Problems in humeral fractures replacement
Complex proximal humeral fractures are still challenging problems to orthopedics surgeons. It is difficult to achieve a satisfactory reduction and an effective fixation for those with severe osteoporosis or significantly comminuted fragments. A humeral head replacement can be a reasonable choice in these situations; however, results are often unpredictable[13-14]. Many authors believed that the nonunion or malunion of the greater tuberosity maybe the most common complications after the prosthetic reconstruction of complex proximal humeral fracture[15-16]. From February 2002 to December 2005, 91 patients with a complex proximal humerus fracture were treated with the humeral head replacement at Department of Sports Medicine, Beijing Jishuitan hospital were analyzed. The overall satisfactory rate was 81%. For patients with a compromised postoperative function, over 80% of them had problems with reconstruction of tuberosities. The reason why the greater tuberosity was not visible on the anterior posterior view of the X-ray film was proved to be the nonunion and posterior migration of the greater tuberosity[17]. Our experiences also suggested that the healing of greater tuberosity in an appropriate position is critical to patients’ functional recovery.

Clinical application significance
The inadequate height of the prosthesis definitely impacts the postoperative function. Thus, theoretically, an overlapping manner of tuberosity reconstruction would benefit the healing process of the tuberosity due to the increased bony contact area between the tuberosity and the humeral diaphysis. However, Boileau[18] suggested that shortening the humerus to no more than 10 mm would not influence the postoperative shoulder function. By reviewing the literatures, we were not able to find any report that compared the biomechanical characteristics between the anatomical and the overlapping reconstruction of the tuberosity. Our hypothesis is that if there is no significant difference in the biomechanical stability between the two methods, then the overlapping reconstruction should be recommended because of the increased bony contact area.
A passive range of motion exercises began from the first postoperative day. For the first two weeks, the range of motion was limited within 90° of forward elevation and 0° to 10° of external rotation. A specifically designed mounting apparatus and loading system was used in our study to simulate this passive forward elevation and external rotation exercise. Our result indicated that no significant difference was found between the two groups when the glenohumeral joint was in the positions of 30° and 60° of forward elevation (accounting for 45° and 90° of forward elevation of shoulder joint). However, a significant difference could be found regarding the greater tuberosity displacement when the shoulder was externally rotated from 40° of internal rotation to 0°. The biomechanical stability was better in the anatomical reconstruction group than in the overlapping group. This suggested that by gaining the increased bony contact area between the greater tuberosity and the diaphysis, the anti-torsion stability of the fixation might also decrease. We could not conclude whether this compromised stability could be compensated by the increment of the bony contact area. The overall effect of the overlapping reconstruction on the healing process could not be solely determined from the information provided by an in vitro cadaveric study. We concluded that the hypothesis prompted previously had not been proved according to our study. Further prospective clinical research is needed to prove that the overlapping reconstruction of the greater tuberosity is better than the anatomical reconstruction regarding the healing process of the tuberosity.
Moreover, a prominent displacement of the greater tuberosity during a passive forward elevation or external rotation could be found with either reconstruction methods. A better fixation of the greater tuberosity should not be expected in a real operation due to the possibly worse conditions of exposure and more complex fractures. To reduce the negative impact of the intro-fragment movement to the healing process of the greater tuberosity, we are considering a change in our postoperative protocol. Postponing the rehabilitation for a period of time to allow partial healing of the surrounding tissue may be helpful to keeping the stability of the greater tuberosity, and therefore increase the chance of a bone union. Now in our clinical practice, the passive range of motion exercise was postponed 2 weeks after the replacement. However, it should be noted that this delay will increase the risk of postoperative stiffness so the overall effect needs to be investigated through further clinical study.

Design of mounting apparatus and measurement of greater tuberosity displacement
As a natural organ, there is neither a definite rotation center in humeral head joints nor a central axis in humeral diaphysis, all of these result in difficulty for humeral loading. Previous studies concerning loading devices neglected unregularity of humerus, which influence the precise of experiment[19]. Here, a special load device was designed to simulate the postoperative passive range of motion exercise. And passive motion we addressed included the external rotation and forward elevation.
Mercury strain gage was used by Frankle[20] in measuring displacement. But, the obtained results were smaller because of the influence of pretensioning tensile. De et al[21] utilized an opto-electronic device to record the measurements, but the precise is unsatisfactory. Authors in this paper applying binocular CCD cameras to obtain two-dimensional images, and to perform three-dimensional reconstruction using related algorithm. 

Limitation of this study
The standard deviation within each group is relatively high due to the limited number of specimens used in this study. Although the rotator cuff muscle tension was simulated by hanging weights on corresponding pulleys, the exact in vivo biomechanical situation could not be exactly reproduced. The biomechanics of the shoulder joint was the result of the cooperation of many shoulder girdle muscles. The effects of many of these muscles could not be simulated during our study. However, the main subject of our study is to evaluate the greater tuberosity displacement happening in the early postoperative period when just a passive range of motion exercises was adopted without any involvement of active muscle traction.
In summary, our study demonstrated that the anatomical reconstruction of the greater tuberosity has a better mechanical stability than the overlapping reconstruction during a passive external rotation to neutral position. This result suggests that although an overlapping reconstruction can increase the bone healing area between the greater tuberosity and the humeral diaphysis, there may be some loss in mechanical stability as the trade-off. According to our data, even though we strictly follow the standardized postoperative rehabilitation protocol, a prominent displacement between the greater tuberosity and the humeral diaphysis was detected. Postponing the rehabilitation for a period of time to allow partial healing of the surrounding tissue may be helpful to keeping the stability of the greater tuberosity, therefore, increase the chance of a bone union.