History

Modern shoulder replacement has developed in two different sources. In Europe it has mainly evolved from massive replacement of the proximal humerus for tumour [1]. Much of this work was done at the Royal National Orthopaedic Hospital in Stanmore in the 1950s and also in Germany [2]. However when the diseased proximal humerus was removed the rotator cuff was often also excised and hence the proximal humeral prosthesis was really acting as a spacer. When attempting to use the arm actively these prostheses would sublux upwards and forwards due to the unopposed action of deltoid. Although functionally limited the long term outlook for these prostheses was remarkably good. When this group at Stanmore wished to develop a shoulder replacement for arthritis it is with this background experience of tumour work that they decided to developed a constrained joint. The Stanmore shoulder was widely used for both osteoarthritis and rheumatoid arthritis. It’s design looked remarkably like the design of hip replacements that were available at that time (The McKee metal on metal prosthesis). Both the humeral and the glenoid side were cemented and the joint was snap fit constrained. To prevent superior subluxation of the humeral component, the glenoid component was built out from the natural glenoid face and the centre of rotation lateralised. Under modern biomechanical principals it would be predicted that this would fail with the forces that would pass through the joint and unfortunately this was true. The glenoid component did loosen and the snap fit constrained design was prone to dislocation. Eventually this constrained type of joint was abandoned. However, at this time other joints were being developed that also pursued the constrained ideal, both the Castle shoulder and the Liverpool shoulder had reverse type of geometry as this was thought to improve range of motion and power by lateralising the centre of rotation.

Unfortunately all these constrained designs had the problem of long term loosening.

In the USA, Dr Charles Neer was developing a shoulder replacement for fractures of the proximal humerus [3]. He had initially been working on the fracture service at Columbia Presbyter and was impressed with the results of femoral hemi-arthroplasty for subcapital fracture. He reviewed the results of severe fractures of the proximal humerus and developed a classification of these, noting that the four-part fracture did particularly badly as the head fragment often underwent avascular necrosis way a the hip the femoral head lost its blood supply because the retinacular vessels were disrupted, the humeral head lost its blood supply. Because the retinacular vessels were disrupted, the humeral head lost its blood supply through fractures of the tuberosities and the muscle attachments. This joint was an unconstrained hemi-arthroplasty on the humeral side and represented a great advance in the treatment of this type of fracture. He always stressed from the beginning that this was a soft tissue operation and the prosthesis was there as a scaffold to rebuild the shoulder around. Having gained good results for the humeral hemi-arthroplasty he then developed a plastic glenoid component for use in degenerative arthritis. The results of this unconstrained prosthesis remained to this day the gold standard by which other shoulder replacements are judged. The development of a non-constrained joint really mirrored development of the hip replacement.

It was not long before modular prostheses were developed and then cementless bony ingrowth prostheses (Copeland, Bi-Modular, McNabb), but essentially these were all the same geometry as the Neer type design. Following the work done by Roberts & Wallace in 1991 [4] and Boileau and Walch in 1997 [5] the wide variation of anatomical morphology at the proximal end of the humerus was noted and prostheses have tried to mimic this. It was noted that the humeral head was offset on the long axis of the humerus and if prosthetic replacement was to mimic anatomy then the prosthetic humeral head must be offset. Therefore any prosthesis that was not made specifically for left or right shoulder could not directly mimic anatomy. The so-called third generation prostheses (eg Tornier) were developed in a modular way to allow for offset version and angulation. But version can vary from -5 to +55 and the equipment to do this and the stock of prostheses required makes this an extremely expensive option.

Hence the stemmed cemented prostheses have really developed in Europe from tumour replacement and in the USA from fracture work. Neither was specifically designed for the treatment of arthritis.

Development 

If a step forward in shoulder replacement is proposed then we have to review what is wrong with what we've got.

1. Problems with cement

1a. Loosening

High incidence of glenoid loosening leaves major problems on the glenoid side with unpredictable bony erosion [6-11]. So much bone may be eroded that revision may be impossible [11, 12]. On the humeral side, although loosening is rare, if it does occur then bony loss and erosion may be catastrophic such that revision may be impossible.

1b. Revision Difficulties

If a stemmed prosthesis is inserted in the wrong degree of version to remove this and revise it is a major surgical procedure, which may fracture the humeral stem and require a longer stem prosthesis to gain stability. When cement is used in the humeral shaft of an elderly rheumatoid patient the shaft may crack on impaction of the prosthesis and leakage of bone cement can cause radial nerve problems. The use of cement without a cement restrictor can allow cement to go all the way down to the elbow and preclude ipsilateral elbow replacement, eg in the rheumatoid [13, 14]. top

2. Problems with a Stem

If an intra-medullary stem is used then a stress riser must arise at the tip of the prosthesis. Shoulder replacement patients are often in the elderly group with fragile osteoporotic bones, particularly in the rheumatoid. If they are elderly and infirm with poor balance then falls are much more likely and hence a fracture at the tip of the prosthesis becomes a much more likely event. This occurs in both the hip and the knee and is now seen in the shoulder in increasing numbers.

Fractures around the stem of a prosthesis are very difficult to treat and may cause major problems with revision of surgery.

If arthritis is secondary to a malunited fracture, sometimes it is impossible to pass a stem down the humeral shaft without corrective osteotomies. This has prognostic implications for the result of the shoulder replacement [15]. On occasion the medulla of the humeral shaft may already be compromised by cement and an intra-medullary stem coming upwards from below ie intra-medullary fixation of fracture or a stemmed elbow replacement.

In cementless stem designs, reaming of humeral shaft can also cause fracture.

We have to question why a stem is used at all to fix the humeral prosthesis in arthritis. The majority of stemmed humeral prostheses extend half way down the length of the humerus. There is no scientific evidence to support why this should be so. If a fracture is not present there is no requirement for such a length of intra-medullary fixation. Obviously if there is a fracture at the proximal humerus as in the four-part fracture then a central intra-medullary stem is required as a scaffold to rebuild the shoulder around.

3. Problems with Materials

We know from long term hip and knee studies that the major long term problem with loosening is due to particulate HDP wear debris causing macrophase response and osteolysis. The same is true in the shoulder. Although the surface area of plastic in the shoulder is smaller, because the shoulder is not a captive joint with a fixed centre of rotation, the range of motion of the humeral head upon it is much greater and translational and sheer forces much more important. If one could avoid using HDP at all this would be of benefit for the long term outlook of the shoulder (i.e. hemi-arthroplasty).