Using GPS to assess glenoid morphology in reverse shoulder arthroplasty: Correlation to manual measurements and effects on final implant selection


Jaicharan Iyengar, MD
Alpine Orthopaedics / Stockton, Calif.


Assessment of glenoid morphology is an important aspect of pre-operative planning in total shoulder arthroplasty.

It has been shown that manual measurement of glenoid parameters such as retroversion and superior inclination are subject to variability and can be enhanced by using computerized CT protocols (1). Accurately assessing these variables is critical in that it can potentially affect the choice of implants, and use of augments, in efforts to achieve optimal implant position. GPS represents an evolution in both assessing glenoid parameters using 3D scapular modeling and simulated implantation of glenoid component placement to optimize position.

Our study (in press) evaluated both the reliability of manual versus GPS computer-generated glenoid morphology parameters, as well as the effects this had on final implant selection in a cohort of primary reverse shoulder arthroplasty patients. We evaluated 68 patients who were prospectively enrolled in the Equinoxe database over a one-year time span. Pre-operative glenoid version and inclination were measured manually by the attending surgeon as well as a trained research assistant (RA) to provide a control, and then compared with measurements generated using the GPS software planning tool. These measurements were then correlated to the final implant/augment selection and position of the implants recorded intra-operatively.

We found that with respect to glenoid retroversion, manual measurements by the surgeon, RA, and GPS had moderate intra-class correlation (ICC2 = 0.50, p < 0.01), but the surgeon and RA manual measurements consistently underestimated the version compared to GPS (mean 7.8o, 6.2o, and 10.6o respectively, n = 67). The significance of this difference was even more pronounced in patients where an 8 degree posterior augment implant was used (9.7o & 8.8o manually vs. 13.2o GPS). The final position of the implanted glenoid overall was within acceptable parameters in both the standard (4.1o retroversion, n = 24) and posterior augmented (5.1o retroversion, n = 39) groups. This suggests that GPS guided pre-operative measurements were an important factor in both accurate assessment of retroversion, guiding the use of posterior augmented implants, and resulted in final implant positioning that was within a desired range.

With respect to superior inclination, the results of our study were less consistent (recall that beta angle is inversely correlated with superior inclination of the glenoid, with higher angles representing less superior tilt). The human measurements in this case (80.7o & 73.5o) consistently overestimated the amount of glenoid superior inclination versus GPS measurements (87.4o) and had poor intra-class correlation with GPS (ICC 2 = 0.27, p < 0.01. Overall, there was minimal correction of native glenoid beta angle from pre-op to post-op in our cohort (87.4o pre-op to 87.0o post-op) and only 4 patients required use of superior augments. This suggests that use of GPS may actually result in less need for correction of apparent superior tilt than manual measurement, and reduced use of superior augments to aggressively downtilt the glenosphere component, which is traditionally taught for Grammont-style implants. Long term follow-up is necessary to determine if this tendency results in any clinical difference in scapular notching or any increased revision risk for a lateralized humeral implant design.

As a final commentary, our data showed that the overall retroversion correction in cases where an 8-degree posterior augment was used was remarkably accurate (13.2o pre-op retroversion vs. 5.1o post-op retroversion, mean difference of 8.1o, n = 39). The consistency of these results, particularly in more difficult cases of increased retroversion certainly suggests that GPS with augmentation produces reproducible correction of glenoid version. In general, standard implants were consistently utilized in cases where both manual measurements and GPS suggested acceptable glenoid version parameters pre-operatively and minimal correction was needed.

In summary, in our single-surgeon cohort of GPS guided reverse arthroplasty cases, computerized assessment of glenoid morphology differed in predictable ways from manual measurements. These systematic biases were correlated to differences in final implant selection and reliable correction of glenoid morphology parameters to acceptable ranges post-operatively. My intuition is that this level of accuracy would be difficult to achieve without computerized pre-operative planning and GPS guided intra-operative navigation. To borrow a golf analogy, GPS does not just help you knowing where to aim (shot selection), it help keeps more of your drives in the fairway (dispersion), and at the end of the day, I think keeps our surgical scores better than par.


  1. Glenoid version by CT scan: an analysis of clinical measurement error and introduction of a protocol to reduce variability.  Fabian van de Bunt,  Michael L. Pearl, Eric K. Lee, Lauren Peng, and Paul Didomenico. Skeletal Radiol. 2015; 44(11): 1627–1635.