Summary

Optimizing function using personalized planning in Reverse Total Shoulder Arthroplasty - Establishing a cadaveric based computational model

Abstract

Background

Reverse Total Shoulder Arthroplasty (RTSA) is the standard of care for combined glenohumeral arthritis and rotator cuff deficiency. While a vast variety of RTSA designs, planning software, and patient specific instrumentation are available, personalizing implant selection and alignment remains a critical challenge to maximize outcomes. The purpose of this study was to assess the predictive capacity of a patient-specific modeling framework to plan RTSA surgeries and investigate the influence of RTSA parameters on range of motion (ROM) and deltoid muscle function.

Methods

Two shoulder cadavers were CT scanned after which RTSA was virtually planned (Arthrex VIP) and surgically performed. Passive adduction, abduction, scaption, and flexion ROM were measured using a KUKA robot for 9 configurations of glenosphere lateralization (0, 4 mm), inferiorization (0, 2.5 mm), and humeral spacer thickness (0, 6, 9 mm). Then an actuated cable system was used to load the deltoid to measure the force required to abduct from 30-50o for 8 configurations of glenosphere lateralization (0, 4 mm) and humeral spacer thickness (0, 6 mm) and inclination angle (135, 155o). Computational models were constructed in OpenSim-JAM for both specimens by segmenting the CT scans and identifying attachment points for nine deltoid paths. Virtual RTSA surgeries were performed where 1000 different combinations of 6 glenosphere and 9 humerus parameters where randomly selected. Passive ROM simulations predicted the elevation angle where bone impingement occurred. Muscle driven simulations predicted the deltoid forces to abduct from 0o to 80o and corresponding glenosphere contact pressures and forces. To evaluate the predictive capacity of the model, the Pearson correlation coefficient (R) was computed between the simulation predictions and cadaver measurements of passive ROM for each RTSA configuration. The influence of RTSA parameters on the predicted ROM and peak deltoid force was ranked by computing the correlation (R) across all 1000 models.

Results

The average errors in the predicted passive ROM were smallest in adduction (2o) and greatest in flexion (17o), with a strong correlation between the simulation predictions and experimental measurements (R=0.97). In the simulations, glenosphere lateralization (R=0.5) and inferiorization (R=0.1), as well as humeral inclination (R=0.5) showed the strongest influence on passive abduction ROM.
The peak predicted deltoid forces varied by more than 50 N across the 1000 virtual RTSA configurations and exhibited similar magnitude to the experiments and the same trend with increasing abduction angle. The peak predicted deltoid force during abduction was strongly negatively correlated with glenosphere inferiorization (R=-0.7) and positively correlated with glenosphere lateralization (R=0.2), humeral spacer thickness (R=0.2), and inclination (R=0.2). The magnitude of the peak glenosphere contact force was also strongly negatively correlated with glenosphere inferiorization (R=-0.4).

Conclusion

The model predictions generally agreed with the cadaver experiments. Glenosphere lateralization and inferiorization increased the ROM in nearly all directions, however inferiorization decreased the deltoid force while lateralization increased the deltoid force. Humeral inclination increased the ROM in abduction and scaption but decreased the ROM in adduction and flexion as well as increased the deltoid force. These tradeoffs agree with the trends found in the cadaver experiments as well as previous cadaver studies.