Biomechanical Testing of Locking and Nonlocking Plates in the Canine Scapula

Introduction

The incidence of scapular fractures in small animal veterinary medicine is low, with reports ranging between 0.5% and 2.4% of all fractures seen in dogs.^1–3 Coaptation is often used in appropriate situations, although unstable scapular body fractures, fractures of the acromion process, scapular neck fractures, and glenoid fractures require internal fixation.^1,4–7

Repair of scapular fractures with Kirshner wires and interfragmentary wires has been described, although internal fixation with plates and screws is also commonly used.^5,6,8 Osteosynthesis with the application of plates and screws can present a challenge because the screws engaging the thin bone of the scapula may not have sufficient holding strength when the plate is compressed to the bone in nonlocking plating methods.^4–7,9 Fractures of the scapular neck and glenoid often require a portion of the plate fixation to engage the supraspinatus or infraspinatus fossa. Mechanical and anatomic studies have been reported to optimize the strength of fixation of the scapula.^6,7,10

Locking plates reportedly have an increased resistance to pullout compared with nonlocking plate systems, particularly in bone of poor quality.^11–15 Additionally, locking plates with divergent screws have an increased resistance to pullout compared with parallel screws.^15–17 Therefore, the use of locking plates with screws placed at divergent angles should increase the resistance to pullout in bone of insufficient quality, such as areas of the scapula. Still, few studies have investigated the pullout strength of locking plate and screw constructs.^16,18 To the authors’ knowledge, no study has investigated biomechanical properties of locking plate constructs in the scapula.

Several studies have shown that bone mineral density correlates linearly with the holding power of screws.^19–21 The holding power of screws is important in nonlocking plates where the plate is compressed to the bone and screw stripping is a mode of failure.^13,22 Increasing the number of screws per fracture segment may diminish the mechanical advantage of locking plate systems in poor quality bone.

The purpose of this study was to compare the pullout strengths of constructs consisting of a 4-hole locking plate system (string of pearls plate [SOP]) and a 4-hole nonlocking plate system (limited contact dynamic compression plate [LC-DCP]), in the suprasinatus fossa of the canine scapula. The null hypothesis was that there would be no difference between the groups. A second objective of this study was to record differences in modes of failure.

Materials and Methods

Plate Application

Eleven scapulae (one from each pair) were assigned to the LC-DCP group and the contralateral scapula of the pair assigned to the SOP group. Plate groups were randomized between left and right scapulae. All plates and screws were applied using standard Association for the Study of Internal Fixation technique. Both the LC-DCP and SOP screws were tightened using three finger maximum torque as previously recommended.⁹ All plates and screws were applied by the same investigator (A.A.).

For the nonlocking plate group, a 6-hole, 2.7 mm, stainless steel LC-DCP^a was centered dorsoventrally and applied to the junction between the scapular spine and the supraspinatus fossa that was previously reported to optimize bone purchase (Figure 1).^4–7 The LC-DCP did not require contouring to sit flush on the bone. The LC-DCP was held against the bone as 2.7 mm conventional cortical bone screws were placed in the second to fifth holes of the plate in a neutral fashion. Screw lengths were chosen to ensure both cortices were fully engaged. The first and sixth screw holes were left open. Screws in the second and fifth holes were placed perpendicular (90°) to the plane of the plate. Screws in the third and fourth holes were offset 7° from the perpendicular using a customized angled drilling guide. The screw in the third hole was 7° caudal and the screw in the fourth hole was 7° cranial to the perpendicular plane of the plate. During application of the LC-DCPs, four segments of 0.6 mm stainless steel wire rope^b were applied under the plates. The wire rope was placed under the noncontact underside of the LC-DCP so that the plate could be compressed to the bone. The position of the plate at the angle of the scapular spine to the scapular blade and the contour of the underside of the LC-DCP allowed the plate to be compressed to the scapula with space between the plate and scapula for passage of the tensioned wires. Two wire segments were applied between the third and fourth screw holes, one segment between the second and third screw hole, and one strip between the fourth and fifth screw hole (Figure 2).

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

The contralateral scapula of the pair was assigned to the locking plate group. A 6-hole, 2.7 mm, locking SOP^c was centered dorsoventrally and applied to the junction between the scapular spine and the supraspinatus fossa (Figure 1). This plate was not contoured, but was held flush against the bone during screw placement. The SOP had four 2.7 mm conventional cortical bone screws placed in the second to fifth screw. Screw lengths were chosen to ensure both cortices were fully engaged. The first and sixth screw holes were left open. The internodes between the second and fifth screw holes were twisted to allow the third and fourth screws to be offset 7° caudal and 7° cranial to the perpendicular plane of the plate, respectively, in a similar orientation to the angle of the LC-DCP screws. Screws in the second and fifth holes were angled perpendicular (90°) to the plane of the plate. During application of the SOP, four segments of 0.6 mm stainless steel wire rope were applied under the plates in a similar manner as the LC-DCP constructs (Figure 3).

View Figure

• Download as Powerpoint
• Download Figure

Testing Apparatus

The scapula and plate-screw constructs were potted in a customized potting system using a hardened polyester resin mixture^d. The dorsal border and the neck of the scapula were potted within customized stainless steel blocks so that the horizontal plane of the plate was perpendicular to the pullout force. A standardized spacer was used between the stainless steel blocks to ensure equal distance was maintained between the blocks. The wire ropes were secured to a customized connecting piece at equal lengths between the plate and the connecting piece using a block spacer to measure the lengths. The wire ropes were tightened with the block spacer in place to ensure equal length and equal tension between wires. The potted constructs were attached to a customized testing jig and tested on a servo hydraulic universal testing machine^e (Figure 4). The top and bottom attachment points to the testing machine consisted of universal joints to allow for rotational freedom of the constructs during testing. The use of the wire ropes also allowed for rotational freedom of the constructs during testing.

View Figure

• Download as Powerpoint
• Download Figure

Mechanical Testing

Pullout forces were tested to construct failure using a rate of 0.5 mm/sec with a data acquisition rate set to 1.0 kHz. The load cell^frange selected was verified to an accuracy of 1% of the expected loads of failure estimated by preliminary testing. A load-displacement curve was created from data recorded by the data acquisition device of the materials testing machine^g. Load at failure was evaluated for each load-displacement curve. Load at failure was defined as the maximum force value prior to any apparent deviation from the linear portion on the load displacement curve.

Mode of failure was determined at the time of testing and confirmed by high-speed video recordings of each trial recorded at 250 frames/sec^h. Failure on the load-displacement curve was correlated with the corresponding video recordings. Failure was categorized into one of four categories as follows: (1) fracture of the bone separate from the plate-screw construct to bone interface; (2) disruption of the plate-screw construct to bone interface by bone coring; (3) disruption of the plate-screw construct to bone interface by bone slicing; and (4) disruption of the plate-screw construct to bone interface by screw stripping.

Bone coring was defined as the construct acutely failing by the screws pulling out along with a section of bone (Figure 5). Bone slicing was defined as construct failure occurring by the screws moving transversely through the bone (Figure 6). Screw stripping was defined as construct failure in which the screws pulled out of the bone, disrupting only the threads cut into the bone during initial screw application. Further characterization of failure was performed by examining the constructs after testing.

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

Results

Mean (standard deviation) load at failure for the SOP group was 1,101 (313) newtons. Mean (standard deviation) load at failure for the LC-DCP group was 1,023 (346) newtons. The load at failure was not significantly different between groups (P = 0.6). The different modes of failure observed in this study have been summarized in Table 1. No significant difference was noted between modes of failure for the two groups (P = 0.053).

Table 1 Modes of Failure after Applying Forces Perpendicular to the SOP and LC-DCPs

LC-DCP, limited contact dynamic compression plate; SOP, string of pearls plate.

View Fullscreen

Discussion

With the lack of a significant difference, the null hypothesis (that there would be no difference between the LC-DCP and SOPs in this testing model using the canine scapula) was accepted. Although increasing the number of specimens may have detected a small difference of mean loads at failure, the similarities of mean loads at failure measured in this study suggest that a large, and therefore clinically important, difference of loads at failure with 11 specimens in each paired group would be unlikely.

The LC-DCP, like other conventional plating systems, relies on the holding power of the screws to maintain the compressive force needed for stability.¹⁶ In bone with limited cortical stock, comminuted fractures, or osteoporotic bone, there may not be sufficient screw torque to resist shear forces. As a result, toggling and motion at the bone-plate interface could occur.^13,22 Locking plate constructs reportedly have a bending stiffness four times stronger than nonlocking plate constructs.^13,23 Previous studies report superior mechanical results for locking plates in bone of inadequate cortical stock. Kim et al. (2007) showed that load at failure for nonlocking plates was dependent on bone mineral density and that load at failure for locking plates was independent of bone mineral density.¹² Similarly, another study demonstrated that the axial load required to displace a locking construct in simulated osteoporotic bone is 82% higher than conventional nonlocking constructs when bicortical screws were used.¹¹

The lack of a significantly different pullout load at failure in this study could have several explanations. Not all biomechanical studies support superior mechanical advantages of locking plates, and some studies even report equivocal results.^22,24 In one such study, biomechanical testing using a fracture gap model yielded only subtle advantages of locking plates compared with nonlocking plates, and many testing parameters yielded similar results.²² It is possible that locking plate systems will not have a biomechanical advantage over a nonlocking plate system when tested in a pullout model where the plates were engaging an optimal area of the scapula at the base of the scapular spine. Another explanation may be the relatively high mean loads at failure. At high loads, it is possible the advantage of a locking plate system would not be noted. It is possible that if a different experimental model was chosen, cyclic loading utilizing lower overall loads (for example), a difference may have been seen. Four screws were used in each construct, screws were placed at divergent angles, and the plates were applied to an optimal position on the scapula. It is possible that an advantage of a locking plate system is less evident when there is a prerequisite material property of bone engaged, a prerequisite number of screws, and/or having screws placed at divergent angles. If fewer number of screws were used (e.g., only two or three in each segment), a difference may have been noted. The clinical relevance of a lack of a significant difference requires further investigation. Further, this study, like others, highlights the fact that the application and limitations of locking plate systems in various clinical situations requires further study.

Stiffness calculations were not reported in this study. Overall stiffness calculated from the load cell and actuator was observed, but would have measured overall stiffness of the entire testing construct and not the interface between the bone and the plate-screw construct. Although two or more extensometers could have been placed between the bone and plate, those stiffness measurements would not likely have reflected as important a measure of the integrity between the plate-screw construct and the bone as the measured load at initial failure.

A second objective of this study was to examine the modes of failure for both nonlocking and locking plates. Though there was no statistical difference between the groups, there was a trend for the locking SOP constructs to fail more often with bone slicing or bone coring and for the LC-DCP to fail more often with screw stripping. Similar modes of failure have been suggested previously.^{11,14–17,22,25} For example, Filipowicz et al. (2009) demonstrated that 66% of locking constructs failed by screws cutting through the bone in a canine humeral fracture gap model. Conversely, 90% of the nonlocking constructs in that study failed as a result of plate subsidence secondary to screw pullout and bending.²⁵ In another study investigating the biomechanical behavior of locking and nonlocking plates in osteoporotic bone, Gardner et al. (2005) found that nonlocking plates showed more evidence of microfracture at the screw-bone interface, whereas locking plates had greater bone disruption.²² The modes of failures seen in the current study support the theory that there are different stress concentrations between locking and nonlocking plates. In order for the locking plates to fail, the entire plate and angle stable screws (i.e., screws in locking plates) must pull out of the bone as a single unit, thus causing more trauma to the bone during failure. Nonlocking plates can fail by motion at each individual screw, causing toggling at the bone-plate interface. The clinical relevance of different failure modes in bone of poor quality is unknown as in vivo studies investigating plate failure modes are lacking.

In locking plates with divergent screws, pullout failure is thought to be less common than failure of the construct by bone slicing, bone coring, or even screw breakage. In the current study, each construct had screws placed at divergent angles. Multiple studies recommend divergent screw angles to maximize the forces required for pullout.^15–17,26 Nonetheless, Dipaola et al. (2008) recently showed that locking plates with parallel screws placed perpendicular to the bone interface had greater pullout strengths compared with angled screws. It was hypothesized that the screw thread angles are at their most effective in this orientation.¹⁸ The comparison of divergent screws versus parallel screws was not examined in the current study. The use of screw divergence requires more investigation before clinical recommendations can be made.

A unidirectional pullout to failure model was chosen to investigate the reported advantages of locking plates and their resistance to pullout.^13,17 The authors of this study aimed to avoid a complex multidirectional force model to simply measure the load required to disrupt the bone at the interface of the bone and plate-screw construct. The use of three locations of wire rope under the plate helped to distribute the pullout force evenly across the plate surface. During the initial experimental design, several options were attempted to optimize pullout forces without disrupting the bone constructs. Six-hole plates were initially chosen as the empty holes at the ends of the plates were to be grasped for the pullout force. Ultimately, the use of stainless steel wire ropes under the plate was shown to distribute the pullout force more evenly. With the use of wire ropes under the LC-DCP, one limitation of this model may be the inability of the plate to compress to the bone. The authors feel the placement of the wire ropes under the noncontact underside of the LC-DCP and the application of the plate to the slightly contoured junction between the scapular spine and supraspinatus fossa allowed for the plates to fully compress to the bone (Figure 2). Attempts to standardize this model led to initial use of an appropriate torque measuring device^j to tighten the screws. Screws were found to be either too loose or stripped in preliminary testing. It was elected to use three finger maximum torque as this produced more consistent and reliable screw tightness. The goal with the customized jig and universal joint attachments was to permit complete rotational freedom during initial settling of the constructs. This freedom to pivot permits the inhomogeneous density of bone to propagate failure from the weakest point along the screw-bone interface.

There are several limitations inherent to this biomechanical model, including the use of an intact scapula compared with a fracture gap model, the use of static loading to failure rather than cyclic loading, and the use of a unidirectional perpendicular pullout to failure model rather than a more complex loading method to duplicate the complex forces placed on the scapula in vivo. The use of an intact scapula (rather than a fracture gap model) is less likely to replicate a clinical situation; however, in clinical situations, such as fractures of the scapular neck or glenoid, the bone of the neck and glenoid is less likely to be an area of fixation concern as the supraspinatus and infraspinatus fossa and the scapular spine. Cyclic loading would be more clinically relevant; however, because the physiologic loads placed on the scapula during ambulation are complex and unknown, an initial study of loads of failure in a simple direction of force should precede cyclic testing. This testing method reduced the variables and complexity of mechanical testing to make simple comparisons between the two plating systems within the limitations of the methods used. These values can be used to develop testing methods of multidirectional and cyclic testing.

Conclusion

The SOP and LC-DCP systems did not have significantly different loads at failure in this specific pullout model in the canine scapula. As the loads of failure were not different in this study, locking plate systems such as the SOP cannot be assured to offer a large difference in plate-screw and bone stability in all situations. The relative advantage of locking plate systems should be more thoroughly defined for specific clinical situations. The SOP system tended to have more failures associated with bone slicing or bone coring compared with the LC-DCP group.