Gerald R. Williams Jr, MD, is the John M. Fenlin, Jr Professor of Shoulder and Elbow Surgery at The Sidney Kimmel Medical College of Thomas Jefferson University in Philadelphia. He is also the Co-Chief of Orthopaedic Surgery at Main Line Health System outside of Philadelphia. Dr. Williams completed his internship and residency in Orthopaedic Surgery and did his Fellowship training in shoulder surgery at the University of Texas Health Science Center in San Antonio with Charles A. Rockwood. His research interest includes reverse and anatomic shoulder arthroplasty.
Shoulder arthroplasty is the fastest-growing segment of the joint replacement market in the United States and probably the world. This is clearly because of the predictable pain relief and improved function following both anatomic and reverse total shoulder arthroplasty. Reconstruction of substantial bone loss increases surgical difficulty and may result in less predictable outcomes. Moreover, bone loss is common in primary cases, both anatomic and reverse, as well as revisions. The prevalence of glenoid bone loss in osteoarthritis, for example, has been estimated to be near 40%.
The types of bone loss vary as a result of the clinical scenario. In the primary anatomic setting, structural humeral bone loss is uncommon. Although central and anterior glenoid bone loss may occur, posterior deficiency is much more common and is associated with posterior humeral subluxation. Substantial humeral deficiency is also uncommon in primary reverse arthroplasty, except in cases involving sequelae of fractures. However, superior glenoid deficiency is common. Revision arthroplasty has a much higher incidence of both humeral and glenoid deficiency, making these cases among the most challenging. Classification systems have been described for all of these scenarios and are beyond the scope of this chapter.
Traditionally, bone insufficiency has been addressed through the use of bone grafting. Cancellous bone has been used in situations in which substantial structural support is not required. The bone in these cases is used to fill a void or to provide a load-sharing role ( Fig. 64.1 ). Cortical or corticocancellous composites have been used when structural or load-bearing support is required because of substantial cortical bone loss. When available, autograft is preferred. The resected humeral head is often the best source. However, in some settings, especially revisions, allograft may be the best or only source of adequate bone.
Specific grafting techniques have been described to address bone loss in primary anatomic replacement. Early arthroplasty techniques included cement fixation on both the humeral and glenoid sides. As the technique evolved, we followed the lead of our hip and knee colleagues and moved toward cementless fixation, especially on the humerus. Long-stem devices that sought to attain diaphyseal fixation resulted in cases of substantial proximal stress shielding. As a result, stems have become shorter and less canal filling. Today cancellous bone from the resected humeral head is commonly used for impaction grafting to improve the press fit between the humeral bone and prosthetic stem. Occasionally, especially with a varus humeral cut, the humeral component may sit too low in the native metaphysis to allow the humeral head to be placed in the center of the humeral cut, even with an offset taper. Under these circumstances, in order to avoid cementing the stem, a piece of corticocancellous bone from the resected head can be placed inferomedially to shift the head superolaterally.
Glenoid grafting has also been used during primary anatomic arthroplasty. Occasionally, large glenoid cysts may require nonstructural cancellous bone grafting from the resected humeral head during glenoid component placement. In addition, posterior glenoid grafting using a structural corticocancellous autograft from the resected humeral head is also performed in cases of severe posterior bone loss and posterior subluxation ( Fig. 64.2 ).
Primary reverse arthroplasty has seen the same trend toward cementless humeral fixation as anatomic arthroplasty. Consequently, the use of nonstructural, cancellous autograft impaction to improve the initial press fit is common. Glenoid deficiency in primary reverse arthroplasty is also managed with structural, corticocancellous grafts from the resected humeral head ( Fig. 64.3 ). The initial fixation is often facilitated by the ability to capture the graft with the metal reverse baseplate and screws.
Structural bone grafts are used much more commonly in revision arthroplasty. When implants fail early from causes such as instability or subscapularis deficiency, bone loss is often mild. However, when implants fail because of aseptic loosening, or late infection, the amount of glenoid and humeral bone loss can be substantial. In order for nonstructural, cancellous bone grafts to be successful during glenoid reconstruction, the glenoid components must be supported on the rim of the remaining glenoid or the floor or walls of the vault and fixation within the remaining bone must be adequate. This allows the defect to be filled with cancellous bone and to be incorporated without having to provide structural support. In many situations, these cavitary defects are complicated by loss of the cortical rim as well as areas for good peripheral or central fixation. Under these circumstances, structural grafts are required to provide support. Autografts from the iliac crest, the distal clavicle, and the coracoid have been described. Allografts from the femoral head, humeral head, and other sources may be required if sources of good native bone are unavailable.
Proximal humeral bone loss is also relatively common in the revision setting. The criteria for the amount of bone loss large enough to require grafting are not clear. However, cases of severe bone loss are often reconstructed with proximal humeral allografts ( Fig. 64.4 ). Various surgical techniques have been described and may differ according to the method of fixation of the graft to the host bone and the presence or absence of rotator cuff attachments. ,
The reported results of bone grafting during shoulder arthroplasty are somewhat mixed. Nonstructural grafts have been almost universally successful. Impaction grafting has resulted in very low rates of humeral loosening, regardless of whether the source is autograft or allograft. Structural grafts in the primary anatomic setting, such as posterior grafting in osteoarthritis, have been shown to provide varying amounts of correction of the subluxation and version abnormalities, but in some series have had a high rate of hardware failure and early glenoid loosening. , One recent series documented excellent correction, low rates of hardware complications, and low rates of loosening at short- to midterm follow-up.
Structural grafts for glenoid deficiency in the primary reverse arthroplasty setting have been mostly successful in short- and midterm follow-up studies. , However, reports of baseplate loosening, signs of graft resorption, and catastrophic baseplate failure have been described in some series ( Fig. 64.5 ). , The same is also true in the revision setting. It is likely that factors such as completeness of graft seating, fixation of the graft, the amount of baseplate support, and the amount and quality of fixation in the native glenoid bone influence failure rates. It is clear that, for a structural graft to succeed, it must heal to native bone and exhibit little graft resorption. In my experience, large structural grafts have failed more often than in most other series. Perhaps some aspect of surgical technique may have had an influence. However, when large structural grafts are used, especially if they are allografts, the time that the construct is strongest is time zero, and it is likely to remain the same or get worse with time.