- Email: [email protected]
rod techniques
The Spine Journal 7 (2007) 194–204
Biomechanical comparison of two-level cervical locking posterior screw/rod and hook/rod techniques Adolfo Espinoza-Larios, MDa, Christopher P. Ames, MDb, Robert H. Chamberlain, MSa,1, Volker K.H. Sonntag, MDa, Curtis A. Dickman, MDa, Neil R. Crawford, PhDa,* a
Spinal Biomechanics, Barrow Neurological Institute, St. Joseph’s Hospital & Medical Center, 350 W. Thomas Road, Phoenix, AZ 85013, USA b University of California–San Francisco, Department of Neurosurgery, 400 Parnassus Ave., San Francisco, CA 94143-0350, USA Received 16 November 2005; accepted 11 April 2006
Abstract
BACKGROUND CONTEXT: Locking posterior instrumentation in the cervical spine can be attached using 1) pedicle screws, 2) lateral mass screws, or 3) laminar hooks. This order of options is in order of decreasing technical difficulty and decreasing depth of fixation, and is thought to be in order of decreasing stability. PURPOSE: We sought to determine whether substantially different biomechanical stability can be achieved in a two-level construct using pedicle screws, lateral mass screws, or laminar hooks. Secondarily, we sought to quantify the differential and additional stability provided by an anterior plate. STUDY DESIGN: In vitro biomechanical flexibility experiment comparing three different posterior constructs for stabilizing the cervical spine after three-column injury. METHODS: Twenty-one human cadaveric cervical spines were divided into three groups. Group 1 received lateral mass screws at C5 and C6 and pedicle screws at C7; Group 2 received lateral mass screws at C5 and C6 and laminar hooks at C7; Group 3 received pedicle screws at C5, C6, and C7. Specimens were nondestructively tested intact, after a three-column two-level injury, after posterior C5–C7 rod fixation, after two-level discectomy and anterior plating, and after removing posterior fixation. Angular motion was recorded during flexion, extension, lateral bending, and axial rotation. Posterior hardware was subsequently failed by dorsal loading. RESULTS: Laminar hooks performed well in resisting flexion and extension but were less effective in resisting lateral bending and axial rotation, allowing greater range of motion (ROM) than screw constructs and allowing a significantly greater percentage of the two-level ROM to occur across the hook level than the screw level (p!.03). Adding an anterior plate significantly improved stability in all three groups. With combined hardware, Group 3 resisted axial rotation significantly worse than the other groups. Posterior instrumentation resisted lateral bending significantly better than anterior plating in all groups (p!.04) and resisted flexion and axial rotation significantly better than anterior plating in most cases. Standard deviation of the ROM was greater with anterior than with posterior fixation. There was no significant difference among groups in resistance to failure (p5.74). CONCLUSIONS: Individual pedicle screws are known to outperform lateral mass screws in terms of pullout resistance, but they offered no apparent advantage in terms of construct stability or failure of whole constructs. Larger standard deviations in anterior fixation imply more variability in the quality of fixation. In most loading modes, laminar hooks provided similar stability to lateral mass screws or pedicle screws; caudal laminar hooks are therefore an acceptable alternative posteriorly. Posterior two-level fixation is less variable and slightly more stable than anterior fixation. Combined instrumentation is significantly more stable than either anterior or posterior alone. Ó 2007 Elsevier Inc. All rights reserved.
Keywords:
Biomechanics; Cervical spine; Pedicle screw; Lateral mass screw; Laminar hook
FDA device/drug status: approved for this indication (lateral mass screw, pedicle screw, anterior plate). Instrumentation and funding were provided by Synthes Spine.
1529-9430/07/$ – see front matter Ó 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.spinee.2006.04.015
* Corresponding author. Spinal Biomechanics, Barrow Neurological Institute, St. Joseph’s Hospital & Medical Center, 350 W. Thomas Road, Phoenix, AZ 85013. Tel.: (602) 406-6652; fax: (602) 406-7197. E-mail address: [email protected] (N.R. Crawford) 1 Deceased January 30, 2005.
A. Espinoza-Larios et al. / The Spine Journal 7 (2007) 194–204
Introduction Several methods are available for fixating the cervical spine to treat instability. These techniques include anterior plating, wiring, lateral mass plating, laminar hooks, and, more recently, transpedicular screw fixation connected between levels with plates or rods. Cervical pedicle screw fixation is currently used most commonly outside the United States and appears to be a promising technique that provides the greatest stability of all posterior fixation devices [1,2]. Although the technique is technically demanding, biomechanical studies evaluating pullout strength have shown that pedicle screw fixation is superior to lateral mass plates [1]. In a calf spine model, transpedicular screw fixation is as stable as fixation with an anterior plate and posterior wiring for the treatment of three-column and multilevel instability [2]. The application of this information to the human spine, however, is limited. The calf spine has larger vertebral bodies, more sagittally oriented facets, and larger pedicles and laminae than the human spine. Our institution previously compared lateral mass fixation with pedicle fixation in a three-column injury model using nonlocking plates. In this study, pedicular fixation in a two-level construct failed to improve rigidity [3]. This finding was unexpected and was suspected to be attributable to the use of a nonlocking plate construct (Medtronic Sofamor Danek, Memphis, TN). An option for posterior cervical fixation that is less challenging than either lateral mass screws or pedicle screws is laminar hook fixation. Little information is available on the stability offered by laminar hooks, which need to be compared directly with screw fixation to understand the stability offered by this device. Through in vitro testing of human cadaveric spines, this study compared the relative stability of a two-level, threecolumn injury (C5–C7) fixated posteriorly with three different screw-rod or screw-hook-rod combinations, anteriorly with a plate, and with combined anterior and posterior instrumentation. The following hypotheses were tested: 1) The order of decreasing stability of the posterior instrumentation is the same as the order of decreasing technical difficulty in placing the hardware (pedicle screwOlateral mass screwOlaminar hook). 2) The addition of anterior instrumentation will decrease the magnitude of any differences observed with posterior instrumentation alone. 3) Posterior techniques of any of the three types studied will outperform anterior plating alone.
Materials and methods Cadaveric specimen preparation and fixation Twenty-one specimens (15 males, 6 females; mean age, 52.3 years; range, 31–66 years) were used in the study. No specimens had a bony malformation or anomaly that
195
indicated a lesion of the ligaments or discs. All specimens were obtained fresh-frozen and thawed in a bath of normal saline at 30 C. All muscles were removed, leaving the ligaments, discs, and joint capsules intact. After the specimen was prepared, three household wood screws were inserted into the exposed articulations of C4 and T1. The screw heads were potted in cylindrical metal fixtures with polymethylmethacrylate at each end. The caudal end was attached to the base of the testing apparatus. The bone mineral density (BMD) of the C5 vertebral body was measured using dual-energy X-ray absorptiometry. Specimens were separated into three groups of seven: Group 1: mean BMD51.1360.16 g/cm2, mean age556 years; Group 2: mean BMD51.0960.27 g/cm2, mean age551 years; Group 3: mean BMD51.0160.24 g/cm2, mean age552 years. One-way analysis of variance (ANOVA) indicated no significant difference in BMD (p5.63) or age (p5.65) among the groups. Specimens were tested in the following five conditions: 1) normal; 2) destabilized; 3) posterior construct; 4) combined construct (posterior hardware and anterior plating); and 5) anterior plating alone. The posterior construct varied by group: Group 1 received lateral mass screws in C5 and C6 and pedicle screws in C7, interconnected by rods (Fig. 1A); Group 2 received lateral mass screws in C5 and C6 and laminar hooks in C7 (Fig. 1B); and Group 3 received pedicle screws from C5 through C7 (Fig. 1C). During testing, specimens were wrapped in saline-soaked gauze to prevent dehydration. Each specimen required 3 to 4 days to complete testing under each condition. Specimens were refrozen at the end of each testing day to preserve their mechanical properties. Surgical procedures After undergoing the normal flexibility test, specimens were destabilized with a simulated three-column injury. Three-column destabilization consisted of incision of the anterior longitudinal ligament, supraspinous ligament, interspinous ligaments, facets capsules, ligamentum flavum, and posterior longitudinal ligament and complete transection of the annulus fibrosus and nucleus pulposus at C5– C6 and C6–C7. Only the fibrous tissues around the uncinate process were left intact. This destabilization left C6 as a ‘‘floating vertebra,’’ which is described as a distractive flexion injury of Grade 4 [4]. The posterior hardware was then attached according to the group. In Groups 1 and 2, lateral mass screws (Starlock, Synthes Spine, Paoli, PA) were placed according to the Magerl technique [5]. Screw pilot holes were drilled using a 2-mmdiameter bit. The screws had a space in the head to receive an interconnecting rod and clamp. The screw-clamp-rod interface became rigid once the nut was tightened. Screws (18 mm long, 3.5 mm diameter) were inserted bicortically. In Groups 1 and 3, pedicle screws (Starlock, Synthes Spine) were placed by following anatomical landmarks.
196
A. Espinoza-Larios et al. / The Spine Journal 7 (2007) 194–204
Fig. 1. Photographs of the three posterior fixation methods used for this study. (A) Group 1: lateral mass screws in C5 and C6, pedicle screws in C7. Note the foraminotomy ‘‘windows’’ in C7 for the surgeon to determine the appropriate trajectory of the pedicle screw. (B) Group 2: lateral mass screws in C5 and C6, and hooks at C7. (C) Group 3: pedicle screws in all the levels. Note foraminotomy windows at each level.
The superior, medial, and inferior walls of the pedicle were palpated by bilateral foraminolaminectomy [6]. Screw pilot holes were drilled using a 2-mm-diameter bit. Screws (26 mm long, 3.5 mm diameter) were inserted to at least the middle of the vertebral body. After insertion, the specimen was inspected carefully to determine whether the screws had violated the pedicle walls. The hooks (Starlock, Synthes Spine) were placed at C7 in Group 2 after excess tissue was removed caudal to the lamina of C7, particularly tissue close to midline such as the excess of ligamentum flavum at this point. Epidural fat was left intact. If there was inadequate space between the adjacent laminae, a small portion of the rostral lamina
of T1 was removed. This resection provided space to attach the upgoing hook (22 mm long from rod attachment end to the most medial point on the curved end, 6 mm wide) and to make the lamina conform to the hook (Fig. 1B). The curved portion of the hook was concave at one end and was 5 mm at its widest point and 4 mm at its narrowest point. The curved end had an angle of 45 degrees to receive the inferior border of the lamina. The other end was higher (7 mm high) and possessed a cylindrical channel where the rod was received and tightened with a set screw. Although only one size hook was available, the size and shape of the hook were adequate for placement in all cases and in no specimen was the hook loose. Connecting rods required
A. Espinoza-Larios et al. / The Spine Journal 7 (2007) 194–204
more angulation in smaller specimens than in bigger specimens to correctly position rods relative to hooks. Titanium rods 3.5 mm in diameter (Starlock, Synthes Spine) were used for all posterior fixations. Their length was determined by the dimensions of each specimen. Overall, Group 1 required a 37-mm rod, Group 2 required a 50mm rod, and Group 3 required a 42-mm rod. Before the rods were attached, they were bent to match the lordotic curvature of each specimen using a rod bender (included in standard tools). Once rods were inserted through the connectors on each screw or hook, the screw heads were compressed toward each other using a compressor tool (included in standard tools). Compression was applied to mimic the procedure often used intraoperatively to improve the environment for fusion across abutting bone masses, and may have slightly increased lordosis across the construct, although curvature was not documented. Compression was moderate, representative of the pressure applied surgically under ‘‘two-finger’’ tightening. While compression was maintained, the small screws located at the top of each clamp were tightened to lock the rods to the connectors. Next, the nut in the head of each screw was tightened while countertorque was applied. This process tightened the connectors to the screws. Anterior interbody grafting and plate fixation were performed at C5–C6 and C6–C7 using standard techniques. The discs were excised and the end plates were superficially flattened using a pneumatic high-speed drill (Medtronic Midas Rex, Fort Worth, TX) before the structural wedge grafts were placed. Wedge grafts were cut from human fibular allograft from another source. Plates (Atlantis, Medtronic Sofamor Danek, Memphis, TN) ranged from 42.5 mm to 50 mm long. Cancellous locking screws (14 mm long, 3.5 mm in diameter) were used to attach plates unicortically.
Biomechanical testing Specimens were tested nondestructively using a standard flexibility testing method [7]. The fixture holding the caudal vertebra was attached to the base of the testing apparatus while loads were applied to the rostral fixture. Nondestructive nonconstraining pure moment loading was applied to each specimen through a system of cables and pulleys in conjunction with a standard servohydraulic test system (MTS, Minneapolis, MN), as described previously [8]. Loads were applied about the appropriate anatomic axes to induce six different types of motion in the following order: flexion, extension, right axial rotation, left axial rotation, right lateral bending, and left lateral bending. For each type of load, three preconditioning cycles were applied at 1.5 Nm for 60 seconds before data were collected. After the third preconditioning cycle, the specimen was allowed to recoil for 60 seconds at zero load. During the data-collection phase, loads were applied in 0.25-Nm increments
197
to a maximum 1.5 Nm. Each load was held for 45 seconds. Data were collected at 2 Hz. Three-dimensional specimen motion in response to the loads was determined using the Optotrak 3020 system (Northern Digital, Waterloo, Ontario, Canada). This system stereophotogrammetrically measures the three-dimensional displacement of infrared-emitting markers rigidly attached in a noncollinear arrangement to each vertebra. Marker position was related to the x (lateral), y (rostrocaudal), and z (anteroposterior) axes of the specimens by identifying anatomic landmarks with a digitizing probe (accessory to Optotrak) and custom software [9]. Custom software also converted the position coordinates to angles about each anatomic axis using a method that models the vertebrae as stacked cylinders [10]. After all nondestructive testing was completed for a specimen, all hardware was removed and the specimen was refrozen. Hardware was reused in subsequent specimens. At a later date, when enough hardware was available, all specimens were thawed and new posterior instrumentation (screws, hooks, and newly contoured rods) was inserted for failure testing (Fig. 2). Specimens were mounted rigidly to side rails and secured in two angle vises, oriented horizontally (prone). Loops of wire were attached to the rostral and caudal ends of the construct. The wire loops were attached to the piston of the servohydraulic test frame using a linkage that does not constrain rotation (rodend). The linkage therefore allowed equal tension to be applied to both loops throughout testing. The construct was pulled out at a constant displacement rate of 5 mm/minute until visible complete failure occurred while monitoring the applied tension at 2 Hz. The first point of failure was also noted. From the quasistatic load-deformation data, the angular range of motion (ROM) and lax zone (LZ) were determined. The LZ is similar to the neutral zone and is the portion of the ROM in which there is minimal ligamentous resistance [11]. The upper boundary of the LZ was calculated by extrapolating the load-deformation slope at data points corresponding to 0.75 Nm, 1.00 Nm, 1.25 Nm, and 1.50 Nm to zero load using the method of least squares. Larger values of LZ or ROM indicate greater instability. The LZ and ROM were quantified at both levels within the construct (C5–C6 and C6–C7), and the percent contribution of each level to the motion across the entire construct was quantified. From load-to-failure data, the ultimate strength was calculated and the mode of failure was tabulated. One-way ANOVA followed by the Holm-Sidak test was used for statistical assessment of the differences among study groups in LZ, ROM, distribution of LZ and ROM between levels of the construct, and failure load. One-way repeated-measures ANOVA followed by the Holm-Sidak test was used to compare LZ and ROM across conditions. P values less than 0.05 were considered significant.
198
A. Espinoza-Larios et al. / The Spine Journal 7 (2007) 194–204
Results Normal and destabilized specimens In the normal condition, averaged for all three groups, the mean unilateral ROM (6 standard deviation) across C5–C7 was 11.862.7 during flexion, 10.562.6 during extension, 9.063.2 during lateral bending, and 7.662.6 during axial rotation. The mean unilateral LZ across C5– C7 was 6.762.3 during flexion or extension, 6.162.7 during lateral bending, and 4.761.9 during axial rotation. There was no significant difference among the three groups in ROM (pO.46) or LZ (pO.37) during any mode of loading. Averaged for all three groups, destabilization increased ROM across C5–C7 by 67631% during flexion, by 85662% during extension, by 33618% during lateral bending, and by 45623% during axial rotation. Destabilization increased LZ across C5–C7 by 136685% during flexion-extension, by 61649% during lateral bending, and by 67636% during axial rotation. After destabilization, the three groups remained equal, with no significant differences in ROM (pO.82) or LZ (pO.88) during any mode of loading. Instrumented flexibility tests Posterior instrumentation reduced ROM to an average of 17% of normal and reduced LZ to an average of 3% of normal. Posterior instrumentation of all three types limited lateral bending and axial rotation better than it limited flexion or extension. There were no significant differences in the ROM or LZ allowed among the three types of posterior instrumentation (Figs. 3 and 4, Tables 1 and 2). The ROM was distributed with a significantly greater percentage of the total motion at C5–C6 in Groups 1 and 3 than in Group 2 during flexion (p!.014), lateral bending (p!.026), and axial rotation (p!.008; Fig. 5A). During extension, the distribution was equivalent in all three groups (p5.635). The LZ was distributed with a significantly greater percentage of the total motion at C5–C6 in Group 3 than in Group 2 during flexion-extension (p5.008; Fig. 5B). Distributions were equivalent among groups during lateral bending (p5.374) and axial rotation (p5.234). Combined instrumentation further reduced ROM to an average of 4% of normal and reduced LZ to an average of 1% of normal. Groups 1 and 2 allowed significantly smaller ROM during axial rotation than Group 3 (pedicle screws only; Fig. 3, Table 1). There were no significant differences in the LZ allowed by the three combined constructs (Fig. 4, Table 2). The ROM with combined hardware in place was significantly smaller than the ROM with posterior hardware alone in place during flexion and extension (all groups, Table 3). When posterior hardware was disconnected leaving only anterior plates, the ROM remained at an average of 31% of normal and the LZ remained at an average of 17% of
normal. There were no significant differences among groups in ROM or LZ (Figs. 3 and 4, Tables 1 and 2). Unlike posterior instrumentation, which limited ROM and LZ better during lateral bending and axial rotation, anterior instrumentation provided approximately equivalent limitation of motion in all directions (Figs. 3 and 4, Tables 1 and 2). During most loading modes in all three groups, the ROM and LZ with anterior hardware alone in place were significantly larger than the ROM and LZ with combined hardware in place (Table 3). In most loading modes, anterior instrumentation allowed a significantly greater LZ and ROM than posterior instrumentation (Table 3).
Failure loading Of the 21 specimens, 19 were loaded to failure with posterior instrumentation only. Two specimens (one from Group 1 and one from Group 3) could not be loaded to failure because a screw had broken off in the bone before failure testing and could not be retrieved. The ultimate strength among groups was not significantly different (Fig. 6; p5.74). Specimens were observed carefully to determine the first point of failure (Table 4). In the combined lateral mass and pedicle screw group (Group 1), failure occurred with equal frequency at rostral and caudal ends of the construct. The first point of failure was usually the rostral (C5)
Fig. 2. Construct pull-off testing was performed with specimens mounted prone with rigidly locked distal ends in side rails. A rod-end attached to two loops of wire allowed both the rostral and caudal ends of the construct to be evenly loaded during pull-off.
A. Espinoza-Larios et al. / The Spine Journal 7 (2007) 194–204
199
Fig. 3. Graphs showing the range of motion summed across C5–C7 in Groups 1, 2, and 3 during each loading mode and in each instrumented condition studied. Error bars show standard deviation. LMS5lateral mass screw; PS5pedicle screw.
screws in the hook group (Group 2). In the pedicle-screwonly group (Group 3), failure usually occurred first at the caudal level (C7).
Discussion Instability model and restabilization models Our three-column discoligamentous injury model simulated a severe flexion-distraction type lower cervical spine
trauma [4,12]. The severity of the injury model should be considered when comparing our results with other published studies that simulated different severities of injury [2,13]. We chose a severe injury model because we believed it would demonstrate the differences among fixation techniques more effectively than if normal ligamentous structures were allowed to contribute to construct stability. Although we inserted lateral mass screws with bicortical purchase, some surgeons prefer unicortical lateral mass screw insertion. We chose the bicortical technique because
Fig. 4. Graphs showing the lax zone summed across C5–C7 in Groups 1, 2, and 3 during each loading mode and in each instrumented condition studied. Error bars show standard deviation. LMS5lateral mass screw; PS5pedicle screw.
200
A. Espinoza-Larios et al. / The Spine Journal 7 (2007) 194–204
Table 1 Comparisons of C5–C7 ROM among groups during different conditions Flexion Condition
Comparison (Groups)
Posterior
1 1 2 1 1 2 1 1 2
Combined
Anterior
vs. vs. vs. vs. vs. vs. vs. vs. vs.
2 3 3 2 3 3 2 3 3
Extension
Lateral bending
Axial rotation
D Angle
p value
D Angle
p value
D Angle
p value
D Angle
p value
0.11 0.51 0.62 0.06 0.15 0.10 0.36 0.48 0.13
(0.23) (0.23) (0.23) (0.51) (0.51) (0.51) (0.87) (0.87) (0.87)
0.18 0.59 0.77 0.01 0.20 0.19 0.09 0.15 0.23
(0.33) (0.33) (0.33) (0.50) (0.50) (0.50) (0.97) (0.97) (0.97)
0.53 0.23 0.30 0.08 0.10 0.02 0.00 1.51 1.51
(0.22) (0.22) (0.22) (0.51) (0.51) (0.51) (0.26) (0.26) (0.26)
0.43 0.23 0.20 0.01 L0.20 L0.19 0.01 1.12 1.13
(0.12) (0.12) (0.12) (0.86) (0.01) (0.02) (0.36) (0.36) (0.36)
ROM5range of motion. For comparison ‘‘x vs. y,’’ D Angle5ROMx – ROMy (in degrees). Differences in boldface are statistically significant (p!.05).
it is reportedly stiffer with a higher resistance to screw pullout [14,15]. We wanted to compare the strongest possible technique of lateral mass screw fixation to pedicle screw fixation. Although unicortical lateral mass screws are less resistant to pullout, further study is needed to determine whether the stability that they offer when used within a construct differs from that of bicortical screws. Seemingly, a biomechanical comparison of lateral mass screws, pedicle screws, and laminar hooks would be more effective if different groups of specimens had each device placed at all three vertebrae. Although Group 3 had pedicle screws at all levels, Group 1 had lateral mass screws only at C5 and C6 and Group 2 had laminar hooks only at C7. These constructs were chosen because they are realistic clinically. Anatomically, C7 is ill suited for lateral mass screws because its lateral masses are narrow and have a steep slope. In contrast, C7 is typically the level of the cervical spine best suited for pedicle screws because the diameter of its pedicle is large. By similar reasoning, hooks were applied only at C7 with lateral mass screws at C5 and C6. Clinically, a hook-only construct would be avoided because it would be expected to dislodge easily. Because the anatomy of C7 is unsuitable for lateral mass screws,
a surgeon might realistically choose laminar hooks rather than the surgically more challenging procedure of placing pedicle screws.
Comparison of posterior instrumentation One of our hypotheses was that the order of decreasing stability would match the order of decreasing technical difficulty of placing the construct. In other words, we expected that specimens with pedicle-screw-only fixation (Group 3) would be most stable, specimens with combined lateral mass screws and pedicle screws (Group 1) would be less stable, and specimens with combined lateral mass screws and laminar hooks (Group 2) would be least stable. This hypothesis, however, was not supported. Instead, no significant differences were observed in LZ or ROM in the various loading modes among the groups. The only exception was that Group 3, presumed to be the most stable, was actually less stable with a combined construct during axial rotation than either Group 1 or Group 2 (Table 1). Compared with the other hardware types, there were no significant differences in ROM and LZ in the group with laminar hooks
Table 2 Comparisons of C5–C7 LZ among groups during different conditions Flexion-Extension Condition
Comparison (Groups)
Posterior
1 1 2 1 1 2 1 1 2
Combined
Anterior
vs. vs. vs. vs. vs. vs. vs. vs. vs.
2 3 3 2 3 3 2 3 3
Lateral bending
Axial rotation
D Angle
p value
D Angle
p value
D Angle
p value
0.02 0.38 0.36 0.03 0.28 0.25 0.08 0.82 0.90
(0.53) (0.53) (0.53) (0.30) (0.30) (0.30) (0.68) (0.68) (0.68)
0.61 0.22 0.39 0.09 0.09 0.00 0.23 2.17 1.94
(0.30) (0.30) (0.30) (0.61) (0.61) (0.61) (0.19) (0.19) (0.19)
0.08 0.04 0.12 0.00 0.06 0.06 0.02 1.43 1.44
(0.13) (0.13) (0.13) (0.32) (0.32) (0.32) (0.30) (0.30) (0.30)
LZ5lax zone. For comparison ‘‘x vs. y,’’ D Angle5LZx – LZy (in degrees).
A. Espinoza-Larios et al. / The Spine Journal 7 (2007) 194–204
201
(Tables 1 and 2). Hooks consistently performed as well or better than screws in resisting flexion and extension. Another way to compare the different posterior techniques is to study the percentage of overall motion resisted at each level within the construct. In Group 1, if more of the C5–C7 ROM is limited at C6–C7, where pedicle screws and lateral mass screws are used, than at C5–C6, where only lateral mass screws are used, then it could be said that pedicle screws provide better stability than lateral mass screws. However, the distribution of ROM at each level in Group 1 closely matched the distribution in Group 3 (Fig. 5A), implying that pedicle screws and lateral mass screws were equivalent. In Group 2, however, a significantly less symmetrical distribution of ROM between levels was observed than in the other two groups. During flexion, lateral bending, and axial rotation (Fig. 5A), more of the C5– C7 motion was allowed at C6–C7 (lateral mass screws and laminar hooks) than at C5–C6 (lateral mass screws only). This finding indicates that laminar hooks contributed significantly less stability than lateral mass screws during these loading modes. Quantitative findings from the load-to-failure tests indicated that ultimate strength (Fig. 6) did not differ significantly among the three groups. Qualitatively, we expected lateral mass screws (at C5) to pull out earlier than pedicle screws (at C7) in Group 1. Instead, C5 failed first in as many cases as C7 (3 each, Table 4). One important finding from Group 2 (lateral mass screws at C5, laminar hooks at C7) was that screws always failed first. In no case did hooks disengage or fail before screws. In one case, the hooks caused the C7 lamina to fracture. Intuitively, this finding seems reasonable. It should be easier to pull a screw out of a hole than to pull a hook through an entire cross-section of bone. Anterior versus posterior hardware We predicted that any of the three posterior techniques would outperform anterior plating. Posterior instrumentation outperformed anterior instrumentation during flexion, lateral bending, and axial rotation, but not during extension. This finding is consistent with expectations based on the design of each device. In resisting flexion, posterior hardware serves as both a tension band and a cantilever. Anterior hardware can serve only as a cantilever. The opposite is true during extension: the anterior plate acts as both a cantilever and a tension band whereas the posterior hardware acts only as a cantilever. Theoretically, with good compression before hardware is tightened, a tension band would be expected to be more effective in limiting motion than a cantilever. During lateral bending and axial rotation, both anterior and posterior devices behaved similarly as cantilevers. However, the anterior plate was much closer to the axis of rotation whereas the lateral mass and pedicle screws were positioned farther laterally to each side of the axis of rotation. Furthermore, the two points of fixation at each level (left and right screws) were positioned farther apart in lateral mass and
Fig. 5. Distribution of the (A) range of motion and (B) lax zone by level in specimens instrumented with posterior hardware. The percentage above the horizontal line is the contribution of C5–C6 to the total (C5–C7) lax zone or range of motion across the construct. The percentage below the horizontal line is the contribution of C6–C7. Asterisks indicate significant differences (p!.05) between groups in the percent contribution of C5–C6 to the total motion across the construct.
pedicle screw fixation than in anterior plate fixation. Therefore, more leverage was available in the posterior constructs to resist lateral bending and axial rotation. Left and right screw separation is unimportant in resisting flexion and extension. Compared with posterior hardware, anterior hardware appeared to be less consistent in the quality of fixation achieved. That is, the standard deviations were larger for specimens with anterior-only hardware in averaged biomechanical parameters than in the corresponding parameters in the same specimens with posterior-only hardware (Figs. 3 and 4). This finding was observed because some specimens showed very good resistance to motion with anterior-only hardware and others showed poor resistance to motion with the same hardware. The quality of vertebral
202
A. Espinoza-Larios et al. / The Spine Journal 7 (2007) 194–204
Table 3 P values from statistical comparisons of ROM and LZ among conditions (ANOVA/Holm-Sidak) Posterior vs. Combined Loading mode ROM Flexion Extension Lateral bending Axial rotation LZ Flexion-Extension Lateral bending Axial rotation
Anterior vs. Combined
Anterior vs. Posterior
Group 1
Group 2
Group 3
Group 1
Group 2
Group 3
Group 1
Group 2
Group 3
0.031 0.001 0.573 0.122
0.026 0.001 0.097 0.062
0.041 0.001 0.767 0.584
0.001 0.001 0.001 0.001
0.001 0.001 0.001 0.001
0.001 0.002 0.002 0.002
0.007 0.652 0.001 0.001
0.001 0.524 0.001 0.031
0.036 0.790 0.005 0.008
0.317 0.784 0.695
0.257 0.163 0.120
0.094 0.887 0.057
0.001 0.001 0.001
0.001 0.001 0.120
0.094 0.012 0.057
0.012 0.001 0.003
0.007 0.022 0.120
0.094 0.016 0.057
ANOVA5analysis of variance; LZ5lax zone; ROM5range of motion. Values in boldface are statistically significant (p!.05).
body bone seemed to be more variable than the quality of lateral mass, lamina, and pedicle bone. Anterior screw fixation was also unicortical whereas posterior screw fixation was bicortical. Combined hardware Finally, we hypothesized that the addition of anterior instrumentation (converting a posterior-only construct to a combined anterior-posterior construct) would decrease the magnitude of any differences observed with posterior
instrumentation alone. There was little opportunity to support this hypothesis because there were few differences among posterior instrumentation groups. In fact, the opposite was true for some testing parameters. For example, with posterior instrumentation only, no difference was observed among groups in the ROM during axial rotation (Fig. 3, Table 1). However, with combined instrumentation, Group 3 allowed significantly more axial rotation ROM than Group 1 or Group 2. The absolute magnitudes of the differences with combined hardware were very small (less than 0.21 across C5–C7), but the data were more consistent than posterioronly data, leading to a statistical significance. The anterior-only condition was studied last after posterior instrumentation was removed. Theoretically, the anterior-only condition might have been expected to be equivalent among groups. In many modes, however, ROM was greatest in Group 3 with anterior-only instrumentation. Therefore, the keyhole foraminotomies for pedicle screw placement or some other aspect of the earlier Group 3 posterior surgery may have left the specimens in a less stable condition than specimens in Group 1 or Group 2.
Comparison to previous literature
Fig. 6. Ultimate strength (maximum load endured) during load to failure. There was no significant difference in strength among groups (p5.74). Error bars show standard deviation.
Kotani et al. [2] demonstrated the stabilizing capabilities of transpedicular screw fixation and combined anterior and posterior fixation in a two-level, three-column instability model under torsional and extension loading. Similarly, in our model in all three groups with combined instrumentation, ROM was significantly decreased from the intact condition during every loading mode. Kotani et al. [2] also reported that during flexion testing, anterior plate fixation was unable to reduce angular displacement below the intact level under a three-column instability condition. Instead, we found that anterior plate fixation reduced motion during all loading modes in all three groups relative to normal. Most of these reductions were statistically significant. However, we did not perform corpectomies in our injury model and we used a human cadaveric model instead of a calf spine model.
A. Espinoza-Larios et al. / The Spine Journal 7 (2007) 194–204
203
Table 4 Load-to-failure qualitative summary Group 1 (LMS, LMS, PS)
Group 2 (LMS, LMS, Hook)
Group 3 (PS, PS, PS)
Spec.
1st point of failure
Spec.
1st point of failure
Spec.
1st point of failure
1a 1b 1c 1d 1e 1f
C5 C7 C5 C5 C7 C7
2a 2b 2c 2d 2e 2f 2g
C7 lamina fracture C5 L screw pullout C5 L screw pullout C5 L screw pullout C5 R & L screw pullout Ligamentous failure C5 L screw pullout
3a 3b 3c 3d 3e 3f
C7 R & L screw pullout C7 R & L screw pullout C5 R & L screw pullout C7 R & L screw pullout Ligamentous failure C7 R & L screw pullout
R screw pullout R & L screw pullout R screw pullout R screw pullout L screw pullout L screw pullout
L5left; LMS5lateral mass screw; PS5pedicle screw; R5right.
Bozkus et al. [12] studied similar hardware by a different manufacturer in a similar injury model using identical testing methods. However, hooks were not studied and posterior screws were inserted by image guidance instead of using keyhole foraminotomies. Their results were highly consistent with the current findings. They also observed that anterior plating provided significantly less stability than posterior hardware; combined hardware provided significantly better stability than anterior or posterior alone; and lateral mass screws performed equivalently to pedicle screws. Koh et al. [13] studied human cadaveric spines instrumented after simulated flexion-distraction injury or burst fracture. They compared an anterior plate construct, a posterior lateral mass screw-plate construct, and a combined construct. In their flexion-distraction model, as in our study, posterior plating constructs with and without anterior fixation significantly reduced flexion, lateral bending, and axial rotation motion compared with that of the intact specimen. During extension, motion decreased significantly from the intact case only with the combined construct. However, anterior plate fixation did not reduce ROM from the intact condition in any mode in their study. In contrast, in our two-level three-column injury model, anterior fixation, posterior fixation, and combined fixation significantly reduced the C5–C7 ROM from the intact condition during all loading modes in all groups except Group 3 anterior fixation during axial rotation. The differences likely occurred because both injury models used by Koh et al. were dissimilar to the injury induced in our study. Adams et al. [16] used a two-level (C4–C5 and C5–C6) three-column in vitro injury to compare the stability offered by anterior plating alone to that offered by combined anterior plating and posterior lateral mass plating. Similar to our results, they found that the two-level anterior plate alone stabilized the spine moderately well (significantly within the ROM of intact specimens). However, the addition of posterior plating further significantly reduced motion during all loading modes. They used the older style posterior nonlocking lateral mass screw-plates, which seemingly would have provided less stability than the rigid system that we studied. Kothe et al. [17] found that three-level injuries stabilized with pedicle screws had a significantly smaller ROM during
lateral bending than constructs stabilized with lateral mass screws. As in the current study, lateral mass screws were inserted bicortically using the Magerl technique. The different findings of our study and that of Kothe may be attributable to their usage of constructs with only one type of screw (not mixed lateral mass and pedicle screws), longer constructs, fewer specimens studied (n54 per group), and usage of image guidance instead of keyhole foraminotomies for pedicle screw insertion. They also found that the pedicle screw construct loosened less with fatigue than the lateral mass screw construct. We did not investigate the effect of fatigue. The study of Schmidt et al. [18] focused on comparing rigid versus nonconstrained posterior systems in a one-level corpectomy model. Although their injury and length of construct were different, the results were consistent with our findings, showing no benefit of pedicle screws over lateral mass screws when used with a rigidly constrained posterior screw-rod system. There was little difference in ultimate strength during failure among the groups with posterior instrumentation only. These findings are inconsistent with those of Jones et al. [1], who showed that pedicle screws pulled out at a significantly higher load than lateral mass screws. Apparently, pullout resistance of individual screws is less important than the angulation and rigid interlocking of an entire posterior construct when considering failure. Clinical considerations We observed no significant difference in stability during any loading mode in specimens incorporating lateral mass screws at two of three vertebrae compared with specimens incorporating pedicle screws at all three vertebrae. Assuming that the amount of immobilization of a construct is directly related to the quality of the environment for fusion, this finding implies that both systems provide equivalent environments for fusion. The two groups also resisted failure equivalently. Together, these findings imply that there is little biomechanical difference between the two fixation types in the cervical spine. Clinically, therefore, the increased biomechanical stability of pedicle screw fixation may not offset the increased risks of the procedure.
204
A. Espinoza-Larios et al. / The Spine Journal 7 (2007) 194–204
A further clinical consideration in choosing pedicle screws or lateral mass screws is that the screw heads are positioned differently. In Group 1 specimens, the heads of the lateral mass screws at C5 and C6 were more medial than the heads of the pedicle screws at C7 (Fig. 1A). Rods must be contoured more in constructs using mixed screw types to create a construct in which the vertebrae are oriented naturally. In contrast, the heads of the screws in Group 2 and Group 3 specimens were better aligned and only required contouring to match the cervical lordosis (Fig. 1B). Laminar hooks appear to be a good alternative to pedicle screws at C7. Overall, they stabilized as well as screws. However, within the Group 2 construct, a greater percentage of the motion was allowed at C6–C7. Hooks were highly resistant to failure. Technically, we found laminar hooks to be time-consuming to attach at C7; bone must be removed carefully to obtain a good interface. Study limitations Our in vitro experimental testing excluded the effects of muscular co-contraction or the weight of the head, which would be assumed to increase stability. Therefore, our results can be considered a worst-case scenario, where only hardware, bones, and ligaments are able to limit motion. Greater stability would therefore be expected in vivo. Also, our injury model was intended to mimic a specific threecolumn injury commonly seen clinically. The results of this study therefore should not be generalized to other injury patterns of the lower cervical spine. In all specimens, posterior instrumentation was tested first, combined instrumentation second, and anterior instrumentation last. The testing order is typically randomized to prevent bias related to specimen degradation. However, we could not randomize the testing order because the anterior instrumentation required disc removal and bone graft placement. Therefore, the poorer performance of the anterior construct could have reflected deterioration of the specimen through the course of testing. A different study design using separate groups of specimens would be needed to determine the magnitude of such an effect. Another uncontrolled factor possibly at work in the degree of stabilization provided by anterior versus posterior instrumentation is the amount of lordosis present. Facet interaction would presumably have been greater in more lordotic specimens, and these specimens could therefore have benefited more from supplementary hardware (anterior or posterior). However, lordosis was not recorded. Further study would be needed to elucidate the importance of lordosis on fixation stiffness using the hardware studied. We investigated immediate postoperative stability and load required for failure. However, we did not study the fatigue endurance of the different cervical fixation constructs. Our study therefore does not address which type of construct is more likely to break with normal usage over time,
which is also a worthwhile consideration when comparing spinal fixation hardware. References [1] Jones EL, Heller JG, Silcox DH, Hutton WC. Cervical pedicle screws versus lateral mass screws: anatomic feasibility and biomechanical comparison. Spine 1997;22:977–82. [2] Kotani Y, Cunningham BW, Abumi K, McAfee PC. Biomechanical analysis of cervical stabilization systems: an assessment of transpedicular screw fixation in the cervical spine. Spine 1994;19:2529–39. [3] Bartolomei JC, Das K, Crawford NR, et al. Cervical pedicle screws/plates or lateral mass screws/plates combined with anterior plating: a biomechanical in vitro study (abstract). 71st Annual Meeting of the American Association of Neurological Surgeons; April 6–11, 2002; Chicago, IL. [4] Allen BL Jr. Ferguson RL, Lehmann TR, O’Brien RP. A mechanistic classification of closed, indirect fractures and dislocations of the lower cervical spine. Spine 1982;7:1–27. [5] Jeanneret B, Gebhard JS, Magerl F. Transpedicular screw fixation of articular mass fracture-separation: results of an anatomical study and operative technique. J Spinal Disord 1994;7:222–9. [6] Abumi K, Itoh H, Taneichi H, Kaneda K. Transpedicular screw fixation for traumatic lesions of the middle and lower cervical spine: description of the techniques and preliminary report. J Spinal Disord 1994;7:19–28. [7] Panjabi MM. Biomechanical evaluation of spinal fixation devices: I. A conceptual framework. Spine 1988;13:1129–34. [8] Crawford NR, Brantley AG, Dickman CA, Koeneman EJ. An apparatus for applying pure nonconstraining moments to spine segments in vitro. Spine 1995;20:2097–100. [9] Crawford NR, Dickman CA. Construction of local vertebral coordinate systems using a digitizing probe. Technical note. Spine 1997;22:559–63. [10] Crawford NR, Yamaguchi GT, Dickman CA. A new technique for determining 3-D joint angles: the tilt/twist method. Clin Biomech (Bristol, Avon) 1999;14:153–65. [11] Crawford NR, Peles JD, Dickman CA. The spinal lax zone and neutral zone: measurement techniques and parameter comparisons. J Spinal Disord 1998;11:416–29. [12] Bozkus H, Ames CP, Chamberlain RH, et al. Biomechanical analysis of rigid stabilization techniques for three-column injury in the lower cervical spine. Spine 2005;30:915–22. [13] Koh YD, Lim TH, You JW, Eck J, An HS. A biomechanical comparison of modern anterior and posterior plate fixation of the cervical spine. Spine 2001;26:15–21. [14] Errico T, Uhl R, Cooper P, Casar R, McHenry T. Pullout strength comparison of two methods of orienting screw insertion in the lateral masses of the bovine cervical spine. J Spinal Disord 1992;5: 459–63. [15] Montesano PX, Jauch E, Jonsson H Jr. Anatomic and biomechanical study of posterior cervical spine plate arthrodesis: an evaluation of two different techniques of screw placement. J Spinal Disord 1992;5:301–5. [16] Adams MS, Crawford NR, Chamberlain RH, Sonntag VKH, Dickman CA. Biomechanical comparison of anterior cervical plating and combined anterior/lateral mass plating. Spine J 2001;1:166– 70. [17] Kothe R, Ruther W, Schneider E, Linke B. Biomechanical analysis of transpedicular screw fixation in the subaxial cervical spine. Spine 2004;29:1869–75. [18] Schmidt R, Wilke HJ, Claes L, Puhl W, Richter M. Effect of constrained posterior screw and rod systems for primary stability: biomechanical in vitro comparison of various instrumentations in a single-level corpectomy model. Eur Spine J 2005;14:372–80.
Biomechanical comparison of two-level cervical locking posterior screw/rod and hook/rod techniques Adolfo Espinoza-Larios, MDa, Christopher P. Ames, MDb, Robert H. Chamberlain, MSa,1, Volker K.H. Sonntag, MDa, Curtis A. Dickman, MDa, Neil R. Crawford, PhDa,* a
Spinal Biomechanics, Barrow Neurological Institute, St. Joseph’s Hospital & Medical Center, 350 W. Thomas Road, Phoenix, AZ 85013, USA b University of California–San Francisco, Department of Neurosurgery, 400 Parnassus Ave., San Francisco, CA 94143-0350, USA Received 16 November 2005; accepted 11 April 2006
Abstract
BACKGROUND CONTEXT: Locking posterior instrumentation in the cervical spine can be attached using 1) pedicle screws, 2) lateral mass screws, or 3) laminar hooks. This order of options is in order of decreasing technical difficulty and decreasing depth of fixation, and is thought to be in order of decreasing stability. PURPOSE: We sought to determine whether substantially different biomechanical stability can be achieved in a two-level construct using pedicle screws, lateral mass screws, or laminar hooks. Secondarily, we sought to quantify the differential and additional stability provided by an anterior plate. STUDY DESIGN: In vitro biomechanical flexibility experiment comparing three different posterior constructs for stabilizing the cervical spine after three-column injury. METHODS: Twenty-one human cadaveric cervical spines were divided into three groups. Group 1 received lateral mass screws at C5 and C6 and pedicle screws at C7; Group 2 received lateral mass screws at C5 and C6 and laminar hooks at C7; Group 3 received pedicle screws at C5, C6, and C7. Specimens were nondestructively tested intact, after a three-column two-level injury, after posterior C5–C7 rod fixation, after two-level discectomy and anterior plating, and after removing posterior fixation. Angular motion was recorded during flexion, extension, lateral bending, and axial rotation. Posterior hardware was subsequently failed by dorsal loading. RESULTS: Laminar hooks performed well in resisting flexion and extension but were less effective in resisting lateral bending and axial rotation, allowing greater range of motion (ROM) than screw constructs and allowing a significantly greater percentage of the two-level ROM to occur across the hook level than the screw level (p!.03). Adding an anterior plate significantly improved stability in all three groups. With combined hardware, Group 3 resisted axial rotation significantly worse than the other groups. Posterior instrumentation resisted lateral bending significantly better than anterior plating in all groups (p!.04) and resisted flexion and axial rotation significantly better than anterior plating in most cases. Standard deviation of the ROM was greater with anterior than with posterior fixation. There was no significant difference among groups in resistance to failure (p5.74). CONCLUSIONS: Individual pedicle screws are known to outperform lateral mass screws in terms of pullout resistance, but they offered no apparent advantage in terms of construct stability or failure of whole constructs. Larger standard deviations in anterior fixation imply more variability in the quality of fixation. In most loading modes, laminar hooks provided similar stability to lateral mass screws or pedicle screws; caudal laminar hooks are therefore an acceptable alternative posteriorly. Posterior two-level fixation is less variable and slightly more stable than anterior fixation. Combined instrumentation is significantly more stable than either anterior or posterior alone. Ó 2007 Elsevier Inc. All rights reserved.
Keywords:
Biomechanics; Cervical spine; Pedicle screw; Lateral mass screw; Laminar hook
FDA device/drug status: approved for this indication (lateral mass screw, pedicle screw, anterior plate). Instrumentation and funding were provided by Synthes Spine.
1529-9430/07/$ – see front matter Ó 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.spinee.2006.04.015
* Corresponding author. Spinal Biomechanics, Barrow Neurological Institute, St. Joseph’s Hospital & Medical Center, 350 W. Thomas Road, Phoenix, AZ 85013. Tel.: (602) 406-6652; fax: (602) 406-7197. E-mail address: [email protected] (N.R. Crawford) 1 Deceased January 30, 2005.
A. Espinoza-Larios et al. / The Spine Journal 7 (2007) 194–204
Introduction Several methods are available for fixating the cervical spine to treat instability. These techniques include anterior plating, wiring, lateral mass plating, laminar hooks, and, more recently, transpedicular screw fixation connected between levels with plates or rods. Cervical pedicle screw fixation is currently used most commonly outside the United States and appears to be a promising technique that provides the greatest stability of all posterior fixation devices [1,2]. Although the technique is technically demanding, biomechanical studies evaluating pullout strength have shown that pedicle screw fixation is superior to lateral mass plates [1]. In a calf spine model, transpedicular screw fixation is as stable as fixation with an anterior plate and posterior wiring for the treatment of three-column and multilevel instability [2]. The application of this information to the human spine, however, is limited. The calf spine has larger vertebral bodies, more sagittally oriented facets, and larger pedicles and laminae than the human spine. Our institution previously compared lateral mass fixation with pedicle fixation in a three-column injury model using nonlocking plates. In this study, pedicular fixation in a two-level construct failed to improve rigidity [3]. This finding was unexpected and was suspected to be attributable to the use of a nonlocking plate construct (Medtronic Sofamor Danek, Memphis, TN). An option for posterior cervical fixation that is less challenging than either lateral mass screws or pedicle screws is laminar hook fixation. Little information is available on the stability offered by laminar hooks, which need to be compared directly with screw fixation to understand the stability offered by this device. Through in vitro testing of human cadaveric spines, this study compared the relative stability of a two-level, threecolumn injury (C5–C7) fixated posteriorly with three different screw-rod or screw-hook-rod combinations, anteriorly with a plate, and with combined anterior and posterior instrumentation. The following hypotheses were tested: 1) The order of decreasing stability of the posterior instrumentation is the same as the order of decreasing technical difficulty in placing the hardware (pedicle screwOlateral mass screwOlaminar hook). 2) The addition of anterior instrumentation will decrease the magnitude of any differences observed with posterior instrumentation alone. 3) Posterior techniques of any of the three types studied will outperform anterior plating alone.
Materials and methods Cadaveric specimen preparation and fixation Twenty-one specimens (15 males, 6 females; mean age, 52.3 years; range, 31–66 years) were used in the study. No specimens had a bony malformation or anomaly that
195
indicated a lesion of the ligaments or discs. All specimens were obtained fresh-frozen and thawed in a bath of normal saline at 30 C. All muscles were removed, leaving the ligaments, discs, and joint capsules intact. After the specimen was prepared, three household wood screws were inserted into the exposed articulations of C4 and T1. The screw heads were potted in cylindrical metal fixtures with polymethylmethacrylate at each end. The caudal end was attached to the base of the testing apparatus. The bone mineral density (BMD) of the C5 vertebral body was measured using dual-energy X-ray absorptiometry. Specimens were separated into three groups of seven: Group 1: mean BMD51.1360.16 g/cm2, mean age556 years; Group 2: mean BMD51.0960.27 g/cm2, mean age551 years; Group 3: mean BMD51.0160.24 g/cm2, mean age552 years. One-way analysis of variance (ANOVA) indicated no significant difference in BMD (p5.63) or age (p5.65) among the groups. Specimens were tested in the following five conditions: 1) normal; 2) destabilized; 3) posterior construct; 4) combined construct (posterior hardware and anterior plating); and 5) anterior plating alone. The posterior construct varied by group: Group 1 received lateral mass screws in C5 and C6 and pedicle screws in C7, interconnected by rods (Fig. 1A); Group 2 received lateral mass screws in C5 and C6 and laminar hooks in C7 (Fig. 1B); and Group 3 received pedicle screws from C5 through C7 (Fig. 1C). During testing, specimens were wrapped in saline-soaked gauze to prevent dehydration. Each specimen required 3 to 4 days to complete testing under each condition. Specimens were refrozen at the end of each testing day to preserve their mechanical properties. Surgical procedures After undergoing the normal flexibility test, specimens were destabilized with a simulated three-column injury. Three-column destabilization consisted of incision of the anterior longitudinal ligament, supraspinous ligament, interspinous ligaments, facets capsules, ligamentum flavum, and posterior longitudinal ligament and complete transection of the annulus fibrosus and nucleus pulposus at C5– C6 and C6–C7. Only the fibrous tissues around the uncinate process were left intact. This destabilization left C6 as a ‘‘floating vertebra,’’ which is described as a distractive flexion injury of Grade 4 [4]. The posterior hardware was then attached according to the group. In Groups 1 and 2, lateral mass screws (Starlock, Synthes Spine, Paoli, PA) were placed according to the Magerl technique [5]. Screw pilot holes were drilled using a 2-mmdiameter bit. The screws had a space in the head to receive an interconnecting rod and clamp. The screw-clamp-rod interface became rigid once the nut was tightened. Screws (18 mm long, 3.5 mm diameter) were inserted bicortically. In Groups 1 and 3, pedicle screws (Starlock, Synthes Spine) were placed by following anatomical landmarks.
196
A. Espinoza-Larios et al. / The Spine Journal 7 (2007) 194–204
Fig. 1. Photographs of the three posterior fixation methods used for this study. (A) Group 1: lateral mass screws in C5 and C6, pedicle screws in C7. Note the foraminotomy ‘‘windows’’ in C7 for the surgeon to determine the appropriate trajectory of the pedicle screw. (B) Group 2: lateral mass screws in C5 and C6, and hooks at C7. (C) Group 3: pedicle screws in all the levels. Note foraminotomy windows at each level.
The superior, medial, and inferior walls of the pedicle were palpated by bilateral foraminolaminectomy [6]. Screw pilot holes were drilled using a 2-mm-diameter bit. Screws (26 mm long, 3.5 mm diameter) were inserted to at least the middle of the vertebral body. After insertion, the specimen was inspected carefully to determine whether the screws had violated the pedicle walls. The hooks (Starlock, Synthes Spine) were placed at C7 in Group 2 after excess tissue was removed caudal to the lamina of C7, particularly tissue close to midline such as the excess of ligamentum flavum at this point. Epidural fat was left intact. If there was inadequate space between the adjacent laminae, a small portion of the rostral lamina
of T1 was removed. This resection provided space to attach the upgoing hook (22 mm long from rod attachment end to the most medial point on the curved end, 6 mm wide) and to make the lamina conform to the hook (Fig. 1B). The curved portion of the hook was concave at one end and was 5 mm at its widest point and 4 mm at its narrowest point. The curved end had an angle of 45 degrees to receive the inferior border of the lamina. The other end was higher (7 mm high) and possessed a cylindrical channel where the rod was received and tightened with a set screw. Although only one size hook was available, the size and shape of the hook were adequate for placement in all cases and in no specimen was the hook loose. Connecting rods required
A. Espinoza-Larios et al. / The Spine Journal 7 (2007) 194–204
more angulation in smaller specimens than in bigger specimens to correctly position rods relative to hooks. Titanium rods 3.5 mm in diameter (Starlock, Synthes Spine) were used for all posterior fixations. Their length was determined by the dimensions of each specimen. Overall, Group 1 required a 37-mm rod, Group 2 required a 50mm rod, and Group 3 required a 42-mm rod. Before the rods were attached, they were bent to match the lordotic curvature of each specimen using a rod bender (included in standard tools). Once rods were inserted through the connectors on each screw or hook, the screw heads were compressed toward each other using a compressor tool (included in standard tools). Compression was applied to mimic the procedure often used intraoperatively to improve the environment for fusion across abutting bone masses, and may have slightly increased lordosis across the construct, although curvature was not documented. Compression was moderate, representative of the pressure applied surgically under ‘‘two-finger’’ tightening. While compression was maintained, the small screws located at the top of each clamp were tightened to lock the rods to the connectors. Next, the nut in the head of each screw was tightened while countertorque was applied. This process tightened the connectors to the screws. Anterior interbody grafting and plate fixation were performed at C5–C6 and C6–C7 using standard techniques. The discs were excised and the end plates were superficially flattened using a pneumatic high-speed drill (Medtronic Midas Rex, Fort Worth, TX) before the structural wedge grafts were placed. Wedge grafts were cut from human fibular allograft from another source. Plates (Atlantis, Medtronic Sofamor Danek, Memphis, TN) ranged from 42.5 mm to 50 mm long. Cancellous locking screws (14 mm long, 3.5 mm in diameter) were used to attach plates unicortically.
Biomechanical testing Specimens were tested nondestructively using a standard flexibility testing method [7]. The fixture holding the caudal vertebra was attached to the base of the testing apparatus while loads were applied to the rostral fixture. Nondestructive nonconstraining pure moment loading was applied to each specimen through a system of cables and pulleys in conjunction with a standard servohydraulic test system (MTS, Minneapolis, MN), as described previously [8]. Loads were applied about the appropriate anatomic axes to induce six different types of motion in the following order: flexion, extension, right axial rotation, left axial rotation, right lateral bending, and left lateral bending. For each type of load, three preconditioning cycles were applied at 1.5 Nm for 60 seconds before data were collected. After the third preconditioning cycle, the specimen was allowed to recoil for 60 seconds at zero load. During the data-collection phase, loads were applied in 0.25-Nm increments
197
to a maximum 1.5 Nm. Each load was held for 45 seconds. Data were collected at 2 Hz. Three-dimensional specimen motion in response to the loads was determined using the Optotrak 3020 system (Northern Digital, Waterloo, Ontario, Canada). This system stereophotogrammetrically measures the three-dimensional displacement of infrared-emitting markers rigidly attached in a noncollinear arrangement to each vertebra. Marker position was related to the x (lateral), y (rostrocaudal), and z (anteroposterior) axes of the specimens by identifying anatomic landmarks with a digitizing probe (accessory to Optotrak) and custom software [9]. Custom software also converted the position coordinates to angles about each anatomic axis using a method that models the vertebrae as stacked cylinders [10]. After all nondestructive testing was completed for a specimen, all hardware was removed and the specimen was refrozen. Hardware was reused in subsequent specimens. At a later date, when enough hardware was available, all specimens were thawed and new posterior instrumentation (screws, hooks, and newly contoured rods) was inserted for failure testing (Fig. 2). Specimens were mounted rigidly to side rails and secured in two angle vises, oriented horizontally (prone). Loops of wire were attached to the rostral and caudal ends of the construct. The wire loops were attached to the piston of the servohydraulic test frame using a linkage that does not constrain rotation (rodend). The linkage therefore allowed equal tension to be applied to both loops throughout testing. The construct was pulled out at a constant displacement rate of 5 mm/minute until visible complete failure occurred while monitoring the applied tension at 2 Hz. The first point of failure was also noted. From the quasistatic load-deformation data, the angular range of motion (ROM) and lax zone (LZ) were determined. The LZ is similar to the neutral zone and is the portion of the ROM in which there is minimal ligamentous resistance [11]. The upper boundary of the LZ was calculated by extrapolating the load-deformation slope at data points corresponding to 0.75 Nm, 1.00 Nm, 1.25 Nm, and 1.50 Nm to zero load using the method of least squares. Larger values of LZ or ROM indicate greater instability. The LZ and ROM were quantified at both levels within the construct (C5–C6 and C6–C7), and the percent contribution of each level to the motion across the entire construct was quantified. From load-to-failure data, the ultimate strength was calculated and the mode of failure was tabulated. One-way ANOVA followed by the Holm-Sidak test was used for statistical assessment of the differences among study groups in LZ, ROM, distribution of LZ and ROM between levels of the construct, and failure load. One-way repeated-measures ANOVA followed by the Holm-Sidak test was used to compare LZ and ROM across conditions. P values less than 0.05 were considered significant.
198
A. Espinoza-Larios et al. / The Spine Journal 7 (2007) 194–204
Results Normal and destabilized specimens In the normal condition, averaged for all three groups, the mean unilateral ROM (6 standard deviation) across C5–C7 was 11.862.7 during flexion, 10.562.6 during extension, 9.063.2 during lateral bending, and 7.662.6 during axial rotation. The mean unilateral LZ across C5– C7 was 6.762.3 during flexion or extension, 6.162.7 during lateral bending, and 4.761.9 during axial rotation. There was no significant difference among the three groups in ROM (pO.46) or LZ (pO.37) during any mode of loading. Averaged for all three groups, destabilization increased ROM across C5–C7 by 67631% during flexion, by 85662% during extension, by 33618% during lateral bending, and by 45623% during axial rotation. Destabilization increased LZ across C5–C7 by 136685% during flexion-extension, by 61649% during lateral bending, and by 67636% during axial rotation. After destabilization, the three groups remained equal, with no significant differences in ROM (pO.82) or LZ (pO.88) during any mode of loading. Instrumented flexibility tests Posterior instrumentation reduced ROM to an average of 17% of normal and reduced LZ to an average of 3% of normal. Posterior instrumentation of all three types limited lateral bending and axial rotation better than it limited flexion or extension. There were no significant differences in the ROM or LZ allowed among the three types of posterior instrumentation (Figs. 3 and 4, Tables 1 and 2). The ROM was distributed with a significantly greater percentage of the total motion at C5–C6 in Groups 1 and 3 than in Group 2 during flexion (p!.014), lateral bending (p!.026), and axial rotation (p!.008; Fig. 5A). During extension, the distribution was equivalent in all three groups (p5.635). The LZ was distributed with a significantly greater percentage of the total motion at C5–C6 in Group 3 than in Group 2 during flexion-extension (p5.008; Fig. 5B). Distributions were equivalent among groups during lateral bending (p5.374) and axial rotation (p5.234). Combined instrumentation further reduced ROM to an average of 4% of normal and reduced LZ to an average of 1% of normal. Groups 1 and 2 allowed significantly smaller ROM during axial rotation than Group 3 (pedicle screws only; Fig. 3, Table 1). There were no significant differences in the LZ allowed by the three combined constructs (Fig. 4, Table 2). The ROM with combined hardware in place was significantly smaller than the ROM with posterior hardware alone in place during flexion and extension (all groups, Table 3). When posterior hardware was disconnected leaving only anterior plates, the ROM remained at an average of 31% of normal and the LZ remained at an average of 17% of
normal. There were no significant differences among groups in ROM or LZ (Figs. 3 and 4, Tables 1 and 2). Unlike posterior instrumentation, which limited ROM and LZ better during lateral bending and axial rotation, anterior instrumentation provided approximately equivalent limitation of motion in all directions (Figs. 3 and 4, Tables 1 and 2). During most loading modes in all three groups, the ROM and LZ with anterior hardware alone in place were significantly larger than the ROM and LZ with combined hardware in place (Table 3). In most loading modes, anterior instrumentation allowed a significantly greater LZ and ROM than posterior instrumentation (Table 3).
Failure loading Of the 21 specimens, 19 were loaded to failure with posterior instrumentation only. Two specimens (one from Group 1 and one from Group 3) could not be loaded to failure because a screw had broken off in the bone before failure testing and could not be retrieved. The ultimate strength among groups was not significantly different (Fig. 6; p5.74). Specimens were observed carefully to determine the first point of failure (Table 4). In the combined lateral mass and pedicle screw group (Group 1), failure occurred with equal frequency at rostral and caudal ends of the construct. The first point of failure was usually the rostral (C5)
Fig. 2. Construct pull-off testing was performed with specimens mounted prone with rigidly locked distal ends in side rails. A rod-end attached to two loops of wire allowed both the rostral and caudal ends of the construct to be evenly loaded during pull-off.
A. Espinoza-Larios et al. / The Spine Journal 7 (2007) 194–204
199
Fig. 3. Graphs showing the range of motion summed across C5–C7 in Groups 1, 2, and 3 during each loading mode and in each instrumented condition studied. Error bars show standard deviation. LMS5lateral mass screw; PS5pedicle screw.
screws in the hook group (Group 2). In the pedicle-screwonly group (Group 3), failure usually occurred first at the caudal level (C7).
Discussion Instability model and restabilization models Our three-column discoligamentous injury model simulated a severe flexion-distraction type lower cervical spine
trauma [4,12]. The severity of the injury model should be considered when comparing our results with other published studies that simulated different severities of injury [2,13]. We chose a severe injury model because we believed it would demonstrate the differences among fixation techniques more effectively than if normal ligamentous structures were allowed to contribute to construct stability. Although we inserted lateral mass screws with bicortical purchase, some surgeons prefer unicortical lateral mass screw insertion. We chose the bicortical technique because
Fig. 4. Graphs showing the lax zone summed across C5–C7 in Groups 1, 2, and 3 during each loading mode and in each instrumented condition studied. Error bars show standard deviation. LMS5lateral mass screw; PS5pedicle screw.
200
A. Espinoza-Larios et al. / The Spine Journal 7 (2007) 194–204
Table 1 Comparisons of C5–C7 ROM among groups during different conditions Flexion Condition
Comparison (Groups)
Posterior
1 1 2 1 1 2 1 1 2
Combined
Anterior
vs. vs. vs. vs. vs. vs. vs. vs. vs.
2 3 3 2 3 3 2 3 3
Extension
Lateral bending
Axial rotation
D Angle
p value
D Angle
p value
D Angle
p value
D Angle
p value
0.11 0.51 0.62 0.06 0.15 0.10 0.36 0.48 0.13
(0.23) (0.23) (0.23) (0.51) (0.51) (0.51) (0.87) (0.87) (0.87)
0.18 0.59 0.77 0.01 0.20 0.19 0.09 0.15 0.23
(0.33) (0.33) (0.33) (0.50) (0.50) (0.50) (0.97) (0.97) (0.97)
0.53 0.23 0.30 0.08 0.10 0.02 0.00 1.51 1.51
(0.22) (0.22) (0.22) (0.51) (0.51) (0.51) (0.26) (0.26) (0.26)
0.43 0.23 0.20 0.01 L0.20 L0.19 0.01 1.12 1.13
(0.12) (0.12) (0.12) (0.86) (0.01) (0.02) (0.36) (0.36) (0.36)
ROM5range of motion. For comparison ‘‘x vs. y,’’ D Angle5ROMx – ROMy (in degrees). Differences in boldface are statistically significant (p!.05).
it is reportedly stiffer with a higher resistance to screw pullout [14,15]. We wanted to compare the strongest possible technique of lateral mass screw fixation to pedicle screw fixation. Although unicortical lateral mass screws are less resistant to pullout, further study is needed to determine whether the stability that they offer when used within a construct differs from that of bicortical screws. Seemingly, a biomechanical comparison of lateral mass screws, pedicle screws, and laminar hooks would be more effective if different groups of specimens had each device placed at all three vertebrae. Although Group 3 had pedicle screws at all levels, Group 1 had lateral mass screws only at C5 and C6 and Group 2 had laminar hooks only at C7. These constructs were chosen because they are realistic clinically. Anatomically, C7 is ill suited for lateral mass screws because its lateral masses are narrow and have a steep slope. In contrast, C7 is typically the level of the cervical spine best suited for pedicle screws because the diameter of its pedicle is large. By similar reasoning, hooks were applied only at C7 with lateral mass screws at C5 and C6. Clinically, a hook-only construct would be avoided because it would be expected to dislodge easily. Because the anatomy of C7 is unsuitable for lateral mass screws,
a surgeon might realistically choose laminar hooks rather than the surgically more challenging procedure of placing pedicle screws.
Comparison of posterior instrumentation One of our hypotheses was that the order of decreasing stability would match the order of decreasing technical difficulty of placing the construct. In other words, we expected that specimens with pedicle-screw-only fixation (Group 3) would be most stable, specimens with combined lateral mass screws and pedicle screws (Group 1) would be less stable, and specimens with combined lateral mass screws and laminar hooks (Group 2) would be least stable. This hypothesis, however, was not supported. Instead, no significant differences were observed in LZ or ROM in the various loading modes among the groups. The only exception was that Group 3, presumed to be the most stable, was actually less stable with a combined construct during axial rotation than either Group 1 or Group 2 (Table 1). Compared with the other hardware types, there were no significant differences in ROM and LZ in the group with laminar hooks
Table 2 Comparisons of C5–C7 LZ among groups during different conditions Flexion-Extension Condition
Comparison (Groups)
Posterior
1 1 2 1 1 2 1 1 2
Combined
Anterior
vs. vs. vs. vs. vs. vs. vs. vs. vs.
2 3 3 2 3 3 2 3 3
Lateral bending
Axial rotation
D Angle
p value
D Angle
p value
D Angle
p value
0.02 0.38 0.36 0.03 0.28 0.25 0.08 0.82 0.90
(0.53) (0.53) (0.53) (0.30) (0.30) (0.30) (0.68) (0.68) (0.68)
0.61 0.22 0.39 0.09 0.09 0.00 0.23 2.17 1.94
(0.30) (0.30) (0.30) (0.61) (0.61) (0.61) (0.19) (0.19) (0.19)
0.08 0.04 0.12 0.00 0.06 0.06 0.02 1.43 1.44
(0.13) (0.13) (0.13) (0.32) (0.32) (0.32) (0.30) (0.30) (0.30)
LZ5lax zone. For comparison ‘‘x vs. y,’’ D Angle5LZx – LZy (in degrees).
A. Espinoza-Larios et al. / The Spine Journal 7 (2007) 194–204
201
(Tables 1 and 2). Hooks consistently performed as well or better than screws in resisting flexion and extension. Another way to compare the different posterior techniques is to study the percentage of overall motion resisted at each level within the construct. In Group 1, if more of the C5–C7 ROM is limited at C6–C7, where pedicle screws and lateral mass screws are used, than at C5–C6, where only lateral mass screws are used, then it could be said that pedicle screws provide better stability than lateral mass screws. However, the distribution of ROM at each level in Group 1 closely matched the distribution in Group 3 (Fig. 5A), implying that pedicle screws and lateral mass screws were equivalent. In Group 2, however, a significantly less symmetrical distribution of ROM between levels was observed than in the other two groups. During flexion, lateral bending, and axial rotation (Fig. 5A), more of the C5– C7 motion was allowed at C6–C7 (lateral mass screws and laminar hooks) than at C5–C6 (lateral mass screws only). This finding indicates that laminar hooks contributed significantly less stability than lateral mass screws during these loading modes. Quantitative findings from the load-to-failure tests indicated that ultimate strength (Fig. 6) did not differ significantly among the three groups. Qualitatively, we expected lateral mass screws (at C5) to pull out earlier than pedicle screws (at C7) in Group 1. Instead, C5 failed first in as many cases as C7 (3 each, Table 4). One important finding from Group 2 (lateral mass screws at C5, laminar hooks at C7) was that screws always failed first. In no case did hooks disengage or fail before screws. In one case, the hooks caused the C7 lamina to fracture. Intuitively, this finding seems reasonable. It should be easier to pull a screw out of a hole than to pull a hook through an entire cross-section of bone. Anterior versus posterior hardware We predicted that any of the three posterior techniques would outperform anterior plating. Posterior instrumentation outperformed anterior instrumentation during flexion, lateral bending, and axial rotation, but not during extension. This finding is consistent with expectations based on the design of each device. In resisting flexion, posterior hardware serves as both a tension band and a cantilever. Anterior hardware can serve only as a cantilever. The opposite is true during extension: the anterior plate acts as both a cantilever and a tension band whereas the posterior hardware acts only as a cantilever. Theoretically, with good compression before hardware is tightened, a tension band would be expected to be more effective in limiting motion than a cantilever. During lateral bending and axial rotation, both anterior and posterior devices behaved similarly as cantilevers. However, the anterior plate was much closer to the axis of rotation whereas the lateral mass and pedicle screws were positioned farther laterally to each side of the axis of rotation. Furthermore, the two points of fixation at each level (left and right screws) were positioned farther apart in lateral mass and
Fig. 5. Distribution of the (A) range of motion and (B) lax zone by level in specimens instrumented with posterior hardware. The percentage above the horizontal line is the contribution of C5–C6 to the total (C5–C7) lax zone or range of motion across the construct. The percentage below the horizontal line is the contribution of C6–C7. Asterisks indicate significant differences (p!.05) between groups in the percent contribution of C5–C6 to the total motion across the construct.
pedicle screw fixation than in anterior plate fixation. Therefore, more leverage was available in the posterior constructs to resist lateral bending and axial rotation. Left and right screw separation is unimportant in resisting flexion and extension. Compared with posterior hardware, anterior hardware appeared to be less consistent in the quality of fixation achieved. That is, the standard deviations were larger for specimens with anterior-only hardware in averaged biomechanical parameters than in the corresponding parameters in the same specimens with posterior-only hardware (Figs. 3 and 4). This finding was observed because some specimens showed very good resistance to motion with anterior-only hardware and others showed poor resistance to motion with the same hardware. The quality of vertebral
202
A. Espinoza-Larios et al. / The Spine Journal 7 (2007) 194–204
Table 3 P values from statistical comparisons of ROM and LZ among conditions (ANOVA/Holm-Sidak) Posterior vs. Combined Loading mode ROM Flexion Extension Lateral bending Axial rotation LZ Flexion-Extension Lateral bending Axial rotation
Anterior vs. Combined
Anterior vs. Posterior
Group 1
Group 2
Group 3
Group 1
Group 2
Group 3
Group 1
Group 2
Group 3
0.031 0.001 0.573 0.122
0.026 0.001 0.097 0.062
0.041 0.001 0.767 0.584
0.001 0.001 0.001 0.001
0.001 0.001 0.001 0.001
0.001 0.002 0.002 0.002
0.007 0.652 0.001 0.001
0.001 0.524 0.001 0.031
0.036 0.790 0.005 0.008
0.317 0.784 0.695
0.257 0.163 0.120
0.094 0.887 0.057
0.001 0.001 0.001
0.001 0.001 0.120
0.094 0.012 0.057
0.012 0.001 0.003
0.007 0.022 0.120
0.094 0.016 0.057
ANOVA5analysis of variance; LZ5lax zone; ROM5range of motion. Values in boldface are statistically significant (p!.05).
body bone seemed to be more variable than the quality of lateral mass, lamina, and pedicle bone. Anterior screw fixation was also unicortical whereas posterior screw fixation was bicortical. Combined hardware Finally, we hypothesized that the addition of anterior instrumentation (converting a posterior-only construct to a combined anterior-posterior construct) would decrease the magnitude of any differences observed with posterior
instrumentation alone. There was little opportunity to support this hypothesis because there were few differences among posterior instrumentation groups. In fact, the opposite was true for some testing parameters. For example, with posterior instrumentation only, no difference was observed among groups in the ROM during axial rotation (Fig. 3, Table 1). However, with combined instrumentation, Group 3 allowed significantly more axial rotation ROM than Group 1 or Group 2. The absolute magnitudes of the differences with combined hardware were very small (less than 0.21 across C5–C7), but the data were more consistent than posterioronly data, leading to a statistical significance. The anterior-only condition was studied last after posterior instrumentation was removed. Theoretically, the anterior-only condition might have been expected to be equivalent among groups. In many modes, however, ROM was greatest in Group 3 with anterior-only instrumentation. Therefore, the keyhole foraminotomies for pedicle screw placement or some other aspect of the earlier Group 3 posterior surgery may have left the specimens in a less stable condition than specimens in Group 1 or Group 2.
Comparison to previous literature
Fig. 6. Ultimate strength (maximum load endured) during load to failure. There was no significant difference in strength among groups (p5.74). Error bars show standard deviation.
Kotani et al. [2] demonstrated the stabilizing capabilities of transpedicular screw fixation and combined anterior and posterior fixation in a two-level, three-column instability model under torsional and extension loading. Similarly, in our model in all three groups with combined instrumentation, ROM was significantly decreased from the intact condition during every loading mode. Kotani et al. [2] also reported that during flexion testing, anterior plate fixation was unable to reduce angular displacement below the intact level under a three-column instability condition. Instead, we found that anterior plate fixation reduced motion during all loading modes in all three groups relative to normal. Most of these reductions were statistically significant. However, we did not perform corpectomies in our injury model and we used a human cadaveric model instead of a calf spine model.
A. Espinoza-Larios et al. / The Spine Journal 7 (2007) 194–204
203
Table 4 Load-to-failure qualitative summary Group 1 (LMS, LMS, PS)
Group 2 (LMS, LMS, Hook)
Group 3 (PS, PS, PS)
Spec.
1st point of failure
Spec.
1st point of failure
Spec.
1st point of failure
1a 1b 1c 1d 1e 1f
C5 C7 C5 C5 C7 C7
2a 2b 2c 2d 2e 2f 2g
C7 lamina fracture C5 L screw pullout C5 L screw pullout C5 L screw pullout C5 R & L screw pullout Ligamentous failure C5 L screw pullout
3a 3b 3c 3d 3e 3f
C7 R & L screw pullout C7 R & L screw pullout C5 R & L screw pullout C7 R & L screw pullout Ligamentous failure C7 R & L screw pullout
R screw pullout R & L screw pullout R screw pullout R screw pullout L screw pullout L screw pullout
L5left; LMS5lateral mass screw; PS5pedicle screw; R5right.
Bozkus et al. [12] studied similar hardware by a different manufacturer in a similar injury model using identical testing methods. However, hooks were not studied and posterior screws were inserted by image guidance instead of using keyhole foraminotomies. Their results were highly consistent with the current findings. They also observed that anterior plating provided significantly less stability than posterior hardware; combined hardware provided significantly better stability than anterior or posterior alone; and lateral mass screws performed equivalently to pedicle screws. Koh et al. [13] studied human cadaveric spines instrumented after simulated flexion-distraction injury or burst fracture. They compared an anterior plate construct, a posterior lateral mass screw-plate construct, and a combined construct. In their flexion-distraction model, as in our study, posterior plating constructs with and without anterior fixation significantly reduced flexion, lateral bending, and axial rotation motion compared with that of the intact specimen. During extension, motion decreased significantly from the intact case only with the combined construct. However, anterior plate fixation did not reduce ROM from the intact condition in any mode in their study. In contrast, in our two-level three-column injury model, anterior fixation, posterior fixation, and combined fixation significantly reduced the C5–C7 ROM from the intact condition during all loading modes in all groups except Group 3 anterior fixation during axial rotation. The differences likely occurred because both injury models used by Koh et al. were dissimilar to the injury induced in our study. Adams et al. [16] used a two-level (C4–C5 and C5–C6) three-column in vitro injury to compare the stability offered by anterior plating alone to that offered by combined anterior plating and posterior lateral mass plating. Similar to our results, they found that the two-level anterior plate alone stabilized the spine moderately well (significantly within the ROM of intact specimens). However, the addition of posterior plating further significantly reduced motion during all loading modes. They used the older style posterior nonlocking lateral mass screw-plates, which seemingly would have provided less stability than the rigid system that we studied. Kothe et al. [17] found that three-level injuries stabilized with pedicle screws had a significantly smaller ROM during
lateral bending than constructs stabilized with lateral mass screws. As in the current study, lateral mass screws were inserted bicortically using the Magerl technique. The different findings of our study and that of Kothe may be attributable to their usage of constructs with only one type of screw (not mixed lateral mass and pedicle screws), longer constructs, fewer specimens studied (n54 per group), and usage of image guidance instead of keyhole foraminotomies for pedicle screw insertion. They also found that the pedicle screw construct loosened less with fatigue than the lateral mass screw construct. We did not investigate the effect of fatigue. The study of Schmidt et al. [18] focused on comparing rigid versus nonconstrained posterior systems in a one-level corpectomy model. Although their injury and length of construct were different, the results were consistent with our findings, showing no benefit of pedicle screws over lateral mass screws when used with a rigidly constrained posterior screw-rod system. There was little difference in ultimate strength during failure among the groups with posterior instrumentation only. These findings are inconsistent with those of Jones et al. [1], who showed that pedicle screws pulled out at a significantly higher load than lateral mass screws. Apparently, pullout resistance of individual screws is less important than the angulation and rigid interlocking of an entire posterior construct when considering failure. Clinical considerations We observed no significant difference in stability during any loading mode in specimens incorporating lateral mass screws at two of three vertebrae compared with specimens incorporating pedicle screws at all three vertebrae. Assuming that the amount of immobilization of a construct is directly related to the quality of the environment for fusion, this finding implies that both systems provide equivalent environments for fusion. The two groups also resisted failure equivalently. Together, these findings imply that there is little biomechanical difference between the two fixation types in the cervical spine. Clinically, therefore, the increased biomechanical stability of pedicle screw fixation may not offset the increased risks of the procedure.
204
A. Espinoza-Larios et al. / The Spine Journal 7 (2007) 194–204
A further clinical consideration in choosing pedicle screws or lateral mass screws is that the screw heads are positioned differently. In Group 1 specimens, the heads of the lateral mass screws at C5 and C6 were more medial than the heads of the pedicle screws at C7 (Fig. 1A). Rods must be contoured more in constructs using mixed screw types to create a construct in which the vertebrae are oriented naturally. In contrast, the heads of the screws in Group 2 and Group 3 specimens were better aligned and only required contouring to match the cervical lordosis (Fig. 1B). Laminar hooks appear to be a good alternative to pedicle screws at C7. Overall, they stabilized as well as screws. However, within the Group 2 construct, a greater percentage of the motion was allowed at C6–C7. Hooks were highly resistant to failure. Technically, we found laminar hooks to be time-consuming to attach at C7; bone must be removed carefully to obtain a good interface. Study limitations Our in vitro experimental testing excluded the effects of muscular co-contraction or the weight of the head, which would be assumed to increase stability. Therefore, our results can be considered a worst-case scenario, where only hardware, bones, and ligaments are able to limit motion. Greater stability would therefore be expected in vivo. Also, our injury model was intended to mimic a specific threecolumn injury commonly seen clinically. The results of this study therefore should not be generalized to other injury patterns of the lower cervical spine. In all specimens, posterior instrumentation was tested first, combined instrumentation second, and anterior instrumentation last. The testing order is typically randomized to prevent bias related to specimen degradation. However, we could not randomize the testing order because the anterior instrumentation required disc removal and bone graft placement. Therefore, the poorer performance of the anterior construct could have reflected deterioration of the specimen through the course of testing. A different study design using separate groups of specimens would be needed to determine the magnitude of such an effect. Another uncontrolled factor possibly at work in the degree of stabilization provided by anterior versus posterior instrumentation is the amount of lordosis present. Facet interaction would presumably have been greater in more lordotic specimens, and these specimens could therefore have benefited more from supplementary hardware (anterior or posterior). However, lordosis was not recorded. Further study would be needed to elucidate the importance of lordosis on fixation stiffness using the hardware studied. We investigated immediate postoperative stability and load required for failure. However, we did not study the fatigue endurance of the different cervical fixation constructs. Our study therefore does not address which type of construct is more likely to break with normal usage over time,
which is also a worthwhile consideration when comparing spinal fixation hardware. References [1] Jones EL, Heller JG, Silcox DH, Hutton WC. Cervical pedicle screws versus lateral mass screws: anatomic feasibility and biomechanical comparison. Spine 1997;22:977–82. [2] Kotani Y, Cunningham BW, Abumi K, McAfee PC. Biomechanical analysis of cervical stabilization systems: an assessment of transpedicular screw fixation in the cervical spine. Spine 1994;19:2529–39. [3] Bartolomei JC, Das K, Crawford NR, et al. Cervical pedicle screws/plates or lateral mass screws/plates combined with anterior plating: a biomechanical in vitro study (abstract). 71st Annual Meeting of the American Association of Neurological Surgeons; April 6–11, 2002; Chicago, IL. [4] Allen BL Jr. Ferguson RL, Lehmann TR, O’Brien RP. A mechanistic classification of closed, indirect fractures and dislocations of the lower cervical spine. Spine 1982;7:1–27. [5] Jeanneret B, Gebhard JS, Magerl F. Transpedicular screw fixation of articular mass fracture-separation: results of an anatomical study and operative technique. J Spinal Disord 1994;7:222–9. [6] Abumi K, Itoh H, Taneichi H, Kaneda K. Transpedicular screw fixation for traumatic lesions of the middle and lower cervical spine: description of the techniques and preliminary report. J Spinal Disord 1994;7:19–28. [7] Panjabi MM. Biomechanical evaluation of spinal fixation devices: I. A conceptual framework. Spine 1988;13:1129–34. [8] Crawford NR, Brantley AG, Dickman CA, Koeneman EJ. An apparatus for applying pure nonconstraining moments to spine segments in vitro. Spine 1995;20:2097–100. [9] Crawford NR, Dickman CA. Construction of local vertebral coordinate systems using a digitizing probe. Technical note. Spine 1997;22:559–63. [10] Crawford NR, Yamaguchi GT, Dickman CA. A new technique for determining 3-D joint angles: the tilt/twist method. Clin Biomech (Bristol, Avon) 1999;14:153–65. [11] Crawford NR, Peles JD, Dickman CA. The spinal lax zone and neutral zone: measurement techniques and parameter comparisons. J Spinal Disord 1998;11:416–29. [12] Bozkus H, Ames CP, Chamberlain RH, et al. Biomechanical analysis of rigid stabilization techniques for three-column injury in the lower cervical spine. Spine 2005;30:915–22. [13] Koh YD, Lim TH, You JW, Eck J, An HS. A biomechanical comparison of modern anterior and posterior plate fixation of the cervical spine. Spine 2001;26:15–21. [14] Errico T, Uhl R, Cooper P, Casar R, McHenry T. Pullout strength comparison of two methods of orienting screw insertion in the lateral masses of the bovine cervical spine. J Spinal Disord 1992;5: 459–63. [15] Montesano PX, Jauch E, Jonsson H Jr. Anatomic and biomechanical study of posterior cervical spine plate arthrodesis: an evaluation of two different techniques of screw placement. J Spinal Disord 1992;5:301–5. [16] Adams MS, Crawford NR, Chamberlain RH, Sonntag VKH, Dickman CA. Biomechanical comparison of anterior cervical plating and combined anterior/lateral mass plating. Spine J 2001;1:166– 70. [17] Kothe R, Ruther W, Schneider E, Linke B. Biomechanical analysis of transpedicular screw fixation in the subaxial cervical spine. Spine 2004;29:1869–75. [18] Schmidt R, Wilke HJ, Claes L, Puhl W, Richter M. Effect of constrained posterior screw and rod systems for primary stability: biomechanical in vitro comparison of various instrumentations in a single-level corpectomy model. Eur Spine J 2005;14:372–80.