While reverse shoulder arthroplasty (RSA) is a reliable treatment option for patients with rotator cuff deficiency, loss of glenoid baseplate fixation often occurs due to screw loosening. We questioned whether an analysis of the trabecular bone density distribution in the scapula would indicate more optimal sites for screw placement. As such, the purpose of this study was to determine the anatomic distribution of trabecular bone density in regions of the scapula available for screw placement in RSA.

Seven cadaveric shoulders were computed tomography (CT) scanned, and then voxels of the scapulae were isolated from the CT volume (Mimics 15.0 Materialise, Leuven, Belgium). Analyses were conducted in a common, 3D coordinate system. Volumetric regions of interest (ROI) within the scapula were identified based on potential baseplate screw sites. ROIs included areas at the base of the coracoid process lateral and inferior to the suprascapular notch, in the posterior and anterior lateral spine and in the anterosuperior and posteroinferior lateral border. Hounsfield Units (HU) were extracted from voxels corresponding to trabecular bone within each ROI. Overall bone density was summarised as the frequency of HU values above 80% of the ROI's maximum density value. Paired, two-tailed t-tests assuming unequal variance were used for pairwise comparisons (P≤0.05). Intra-region analyses compared two ROIs within the same broad anatomical structure; inter-region analyses compared ROIs between anatomical structures.

Areas of the spine and lateral border of the scapula appeared to be denser than the coracoid process. Intra-region comparisons indicated no significant differences within ROI: coracoid P=0.43, spine P=0.95, lateral border P=0.41. ROI inferior to the suprascapular notch were on average 3.78% (P=0.08) and 6% (P=0.04) less dense than the anterosuperior and posteroinferior lateral border and 7.59% (P=0.006) and 7.72% (P=0.01) less dense than the anterior and posterior lateral spine. ROI lateral to the suprascapular notch were 6% (P=0.05) and 8.21% (P=0.02) less dense than the anterosuperior and posteroinferior lateral border and 9.8% (P=0.006) and 9.94% (P=0.008) less dense than the anterior and posterior lateral spine. There was no significant difference between the anterior spine and anterosuperior and posteroinferior lateral border (P=0.12, P=0.58), nor between the posterior spine and anterosuperior and posteroinferior lateral border (P=0.14, P=0.57).

Results from this study indicate that the spine and lateral border of the scapula contain denser trabecular bone relative to regions in the coracoid. The higher quality bone of the spine and lateral border should be favoured over the coracoid process when fixing the glenoid baseplate in RSA. Further research may support the redesign of the glenoid baseplate geometry to better integrate the anatomy of the scapula and improve implant survival.