Surgeons have always preferred limb salvage surgery for patients with extremity tumors, and even before adjuvant chemotherapy and the use of irradiation in association with surgery, approximately a quarter of patients with an extremity tumor requiring surgical resection had a limb salvage operation. Before the 1970s, almost all osseous reconstructions involved the use of autogenous bone grafts and the associated joint was either managed with arthrodesis or left flail. During the 1970s, as total joint replacements became functional, custom endoprostheses were used to replace resected bone and restore joint function. The initial endoprostheses were not well designed and performed poorly. It was not until the mid-1980s that endoprostheses with improved mechanics and modularity were available. These authors have collected follow-up data from patients treated with the early custom endoprostheses and from those treated with the modern modular endoprostheses.
To borrow a principle of internal fixation of fractures, the success of an endoprosthesis is a race between its failure and the patient's death. Most endoprostheses eventually fail because the patients are, for the most part, young and active, but we expect the endoprostheses to function for at least fifteen years and, we hope, longer. That only a quarter of the endoprostheses in this review failed is due to a combination of patients dying of their disease before failure and insufficient time since implantation for failure to occur in others. Despite this, having a better understanding of how the endoprostheses fail is important. A classification system for failures is valuable so endoprosthetic failures can be recorded in a similar manner. I think the suggested classification system is acceptable and recommend its use, although I have some suggested changes.
One change is to eliminate any mention of the two major categories: mechanical and nonmechanical. These are unnecessary and confusing. The use of these two categories results in soft-tissue failure (Type-1 failure), aseptic loosening (Type-2 failure), and structural failure (Type-3 failure) being lumped together into the mechanical category. Joint instability, the symptom of soft-tissue failure, is not a variable associated with the endoprosthesis. It is the consequence of inadequate surgery, particularly in the case of proximal humeral replacements. Without soft-tissue augmentation to keep the prosthetic humeral head located, virtually all prosthetic shoulders will be unstable. The unstable reconstructed shoulder was accepted for many years; however, for approximately a decade, most surgeons have used synthetic materials to secure the humeral head into the glenoid to eliminate instability1. Aseptic loosening (without occult infection) (Type-2 failure) is also more of a surgical technique failure than endoprosthetic failure, which is what the data from our total joint replacement colleagues suggest2. Local recurrence (Type-5 failure), also not an endoprosthetic failure or a surgical failure, more likely is a biologic phenomenon.
A second suggestion is to separate periprosthetic fractures and prosthetic fractures. They probably have different causes and different treatments. Endoprosthetic fractures were not uncommon with early endoprostheses but now are rare. Type-4 failure could be periprosthetic fracture, and Type-6 failure could be endoprosthetic fracture.
A third issue is related to a type of failure the authors excluded and did not classify as a failure, namely, planned minor revisions of endoprostheses. Most of the modern endoprostheses have polyethylene bushings and bumpers that wear and/or fail (Figs. 1 and 2). These can be replaced but require an open arthrotomy. These types of failures, which would be a Type-7 failure, should be included for a complete understanding of endoprosthetic failures.
The value of the authors’ failure data is problematic. Even though the authors collected failure data on a large number of patients from different centers, the patient population is too diverse to draw meaningful conclusions. Included are patients with benign osseous tumors, primary malignant tumors, and metastatic tumors to bone. Some patients were treated with systemic chemotherapy, some with local irradiation, and some with neither. Many of the patients were young adults, while others were senior citizens. All of these variables influence endoprosthetic failure. The elderly patient with metastatic carcinoma who had progression of disease after internal fixation and irradiation who then has an endoprosthesis should not be compared with an eighteen-year-old patient who has an endoprosthesis for an osteosarcoma. This is not to say that the information the authors collected does not have value, but it must be interpreted with caution.
The authors suggest that the results for proximal femoral and proximal humeral endoprostheses differ from those for a endoprosthesis around the knee because the hip and shoulder are polyaxial and the knee is uniaxial. This may be true, but I am not convinced. There are other differences between the shoulder and hip and the knee. There are also differences between patients needing replacement of the proximal part of the femur or humerus and the distal part of the femur or proximal part of the tibia. These differences influence the failure of endoprostheses. For example, patients needing a proximal femoral replacement or proximal humeral replacement are more likely to have metastatic carcinoma than a patient needing a distal femoral or proximal tibial replacement. The hip and shoulder have better soft tissues covering them compared with the knee.
The authors are to be congratulated for the major accomplishment of collecting and analyzing data from five separate hospitals and for suggesting a classification system for endoprosthetic failures. We need to understand how, why, and when endoprostheses fail, and a classification system allows better and more accurate reporting so we can advise patients who require these implants.