Instrumented Trial Prosthesis for Intraoperative Measurements of Joint Reaction Forces during Reverse Total Shoulder Arthroplasty

Although soft tissue tension is an important factor in the clinical performance of reverse total shoulder arthroplasty (RTSA), this tension has not been quantified intraoperatively. Knowledge of the shoulder joint reaction forces during RTSA could facilitate the optimal placement of implant components to minimize the risk of both intraand postoperative complications. We developed a strain gauge instrumented trial glenosphere to measure shoulder joint reaction forces during RTSA. The strain gauges and their connections were hermetically sealed against body fluids by biocompatible materials. All materials in direct body contact were biocompatible. In this study, we introduce the structure and calibration results of the instrumented prosthesis. We also demonstrate the practical use of the prosthesis on a cadaveric shoulder. The instrumented prosthesis showed mean measurement errors of approximately 3.4% for forces up to 400 N. A maximum joint reaction force of 132 N was observed during abduction in a single cadaver specimen. This sensor will be useful for quantifying soft tissue tension during RTSA surgery.


Introduction
Reverse total shoulder arthroplasty (RTSA) has been used to treat selected complex shoulder problems as an alternative to standard total shoulder arthroplasty (TSA). (1)(2)(3) However, persistent problems and relatively high complication rates with this procedure are reported to include scapular notching, infection, instability, dislocation, intraoperative fracture, postoperative fracture, brachial plexopathy, and glenoid component loosening. (4)(5)(6) A 5% rate of instability at two years (7) and a 7.5% rate of dislocation at three years (8) indicate that achieving appropriate or optimal soft tissue tensioning during surgery is a major challenge. (6) The joint center of rotation can be moved to lengthen the humerus and increase stability, (9) but this increases the risks of stress fracture, brachial plexopathy, deltoid overtensioning, and loss of motion. (2) Achieving proper intraoperative soft tissue tension is an obvious surgical goal and surgeons have the ability to choose different implant designs, sizes, placements, and orientations to achieve it. However, methods for measuring soft tissue tension during RTSA have not been reported, and these tension values remain unknown.
There are numerous reports of direct measurements and mathematical calculations of shoulder forces during daily activities using TSA. (10)(11)(12) Although RTSA has been used for several years, few studies have focused on RTSA biomechanics. (11,13) Since the use of RTSA was approved in the United States in 2004 and in Japan in 2014, we lack long-term clinical outcomes. Soft tissue tension can be quantified by measuring shoulder joint reaction forces as counteraction forces during RTSA. Therefore, we have developed a strain gauge instrumented RTSA to measure shoulder joint reaction forces intraoperatively. In this report, we describe the design of the instrumented prosthesis and the accuracy of its measurements during RTSA. Proof of feasibility using this device in a cadaveric shoulder is also reported.

Materials and Methods
The force sensor is a custom-instrumented trial implant that can be used with an existing RTSA system (EQUINOXE, Exactech Inc, Gainesville, FL), just as a standard trial implant is used. We designed a new trial glenosphere and adapter that easily attach by screw connections to the standard glenoid baseplate (Fig. 1). Since the trial is going to be used in the human body, all materials in direct body contact were biocompatible. The geometries of the commercial glenosphere can be reproduced using the following design combinations: a diameter of 38-46 mm, a lateral offset of 0-4 mm, and a distal offset of 0-2 mm. The outer dimensions of the instrumented trial prosthesis are identical to those of the prosthesis that is used clinically. (14) Four uniaxial foil strain gauges (QFLG-02-11-3LJB, Tokyo Sokki Kenkyujo Co., Ltd., Japan) were placed on the side faces of a rectangular post connecting the glenosphere to the baseplate using cyanoacrylate adhesive. Four strain gauges in quarter-bridge configurations were placed  parallel to the longitudinal axis of the post. All strain gauges, including a base sheet and wire connection terminals, were protected by a polychloroprene coating. Assuming that the articular friction between the glenosphere and the wetted plastic humeral tray is negligible during intraoperative use, the four strain gauges provide a redundant system to quantify three joint reaction force components. The sensor was calibrated using a standard glenoid baseplate fixed to a custom calibration jig. The instrumented prosthesis was attached by screw connections to the glenoid baseplate. The calibration jig was fixed on an angle vice to permit the change in force directions. The calibration jig with the angle vice was placed on top of a low-friction x-y translation table that eliminated horizontal constraint forces. Known forces (F) were applied by a standard material testing machine (852 Mini Bionix, MTS, Eden Prairie, MN, USA) through the polyethylene humeral insert (Fig. 2). Applied forces were increased from zero to each verification force level. The verification force values were determined by the guide of "Standard Practices for Force Verification" according to the American Society for Testing and Materials (ASTM) E-4. (15) The selected calibration ranges for all force components (absolute values, polar angles, and cone angles) are given in Table 1. Load components (F x , F y , and F z ) were defined by combinations of these components. For example, if a 100 N load was applied from an angle (φ = 45°, θ = 60°),   (Fig. 3). Identical loads were applied three times and the outputs were subsequently averaged. The external force vector (F) can be expressed in terms of the strain gauge outputs as follows: where T is a calibration matrix, and S i (i = 1 to 4) corresponds to the outputs of the four strain gauges. Calibration errors were calculated according to well-established methods. (16,17) The calibration procedures conformed to ASTM E-4 standards. (15) In order to model surgical use, sensors were washed in alcohol, packaged, and sent to an external vendor for gamma sterilization prior to the testing. We performed a single cadaver trial to demonstrate that the instrumented prosthesis could be used intraoperatively. Two experienced shoulder surgeons performed the cadaver evaluation. Following standard surgical protocols for the exposure and placement of the glenoid baseplate, the instrumented trial prosthesis was easily assembled within the joint. The humeral trial component was placed, the joint was reduced, and joint reaction forces were recorded during the cyclic abduction/adduction of the shoulder. Following measurements, the instrumented trial prosthesis was removed and the procedure could have concluded with the normal placement of a permanent prosthesis.

Results
When the errors were calculated under all applied forces, the average absolute errors of the force components were 3.5% for F x , 3.1% for F y , and 3.7% for F z . The highest single-frame error was 8.1% for F x . An example (φ = 45°, θ = 80°) of the applied and measured forces, and the calculated errors of the three force components are shown in Fig. 4. In this loading configuration, all error values for F x , F y , and F z were under 4% or 16 N.
The instrumented trial prosthesis was used in a single cadaver shoulder specimen, successfully demonstrating the assembly of the prosthesis within the shoulder joint, acquisition of joint reaction force data, and removal without significantly affecting the normal RTSA procedure. The recorded joint forces during seven cycles of abduction/adduction are shown in Fig. 5. A passive static resultant force of 13.5 N and a maximum resultant force of 132 N were recorded during this demonstration (Fig. 5).

Discussion
Intraoperative soft tissue tension is important for the long-term functional and clinical performance following RTSA. We have developed and characterized an instrumented trial glenosphere component that can be used in standard RTSA surgical procedures to quantify the joint reaction forces during the passive manipulation of the arm.
We found average force measurement errors of 2-4% of full-scale load or 8-16 N. Although these errors are somewhat higher than those of a permanently implanted instrumented shoulder prosthesis, (18) the accuracy of our single-use trial sensor is adequate to provide some initial values and ranges for joint reaction forces during RTSA. The instrumented post-sensor configuration results in greater sensitivity for shear forces (F x and F y ) than axial forces (F z ), so the measurement resolution and accuracy for small axial forces (e.g., forces below 50 N) are relatively poor. For example, if measurement errors are calculated according to ASTM standards, (15) which report average relative errors over all loads, calibration errors were 16% for F x , 15% for F y , and 48% for F z . If the ASTM method is restricted to loads greater than 50 N, the calibration errors reduce to 11% for F x , 8% for F y , and 11% for F z . These errors are similar to the values found using the calibration methods of Westerhoff et al., (18) where errors are assessed relative to the maximum applied load. Based upon preliminary intraoperative use, joint reaction force magnitudes are typically 50-300 N, so the method of Westerhoff et al. provides the most relevant assessment of joint reaction force accuracy. Nevertheless, caution should be exercised in interpreting results with force magnitudes below 50 N.
Three independent strain gauge outputs are sufficient to calculate three independent unknown force components with our sensor. Thus, there is redundancy in using outputs from four strain gauges. The errors calculated using only three outputs (channels 1, 2, and 3) were 4% for F x , 4% for F y , and 5% for F z . If another combination (channels 2, 3, and 4) were used, the errors were 3% for F x , 4% for F y , and 4% for F z . All 3-channel calibrations provide results similar to the case when all four channels were used. Having the extra strain gauge is useful because the wired connections are subject to damage in intraoperative use.
We used a limited number of sensor orientations (angles θ and φ) for calibration. If other angles (φ = ±30, ±60, ±120, and ±150°) were added, the error values changed to 3% for F x , 2% for F y , and 4% for F z . The slight change in calibration accuracy does not justify the much more time-consuming calibration procedure, especially in the context of having to calibrate many of these single-use sensors during a study.
Intraoperative use of the force-sensing trial glenosphere merits three considerations. First, the current device uses wired strain gauges. Care has to be exercised to maintain sterility within the operative field, similar to any wired surgical hand tool. The wires and their connections can sustain damage if tensioned or severely kinked within the operative field. Second, the use of the force-sensing trial and recording of joint reaction force data will add ten minutes to the RTSA procedure. Finally, cleaning and sterilization of the sensor may affect the measurement performance, so post-use calibration will be performed in future clinical studies.

Conclusions
We developed a strain gauge instrumented trial glenosphere to measure shoulder joint reaction forces during RTSA. Calibration results of the instrumented prosthesis showed mean measurement errors of approximately 3.4% for forces up to 400 N. We also demonstrated practical use on a cadaveric shoulder. A maximum joint reaction force of 132 N was observed during abduction in a single cadaver specimen. Measurements of shoulder joint reaction forces during RTSA will allow surgeons to place and align implant components more objectively. This may lead to increasingly predictable and durable RTSA results. The strain gauge instrumented glenosphere trial implant represents a first step towards the objective intraoperative assessment of shoulder joint tissue tension.