Introduction. Arthritis of the glenohumeral joint is usually associated with erosion and flattening of the articular surfaces. The aim of this study was to evaluate the influence of the articular flattening on the joint reaction forces and the humeral head translations during abduction and rotation.

Method. Analysis was conducted with a 3D finite element model of the shoulder, including the scapula, the humerus and 6 muscles: middle, anterior and posterior deltoid, supraspinatus, subscapularis, and infraspinatus. Both the glenoid and humeral head were eroded to artificially reproduce the flattening of an arthritic joint. Two situations were studied:

1) an intact joint with a radius of curvature of 24mm for the humeral head and 26mm for the glenoid;

Movements of external rotation (0–45°) and abduction (0–150°) were performed by muscles’ activation. Contact forces caused by muscles wrapping on bony surfaces were accounted for. Joints forces, glenohumeral contact point locations and humeral head translations were calculated for the intact and eroded joint.

Results: For the eroded joint, articular forces were up to seven times higher during rotation and five times higher during abduction. For the intact joint, the glenohumeral contact point and humeral head remained centred. On the other hand, for the eroded joint, eccentric contact points with large antero-posterior and supero-inferior humeral head translations were observed. Animated views showed that this fact was clearly related to the rocking-horse effect.

Conclusions: This study showed that flattening of the glenohumeral joint due to osteoarthritis increases dramatically the articular forces and humeral head translations. This phenomenon is by itself responsible for progression of the joint’s erosion and flattening and acts as a vicious cycle. It also partly explains the reduced range of motion observed clinically. Accordingly, to limit the risks of rocking-horse effect after shoulder arthroplasty, the joint’s reconstruction should restore a natural articular radius of curvature, with a centre of rotation in the middle of the humeral head.

Bone tissue is known to adapt to a stress change with some time delay. In vivo experimental studies were conducted for measuring the effects of mechanical loading on bone remodelling. In parallel, numerous models were developed for simulating the long-term bone response to various physical activities. However, most of models neglected the delay of bone response and they were not fully identified with corresponding experimental measurement. The purpose of this work was to develop a model describing the delay between stress change and cortical bone response.

A mathematical model was developed, accounting for the delays for bone response to stress. For in vivo experiment, 80 female Wistar rats (9-week old) were randomly divided into a running and a control group. First group regimen consisted of treadmill running program: 1 hr. per day, 6 days a week during first 15 weeks (treadmill speed 1.6 km/h). At week 15, the running group rats were returned to normal activity (sedentary state in cages), during last 15 weeks. Rats of the control group were subjected to normal activity for each period. At week 0, 3, 7, 15 (end of running period), 16, 18, 22 and 30 (end of experiment), 5 rats of each group were sacrificed for measuring the bone relative density via micro-hardness measurement on the left tibia (60 points per tibia).

Bone density of running group increased asymptotically during the first 15 weeks. An abrupt decrease of density occurred when rats returned to sedentary state at week 15. The densification rate is ten times lower than the rate whereas bone formation delay (13 days) is greater than bone resorption delay (1 day). These delays were related to the delays of bone cells activities with mineralisation process in reaction to physical activities.