Patients
The institutional review board approved this study. Our Total Joint Database was used to identify all patients who were treated with a permanent resection following elbow arthroplasty. Between 1975 and 2005, 1637 primary and revision total elbow arthroplasties were performed at our institution, and 141 were identified as having a deep infection (an 8.6% infection rate). Of the elbows that had a deep infection following total elbow arthroplasty, fifty-one elbows in fifty patients underwent elbow resection (a 36% resection rate). The patients were contacted by mail or telephone for participation in this study. Sixteen patients had died at the time of follow-up, and five patients declined participation in this study. The remaining twenty-nine patients (thirty elbows) were evaluated at an average follow-up time of eleven years (range, 2.7 to twenty-eight years). All fifty-one elbows were included in the analyses of complications, eradication of infection, and early outcome. Long-term clinical outcome is reported only for the twenty-nine patients available at the time of the most recent review.
There were thirty-two female and eighteen male patients with a mean age (and standard deviation) of 52 ± 15 years (range, seventeen to eighty-nine years) at the time of the index elbow arthroplasty. The mean age at the time of resection was fifty-nine years (range, twenty-five to ninety years). The time elapsed between the index elbow arthroplasty and the resection averaged 6.6 years (range, one month to twenty years). The average number of revisions prior to resection was 0.7 (range, zero to five revisions), and the average number of revisions after resection was 1.5 (range, zero to eleven revisions).
Forty-five of the index elbow arthroplasties had been performed at our institution, and six had been done elsewhere. The definitive resection was performed by one of seven surgeons. The initial diagnosis leading to the index elbow arthroplasty included rheumatoid arthritis in twenty-seven elbows, juvenile rheumatoid arthritis in three elbows, posttraumatic arthritis in fifteen elbows, osteonecrosis of bone in two elbows, osteoarthritis in two elbows, Charcot arthropathy secondary to a syringomyelia in one elbow, and postradiation Ewing sarcoma in one elbow.
The implants removed included twenty-four linked and twenty-six unlinked prostheses. The linked prostheses included Coonrad-Morrey implants (Zimmer, Warsaw, Indiana) in fifteen elbows, Coonrad (Zimmer) in seven elbows, and GSB (DePuy, Warsaw, Indiana) in two elbows. The unlinked prostheses included Pritchard-Walker implants (DePuy) in ten elbows, Mayo (Howmedica, Rutherford, New Jersey) in nine elbows, Capitellocondylar (Codman, Randolph, Massachusetts) in five elbows, and Latitude (Tornier, Saint-Ismier, France) in two elbows. One of the implants could not be identified.
All elbows included in the study had a deep infection at the time of resection. Deep infection was defined as a positive intraoperative culture and/or gross purulence with histological evidence of acute inflammation. The infection may have involved soft tissue or bone and was either acute or chronic. Table I summarizes the different microorganisms cultured. Staphylococcus species (twenty-eight elbows; 55%) was the most common bacterial organism cultured. Wound-healing problems were also reported in twelve elbows (24%) prior to resection.
Resection was considered when deep infection was resistant to medical treatment with intravenous antibiotics and surgical irrigation and debridement. Elbows were also resected if patients were unable or unwilling to undergo multiple surgical procedures including reimplantation. Twenty-nine elbows (57%) had an elbow resection following an infection at the site of a primary total elbow arthroplasty, while twenty-two elbows (43%) had at least one revision (range, one to five revisions) prior to elbow resection. Thirty-seven elbows (73%) were diagnosed with a deep infection after the primary arthroplasty and were either resected at the time of the first reoperation or eventually resected after failed irrigation and debridement or failed component exchange. Fourteen elbows (27%) underwent a revision surgical procedure for an indication other than infection and eventually became infected, requiring elbow resection as a salvage procedure. The indications for revision surgery in these fourteen elbows included periprosthetic fracture (five elbows), aseptic loosening (five elbows), instability (three elbows), and polyethylene wear (one elbow).
Elbow Resection: Surgical Technique
The goals of resection as a definitive treatment for infection at the site of an elbow arthroplasty include removal of all infected tissue and implant material including cement, preservation of as much bone as possible, protection of neurovascular structures, and careful soft-tissue closure. We believe that preservation of the medial and lateral condyles is of paramount importance to achieve a stable elbow resection.
The previous skin incision is most commonly used for exposure. The wound edges are resected only when devitalized. The ulnar nerve (commonly transposed in this setting) is identified and protected, but it does not need to be circumferentially dissected and retracted unless symptomatic or when necessary for a safe exposure. We prefer to leave the triceps attached to the ulna whenever possible. The medial and lateral aspects of the triceps are identified and elevated from the posterior aspect of the humerus. The locking mechanism of the Coonrad-Morrey arthroplasty implant is easily uncoupled by removing a limited amount of bone from the anterior aspect of the condyles. Uncoupling other implants may require removing more bone or splitting the triceps.
The humeral and ulnar components are easily removed when loose, but extraction of well-fixed components and/or cement requires special attention and careful technique in order to avoid intraoperative fracture. During component and cement removal, care must be taken to avoid violation of the humeral and ulnar shafts. In addition, care should be taken to avoid fracture of the humeral condyles when present. Removal of all intramedullary cement should be attempted with care so as not to compromise the host bone, as it may act as a residual source of infection.
On the humeral side, our preferred technique to remove well-fixed implants and cement involves creation of a posterior humeral window. The window is trapezoidal in shape and wider distally than proximally. The size and shape of the window are designed to minimize weakness of the medial and lateral condyles, and it extends approximately two-thirds of the length of the stem (Fig. 1). After removal of the component and cement, the posterior window is secured with number-5 monofilament absorbable suture. On the ulnar side, cement is preferentially removed with a high-speed burr and flexible reamers. Occasionally, a longitudinal osteotomy may be created along the subcutaneous border of the ulna to extract well-fixed implants and cement.
Once the implants and cement are removed, the ulna is stabilized with respect to the distal end of the humerus. Adequate soft-tissue releases are necessary in order to balance the proximal part of the ulna within the confines of the distal end of the humerus. In the ideal situation, osseous stability is achieved by articulation of the condyles within the olecranon fossa (Fig. 2). In the absence of sufficient osseous stability, the ulna and the humerus may be stabilized with heavy sutures through bone. When accidentally fractured, the condyles may also be stabilized with sutures or Kirschner wires.
Postoperative Protocol
Postoperatively, the goal is to facilitate mature scar formation to provide stability between the distal end of the humerus and the proximal part of the ulna. After surgery, the elbow is placed in a cast in 90° of flexion for six weeks. Additional bracing may be required when sufficient stability has not been achieved, and some patients use a brace for life (see Appendix).
Clinical Evaluation
The records of all fifty patients were reviewed to collect demographic data, the specifics of the operative procedures, and preoperative and postoperative Mayo Elbow Performance Scores. The Mayo Elbow Performance Score in this study is made up of four components, each weighted with a maximum number of points for a total score of 100: pain (45 points), range of motion (20 points), stability (10 points), and function (25 points). Stability is further graded as stable (10 points), moderately unstable (5 points), and grossly unstable (0 points). Classification of outcome was based on the total number of points achieved, with =90 points indicating excellent; 75 to 89 points, good; 60 to 74 points, fair; and <60 points, poor results. A postoperative DASH score was also calculated. No preoperative DASH score was available in this review. The DASH outcome measure is a thirty-item, self-reported questionnaire designed to measure physical function and symptoms in people with musculoskeletal disorders of the upper extremity. The DASH scoring system is measured on a 100-point scale, where 0 represents no disability and 100 represents total disability.
The demographic information collected included sex, age, handedness, surgical side, type of initial implant (linked or unlinked), longevity of the total elbow replacement, number of operations before resection, and complications after resection. All twenty-nine patients who had long-term follow-up were evaluated by one of two authors (B.C. or R.A.), either in the clinic (twenty-four patients), by telephone (three patients), or by mail (two patients). All patients had a radiograph of the affected elbow made at the time of review.
Statistical Analysis
Descriptive statistics are reported as the mean, with the range in parentheses, for continuous measures and as the number, with the percentage in parentheses, for discrete assessments. A paired t test was used to compare preoperative, postoperative, and long-term changes for the Mayo Elbow Performance Score and the individual components of this score (pain, arc of motion, stability, and function). Outcomes assessed included any change from preoperative to early postoperative and long-term follow-up measures for the Mayo Elbow Performance Score and its individual components. Correlation of elbow stability with the Mayo Elbow Performance Score and the DASH score was performed with a nonparametric Kruskal-Wallis correlation test. All statistical tests were two-sided, and p values of <0.05 were considered significant.
Source of Funding
No external source of funding was used for any aspect of this study.
Clinical Outcome
Implant resection resulted in a significantly improved Mayo Elbow Performance Score (Table II and Appendix). The average Mayo Elbow Performance Score improved from a preoperative value of 37 points to an early postoperative value of 59 points (p < 0.05) for all patients. This score was maintained at 60 points (range, 30 to 85 points) for the twenty-nine patients contacted at the completion of this study (p < 0.05). The Mayo Elbow Performance Score was not significantly different in the long term compared with the early postoperative score (p = 0.46). The DASH score for the twenty-nine patients at the time of the final follow-up averaged 71 points (range, 51 to 91 points).
Elbow stability was graded as stable (10 points) in five elbows, moderately unstable (5 points) in nine elbows, and grossly unstable or flail (0 points) in sixteen elbows at the time of the long-term follow-up (see Appendix). The elbow stability score was associated with the Mayo Elbow Performance Score at the latest follow-up evaluation (p < 0.05). Five elbows that were evaluated as stable (a score of 10) at the time of the long-term evaluation had a good result (a Mayo Elbow Performance Score of 75 to 89 points).
The long-term results were good for eight elbows, fair for eleven, and poor for eleven on the basis of the Mayo Elbow Performance Score. Five patients expressed interest in reimplantation but were discouraged by the high-risk nature of the procedure. Six patients with a lower-risk medical profile were offered reimplantation but declined. Functional bracing was believed to improve function for twenty-one of the fifty-one elbows as self-reported by patients when asked if they found bracing beneficial. The Mayo Clinic Elbow Brace (Aircast; DJO, Vista, California) was the most commonly used brace in this study after a period of casting if the elbow remained unstable. The patient often decided when they would use the brace depending on the degree of instability of the elbow and the type of activity performed.
Complications and Reoperations
Twenty-four elbows (47%) required additional surgery to treat a persistent infection after resection arthroplasty. There were wound-healing problems in twelve elbows (24%) after resection. The average number of reoperations following elbow resection for either debridement or definitive wound closure was 2.7 (range, zero to eleven reoperations). Ultimately, there were no refractory infections or chronic draining wounds. There was a high rate of other intraoperative and postoperative complications. One patient sustained an iatrogenic vascular injury requiring amputation. Intraoperative fractures occurred in 35% (eighteen) of the fifty-one elbows and involved the humerus (fifteen elbows), ulna (one elbow), or both (two elbows). A transient or permanent nerve injury was seen in nine elbows (18%). There were six transient (four radial and two ulnar) and three permanent (two ulnar and one radial) nerve injuries.
Resection of a failed total elbow replacement is a salvage procedure that should be considered only as a last resort when an infection at the site of a total elbow arthroplasty is refractory to medical and surgical treatment. The infection rate after the initial elbow arthroplasty during the study period was similar to other reported rates in the literature at 8.6%1-3. Of the elbows that had an infection at the site of a total elbow arthroplasty, 36% required elbow resection for definitive treatment. The misconception that elbow resection leads inevitably to a flail elbow with poor function is not supported by the results of this study. Elbow stability following elbow resection decreased, from a mean score of 6 preoperatively to 3 at the time of the long-term follow-up (p < 0.05). Of the thirty elbows evaluated after long-term follow-up, five were stable and nine elbows were moderately unstable, suggesting that nearly half of the elbows in this study achieved some degree of stability despite the absence of articular congruity. The results of our study show that resection arthroplasty, while not ideal, is still a reasonable salvage procedure to consider when all other options have been exhausted. The main benefits of elbow resection in this study are the eradication of the infection and the improvement in pain score compared with preresection scores. The pain component of the Mayo Elbow Performance Score improved from a mean score of 11 (moderate) preoperatively to 33 (mild) in the long term (p < 0.05).
There is limited information on the outcome of resection arthroplasty in the English-language literature. Figgie et al. reported their experience in the treatment of eleven patients who underwent resection for infection (seven elbows), implant fracture (three elbows), or recurrent dislocation (one elbow)11. External fixation was used in ten patients, and arthrodesis was attempted in one. There were seven successful outcomes attributed to an "anatomic arthroplasty," whereby the ulna was contained by the elbow condylar remnants12. In the study on five elbow resections for infection, Yamaguchi et al.2 reported a mean Mayo Elbow Performance Score postoperatively of 49 points compared with a score of 60 points in our study. It is unclear why our patient cohort had superior results, except to say that our larger sample size may be more representative of this study population.
Several conclusions may be derived from our study. Removal of the failed elbow prosthesis does not guarantee resolution of the infection in a single procedure, and often requires additional surgery before wound-healing and infection eradication occur. Implant removal is associated with a high rate of complications, namely, intraoperative fracture and nerve injury. Elbow resection results in significant improvements in pain, but many patients report high disability scores and poor function following this procedure. Approximately half of our patients required use of a brace in order to improve stability and function of the elbow.
Every effort should be made to preserve as much bone as possible at the time of resection to prevent shortening of the affected extremity and facilitate achievement of a stable resection arthroplasty. The rationale for preserving bone stock at the time of resection is also justified if reimplantation or elbow arthrodesis is to be considered in the future. On the basis of our findings, we approach every resection arthroplasty with the goals of removing all infected and foreign material, protecting the neurovascular structures, avoiding intraoperative fractures and perforations, preserving as much bone stock as possible, and achieving a stable elbow resection by combining adequate intraoperative osseous apposition, a careful soft-tissue repair, and a minimum of six weeks of postoperative immobilization. Elbow stability following resection is likely achieved by a combination of both soft-tissue scar and osseous articulation, and both casting and splinting can be extremely helpful.
Our study has several limitations. It is a retrospective study in a very challenging group of patients with complex elbow problems managed over the course of thirty years. The use of the Mayo Elbow Performance Score in this study may be misleading as stability in this scoring system was only allocated a maximum of 10 points of a possible 100. The improvement in outcome following resection is attributed largely to an improvement in pain (a maximum score of 45) and not necessarily stability. A flail elbow, for example, may in fact score reasonably well with the system of the Mayo Elbow Performance Score on pain, range of motion, and function even though the elbow is completely unstable. The DASH score, which was generally poor at an average of 71 points, is likely a better measure of the results in this study. Despite these limitations, this study, to the best of our knowledge, represents the largest series to date of patients requiring resection as a salvage procedure after an infection at the site of a total elbow replacement.
In summary, resection arthroplasty should be considered as a salvage procedure for patients with an infection at the site of a total elbow replacement. There is significant improvement in the pain component of the Mayo Elbow Performance Score, while improvements in function depend to a significant degree on the stability achieved after resection. Patients should understand before resection surgery the relatively high risk of intraoperative fractures and nerve injuries. An elbow resection is recommended when no other surgical procedure seems appropriate.