above [43]. In general, it would be reasonable to expect that the BMD of the thoracic spine would be similar to or slightly larger than the lumbar spine. One study in particular evaluated this relationship in cadaveric spines and determined a linear relationship where lumbar spine BMD values were approximately 85% of thoracic spine BMD values [44]. Taking this approximation into account, the BMD values of the lumbar spines of the cadavers used in this study would be about 0.7 g/cm2, which is still above the AIS lumbar spine BMD values reported in the literature. Therefore, the BMD of the specimens in the present study was approximately in the same range as that in AIS patients. Despite the complexity involved in developing the model, the present study addressed only the strength of the thoracic spine under the specific condition of a simulated direct vertebral rotation device mounted on four adjacent vertebrae, testing the strength of a destabilized T10eT11 FSU. Because of larger size, T10eT11 FSUs generally have higher strength than upper levels of the thoracic spine. Therefore, the results of the present study may provide an upper limit of approximation for the strength of thoracic spine FSUs under similar destabilized conditions. Future studies could provide useful information testing the strength of destabilized FSUs at different levels of the thoracic spine, particularly because maximum scoliosis curvatures commonly occur at different levels. Additional studies could also measure and document the strength of intact thoracic spine levels, under torsional loading as well

as flexion-extension or lateral bending. Collectively, such studies could provide useful information for spine deformity surgeries. In summary, for the lumbar spine, throughout all of the in vitro studies in the literature, failures have been observed at substantially smaller torque magnitudes than those reported clinically. However, for the thoracic spine, no comparable data have been published. That is, despite the intraoperative failures [9,10,12], including pedicle screw plow, aortic abutment, and pedicle fracture, the safety limits of the DVR maneuvers on thoracic spines have not yet been established. Measurement of the strength of the thoracic spine in vitro represents the first step toward establishing such safety guidelines. To our knowledge, this is the first study to report the strength of the thoracic spine under a simulated specific scoliosis correction maneuver, commonly used for surgical deformity correction. References [1] Hackenberg L, Link T, Liljenqvist U. Axial and tangential fixation strength of pedicle screws versus hooks in the thoracic spine in relation to bone mineral density. Spine 2002;27:937e42. [2] Liljenqvist U, Hackenberg L, Link T, Halm H. Pullout strength of pedicle screws versus pedicle and laminar hooks in the thoracic spine. Acta Orthop Belg 2001;67:157e63. [3] Chang MS, Lenke LG. Vertebral derotation in adolescent idiopathic scoliosis. Operative Tech Orthop 2009;19:19e23. [4] Lee SM, Suk SI, Chung ER. Direct vertebral rotation: a new technique of three-dimensional deformity correction with segmental pedicle screw fixation in adolescent idiopathic scoliosis. Spine 2004;29:343e9. [5] Asghar J, Samdani AF, Pahys JM, et al. Computed tomography evaluation of rotation correction in adolescent idiopathic scoliosis: a comparison of an all pedicle screw construct versus a hook-rod system. Spine 2009;34:804e7. [6] Hwang SW, Samdani AF, Gressot LV, et al. Effect of direct vertebral body derotation on the sagittal profile in adolescent idiopathic scoliosis. Eur Spine J 2012;21:31e9. [7] Kadoury S, Cheriet F, Beausejour M, et al. A three-dimensional retrospective analysis of the evolution of spinal instrumentation for the correction of adolescent idiopathic scoliosis. Eur Spine J 2009;18: 23e37. [8] Samdani AF, Hwang SW, Miyanji F, et al. Direct vertebral body derotation, thoracoplasty, or both: which is better with respect to inclinometer and scoliosis research society-22 scores? Spine 2012;37: E849e53. [9] Wagner MR, Flores JB, Sanpera I, Herrera-Soto J. Aortic abutment after direct vertebral rotation: plowing of pedicle screws. Spine 2011;36:243e7. [10] Di Silvestre M, Parisini P, Lolli F, Bakaloudis G. Complications of thoracic pedicle screws in scoliosis treatment. Spine 2007;32: 1655e61. [11] Faraj AA, Webb JK. Early complications of spinal pedicle screw. Eur Spine J 1997;6:324e6. [12] Suk SI, Kim WJ, Lee SM, et al. Thoracic pedicle screw fixation in spinal deformities: are they really safe? Spine 2001;26: 2049e57. [13] Sucato DJ, Duchene C. The position of the aorta relative to the spine: a comparison of patients with and without idiopathic scoliosis. J Bone Joint Surg Am 2003;85-A:1461e9. [14] Sandhu HK, Charlton-Ouw KM, Azizzadeh A, et al. Spinal screw penetration of the aorta. J Vasc Surg 2013;57:1668e70.

S.L. Borkowski et al. / Spine Deformity 4 (2016) 85e93 [15] Mac-Thiong JM, Parent S, Poitras B, et al. Neurological outcome and management of pedicle screws misplaced totally within the spinal canal. Spine 2013;38:229e37. [16] Suk SI, Kim JH, Kim SS, et al. Thoracoplasty in thoracic adolescent idiopathic scoliosis. Spine 2008;33:1061e7. [17] Lowenstein JE, Matsumoto H, Vitale MG, et al. Coronal and sagittal plane correction in adolescent idiopathic scoliosis: a comparison between all pedicle screw versus hybrid thoracic hook lumbar screw constructs. Spine 2007;32:448e52. [18] Vora V, Crawford A, Babekhir N, et al. A pedicle screw construct gives an enhanced posterior correction of adolescent idiopathic scoliosis when compared with other constructs: myth or reality. Spine 2007;32:1869e74. [19] Mladenov KV, Vaeterlein C, Stuecker R. Selective posterior thoracic fusion by means of direct vertebral derotation in adolescent idiopathic scoliosis: effects on the sagittal alignment. Eur Spine J 2011;20:1114e7. [20] Demura S, Yaszay B, Carreau J, et al. Maintenance of thoracic kyphosis in the 3D correction of thoracic adolescent idiopathic scoliosis using direct vertebral derotation. Spine Deform 2013;1: 46e50. [21] Cidambi KR, Glaser DA, Bastrom TP, et al. Postoperative changes in spinal rod contour in adolescent idiopathic scoliosis: an in vivo deformation study. Spine 2012;37:1566e72. [22] Hwang SW, Samdani AF, Cahill PJ. The impact of segmental and en bloc derotation maneuvers on scoliosis correction and rib prominence in adolescent idiopathic scoliosis. J Neurosurg Spine 2012;16: 345e50. [23] Hwang SW, Samdani AF, Lonner B, et al. Impact of direct vertebral body derotation on rib prominence: are preoperative factors predictive of changes in rib prominence? Spine 2012;37:E86e9. [24] Kuklo TR, Potter BK, Polly Jr DW, Lenke LG. Monaxial versus multiaxial thoracic pedicle screws in the correction of adolescent idiopathic scoliosis. Spine 2005;30:2113e20. [25] Chang KW, Leng X, Zhao W, et al. Broader curve criteria for selective thoracic fusion. Spine 2011;36:1658e64. [26] Lonner BS, Auerbach JD, Boachie-Adjei O, et al. Treatment of thoracic scoliosis: are monoaxial thoracic pedicle screws the best form of fixation for correction? Spine 2009;34:845e51. [27] Lonner BS, Auerbach JD, Estreicher MB, Kean KE. Thoracic pedicle screw instrumentation: the learning curve and evolution in technique in the treatment of adolescent idiopathic scoliosis. Spine 2009;34: 2158e64. [28] Parent S, Odell T, Oka R, et al. Does the direction of pedicle screw rotation affect the biomechanics of direct transverse plane vertebral derotation? Spine 2008;33:1966e9.

93

[29] Cheng I, Hay D, Iezza A, Lindsey D, Lenke LG. Biomechanical analysis of derotation of the thoracic spine using pedicle screws. Spine 2010;35:1039e43. [30] Bisschop A, van Dieen JH, Kingma I, et al. Torsion biomechanics of the spine following lumbar laminectomy: a human cadaver study. Eur Spine J 2013;22:1785e93. [31] Miller JA, Schultz AB, Warwick DN, Spencer DL. Mechanical properties of lumbar spine motion segments under large loads. J Biomech 1986;19:79e84. [32] Wilke HJ, Wenger K, Claes L. Testing criteria for spinal implants: recommendations for the standardization of in vitro stability testing of spinal implants. Eur Spine J 1998;7:148e54. [33] Panjabi MM, Hausfeld JN, White 3rd AA. A biomechanical study of the ligamentous stability of the thoracic spine in man. Acta Orthop Scand 1981;52:315e26. [34] Oda I, Abumi K, Cunningham BW, et al. An in vitro human cadaveric study investigating the biomechanical properties of the thoracic spine. Spine 2002;27:E64e70. [35] Borkowski SL, Sangiorgio SN, Bowen RE, et al. Flexibility of thoracic spines under simultaneous multi-planar loading. Eur Spine J 2014. http://dx.doi.org/10.1007/s00586-014-3499-0. [36] Mimura M, Panjabi MM, Oxland TR, et al. Disc degeneration affects the multidirectional flexibility of the lumbar spine. Spine 1994;19: 1371e80. [37] Lane NE, Nevitt MC, Genant HK, Hochberg MC. Reliability of new indices of radiographic osteoarthritis of the hand and hip and lumbar disc degeneration. J Rheumatol 1993;20:1911e8. [38] Pfirrmann CW, Metzdorf A, Zanetti M, et al. Magnetic resonance classification of lumbar intervertebral disc degeneration. Spine 2001;26:1873e8. [39] Tan TL, Borkowski SL, Sangiorgio SN, et al. Imaging criteria for the quantification of disc degenaration. J Bone Joint Surg Rev 2015;3:e2. [40] Geck MJ, Macagno A, Ponte A, Shufflebarger HL. The Ponte procedure: posterior only treatment of Scheuermann’s kyphosis using segmental posterior shortening and pedicle screw instrumentation. J Spinal Disord Tech 2007;20:586e93. [41] Sangiorgio SN, Borkowski SL, Bowen RE, et al. Quantification of increase in three-dimensional spine flexibility following sequential Ponte osteotomies in a cadaveric model. Spine Deform 2013;1:171e8. [42] Sangiorgio SN, Sheikh H, Borkowski SL, et al. Comparison of three posterior dynamic stabilization devices. Spine 2011;36:E1251e8. [43] Cheng JC, Tang SP, Guo X, et al. Osteopenia in adolescent idiopathic scoliosis: a histomorphometric study. Spine 2001;26:E19e23. [44] Budoff MJ, Hamirani YS, Gao YL, et al. Measurement of thoracic bone mineral density with quantitative CT. Radiology 2010;257: 434e40.