Radiological and biomechanical assessment of displaced greater tuberosity fractures

For the purpose of simplification, we assumed that the humeral head is spherical, that the thickness of the articular cartilage is constant, and that the head remains centered on the glenoid cavity in the mid-range of motion. With these assumptions, the geometric center of the articular surface corresponds to the center of rotation, and all points on the humeral head move on concentric spheres. During normal glenohumeral flexion and abduction, the greater tuberosity passes under the acromion. A subacromial conflict must be suspected, if the sphere, which is tangent to the outmost point of the greater tuberosity, touches the undersurface of the acromion. Determining the center of rotation and drawing a sphere through the outmost point of a displaced greater tuberosity fragment should therefore allow predicting a possible subacromial conflict. For simplification and because CT scans are not always available for decision-making of greater tuberosity fractures, we assessed the displacement of the fragment on two-dimensional x-ray pictures rather than three-dimensional reconstructions of CT scans.

Three concentric circles were therefore drawn on standardized anteroposterior radiographs and the corresponding radii were measured. The inner circle (radius Rh) was drawn in such a way that it perfectly matched the subchondral bone of the humeral articular surface. Its centre corresponded to the geometric centre of the humeral head. The middle circle (radius Rt) was tangent to the greater tuberosity, and the outer circle (radius Ra) touched the undersurface of the acromion (Fig. 2).

In order to describe how much the greater tuberosity projects above the articular surface and how much room it occupies in the subacromial space, the relationships Rt/Rh (Greater Tuberosity Index) and (Rt-Rh)/(Ra-Rh) (Impingement Index) were calculated. Using relationships rather than absolute values enabled us to eliminate possible magnification errors on the x-ray images.

In a first step, we assessed the influence of rotation and elevation of the arm on the Greater Tuberosity Index. This was done with use of five dry humeri without osseous pathologies obtained from our Institute of Anatomy. All bones were from the right side. They were fixed one after the other on a specially designed frame, first in a vertical position and then in 30° of abduction. In both positions, the humeri were turned around their shaft axis in steps of 10° from 60° of external to 70° of internal rotation (Fig. 3). In order to compare the results with standardized radiographs taken in the clinic, neutral rotation was defined as the position, in which the epicondylar axis was 30° externally rotated relative to the X-ray cassette. In each position, a radiograph was made with the X-ray beam downwards-tilted 20° and the Greater Tuberosity Index was determined. Descriptive statistics were made using R (R Foundation for statistical computing, Vienna, Austria).

In a second step, we searched our database and determined the Greater Tuberosity Index and the Impingement Index on 80 anteroposterior radiographs of patients treated in our clinic for different shoulder pathologies (AC joint disease n = 20, frozen shoulder n = 20, calcific tendinopathy n = 20, and subacromial bursitis n = 20). Only radiographs of adult patients without previous operations and without fractures were included. There were 36 men and 44 women and 47 right and 33 left shoulders. All pictures needed to be made with the arm in neutral rotation and the glenoid perpendicular to the x-ray cassette. The humeral head needed to be perfectly centered on the glenoid cavity. All images had been made for diagnostic purposes; no radiographs of healthy volunteers were included. In order to determine the intra- and inter-rater reliabilities, two independent observers analyzed 50 radiographs at two different times. A multivariate analysis was made using R (R Foundation for statistical computing, Vienna, Austria), and intraclass correlation coefficients were determined with the SPSS software (IBM, Armonk, USA).

Ten Thiel-embalmed [13] upper extremities of unknown age and sex were obtained from our Institute of Anatomy and used according to the Guidelines of the Swiss Academy of Medical Sciences. Donors had formally agreed the use of body parts for research purposes by signing the donation forms. The skin and the deltoid muscle were removed; the coracoacromial ligament was carefully preserved. Inspection of the dissected shoulders showed a rotator cuff tear or osteoarthritis in five cases. These specimens had to be excluded. The scapulae of the remaining five specimens (one right, four left) were rigidly fixed one after the other on a specially designed jig, with the glenoid surface in a vertical plane (Fig. 4). The humeral head was manually pressed into the glenoid cavity, with the arm in a vertical position and in neutral rotation. A true anteroposterior X-ray image was then made under fluoroscopy (Ziehm vision, Ziehm Imaging GmbH, Nürnberg, Germany) with the X-ray beam downwards-tilted 20°. If the humeral head was not perfectly centered on the glenoid, the positioning procedure was repeated and the adequate image was stored as a jpeg file for further assessment. The arm of the specimen was then passively elevated and rotated while keeping the humeral head centered on the glenoid cavity. During these movements, particular attention was paid to the friction between the undersurface of the acromion and the bursal side of the rotator cuff. If the arm could be elevated without any resistance and without inferior subluxation of the humeral head on the glenoid, we considered that there was no impingement. If the bursal side of the rotator cuff got caught under the acromion or the coracoacromial ligament, a subacromial conflict was noted. Every movement was repeated several times, until the examiner (RWN) was sure if there was friction or not. Once the experiments were done with the intact specimen, an osteotomy of the greater tuberosity was made in order to simulate an isolated greater tuberosity fracture [14]. Care was taken to not damage the insertion of the rotator cuff. In each specimen, four different malpositions of the greater tuberosity fragment were examined. The fragment was either displaced 2 to 5 mm superiorly or 2 to 10 mm posteriorly and fixed to the humeral head with a screw. The displacement was measured at the tip of the fragment on the lateral side of the humeral head with use of a caliper (Fig. 5). For each malposition, an anteroposterior radiograph was made with the arm in neutral rotation (Fig. 6), and the impingement tests were repeated. All images were analyzed according to the method described above, and the results were correlated with the presence or absence of subacromial impingement.

The shape of the humeral head changed as the arm was rotated. On the x-ray pictures, the greater tuberosity was most prominent in neutral rotation and less salient in internal and external rotation (Fig. 3). Internal rotation exposed the posteroinferior aspect and no longer the highest point of the greater tuberosity, which is responsible for subacromial impingement. We therefore decided to determine the Greater Tuberosity Index and the Impingement Index in neutral rotation. In this position, the Greater Tuberosity Index of the dry humeri averaged 1.15 (range 1.11 to 1.20, SD 0.04). It did not significantly differ between 0 and 30° of abduction (p = 0.47).

The measurement technique proved to be reproducible. The intra- and inter-observer agreements were good to excellent [15] with intraclass correlation coefficients of 0.89 and 0.88 for the Greater Tuberosity Index and 0.89 and 0.85 for the Impingement Index. The results of the radiographic study are reported in Fig. 7a and Table 1. There was a significant positive but only moderate correlation between the Greater Tuberosity Index and the Impingement Index (Pearson correlation coefficient R = 0.699). Female patients had significantly higher values than male patients, for both the Greater Tuberosity Index (1.17 vs 1.13; p = 0.0002) and the Impingement Index (0.48 vs 0.46; p = 0.02). There was a non-significant positive correlation between the Greater Tuberosity Index and age (Pearson correlation coefficient 0.183) and a significant positive but poor correlation between the Impingement Index and age (0.255). Neither the diagnosis nor the side had a significant influence on these indices.

The Greater Tuberosity Index and the Impingement Index of the five cadaver shoulders averaged 1.16 (SD 0.07; range 1.08 to 1.23) and 0.51 (SD 0.15; range 0.35 to 0.66), respectively. In none of the intact specimens, a subacromial impingement could be detected. A superior displacement of the greater tuberosity fragment of 3 mm resulted in a subacromial impingement in all specimens. A posterior displacement of 5 mm caused a subacromial impingement in half of the specimens, whereas a posterior displacement of 10 mm resulted in an impingement during abduction with internal rotation in all cases. The relationship between Impingement Index and Greater Tuberosity Index is shown in Fig. 7b. If the Impingement Index was less than 0.71 and/or if the Greater Tuberosity Index was less than 1.17, no impingement could be detected. A displacement of the fragment corresponding to an Impingement Index of more than 0.75 and/or a Greater Tuberosity Index of more than 1.26 was always associated with subacromial impingement. Between these values, subacromial impingement could be present or not. The subacromial conflict occurred at very low abduction angles, typically at 20 to 30° of glenohumeral abduction.