Patients
After institutional review board approval, a search of our institutional Total Joint Registry, Surgical, and Medical databases identified 244 periprosthetic elbow fractures that had been referred to or diagnosed at our institution from 1980 to 2008. In the present study, a Mayo type-U-II fracture was defined as a circumferential bone discontinuity around the ulnar stem, located between the metaphyseal-diaphyseal junction proximally and one ulnar bone diameter distal to the tip of the stem. Cortical perforations without a complete fracture line were excluded. Sixty-five fractures affected the ulna in different locations: at the olecranon (type U-I) (nineteen elbows), around the stem (type U-II) (thirty-three elbows), and distal to the stem (type U-III) (thirteen elbows). Of the thirty-three elbows with a type-U-II periprosthetic ulnar fracture, the stem was well-fixed in two elbows (type U-II-1) and occurred in an unlinked implant in one. The remaining thirty elbows (fourteen type U-II-2 and sixteen type U-II-3) form the basis of this study.
All loose ulnar components were revised, including twenty Coonrad-Morrey components (Zimmer, Warsaw, Indiana), seven Pritchard-Walker components (DePuy, Warsaw, Indiana), two Pritchard Mark II components (DePuy), and one Solar component (Stryker, Kalamazoo, Michigan).
The study group included twenty-three women and seven men with a mean age (and standard deviation) of 64 ± 12 years (median, sixty-five years; range, twenty-four to eighty-four years) at the time of the fracture. The mean age at the time of the index arthroplasty was 55 ± 13 years (median fifty-six years; range, twenty-three to seventy-seven years). The mean interval between the index arthroplasty and the periprosthetic fracture was 8.3 ± 6 years (median, eight years; range, 0.3 to twenty-one years). The ulna that fractured at four months had no bone loss, but the spiral fracture around the stem loosened the implant. The right elbow was affected in twelve patients, and the left elbow was affected in eighteen patients. Twenty patients (66%) had elbow replacements performed elsewhere and were referred to our institution for the treatment of the periprosthetic ulnar fracture.
One patient died within the first three months after surgery for reasons unrelated to the surgical procedure, one elbow underwent resection arthroplasty because of deep infection eight months after implant revision alone, and seven additional patients were lost to follow-up. The remaining twenty-one elbows were followed for an average of 4.9 ± 2.6 years (range, two to 10.1 years). The mechanisms of fracture and other fracture features, the surgical procedure performed, and the complications that occurred are described for the whole group of thirty elbows. Clinical outcomes are reported for the twenty-one patients who were followed for at least two years (see Appendix).
The underlying diagnosis included rheumatoid arthritis (fourteen elbows), posttraumatic osteoarthritis (ten elbows), juvenile rheumatoid arthritis (one elbow), distal humeral nonunion (one elbow), previous septic arthritis (one elbow), and massive structural bone loss after a gunshot wound (one elbow). The diagnosis could not be determined for two elbows. Eighteen fractures followed a primary total elbow arthroplasty, and twelve followed a revision arthroplasty (the first revision in ten elbows, the second revision in one elbow, and the third revision in one elbow). The indications for revision arthroplasty included implant loosening (five elbows), bushing wear (two elbows), deep infection (one elbow), instability (one elbow), and periprosthetic fracture (one elbow). The diagnosis was unknown for two elbows. The elapsed time between elbow arthroplasty and subsequent revision for fracture treatment was similar for elbows with fractures that occurred after primary and revision arthroplasties.
Seventeen patients (eight of whom had sustained the fracture after a primary procedure and nine of whom had sustained the fracture after a revision procedure) were unable to identify the specific moment when the fracture occurred and complained of worsening pain. Six patients reported a cracking sensation when the fracture occurred during normal daily activity, and seven fractures were the result of a fall.
All of the authors involved in the present study reviewed the preoperative, immediate postoperative, and latest follow-up radiographs for all elbows to confirm the fracture type, to grade the severity of bone loss, and to evaluate fracture-healing and component fixation status. This information was combined with the surgical findings described in the operative report. The ulnar component was confirmed to be loose in all surgically treated elbows. Bone loss was considered moderate (not enough to require bone stock augmentation) in fourteen cases (fracture type U-II-2) and severe in sixteen elbows (fracture type U-II-3). Additional findings included associated periprosthetic humeral fractures in two elbows and associated olecranon fractures in four elbows.
The ulnar component that was implanted at the time of surgical treatment of the periprosthetic fracture was a Pritchard Mark II implant in one elbow and a Coonrad-Morrey implant in twenty-nine elbows.
Surgical Techniques
The exposure was performed through a previous posterior skin incision whenever possible. The ulnar nerve was identified and was transposed anteriorly if in its anatomic location. Previously transposed ulnar nerves were not surgically exposed. The deep exposure was through a Bryan-Morrey approach8 in twenty elbows, a triceps split in three elbows, an olecranon fracture or nonunion in four elbows, and a partially or totally torn deficient triceps tendon in three elbows. After the loose ulnar component was removed, the fracture site was identified and the ulnar canal was thoroughly cleaned of membranes, cement, and debris. Tissue samples were routinely sent for culture and intraoperative histological analysis to look for signs of acute inflammation.
The surgical reconstruction technique in each case was based on the severity of bone loss. Bone loss was considered to be severe when the cortical bone around the ulnar stem was too thin, brittle, or even absent, such that bone stock augmentation by means of impaction grafting9 or an allograft-prosthetic composite7 (a hybrid formed by a prosthetic component partially cemented in a structural bone allograft, leaving a nude part of the component available for cementation into the native bone) was necessary. Bone loss was considered to be moderate when techniques to augment the bone stock were not needed.
Fracture fixation was achieved with the implanted component alone in two elbows, bypassing the fracture site by at least two ulnar diameters with the stem.
Strut allograft fixation was used in twenty elbows to stabilize frail osteolytic host bone fragments, establishing a competent and strong continuity between the proximal metaphyseal-epiphyseal bone and the distal ulnar diaphysis. Struts were fixed with circumferential 18-gauge wires providing structural support and avoiding the need for plates and screws, which could compromise the thin and brittle host ulnar bone (Figs. 1-A and 1-B). The details of allograft selection with respect to thickness and length have been described elsewhere6. The allograft length spanned two ulnar diameters distal and proximal to the fracture site. The strut allografts were either fresh-frozen or freeze-dried (AlloSource, Centennial, Colorado) and included seven femoral, three ulnar, five fibular, four humeral, and one rib cortical struts. Impaction grafting was used to restore insufficient cortical bone in expanded ulnae in eight of the fractures that were also fixed with strut allografts as well as in three additional elbows as an isolated revision technique (Figs. 2-A and 2-B)9. The source of the allograft was usually a femoral or humeral head. In these three elbows without additional strut allograft, the ulnar trial component was used to maintain provisional fixation during the impaction of bone graft, and fracture stability was obtained by the combination of tightly packed graft and the cemented long-stem implant extending at least two ulnar diameters distal to the fracture site.
Allograft-prosthetic composites were used in five elbows with severe bone loss such that no cortical strut allograft augmentation could restore the proximal part of the ulna and securely contain a new ulnar component (Figs. 3-A and 3-B). Ulnar allografts were used in two elbows, and proximal fibular allografts were used in three elbows.
Graft-host fixation varied. One ulnar allograft-prosthetic composite was telescoped inside the native ulna and was stabilized with wires. In the second ulnar allograft-prosthetic composite, the distal ulnar allograft was step cut and was fixed with a plate to the distal end of the native ulna. In two cases, a fibular allograft-prosthetic composite was placed along the native bone, obtaining side-to-side contact, and was stabilized with 18-gauge cerclage wires. In the third case, the fibular allograft-prosthetic composite was telescoped inside the native ulna and was stabilized with wires.
Clinical Evaluation
At our institution, all patients requiring implant revision are followed prospectively at regular intervals one, two, and five years after surgery, and every five years thereafter. Patients are offered a return appointment for a physical examination or, if unable to return, are asked to complete a questionnaire that allows calculation of the Mayo Elbow Performance Score (MEPS). Twenty-one patients were available after a minimum duration of follow-up of two years. An orthopaedic surgeon completed the latest follow-up evaluation at our institution for twelve patients. The latest follow-up evaluations for the other nine were conducted by means of a telephone call or a mailed patient questionnaire; an outside orthopaedic surgeon evaluated three of these nine patients. The mean duration of follow-up at our institution was 3.3 ± 3.5 years (median, 1.9 years; range, zero to eleven years). Outside evaluations were performed at a mean of 5.6 ± 2.6 years (range, 2.5 to ten years). Charts were reviewed by one of the authors (A.M.F.) to collect clinical data and information on complications.
Statistical Analysis
Quantitative variables were tested for normal distribution with the Kolmogorov-Smirnov test. Differences in quantitative variables were tested with the Mann-Whitney test. Differences in qualitative variables were tested with the chi-square test with use of raw data and with use of the Fisher exact test when the criteria were not met for chi-square analysis.
Source of Funding
There was no external funding for this study.
Clinical and Radiographic Outcome
Pain was the most common complaint when patients were initially evaluated for the treatment of the periprosthetic fracture. Instability depended on the severity of bone loss. Because of the associated patient discomfort, a formal evaluation of elbow range of motion was not consistently gathered.
Twenty-one patients had complete data at the time of the most recent follow-up (see Appendix). Of those, ten patients had rheumatoid arthritis, nine had posttraumatic osteoarthritis, one had a distal humeral nonunion, and one had previous septic arthritis. Eighteen patients were female, and three were male. The mean age at the time of the periprosthetic fracture was 65 ± 10 years (range, fifty-one to eighty-four years), which was a mean of 9 ± 6 years (median, nine years; range, one to twenty years) after the first elbow replacement. The periprosthetic fracture occurred after primary arthroplasty in twelve and after revision arthroplasty in nine. Nine patients (seven of whom sustained the fracture after a primary arthroplasty and two of whom sustained the fracture after a revision arthroplasty) had a loose ulnar component with sufficient cortical bone stock (fracture type U-II-2), and twelve (five of whom sustained the fracture after a primary arthroplasty and seven of whom sustained the fracture after a revision arthroplasty) had loosening and insufficient cortical bone (fracture type U-II-3). Table I shows the reconstructive procedures performed in this group.
At the time of the most recent follow-up, at a mean of 4.9 ± 2.6 years (range, two to ten years), all twenty-one fractures were healed (Figs. 1-B, 2-B, and 3-B).
Twelve patients reported no pain, four reported mild pain with activity, two reported mild pain at rest, and three reported moderate pain. The mean extension was 26° ± 22° (median, 30°; range, 0° to 60°), and the mean flexion was 134° ± 17° (median, 135°; range, 100° to 150°), with a mean arc of motion of 112° ± 32° (median, 120°; range, 40° to 150°). The mean pronation was 72° ± 18° (median, 77.5°; range, 30° to 90°). The mean supination was 64° ± 27° (median, 75°; range, 0° to 90°). Table II shows the ability to perform different activities with the affected upper extremity. Fifteen elbows were rated as stable, two were rated as mildly unstable, and four were rated as moderately unstable. The most recent mean MEPS was 81 ± 14 (median, 85; range, 40 to 100). According to the MEPS, the result was graded as excellent for ten elbows, good for five, fair for four, and poor for two. When compared with the status prior to fracture treatment, twelve patients were much improved at the time of the latest follow-up, six were somewhat improved, two were the same, and one was worse.
With the number of patients available, no differences were found in terms of outcome parameters (mobility, pain, MEPS) between patients with U-II-2 and U-II-3 fractures or between patients who had sustained the fracture after primary and revision procedures because of a lack of statistical power. There was a tendency for patients who had sustained the fracture after a revision procedure to have less forearm rotation and a lower MEPS score and for patients who had Mayo U-II-3 periprosthetic elbow fractures to have more pain (Table III).
In addition to the ulnar U-II periprosthetic fractures, there were four olecranon fractures requiring fixation with cerclage wiring. Of these four elbows, three had the humeral component revised: two to treat an associated humeral periprosthetic fracture (also augmented with allograft struts), and the other to change the implant design to a Coonrad-Morrey prosthesis. Five additional elbows had the humeral component revised to change the design to a Coonrad-Morrey implant. A table in the Appendix shows results for patients who required additional procedures and those in whom only the Mayo U-II periprosthetic fracture was addressed. Absence of statistical power precluded further analysis.
Complications
No patient had signs of acute inflammation on intraoperative pathological analysis. One patient had a positive culture for coagulase-negative Staphylococcus at the time of revision surgery. She received chronic antibiotic suppression treatment and was lost to follow-up 2.2 years after surgery without clinical signs of infection. There were three cases of deep infection; two were diagnosed one month after revision surgery, and the third was diagnosed 3.5 years after fracture reconstruction. The first two patients were managed with multiple irrigation and debridement procedures and chronic antibiotic suppression. One of these two patients was managed with conversion to a resection elbow arthroplasty eight months after fracture reconstruction (and was not included in the group of the twenty-one patients who were followed for at least two years). The third patient died three weeks after the diagnosis of the infection as a result of unrelated causes. One superficial wound infection was treated with oral antibiotics without further complications.
One patient developed aseptic loosening and olecranon fracture one year after surgery. After revision surgery and impaction grafting with a long-stem ulnar component, there was a good clinical result at 2.5 years of follow-up. One patient developed a periprosthetic humeral fracture after two years of follow-up and had revision arthroplasty. Neurological complications included one case of transient median and ulnar nerve dysfunctions with electrodiagnostic testing changes that resolved spontaneously.
Periprosthetic fractures of the ulna are a relatively uncommon mode of failure after elbow arthroplasty3. The specific rate of revision surgery for the treatment of a periprosthetic ulnar fracture around the stem of a loose ulnar component at our institution was relatively low. Between 1980 and 2008, 1251 primary and 550 revision total elbow replacements were performed at our institution, including only thirty-one revision arthroplasties (6%) to treat this complication.
Although the general features of periprosthetic elbow fractures are well-recognized3, there is very limited information about the outcome of revision surgery for the treatment of periprosthetic ulnar fractures around loose ulnar components. Our results suggest that periprosthetic ulnar fractures may be addressed with revision surgery and bone augmentation techniques with a moderate complication rate, related mostly to infection.
Our joint registry identified only two periprosthetic fractures around a well-fixed ulnar stem (U-II-1), and they were treated conservatively and were not included in this study. These fractures were located near the tip of the stems, and there was good bone stock in both cases. In our study, ulnar loosening was commonly found at the time of surgical treatment of the ulnar fracture. In most instances, loosening had likely happened prior to the fracture. In some instances, the fracture may have contributed to loosening by compromising the ulnar bone around the prosthesis or disrupting the cement mantle.
Eighty percent of these fractures occurred in patients who had had an elbow replacement at least five years earlier, and there was severe bone loss that made them prone to this complication. The mechanism of fracture represents a problem as 78% of the patients did not recall when they sustained the fracture (56%) or sustained the fracture during normal activities (20%). The role of osteolysis in the pathogenesis of this condition is clear. A polymethylmethacrylate (PMMA) precoat implant surface finish has been associated with bone osteolysis10. Precoating consists of application of a thin layer of PMMA on the surface of the implant with the intention of reinforcing the implant-cement interface. In nineteen cases, the implant that was removed was a Coonrad-Morrey design. Of those, four had a beaded surface, fourteen had a precoat surface, and one had a titanium plasma spray surface. Eight (57%) of fourteen fractures that occurred around a precoat component had severe bone loss (U-II-3), and six (43%) had good bone stock. After the substitution of precoat implants with titanium plasma spray implants in the year 2000, only three cases of periprosthetic Mayo U-II fractures have been treated at our institution, including two Mayo U-II-1 fractures without implant loosening that were suitable for nonoperative treatment and one Mayo U-II-2 fracture in a patient with good bone stock. Improvement of implant design and finish may be of great value in the prevention of this complication.
The surgical technique to treat this periprosthetic fracture is highly demanding and includes a number of options for bone stabilization and reconstruction. When treating this fracture, several questions need to be addressed: Is the ulnar component loose and does it need revision? How will fracture stability be achieved? Is the ulnar bone stock sufficient to perform a revision arthroplasty alone, or is bone augmentation needed? Is the structural integrity and function of the elbow extensor mechanism preserved? Is there any problem in the humeral component that needs to be addressed at the same time? Will infection or other external factors complicate treatment (Fig. 4)?
Despite the complexity involved in the treatment of periprosthetic fractures around an ulnar component, a good result can be achieved in the majority of cases. Complications can be expected, with deep infection being the most frequent and devastating. Loosening of the revised ulnar component occurred in one (5%) of the twenty-one cases in the present study. None of the four allograft-prosthetic composite reconstructions were associated with any major complications requiring revision surgery after a mean duration of follow-up of 3.1 years (range, two to 6.2 years).
The limitations of the present study include its retrospective nature, the fact that nine (29%) of thirty patients were lost to follow-up, and nine (43%) of twenty-one patients had their latest follow-up at other institutions or through mailed questionnaires. These limitations were due to the low frequency of this complication and the fact that twenty-eight years were required to collect enough cases to provide this information. Other potential limitations were the variety of reconstructive procedures that were required to treat the different fracture types and the finding that some patients had other concomitant fractures involving the olecranon and humerus that could have compromised the results.
Periprosthetic fractures around the stem of the ulnar component are a therapeutic challenge. Implant revision usually is required because of implant loosening, and bone augmentation with use of struts, impaction grafting, or an allograft-prosthetic composite is necessary because bone loss is usually present. Overall results can be satisfactory, although deep infections and loosening following revision remain serious complications.