4.9 Stemmed total shoulder replacement

Gluck of Berlin was probably the first surgeon to implant a humeral head made from ivory. However, most people recount the story of Emile Pean and how he replaced the destroyed shoulder of a young Parisian in 1893. The implant was made of platinum and vulcanized rubber. The implant had no hope of survival in the face of active tuberculosis and was removed 2 years later, the patient surviving both operations.

Credit must be given to Charles Neer for giving us the first successful shoulder replacement. The Neer prosthesis was designed for the management of complex fractures and first used in 1955. Later Neer extended its use to rheumatoid and osteoarthritis of the shoulder. A high-density polyethylene glenoid component was soon added to the metal humeral implant. The original Neer prosthesis was so well designed that for three decades it remained the gold standard.

During the 1980s surgeons were realising that a greater range of implants was needed to replicate the patient’s own anatomy. In particular, the diameter of the humeral head (at 48mm or 50mm according to manufacturer) was noted to overstuff the shoulder of small women whose native head size was 38–43mm. Thus the mono-block Neer implant was replaced by a number of modular implants. Specifically the stem and head of the humeral components were interchangeable, allowing a better fit to the patient’s normal anatomy. However, the large head diameter was kept, the neck length being reduced, and this failed to solve the problem.

A decade later careful anatomical studies by Iannotti and co-workers in the United States, Roberts and Wallace in the United Kingdom, and Boileau and Walch in France led to the understanding of variable anatomy. In particular it was noted that the head diameter varied from 38–50mm, that the version of the shoulder was peculiar to the individual (and not a constant 30 degrees), but most importantly that the centre of the humeral head was not directly in line with the centre of the humeral shaft, giving us the concept of posterior and medial offset (Figure 4.9.1). A third generation of prostheses now came on the market with variable head diameters and the ability to offset the head on the prosthetic shaft, thereby replicating the patient’s own anatomy more accurately. The price paid for this accuracy was increased complexity of the prosthesis (Figure 4.9.2).

There is no doubt that the philosophy of surgeons undertaking total shoulder replacement has changed over the years. Surgeons now understand that shoulder replacement is a soft tissue procedure as much as a bony one. A‘soft tissue procedure’ means optimizing rotator cuff function, releasing the contracted capsule, and repairing the subscapularis strongly enough to allow early movement with no risk of retear.

The aim of shoulder replacement has always been to alleviate pain, improve movement, increase function, and allow the patient to maintain their own independence. These benefits should be expected to last for a minimum of 10–15 years.

The implants used currently have a number of similarities and none can claim that their clinical outcome is any better than another. With regard to implant design there is now a move towards perfect anatomical replacement of the humeral head. This is possible with a surface replacement and increasingly it is so with stemmed implants. Whilst to date there is no evidence that this drive towards anatomical replacement has had any effect on clinical outcome, it has a beneficial effect on any glenoid component in so much that it seems to reduce migration and loosening and, as a consequence, improve longevity.

Whether a glenoid component should be inserted or not remains contentious. What can be said from a review of the literature is that any series comparing hemiarthroplasty (HA) to total shoulder replacement (TSR) does seem to indicate that total shoulders do better, at least in the short to medium term. With regard to whether a surface replacement or a stemmed humeral component should be used, there is no comparative data. It does not appear that humeral loosening of either a stemmed or a surface implant is significant. The only potential problem is the difficulty of access to insert a glenoid component with the head still in place if resurfacing is used (Figure 4.9.3).

Finally, the role of postoperative mobilization and therapy cannot be understated. It is crucial for any surgeon undertaking this type of surgery that they have access to an experienced network of therapists such that early and appropriate mobilization can start immediately after surgery.

The deltopectoral approach is standard for stemmed shoulder replacements. A skin incision is made over the deltopectoral groove along the line connecting the coracoid to the midpoint of the lateral aspect of the humerus (Figure 4.9.4). Three structures cross the deltopectoral interval: two arteries and one vein. The arteries are branches of the thoracoacromial trunk: the deltoid artery and the acromial artery. The acromial artery can be preserved, but the deltoid artery must be ligated and divided. The deltoid artery has two patterns. The commonest pattern gives off a large pectoral branch as it emerges from the thoracoacromial trunk, and then passes directly into the bulk of the deltoid muscle. The less common variant runs alongside the cephalic vein giving off a series of much smaller pectoral branches. This second pattern confuses the novice shoulder surgeon who incorrectly thinks that these are tributaries of the cephalic vein when actually they are branches of the variant deltoid artery. The deltopectoral interval is always developed medial to the cephalic vein, with preservation of its major tributaries leaving it on the deltoid muscle. The cephalic vein must never be taken off the deltoid. The cephalic vein should be preserved if possible. There is usually no need to release the deltoid muscle either proximally or distally although it is useful to free the undersurface of the muscles from the humerus and rotator cuff so that the head can be pushed back freely when it comes to exposing the glenoid. Be careful for the axillary nerve runs on the deep surface of the deltoid and can be entrapped in scar, particularly following fractures of the humeral neck. The terminal branch of the posterior circumflex artery sends a small anastomosing branch that enters the bone lateral to the insertion of pectoralis major, a vessel that can easily be torn whilst mobilizing deltoid.

Incising the clavipectoral fascia at the lateral edge of the conjoined tendon up to, or through, the coracoacromial ligament exposes subscapularis and the anterior capsule. It is often helpful to release the top half of the insertion of pectoralis major in order to gain a better exposure of the inferior capsule.

The subscapularis is incised at its insertion on the lesser tuberosity along with the subjacent capsule. The junction of the middle and lower third of the subscapularis is demarcated by the anterior circumflex vessels. These should be ligated or coagulated and divided. It is also important to leave a 1-cm stump of subscapularis/capsule for reattachment of the muscle to the humerus. Alternatively the lesser tuberosity can be osteotomized from the humerus, being reattached at the end by sutures or anchors (Figure 4.9.5).

Finally, as with all joint replacements, it is important to release contractures, particularly of the anterior and inferior capsule. Patients suffering with arthritis often suffer with significant stiffness resulting in marked contracture. It is only if these contractures are released that a significant improvement in range of motion be obtained. Plainly care must be taken when releasing the inferior capsule as this could result in damage to the axillary nerve. The latter will result in a deltoid paralysis, which is a catastrophe. Some surgeons actually dissect the quadrilateral space and place a silastic sling around the axillary nerve and its accompanying posterior circumflex artery and vein, whereas other surgeons make a point of seeing the neurovascular bundle so that it can be protected throughout the procedure. With the axillary nerve seen or placed in a protective sling the inferior capsule can be released, and this is the key to glenoid exposure and regaining a good range of movement postoperatively.

The superior capsule must now be released, including the coracohumeral ligament, and if necessary the long head of biceps (Figure 4.9.6).

The principal indications for joint replacement in the shoulder are inflammatory/rheumatoid arthritis, degenerative joint disease, secondary osteoarthritis, avascular necrosis, and trauma. Added to this in a special category is cuff tear arthropathy and this is discussed in Chapter 4.2.

In rheumatoid or inflammatory arthritis the disease process is typically erosive with loss of bone and cyst formation. In this condition the glenoid can be markedly eroded anteriorly and superiorly. In addition there is often thinning of the rotator cuff with superior subluxation of the humeral head. Erosion of the glenoid may be of such severity that it is insufficient to support a glenoid component. Any bone cysts should be curetted and filled with either bone graft or cement. Again any soft tissue contracture should be released. Repair of the rotator cuff may not be feasible because the tendon tissue is so poor.

Degenerative joint disease can present with significant deformity of both the humeral head and glenoid. The glenoid is eroded posteriorly with posterior subluxation of the humeral head. The latter is often flattened anteriorly. Added to this there can be a number of large osteophytes surrounding the whole of the humeral head with loose bodies either inferiorly or posteriorly. Capsular contractures are common and will need to be released at surgery.

The treatment of patients with avascular necrosis of the humeral head can be rewarding at least in the short term. In most cases a HA is all that is required and good results can be expected. Patients are often relatively young and the soft tissues around the shoulder are well maintained. Because of their young age these patients may need regular and long-term follow-up as at some stage revision is likely.

Joint replacement in post-traumatic arthritis is probably the most difficult and is often unrewarding. The normal anatomy can be so distorted that the surgeon can experience significant difficulties navigating the joint. In these cases the glenoid may not be damaged and may not require replacement. Paramount to the success of the operation is reconstruction of the fractured tuberosities. These must be re-attached strongly to the humeral shaft in the correct anatomical location allowing movement to resume as soon as possible after surgery. If the tuberosity has to be osteotomized and reattached this can be both difficult and lead to diminished function.

The role of shoulder arthroplasty in acute trauma falls outside the remit of this chapter. Plainly an implant should only be used in cases that cannot be treated by open reduction and internal fixation or where there is a significant risk of avascular necrosis. Surgical techniques used are significantly different from those used in arthritis. Obtaining correct humeral length, alignment, and version are critical to success. Further difficulties arise from the fixation of the greater and lesser tuberosities to the prosthesis and humeral shaft. Bony union can also be unpredictable. Implant manufacturers have responded to these difficulties and their implants have been modified for use in the trauma scenario.

Shoulder arthroplasty is indicated in all the earlier discussed clinical situations, in patients who are suffering sufficient pain, loss of movement, weakness, and reduced function. Pain remains the principal indication particularly if the patient is having difficulty sleeping and requires strong analgesia on a regular basis. Patients who undergo total shoulder arthroplasty do improve the range of motion of their shoulder, but surgeons should beware operating for this symptom in isolation. A patient with a stiff yet pain-free shoulder will not do as well as one with a painful yet mobile joint.

Before embarking on shoulder arthroplasty it is important to assess the status of the bone stock, not only of the humerus but especially the glenoid. Is there sufficient bone stock to support a glenoid component? This can be assessed by radiographs, particularly an axillary view, or by computed tomography scan. It is important to allow for any correction of deformity. Any deficit in the rotator cuff should also be assessed as this may affect outcome.

Patient motivation is an important factor in the success of the operation. The patient has to be committed to a surgical procedure and the lengthy postoperative rehabilitation programme. The patient must have a realistic expectation and an ability to follow advice as appropriate. The experience of the surgeon and supporting team, particularly physiotherapists, are vital factors. Finally, the least significant factor is the prosthesis itself.

The principal contraindication to surgery is lack of bone stock to support the prosthesis. This usually affects the glenoid side. If the deltoid muscle is not functioning then any arthroplasty will be unsuccessful. The better option here would be an arthrodesis.

Rotator cuff deficiency may cause difficulties. If the shoulder is concentric on the glenoid (Seebauer I) with a small tear (less than 2cm) a standard shoulder replacement is effective, but if the tear is large then a CTA head or a reverse shoulder can be used. If, however, the shoulder is subluxed upwards (Seebauer II) then a reverse implant should be used (Figure 4.9.7).

Infection is a definite contraindication. Infection will usually be encountered in the revision situation. Again this contraindication is not absolute in so much that two-stage surgery can often be successful. The first stage involves the removal of any metalwork or prosthesis and the debridement of soft tissues and bone. The second stage involves insertion of the prosthesis covered by appropriate antibiotics.

Finally, a neuromuscular disorder can make the outcome of TSR unpredictable. With good medical management in conditions such as epilepsy and Parkinson’s disease the results of surgery can often be quite gratifying.

The insertion of a glenoid component is still controversial (Figure 4.9.8). Long term, the problems are that HA may lead to glenoid erosion and TSR may lead to wear or loosening of the glenoid component.

In the interim the patient wants to have the best result in terms of pain, movement and function. There is no doubt that TSR wins out over HA in the short to medium term. There are four level-one evidence studies (prospective randomized controlled studies) that show this. Gartsman’s study (2000) on 51 randomized patients, operated on by the same surgeon, using the same prosthesis and followed for 3 years with UCLA (University of California–Los Angeles) and ASES (American Shoulder and Elbow Surgeons) scores showed that the TSR had better pain relief, better patient satisfaction, better function, and better strength. Three of the HAs came to revision for painful glenoid erosion during the study, but none of the TSRs came to revision. Kirkley’s study (2005) on 41 randomized patients also had three HAs come to revision for painful glenoid erosion during the study, but none of the TSRs were revised. The study by Sandow et al. (1998) on 32 randomized patients showed that the TSR group had significantly less pain, and the study by Jónsson (1998) on 49 randomized patients showed significantly better Constant scores for the TSR group. Of course all these studies suffered from low numbers so Kirkley did a meta-analysis (2005) combining these randomized studies and showed that the TSR group had significantly better pain relief, significantly better function, and better movement than the HA group (Figure 4.9.9).

There is also a lot of level 2 and 3 evidence (not randomized) that TSR fares better than HA. In our study (Haines and Trail 2006) on 124 patients, both groups improved significantly and equally, with a small trend towards better results for pain and movement in the TSR group but the major cause for revision was glenoid pain due to erosion in 12% of the HA group. The Tornier group (2003), in a study of 601 TSR compared to 89 HA, showed good to excellent results in 94% of the TSR group compared to 86% of the HA group, with the TSR group having better scores for pain, mobility, and activity, and their conclusion was that TSR was better than HA. Finally Bigliani and co-workers (2007) did a meta-analysis that included 23 non-randomized studies with 1952 patients and showed that the TSR group had significantly better pain relief, significantly better elevation, significantly better gain in elevation, significantly greater gain in external rotation, and significantly better patient satisfaction. There is no more to be said.

So what about the late results? Well there is no doubt about painful glenoid erosion following HA. It was bad enough for revision surgery in 12% of our study, 12% of Gartsman’s study, and 12% of Kirkley’s study. In Sperling and Cofield’s long-term study 72% of the HA group had glenoid erosion on radiographs at 10 years.

And what of glenoid revision following TSR? Sperling and Cofield had a 97% survival of their glenoids at 10 years. In the Radnay and Bigliani meta-analysis of 1952 patients only 1.7% of sockets needed revision. In Bunker’s study with a minimum 10-year follow-up there was a 94% survival for the glenoid. The TSR versus HA debate will continue to rumble on but evidence is now stacking up for the superiority of TSR in the short to medium term, and greater survival of TSR in the long term.

What is also of note is that improvements in glenoid component design, preparation of the glenoid, fixation including cementing techniques, and, interestingly enough, an anatomical humeral head replacement have resulted in a lower incidence of migration and loosening and potentially better long-term survival.

Whilst it is difficult to prove, I have no doubt that the results of shoulder replacement continue to improve and certainly have become more reproducible. This is a testament to all of the surgeons who have helped develop prostheses as well as the techniques of surgery. With regards to the results themselves, these are now being reported from all corners of the globe. Generally these are extremely favourable, although interpretation can be difficult given the lack of standardization of outcome evaluation. In Europe the Constant score, and more recently the Oxford score, and in the United States the ASES score have been used. Unfortunately these are not interchangeable. As such it is probably more sensible to break the scores down into their various components, particularly pain relief, range of motion, strength, and finally function.

It is not possible to report on every single article that has been written on shoulder arthroplasty. As such I have chosen six with the largest numbers and the longest follow-up. These have been summarized in Table 4.9.1. As can be seen in most series the authors report extremely good pain relief and an average range of motion of between 100–110 degrees of abduction and 110–120 degrees of forward flexion; internal rotation to the buttock or lower lumber spine and an external rotation of 40–50 degrees. As a result of this there is a significant improvement in function. Perhaps more importantly authors are now reporting that these improvements continue for 10 or 15 years.

As with all joint replacements, whilst most patients can expect a good to excellent result some will get a complication. These can arise during the procedure, immediately afterwards and in the long term (Box 4.9.2).

Peri-operative complications include nerve or blood vessel injury, peri-prosthetic fractures, and, on occasion, implant-related problems.

The axillary and musculocutaneous nerves are always at risk during shoulder replacement. At a recent audit at Wrightington Hospital (Wigan, United Kingdom) we were able to identify 17 instances of neuropraxia of various nerves after shoulder arthroplasty. As all these patients had concomitant supraclavicular nerve blocks it was not possible to know clearly whether these injuries had occurred as a direct result of surgery or the block itself. Fortunately all these recovered although one took 12 months. More significant and permanent injuries, however, can occur, particularly to the axillary and musculocutaneous nerves. The axillary nerve can be damaged during an inferior capsular release. The musculocutaneous nerve lies on the undersurface of the short head of biceps and can be damaged when this muscle is retracted. Damage to either nerve can leave the patient with a significant functional deficit. It is therefore important that any damage should be identified at the time of surgery and rectified. Postoperative neuropraxia should be monitored, and if at 6 weeks there has been no improvement then electromyographs obtained. The latter will often distinguish between neuropraxia and definitive nerve damage. For the latter it is important that further exploration and nerve repair is undertaken as appropriate. Finally, later management might involve some form of tendon transfer.

Fortunately vascular injuries following TSR are rare. The author has seen two in over 17 years of consultant practice. As would be expected, diagnosis is relatively straightforward. Indeed if prompt help is summoned by way of vascular expertise including an arteriogram and bypass grafting, then a good outcome can be expected. It should also not be assumed that this compromise has occurred as a direct result of surgical trauma. In one case in the author’s practice intraluminal occlusion was found quite distal to the surgical site. It was assumed that this was ‘an accident waiting to happen’ perhaps precipitated by external rotation of the arm. Whatever the circumstances the surgeon should not hesitate to seek urgent advice and must not prevaricate in the vain hope that things will improve spontaneously.

Periprosthetic fractures can include fractures of the glenoid but more commonly the humerus, the latter as a result of overzealous manipulation of the upper extremity during exposure, particularly external rotation, but also inadvertent reaming and finally impaction of the prosthesis. The commonest follows external rotation and takes the form of a spiral fracture. Plainly great care should be taken when the humeral head is exposed, particularly in patients with fragile bone. Management of this problem is fortunately relatively straightforward and involves the insertion of a longer-stemmed humeral component and the application of cerclage wires (Figure 4.9.11). It is, however, important that the stem of the humeral component extends beyond the end of the fracture by at least two diameters of the radius of the humerus. Interoperative fracture of the glenoid, however, would probably preclude the insertion of a glenoid component. Fixation should be attempted. In most circumstances this can be undertaken by the use of a lag screw.

Postoperatively there are two major complications: infection and instability. Infection after TSR is relatively uncommon with most published series reporting an incidence of between 1–2%. As with all arthroplasty, one must assume that the implant was contaminated at the time of surgery. Infection can present immediately after surgery or somewhat later (12 months after surgery). A number of factors increase the likelihood of infection, particularly previous surgery, diabetes mellitus, and systemic corticosteroids. Multiple organisms have been described, particularly Propionibacterium acnes, Staphylococcus aureus, and Staphylococcus epidermidis and even Candida. P. acnes is an aerotolerant anaerobic Gram-positive rod that is sensitive to clindamycin, erythromycin, tetracyclines, piperacillin, and penicillin G. It is sensitive to ultraviolet light. It is a commensal in the skin of the face and shoulder girdle, accounting for the incidence of acne in these sites. It is therefore a specific risk to shoulders and not to the hip and knee and is difficult to grow in the laboratory. In a significant number of cases no organism can be identified. Continuing antibiotic use or poor bacteriological techniques may account for this. The C-reactive protein (CRP) and white cell count (WCC) can be normal in 40% of revision cases where infection is later confirmed by biopsy. When infection has been confirmed either by clinical means, cultures, or blood tests, then two-stage revision surgery is the only remedial treatment. The first stage involves removal of the implant and the insertion of an antibiotic-loaded cement ball and stem which fits down the humerus like a HA or packing the space with gentamicin beads. The second stage is removal of the antibiotic-impregnated cement spacer and the reinsertion of a new prosthesis. In the author’s experience this type of surgery is relatively successful. It seems to result in a resolution of the infection leaving the patient with a relatively pain-free and mobile shoulder joint.

The time interval between the two stages of surgery depends on the clinical situation but should be a minimum of 3 months. Prior to the second stage it is crucial that the patient’s wounds have healed and that there has been no recurrence of infection. Added to this, blood parameters including erythrocyte sedimentation rate, CRP, and WCC should all be normal. If there is any uncertainty then an open biopsy of soft tissues and bones should be undertaken to exclude infection prior to the second stage.

The incidence of dislocation varies widely (Figure 4.9.10). In most cases it arises as a result of poor surgical technique, that is, malalignment of the components, particularly excessive humeral retroversion, excessive soft tissue release, or poor reconstruction. Version is variable but the average humeral retroversion is 21 degrees. Glenoid version has to be performed by naked eye, for it is difficult to measure even radiographically.

With regard to the soft tissues, release of contractures is key to the procedure but must not lead to damage of the rotator cuff or subscapularis. At the end of the procedure the subscapularis muscle should be firmly attached to the lesser tuberosity. Thereafter, therapy whilst allowing movement should protect this repair, at least in the early 4–6-week period after surgery.

In the author’s opinion instability can be extremely difficult to treat. If revision surgery is contemplated this should involve an assessment of the alignment of the prosthesis and muscles around the shoulder. Whilst realignment of the humeral component is relatively straightforward, that of the glenoid can be more difficult. The removal of a cemented glenoid component can result with insufficient bone being left for the insertion of a second. Treatment is very difficult; the alternatives include insertion of a corticocancellous bone graft or impaction grafting technique where appropriate. The results of these are unpredictable and leaving the patient with a HA will not correct instability.

Instability can occur after failure of the subscapularis repair. Mobilization and reinsertion of the subscapularis can be successful. This can be reinforced by a pectoralis major transfer. Of all revision surgeries for instability however, in the author’s experience, the most successful has been the revision of an anatomical shoulder replacement to a reverse implant. This can stabilize the shoulder but it relies on the presence of good glenoid bone stock to support the glenosphere.

Failure of the rotator cuff is common in the elderly patient a decade on from surgery. This can occur after minor trauma and result in the patient presenting with a sudden loss of movement, particularly abduction and external rotation. In the author’s experience the clinical picture is usually typical. In the first instance treatment is often a course of physiotherapy. Cuff repair in the degenerate cuff of the 80-year-old following shoulder replacement is doomed to failure, and one way out may be revision to a reverse polarity replacement.

Loosening of the glenoid component, may be the current rate-limiting step in TSR (Figure 4.9.11). Glenoid lucent lines remain a concern. Historical papers report a high incidence of lucent lines, but these could be detected in the recovery room and were thus a sign of poor surgical technique. However, studies using modern cement techniques are also of concern. Mansat had lucent lines in 67%, but complete lucent lines in only 35%, yet had no case of glenoid loosening and no revision for loosening. The Tornier multicentre group study had lucent lines detectable on day 1 in 60% of cases (a sign of poor technique) and 25% were progressive. The good news is that despite these high figures for lucent lines, glenoid revision rates are low with a figure of 1.7% in the meta-analysis of 1952 cases undertaken by Bigliani and co-workers.

When the glenoid loosens, management means revision surgery; the principal indication being pain. With the newer modular implant the ease of surgery has increased because the head can be separated from the stem of the humerus; this makes any approach to the glenoid easier to undertake. The removal of cemented mono-blocks from the humerus, however, is notoriously difficult. Reconstruction of a glenoid relies on either the insertion of a corticocancellous graft or impaction grafting techniques. These are highly specialized and should only be undertaken in units familiar with these techniques. This type of surgery involves a significant amount of time, resource, and equipment. If reconstruction of the glenoid is not possible then the patient will have to be left with a HA.