Journal of Spine & NeurosurgeryISSN: 2325-9701

All submissions of the EM system will be redirected to Online Manuscript Submission System. Authors are requested to submit articles directly to Online Manuscript Submission System of respective journal.

Research Article, J Spine Neurosurg S Vol: 3 Issue: 0

Long Construct Pedicle Screw Reduction and Residual Forces are Decreased Using a Computer-assisted Rod Bending System

Antoine G Tohmeh1*, Robert E Isaacs2, Zachary A Dooley3 and Alexander WL Turner3
1Northwest Orthopaedic Specialists, Spokane, WA, USA
2Department of Surgery, Division of Neurosurgery, Duke University Medical Center, Durham, NC, USA
3NuVasive Inc., San Diego, CA, USA
Corresponding author : Antoine G Tohmeh
Northwest Orthopaedic Specialists, 212 East Central Avenue, Suite 140, Spokane, WA 99208, USA,
Tel: 509-465-1300
[email protected]
Received: July 16, 2014 Accepted: July 30, 2014 Published: August 04, 2014
Citation: Tohmeh AG, Isaacs RE, Dooley ZA, Turner AWL (2014) Long Construct Pedicle Screw Reduction and Residual Forces are Decreased Using a Computer-assisted Rod Bending System. J Spine Neurosurg S2. doi:10.4172/2325-9701.S2-002


Long Construct Pedicle Screw Reduction and Residual Forces are Decreased Using a Computer-assisted Rod Bending System

Previous biomechanical studies have shown that reduction loads placed on pedicle screws during assembly of the construct, with a rod that does not adequately fit the screw locations, can reduce the strength of the screw-bone interfaces. In this bench top study, axial pedicle screw forces on a unilateral 7-level construct were evaluated for 2 rod bending techniques: manual and computerassisted.

Keywords: Pedicle screw; Rod bending; Computer-assisted surgery; Rod reduction; Screw pullout; Screw loosening


Pedicle screw; Rod bending; Computer-assisted surgery; Rod reduction; Screw pullout; Screw loosening


Minimally invasive pedicle screw stabilization of reconstructive spine surgery procedures is believed to reduce blood loss, decrease paraspinal muscle damage, lower infection risk, and lead to faster return to work compared with conventional open techniques [1,2]. Minimally invasive spinal instrumentation has recently expanded to include complex deformity/multilevel cases [3-6]. Those increasingly complicated cases have the added difficulty in shaping and reducing the long rod into pedicle screws without direct visualization, which could result in inability to capture all screws, need for forceful reduction maneuvers via towers, potential screw loosening or even pull-out, and longer surgery time.
Rod bending is typically performed manually using a French bender prior to insertion. Additional bending may be achieved using in situ benders. The accuracy with which the rod fits the screw locations has not been characterized, however inability to engage a lock screw with the screw tulip is indicative of a gap between the rod and the screw, and reduction of the screws to the rods is then required using additional instruments. Biomechanical studies investigating rod reduction are limited, however Paik et al. [7] showed that reduction loads placed on pedicle screws during assembly of the construct with a rod that does not adequately fit the screw locations can reduce the strength of the screw-bone interface. In that study, a relatively small 5 mm distance between the rod and pedicle screw caused outright screw pullout failure (visible screw displacement from the bone) in 44% of normal and 50% of osteoporotic specimens. They also reported significantly decreased screw pullout strength (495 ± 379 vs. 954 ± 237 N, p<0.05), and significantly decreased work to failure (3204 ± 3356 vs. 5414 ± 2332 N•mm, p<0.05) with reduced screws compared with screws not subject to reduction. Reduction forces for the same screw-rod reduction distance may be increased with stiffer rod materials such as cobalt chrome and stainless steel [8].
By more precisely bending the rod to match the screw locations, it may be possible to reduce both the loads placed on the screwbone interface during construct assembly and the residual postoperative loads. To this end, a computer-assisted rod bending system (Bendini, NuVasive, Inc., San Diego, CA) was developed to provide more exact rod bending. This system may decrease the need for reduction maneuvers by minimizing the screw-rod distances, thereby decreasing the loads placed on the screw-bone interfaces, and potentially reducing likelihood for screw pullout or loosening. To assess the influence of rod bending technique on screw loading, axial pedicle screw forces on a simulated unilateral seven-level construct were measured during rod reduction and final tightening for 2 rod bending methods: (1) manual, and (2) computer-assisted.

Materials and Methods

Test apparatus
A custom test fixture (Figure 1) was created with 8 unilateral pedicle screws, the locations of which were based on a clinical deformity case. The fixture consisted of a large supported rectangular polyacetal block with cylindrical holes. Screw caudo-cephalad spacing was controlled by the hole spacing in the block. Screw height was controlled by adjusting the position of polyacetal (low friction) cylinders which were able to translate through the large block. Pedicle screws (Precept, NuVasive, Inc.) were located in the polyacetal cylinders with percutaneous towers attached. The most caudal screw was fixed while the remaining 7 screws were allowed to translate in the anterior-posterior (A-P) direction, with other linear and rotational displacements constrained (rotational displacements were allowed by the polyaxial screw). Each of the translating screws was attached to a cantilever spring with stiffness of 350 N/mm based on the average A-P shear stiffness of the spine obtained from the literature [9-14] Axial forces at the 8 screws were independently measured via 2250 N load cells (Interface Inc., Scottsdale, AZ) with data acquisition performed via in-line signal conditioners (Interface Inc.), a high-speed 16-bit USB data acquisition board (Measurement Computing Corp., Norton, MA) and a custom software interface (Lab View, National Instruments Corp., Austin, TX).
Figure 1: Test setup: (A) mounting block, (B) A-P translating cylinder, (C) load cell to measure A-P load, (D) cantilever spring to provide A-P shear stiffness, (E) pedicle screw and reduction tower.
Computer-assisted rod bending system description
The computer-assisted rod bending system consists of 4 core components (Figure 2): (i) an infra-red motion tracking camera, (ii) a digitizer with an array of retro-reflective markers, (iii) an NVM5 computer unit, and (iv) a proprietary rod bender. The digitizer is used to locate the screw heads and the positions are captured by the infrared camera. The computer unit generates bend instructions (distance, rotation, bend angle) that are carried out using the rod bender.
Figure 2: Computer-assisted rod bending system components: (A) motion tracking camera and digitizer, (B) NVM5 computer unit, (C) proprietary rod bender, (D) example computer screenshot of digitized screw positions and calculated rod shape fit to screws, (E) example computer screenshot of bend instructions to perform rod contouring.
Radiographs in Figures 3 and 4 were obtained following a minimally invasive deformity case and show clinical use of the computer-assisted rod bending system. The patient was a 74 year old female with thoracolumbar kyphoscoliosis with a 2 year history of back and leg pain treated with minimally invasive interbody fusion and multilevel posterior fixation using the computer-assisted rod bending system.
Figure 3: (A, B) Antero-posterior and lateral pre-operative radiographs showing thoracolumbar kyphoscoliosis including severe L3-4 lateral listhesis, (C) axial MRI showing severe L3-4 stenosis and facet disease. Patient underwent staged thoracolumbar reconstruction using a minimally invasive lateral interbody fusion technique preceded by facet release and minimally invasive L3-4 decompression under the same anesthetic. Second stage surgery was done 4 days later and consisted of minimally invasive TLIF at L5-S1 and minimally invasive posterior Instrumentation from T8 to pelvis including percutaneous S2 alar-iliac screws.
Figure 4: (A, B) Antero-posterior and lateral radiographs of patient in figure 3 at 1 year postoperative showing correction of sagittal and coronal plane deformities. The rod was inserted from cephalad to caudad through a limited thoracic midline incision (T8-T12) and percutaneous lumbar, sacral and iliac screws. The computer-assisted rod bending system facilitated shaping and passing the rod across the lumbosacral junction and capture of the iliac screws percutaneously without resorting to an open iliac technique and connectors. (C) CT scan obtained after 1 year showed solid fusion at all levels without loosening of the hardware.
Test procedure
Seven surgeons first manually bent a rod using a French bender, inserted the rod through the towers, reduced as needed, and locked the rod in place to the torque prescribed by the manufacturer while force data was acquired from each screw (Figure 2). The rod was removed and the test was repeated with the computer-assisted (Bendini) system. Each screw location was first digitized and then the rod was bent using the proprietary rod-bender per computer-generated bend instructions. The rod was inserted, reduced, and locked in place while force data was recorded. For both rod bending methods, the order of screw lockdown was left to the discretion of the surgeon.
Data analysis
The maximum tensile forces during screw reduction/lock down, maximum compressive forces during screw reduction/lock down, and residual forces (absolute values of compressive/ tensile forces at end of test) for each screw were averaged and compared between rod bending methods with a paired t-test, with p<0.05 considered significant. The number of screws where the magnitude of the force during the test was greater than 300 N and 500 N were also assessed and calculated as a proportion of all screws in the experiment. These force thresholds were determined based on cadaveric pullout force data. High magnitudes and frequencies of force peaks are potentially associated with intraoperative reduction of the screw-bone interface strength, while high residual forces are may be associated with postoperative loosening.


Force versus time graphs from a single surgeon with average results for each bending method is presented in Figure 5 to illustrate the differences in screws forces between the bending methods. Screw forces tend to increase and decrease as adjacent screws are reduced or locked down on to the rod. On average, the residual force was 60% lower (p=0.002) for the computer-assisted rod (93 ± 78 N) versus the manually bent rod (233 ± 186 N). The average maximum tensile force was 49% lower with the computer-assisted technique (Bendini: 275 ± 90 N, manual: 537 ± 211 N; p=0.005) and the maximum compressive force was 57% lower (Bendini: 206 ± 37 N, manual: 478 ± 190 N; p=0.005) (Figure 6).
Figure 5: Representative axial force vs. time screw data acquired during testing with a single surgeon: (A) manual rod bending test data, (B) computerassisted rod bending test data. Tensile screw forces are positive; compressive screw forces are negative.
Figure 6: Test results for each rod bending method: (A) average peak screw forces during rod reduction and set screw lock down, (B) average residual forces after set screw lock down, and (C) threshold screw forces during rod reduction and set screw lock down.
The total proportion of screws with load peaks exceeding 500 N was 20% for the manual rod and 0% for the computer-assisted method, while loads greater than 300 N were over seven times more likely (39% vs. 5%) with manual rod bending.


Computer-assisted rod bending with Bendini may improve the repeatability of what has traditionally been an empirical technique. Conventional “open” spinal surgery techniques offer direct visualization of screw heads to guide the surgeon in fashioning a fitting rod. However, in deformity cases this technique not only requires surgeon experience in addressing rod shape in all 3 planes but also may be inaccurate, resulting in significant residual forces beyond what is required for deformity correction. This compels the surgeon to use powerful instruments to be able to seat the rods into the screw heads, potentially increasing stresses along the screwbone interfaces. As minimally invasive techniques have expanded to include multi-planar deformity cases, the requirement for a wellfitting rod has become more prevalent as the use of pre-bent rods in only the sagittal plane no longer circumvents the problem. Proper bends in those cases relies almost purely on intra-operative X-rays and guesswork, thereby requiring powerful reduction instruments and maneuvers to overcome the forces between components, that when finally reduced with the rod seated into the tulip, could stress the screw-bone interface as well as the metal components. Furthermore, it is very difficult to pass a non-fitting rod through multiple-screw constructs, especially in a percutaneous fashion, and if the rod is not appropriately contoured, it may need to be removed, rebent and passed multiple times. In addition to the added time this imparts on the surgery, rebending directly weakens the construct through metal fatigue [15].
In this study, Bendini demonstrated significant reductions in peak axial screw loads during rod reduction and residual screw loads after final tightening by creating a rod that better conformed to the pedicle screw locations than a rod bent manually. The literature is replete with publications describing axial screw pullout forces. For example, Takigawa et al. [16] recorded screw pullout forces of 346 ± 172 N. Halvorson et al. [17] measured pullout forces of 1540 ± 361 N in normal bone and 206 ± 159 N in osteoporotic bone. Jacob et al. [18] compared the maximum pullout force for dual-lead screws (533.9 ± 285.7 N, range 126.7–1210.8 N), with single single-lead screws (524.9 ± 311.6 N, range 88.8-1203.2 N). These studies illustrate a wide variation in the screw pullout forces, likely attributable to individual specimen density and morphometry, and screw threadform geometry and trajectory. These screw pullout forces are also in the range of the forces seen in this study; especially for the manuallybent rods where forces greater than 300 N were recorded in 39% of all screw, compared with only 5% using the computer-assisted system. The relevance of axial screw pullout testing to intraoperative weakening of the screw-bone interface is readily apparent as this loading mode occurs during reduction maneuvers to seat the rod in the screw. In vivo, caudo-cephalad toggling is thought to be the more relevant failure mode for the pedicle screw-bone interface [19]; however, pullout may be relevant to screw loads at the ends of long instrumentation and screws used in the reduction of deformity. In addition, residual axial screw forces may contribute to accelerated screw loosening in vivo after repeated loading during activities of daily living.
Other techniques to reduce screw-bone interface forces have been investigated; however these have focused on the pedicle screws rather than the rod bending. Wang et al. [20] used anatomic computational models to evaluate four different screw types and found the bonescrew loads were significantly different. Screws with more degrees of freedom (both rotational and translational) were found to reduce screw loads and these were recommended for patients with large and stiff spinal deformities and patients with compromised bone quality in order to reduce bone-screw connection failure. Similarly, Driscoll et al. [21] compared polyaxial screws with favored-angle screws that incorporated greater angulation and found the favored-angle screws reduced peak pullout and lateral forces by 27% and 35%, and mean pullout and lateral forces by 48% and 40%. While addressing the rod-screw mismatch issues through novel screw designs has been attempted since the advent of polyaxial instrumentation, ultimately there are inherent limitations in screw formulation that can be used to fix the issue without looking to contend with the rod.
Limitations for the study include the simplified geometry and constraints applied to the pedicle screws. Screws were only allowed to translate perpendicular to the coronal plane. In vivo, each screw would be inserted in a direction that was generally more perpendicular to the curvature of the spine and therefore reduction forces would also be applied in a similar direction. In addition, the natural spine would provide less constraint to the screw positions. In this study, screws were only allowed to translate in the A-P direction while lateral and caudo-cephalad translations were prevented. In vivo, the screws could potentially translate and rotate about other axes to allow reduction of the screws to the rods with resistance provided by the anatomical structures. To minimize the effect of additional constraint, the screw positions were simplified to all be aligned in the sagittal plane which also limited the potential screw-rod mismatch encountered by the surgeons to a single plane problem. The screw constraints used in this study may have resulted in higher screw forces than seen clinically; however, both rod bending methods were subjected to the same conditions and therefore the relative differences appear valid.
In conclusion, the computer-assisted rod-bending system provides the surgeon with an intra-operative global positioning system for screw heads that is independent of extent of exposure, patient size, screw position and severity of deformity. In this study, the computerassisted system produced significantly lower peak and residual screw loads during rod-screw construct assembly than the manuallybent rod. Since forces as low as 300 N can result in screw pullout [8], especially with compromised bone quality, the imprecision of manual rod bending can have significant clinical implications. These include potentially more frequent intra-operative and postoperative screw loosening and pull-out as well as more challenging rod passage, especially in percutaneous techniques. Future advancements in software will potentially offer added benefits in planning deformity correction to optimize sagittal alignment and balance.


Research materials were provided by NuVasive, Inc.


Track Your Manuscript