Assessment of glenoid baseplate initial micromotion and fixation strength in reverse total shoulder arthroplasty designs using a direct shear force methodology

Abstract

Background

In a reverse total shoulder arthroplasty, the altered glenohumeral joint center of rotation subjects the glenoid baseplate to increased shear forces and potential loosening.

Methods

This study examined glenoid baseplate micromotion and initial fixation strength with the application of direct shear force in a Sawbone model. The reverse total shoulder arthroplasty systems examined were the DJO Reverse® Shoulder Prosthesis, the Exactech Equinoxe® Reverse System, and the Tornier Aequalis ^TM Reverse Shoulder Prosthesis. Specimens were cyclically tested with increasing shear loads until 150 µm of displacement between the implant and glenoid was achieved, and subsequently until failure, classified as either 1 cm of implant/glenoid displacement or fracture.

Results

The average load withstood for the 150 µm threshold for DJO, Tornier, and Exactech was 460 ± 88 N, 525 ± 100 N, and 585 ± 160 N, respectively. The average total load at device failure for DJO, Tornier, and Exactech was 980 ± 260 N, 1260 ± 120 N, and 1350 ± 230 N, respectively.

Discussion

The Exactech implant design trended toward requiring more load to induce micromotion at each threshold and to induce device failure, most commonly seen as inferior screw pull out. This study proposes design features that may enhance fixation and suggests little risk of initial micromotion or failure during initial post-operative recovery.

Introduction

In cases of irreparable rotator cuff damage and severe rotator cuff disease, the reverse total shoulder arthroplasty (RTSA) is biomechanically beneficial. The main purpose of the RTSA design is to transfer loading from the rotator cuff to the deltoid muscle for activities such as arm abduction and elevation. ¹ The RTSA alters the center of rotation to increase deltoid muscle fiber recruitment and generate greater forces and a larger range of motion for the shoulder. ² An unintended consequence of altering the center of rotation is that torque and shear forces increase at the glenoid baseplate, which increases the probability of post-operative implant loosening. ³ Loosening is the leading cause for revision and complications (14%–75%) of the glenoid baseplate.^1,4–6

To achieve baseplate stability and fixation post-operatively as well as long-term osseointegration, initial baseplate fixation with screws and/or cement is imperative. ⁴ Each unique implant design must resist initial micromotion while also providing structural resistance to loading failure. During the post-operative healing period, ingrowth is the most successful if initial baseplate sliding motion is held below 150 µm relative to the glenoid.^7–9 This potential for sliding movement (or shearing) is the direct result of shear forces that occur during normal function or during abrupt incidences of trauma, falls, or dislocation.

In addition to increased superior joint shear force, reverse shoulder arthroplasty results in reduced joint compressive force. Compressive forces on the glenoid baseplate assist in compaction and stabilization, but shear forces can occur in isolation.^10,11 During the post-operative recovery period, patients are more likely to subject the baseplate to isolated shearing forces as they adapt to the altered use of the deltoid muscle, altered compressive contraction mechanics, and increased range of motion as a result of RTSA.^10–12 Limited range of motion exercises are encouraged during the initial rehabilitation period and compressive forces are low, placing enhanced importance on initial resistance to shear loading in the absence of significant compressive loading. ¹³ Most studies focus on the application of cyclic shear and compressive loading while measuring pre-cyclic and post-cyclic displacement.^14,15 There have been few studies that have quantified the direct shear forces needed for baseplate loosening and absolute failure without using a compressive force.

This study applied isolated shear force to the baseplate to examine shear-induced glenoid baseplate micromotion and prosthesis ultimate failure. The aims of this study were to quantify the amount of cyclic shear force needed to induce 150 µm of initial baseplate motion, and then the amount of ultimate force needed to cause failure of the glenoid component.

Methods

For this study, three popular RTSA systems were examined: the DJO Reverse® Shoulder Prosthesis (RSP) (DJO®, Lewisville, TX), the Exactech® Equinoxe Reverse System, (Exactech®, Gainesville, FL) and the Tornier Aequalis ^TM Reverse Shoulder Prosthesis (Tornier, Bloomington, MN). The DJO glenoid baseplate had a diameter of 26 mm, four peripheral (5.0 mm diameter and 30 mm length) locking bone screws, and a central (6.5 mm diameter and 30 mm length) lag cancellous bone screw. The Exactech glenoid baseplate had a diameter of 38 mm and contained a bone cage central peg (25.4 mm length); it also had six peripheral screw holes for four screws that allowed for screw location variability as well as 30° of orientation variability. The Tornier glenoid baseplate had a 29 mm diameter press-fit design with a central peg (15 mm length) and four (4.5 mm length) peripheral screws that had 30° of orientation variability (Figure 1). Five samples of each design were obtained from the respective manufacturers and set up for mechanical testing.

Figure 1.

Glenoid baseplate design details: DJO (left); Exactech; (center) Tornier (right) reprint pending permission.

Each of the 15 reverse shoulder prostheses was implanted into a #3413-2 Sawbone® scapula (Sawbone®, Vashon, WA) with a 12 pcf solid foam core, which mimicked osteopenic bone. Each specimen was radiographed, and the image was used in conjunction with a custom MATLAB program to calculate and verify the implantation angles in three planes: superior/inferior, anterior/posterior, and the version of the glenoid baseplate (Figure 2).

Figure 2.

Radiograph implantation angle verification from the anterior to posterior view (left) and the superior to inferior view (right).

To verify that there were no significant differences between the alignment parameters, a t-test of the average angles was calculated. Next, the specimen was potted in a 5″ by 2″ metal frame using a FastCast™ (Goldenwest Manufacturing, Grass Valley, CA) epoxy resin with the glenoid surface aligned perpendicular to the block (Figure 3).

Figure 3.

Prepared specimen from the anterior to posterior view (left) and the inferior to superior view (right).

Current ASTM standards (F2028) for dynamic glenoid loosening or dissociation consist of measuring displacement of a glenoid component in a bone substrate material under both an axial and shear load; these standards were followed to achieve ultimate shear loading but without the inclusion of compressive force, which differs from the standard. Each specimen was subjected to both a cyclic testing and a progressive loading until failure procedure. An Instron® servo-hydraulic test system (Model 8874, Instron®, Norwood MA) with a 25 kN load cell (Dynacell, Instron®, Norwood MA) applied the progressive loading of the specimens via a cylindrical anvil (8 mm diameter) that contacted the glenoid components as close to the bone–implant interface as possible to apply a shear force in the inferior to superior direction. The potted Sawbone® scapula was placed in an inverted orientation on the test system so as to apply the inferior to superior loading (Figure 4).

Figure 4.

Instron and LVDT testing set-up including load cell, threaded rod, specimen holder, clamps, aluminum blocks, and plunger (from top to bottom).

Micromotion was measured directly between the scapula and the glenoid using a linear variable transformer (LVDT) (National Instruments, Austin, TX). The LVDT was attached to the specimen at the acromion and coracoid process, with the plunger directly contacting the glenosphere at the interface between the implant and the Sawbone® and was capable of measuring ± 2 µm of repeatable movement. A WaveMatrix™ (Version 1.5, Norwood, Massachusetts) program was created to run the cyclic displacement and failure testing. For the first test of cyclic loading procedure, the glenoid was forcibly displaced using a 1 Hz sinusoidal displacement profile for 30 cycles, starting with a 50 µm of actuator amplitude and increasing by increments of 50 µm. These increasing whole-implant movements resulted in increasing micromotion between the baseplate and glenoid, and these were monitored until 150 µm of baseplate micromotion was recorded between the glenoid baseplate and the scapula (Figure 5). During each test, the Instron® time, displacement, and loading inputs, and the LVDT micromotion were recorded. Following the observed 150 µm of baseplate micromotion, a second test was conducted wherein the glenoid of each implant was subjected to 10 mm of direct shear displacement with a ramp rate of 0.5 mm/s from the inferior to superior direction in order to cause failure of the implant between the implant and scapula. Each implant failure was videoed and photographed for classification of fracture type.

Figure 5.

Cyclic loading process to reach 150 µm of baseplate motion.

For each cyclic test, both baseplate micromotion and Instron® loading data were examined. The averages of micromotion minimums and maximums were calculated from the 30 cycles of implant loading. Average micromotion was calculated as the difference between the average maximum and minimum. Plastic deformation of the implant baseplate with respect to the glenoid was recorded from changes in cyclic minimum values.

For the tests to failure, micromotion data and loading data were plotted against time to visualize the deformation profile of the baseplate and the loading profile, respectively. The failure point was determined by a breakpoint threshold or the sudden decrease in loading associated with continued displacement. The maximum values for both the compressive load and baseplate micromotion were calculated from the collected data and associated with the mode of failure and description of the failure for each specimen.

Analysis of variance statistical significance tests were conducted at a 95% confidence interval, to compare each of the designs with respect to the load at 150, 100, and 50 µm of baseplate micromotion, the Instron® displacement to cause 150, 100, and 50 µm of baseplate micromotion, the load at failure, and the amount of micromotion at failure.

Results

The three implant designs were compared at three critical points (50, 100, and 150 µm of micromotion) during the cyclic testing process. The Exactech design consistently withstood more load than the other designs for each increment of micromotion testing.

Each implant that was able to reach 50 µm of baseplate motion was within a similar loading range of approximately 250 N; the Exactech required about 10 N more load than both the Tornier and DJO designs. At 100 µm of baseplate micromotion, the Tornier and DJO designs displayed similar values of around 360 N. The Exactech withstood approximately 60 N more load than both the Tornier and DJO or about 430 N total load. At 150 µm of motion, the first variances between DJO and Tornier emerged. DJO was the first to experience 150 µm of micromotion, with Tornier and Exactech requiring about 85 N and 125 N more load than DJO, respectively. The average total load withstood for the 150 µm threshold for DJO, Tornier, and Exactech was 460 ± 100 N, 525 ± 88 N, and 585 ± 160 N, respectively (Figure 6). Although not statistically significantly different, the data showed the Exactech implant design trended toward requiring more load to cause micromotion between the baseplate and the bone interface. This design continued to demonstrate stiffer behavior when compared to the others as loading was increased.

Figure 6.

Average loads withstood by each RTSA design before the micromotion threshold was reached (50 µm – Exactech: 260 ± 105, Tornier: 250 ± 40 N, DJO: 250 ± 70 N; 100 µm – Exactech: 430 ± 130, Tornier: 360 ± 70 N, DJO: 360 ± 90 N; 150 µm – Exactech: 585 ± 160 N, Tornier: 525 ± 90 N, DJO: 460 ± 100 N).

The specimen was considered to have failed after either the baseplate experienced 1 cm (10,000 µm) of displacement or the scapula fractured. All the specimens either experienced failure by fractures at the anterior and posterior screws or by fracture of the lateral border of the scapula (Figure 7). Most failures at the anterior and posterior screws were associated with varying levels of inferior screw pull out (Table 1). The average load to failure was the highest for the Exactech design at 1350 ± 230 N. Tornier displayed the next largest load to failure at 1260 ± 120 N. The DJO specimens resulted in the lowest average load to failure of 980 ± 260 N (Figure 8). While the difference in these values was not statistically significantly, there was trending evidence that the DJO implants might experience failure at lower loads compared to both the Exactech and Tornier designs. This comparison also supported the observation that the Exactech implant design had a stiffer response and required higher loads to achieve micromotion.

Figure 7.

Fracture failure at the anterior posterior peripheral screws (left) and the lateral border of the scapula (right).

Table 1.

Mode of failure for each specimen.

Fracture at anterior/posterior screws		Fracture at lateral border of scapula
Inferior screw pull out	No visible pull out
E1, E2, E4, T1, T2, D3, D5	T5, D1	E3, T3, T4, D2, D4

E: Exactech; T: Tornier; D: DJO.

Figure 8.

Loads withstood before failure for each trial and averaged for each of the three RTSA designs (average load until failure – Exactech: 1350 ± 230 N, Tornier: 1260 ± 120 N, DJO: 980 ± 260 N).

Discussion

Surface area, alignment, and screw fixation are key design variables that can be attributed to initial fixation of glenoid baseplates. Initial results suggest that the Exactech Equinoxe reverse shoulder system tended to withstand higher loads before reaching micromotion thresholds or failure when compared to both the DJO RSP® and the Tornier Aequalis ^TM RTSA systems (although this was not statistically significant). The Exactech design tended to require a larger amount of cyclic load for 150 µm of baseplate micromotion and for device failure when testing cyclic displacement and failure mechanics. The overall shape of the Exactech baseplate was oval, whereas DJO and Tornier were circular. This supports the observation that oval baseplates result in greater fixation strength when compared to circular baseplates.^14,15 The oval shape of the baseplate and the larger size of the baseplate gave the Exactech a larger contacting surface area than the DJO and Tornier baseplates. This would agree with previous studies that increased baseplate surface area results in increased fixation strength of the baseplate and a decrease in micromotion.^15,16

Screw fixation, including number of screws, the use of central pegs, screw orientation, and screw cortical engagement, are all design variables that can affect initial stability. Several RTSA designs incorporate strategies to minimize baseplate micromotion, such as hydroxyapatite coatings, locking screws, a central screw, and refinements in baseplate location and placement.^3,4 The Exactech baseplate design featured six peripheral screw locations with 30° of screw angle variability. In contrast, the Tornier baseplate utilized a central peg and had four peripheral screw locations with 30° of screw angle variability, while DJO had a central screw and four peripheral screws with minimum variability. Increased screw engagement and purchase into the best available bone quality are important factors relating to the fixation strength of the baseplate.^11,15 The Exactech baseplate can allow the surgeon the most adjustment options for screw location and angle so that ideal screw purchase within the available bone of the glenoid is attained. This feature is particularly ideal for clinical situations where patients may have limited healthy bone or where the placement of the baseplate may not be in the most ideal location. The placement of the inferior screw is the most important, compared to the superior, anterior, and posterior screws, because most of the loading from the humeral component occurs in the inferior-to-superior direction. ¹¹ The improved fixation strength of the baseplate for the Exactech over the Tornier and the Tornier over the DJO for cyclic displacement testing and for failure testing may be attributed to the additional screw placement options provided by the design allowing for ideal bone purchase at the glenoid site.

Even with specific design differences, all three implant systems performed similarly and showed comparable micromotion resistance to initial cyclic shear loading and strength of fixation during maximal shear loading. This study suggests that these designs are at little risk of initial micromotion or failure during initial post-operative recovery and light manipulation or therapy, prior to significant bone integration. The magnitudes of shear forces need to induce significant micromotion ( > 150 µm) ranged from 460 to 585 N, which would require a significant and repetitive manipulation or muscular effort to induce.^13,17 In addition, these types of movements would also include significant joint compressive loads which would further stabilize the baseplate. However, this study has highlighted that these devices are still at risk of micron-level movement from light rehabilitative-level shear loading and from shear load failure that could be associated with abrupt shear loading such as might occur during falls or trauma. The severity of these failures is associated with the screw and fixation methodologies that are used in these designs.

There were several limitations to the work that should be considered in translating our results. The study was limited in sample size due to funding. Given the trends observed in the data, a larger sample size might have increased the significance of the data comparisons, but it does not appear that the differences in magnitudes would be significantly altered. The implantation process was heavily reliant on the surgeon. Each device was implanted by the same surgeon, but there is still potential for implantation variability. Loading to failure could have been influenced by the fixation method of the scapula and is not intended to represent any true clinical failure mode. Most of the fractures continued through the scapula to the superior side, near the acromion and coracoid process. The abrupt fractures of the scapula can be credited to the direct shear force on the baseplate; this may have resulted in an overestimate of anatomical loading. The Sawbone® model mimicked the material properties of osteopenic bone in this study because it has been proven to accurately represent the biomechanical responses of human bone under bending, axial, and torsional loads. ¹⁸ Composite bone has the ability to be more consistent with strength and anatomical properties ¹⁹ and the three RTSA systems were implanted into the same medium so that they could be directly compared to one another. However, composite bone, such as Sawbone®, is not an exact replica of osteopenic bone and this may have caused the magnitude of loading to be an underestimate. The baseplates of RTSA systems are designed to allow and promote bone ingrowth to stabilize the implant; this was not possible to simulate without a living tissue model. Design features intended to promote osseointegration (other than inherent roughness) were not included in baseplate performance during testing.