Anatomical shape of the shoulder as a predictor of rotator cuff injury

INTRODUCTION: Humans exhibit remarkable variability in the skeletal shape and configuration of the shoulder (FIG.A), but the relevance of individual differences for normal function and disease incidence is both poorly understood and controversial (e.g., Moor et al., 2014). In our previous comparative analyses of human shoulder evolution we noted the axis of greatest shape variation within humans is aligned with the evolutionary trajectory linking humans to our last common ancestor with the African apes (Young et al., 2015). This vector describes changes to the shape of the scapular blade and the orientation of the scapular spine and glenoid to more lateralized configurations (FIG.B). We hypothesized that this vector is the result of selection against overhead behaviors (e.g., climbing). We further hypothesized that the functional flexibility of the upper limb results in the retention of significant variability in morphology, which reflects in part tradeoffs between motions and joint reaction forces at the shoulder joint. This model predicts that individuals with extreme shape configurations have a higher likelihood of experiencing a rotator cuff injury (RCI) and this risk is higher earlier in life. METHODS: As a preliminary test of our model predictions we obtained computed tomography (CT) data of the scapula from individuals with a clinical diagnosis of rotator cuff tear (n=20, Mean age = 58.7 years) and age-matched controls with no history of injury (n=28, Mean age = 60.2 years). To characterize shape, we collected 3D coordinate data from 28 landmarks representing the shape of the blade, glenoid, spine, acromion and coracoid. We next performed a Procrustes superimposition to remove the effect of orientation and scale from the data, and performed a Canonical Variates Analysis (CVA) to identify shape differences that maximized between group variation. We used ANOVA to test for differences in mean shape as measured by Procrustes distance. We visualized shape differences by warping the average shape to the extremes of the observed values via the associated eigenvector (FIG.C). RESULTS: Rotator cuff tear and control populations exhibit a significant difference in shape (F=1.77, df=77, p\u3c0.0001). Shape differences are consistent with the principal axis of variation within humans, and indicate that individuals with a tear have a more cranial orientation of the glenoid and scapular spine, a reduced supraspinatus attachment area, and a smaller attachment and altered orientation of the teres major (cf. FIG.A-C). DISCUSSION: Results are consistent with the hypothesis that anatomical variation of the scapular blade impacts the incidence of shoulder dysfunction. Incorporation of shape variation into in vivo testing and in silico shoulder motion models is necessary to validate whether, how, and to what extent the identified anatomical variation directly contributes to function and dysfunction of the shoulder