Failure of metal radial head replacement

Replacement of the head of the radius may be used to treat fractures which cannot be reconstructed. The radial head acts as a stabiliser of the elbow in valgus¹ and rotation.² In the presence of other lesions, commonly associated with fractures of the radial head,^3–5 replacement arthroplasty can contribute substantially to the stability of the elbow and forearm. Replacement has been shown to be superior to open reduction and internal fixation when the head is fractured into more than three displaced fragments.^6,7 Replacement prostheses may also be used in chronic conditions,^8,9 including nonunion, malunion, post-traumatic osteoarthritis, and when symptoms arise following previous resection of the head.

Most reports oof replacements are in the short- or medium-term,^9–13 and there are very few long-term follow-up studies.¹⁴ Failure of replacement leading to revision surgery is mentioned in several studies, but the specific mechanisms of failure and the results of re-operation are currently unknown. The aim of this study was to examine the mechanism of failure requiring revision surgery, and to document the findings at the time of re-operation.

Patients and Methods

Following approval of the institutional review board, a joint registry database search was set up to identify all patients who had a radial head replacement, or removal or revision of a replacement, at our institution between 1998 and 2008. Patients who had a failed silastic prosthesis or failure of a radial head component as part of a total elbow or radiocapitellar arthroplasty were excluded from the study. A total of 47 radial head replacements (44 patients) which had been removed for various reasons were identified and formed the basis of this study. There were 25 males and 19 females and the mean age at the time of the initial operation was 48 years (16 to 70). The dominant arm was involved in 28 patients. The most common indication for the initial operation was a fracture of the radial head in 42 patients (45 elbows). The indication was rheumatoid arthritis in one patient (one elbow), and osteoarthritis with a previous radial head resection one year earlier in one patient (one elbow). A second revision to a metal prosthesis, was performed in three patients (three elbows) due to loosening of the index prosthesis.

In the 42 patients (45 elbows) requiring radial head replacement after trauma, the initial operation had been undertaken at a mean of 17 months after injury (0 to 230). It was introduced acutely (less than one week after trauma) in 13 patients (32%), subacutely (one to six weeks) in seven (17%), and in a delayed fashion (six weeks or more in 22) (52%). The mean delay between fracture and replacement in the delayed group was 32 months (6 to 230).

Radial head replacement was the initial treatment in 11 patients (11 elbows). The remaining 31 patients (34 elbows) underwent replacement after failure of internal fixation, resection or arthroplasty. A total of 15 patients (18 elbows) had undergone more than one previous surgical procedure. The radial head was implanted after failure of open reduction and internal fixation (ORIF) in ten patients (11 elbows), and for failure of radial head resection in 15 patients (17 elbows). Resection had been performed after failed ORIF in three of these 15 patients (17 elbows). Two patients (two elbows) had undergone ORIF for an ulnar fracture without surgery on the associated fracture of the radial head. The radial head implant had already been revised previously in seven patients (seven elbows).

The notes and radiographs of these patients (47 elbows) were reviewed retrospectively. The information obtained included the indications for and timing of the initial operation, post-operative complications, mechanisms of failure, and the type of revision surgery. The design and fixation used in both the primary and revision procedure were also recorded. Radiographs were analysed for loosening, instability and radial over-lengthening relative to the lesser sigmoid notch of the ulna,¹⁵ so-called joint ‘over-stuffing’ leading to asymmetry of the ulnohumeral joint. If over-lengthening of the radial head replacement was suspected radiologically,^16–18 it was cross-checked with the operative notes for clinical signs of over-stuffing.¹⁸ Any degenerative changes were graded using the Broberg and Morrey osteoarthritis (OA) scale,¹⁹ where 0 represents a normal elbow, and 1, 2 and 3 represent mild, moderate and severe degenerative changes, respectively.

Statistical analysis.

Separate analyses were performed on implants categorised by their design and the timing of implantation following the initial trauma. Student’s t-test was used to analyse differences in range of movement, the timing of initial implantation, radiological signs of degenerative changes, and time to failure between patients who had received different prosthetic designs, as well as between those in whom the radial head had been placed in an acute, subacute or delayed fashion. Differences in the range of movement were compared between patients in whom the radiocapitellar joint had been over-stuffed, and those in whom the length of the radial head was judged to be correct. The significance level was set at p = 0.05.

Results

Design and fixation of the index arthroplasty.

The radial head was fixed with cement in nine and without in 38 elbows, including 13 uncemented polished stems. A bipolar design was used in eight elbows, four of them in seven patients who had already had their index radial heads revised prior to further revision. Details of the failed prostheses in situ, before further surgery, and designs used for revision are shown in Table I.

Pre-operative evaluation and reasons for failure.

Prior to revision, the mean flexion-extension arc of movement was 86° (30° to 150°), with a mean flexion of 117° (60° to 150°) and a mean extension deficit of 30° (0° to 65°). A total of 18 patients (20 elbows) had a functional range of movement of more than 100°, from at least 130° of flexion to < 30° fixed flexion deformity.²⁰ The pronation-supination arc was a mean of 110° (0° to 185°), with a mean pronation of 58° (−15° to 95°) and a mean supination of 52° (0° to 90°). A total of 28 patients (30 elbows) had a functional arc of rotation from at least 50° of pronation and 50° of supination.¹⁹

The radial head replacements had failed for a variety of reasons at a mean of 23 months (0 to 180) following insertion. The clinical reasons for failure leading to revision are listed in Table II. The most common symptom was pain, followed by stiffness.

Radiological analysis.

Table III summarises the radiological findings. Pre-operative radiographs showed evidence of implant loosening in 31 elbows. Only three radiologically loose implants had been fixed with cement (Fig. 1), whereas the remaining loose implants were uncemented (Fig. 2). Of the 28 loose uncemented implants, 11 had been specifically designed to be placed loose in the intramedullary canal of the radius (Fig. 3). Signs of instability were found in 11 elbows (11 patients). There was subluxation in the radial head in five (Figs 2 and 4), a complete dislocation in three, and dissociation of the radial head from the stem in three (Fig. 5). There were radiological signs of overlengthening and joint over-stuffing in 11 elbows (Fig. 6). These findings were confirmed at surgery. However, there was no significant difference in the range of flexion, extension, pronation and supination between patients in whom the radiocapitellar joint was over-stuffed and those where the prosthesis was of the correct length (student t-test; flexion p = 0.38, extension p = 0.45, pronation p = 0.79, supination p = 0.65).

Further procedures.

In 17 patients the radial head replacement was removed and not replaced. The radial head component from a bipolar prosthesis which had become dissociated from the stem was removed in one patient, leaving the stem in place (Fig. 6). In 23 patients (24 elbows) the radial head replacement was removed and replaced. In 17 patients (18 elbows) the revision stem was cemented into place. A bipolar design was used in 17 patients (17 elbows). In two patients (two elbows) only the radial head component of a bipolar design was replaced, and the stem was left in situ. In one of these the metal radial head was replaced with a polyethylene component and a capitellar replacement was inserted at the same time. In the other patient (one elbow), the radial head component was replaced because of dissociation of this component from the stem. In three patients (three elbows) the radial head was removed and a total elbow replacement was implanted. These three patients (three elbows) are not included in the data in Table IV. Severe erosion of the articular cartilage of the capitellum was noted at the time of further operation in 18 patients (20 elbows).

Effect on failure of the design and the timing of the index replacement of the radial head.

The clinical, radiological and pre-operative findings are shown in Tables IV and V. The radial head was replaced for osteoarthritis in one patient (one elbow) and for rheumatoid arthritis in another. These two patients were not included in the data in Tables IV and V. We found no significant difference for flexion, extension, pronation and supination excursion between designs (p > 0.05).

Bipolar implants had been inserted at a significantly longer interval after trauma than the loose stem unipolar designs (p = 0.037). There was no significant difference in this interval between the loose-stem unipolar and the fixed-stem unipolar, nor between the fixed-stem unipolar and the fixed stem bipolar designs. There were no significant differences between designs for time to failure or osteoarthritis scores (p > 0.05). Loosening was less common in fixed-stem bipolar implants. Five of eight fixed-stem bipolar implants had been cemented in place, and three were uncemented. Both loose fixed-stem bipolar implants had been implanted without cement. Radiocapitellar subluxation did not occur in any of the bipolar implants, but capitellar erosion was more prevalent in this group. Relatively more bipolar prostheses were used in reconstructive procedures delayed more than six weeks after trauma. The difference in the radiological osteoarthritis score was significant between patients who had acute implantation (less than one week) of the radial head replacement and those who had delayed implantation (more than six weeks) (p = 0.03). We also found a significantly increased fixed-flexion deformity in patients in whom the radial head had been implanted at between one to six weeks, compared to acutely (p = 0.03). Over-lengthening, radial head subluxation, component dissociation and instability all occurred more often in the one- to six-week group. Capitellar erosion occurred more often when the implant was inserted more than six weeks after trauma.

Discussion

Replacement arthroplasty of the head of the radius is commonly used in the treatment of radiocapitellar disorders. However, there is little published information regarding the complications and rate of revision of this procedure.

Painful loosening,²¹ radioulnar synostosis,^21,22 dissociation of the components²³ and deep infection²⁴ have been reported. In this study, implant loosening was the most common radiological finding prior to further surgery in 29 patients (31 elbows). In contrast with the findings of other authors who found a high rate of radiolucent lines in asymptomatic patients,^{14,21,24–27} all 29 patients (31 elbows) with a radiologically loose implant had pain, which was moderate or severe in 28 patients (29 elbows). Loosening of a fixed-stem bipolar implant was less common, possibly because the stresses at the radial head component are transmitted less, protecting the stem interface.

Reduced movement was the most common indication for further surgery, which is also reflected in the literature.^11,25,28 Only 18 elbows in our group had a functional flexion arc. Stiffness may reflect the severity of the initial injury, but may also be secondary to pain and/or overstuffing of the radiocapitellar joint.²⁹ With the numbers available there was no statistical difference between the patients in whom the radius was judged to be over-lengthened and those in whom the radial head was the correct length. Interestingly, although it was not reflected by objective measurements, a higher percentage of patients in the acutely reconstructed group complained of stiffness in their pre-operative clinical examination (Table IV). By contrast, patients in whom the radial head replacement had been inserted between one and six weeks had significantly reduced extension compared with the acute group (p = 0.03). Other technical issues, such as over-lengthening, radial head subluxation, component dissociation²³ and instability, all occurred more frequently in the one- to six-week group, suggesting that radial head replacement during this period may be technically more challenging.

Our analysis also showed that timing was the most important factor in predicting the prevalence of capitellar erosion. In 12 of 22 elbows (55%) who had a radial head replacement inserted more than six weeks after trauma there were clear signs of capitellar erosion. This constituted 63% of all 19 (21 elbows) with capitellar erosion, and may be due to disuse osteopenia as previously suggested.³⁰ Only four of 11 patients (11 elbows) with an over-lengthened radial head had severe capitellar erosion at a mean of 13 months (1 to 53) following the initial replacement, suggesting that over-stuffing may be only a secondary factor.

With regard to the different design philosophies, we made several observations. No radiocapitellar subluxation was found where a fixed-stem bipolar design had been used. Surprisingly, in two of 13 patients (2 of 13 elbows) where the loose-stem unipolar prosthesis was used, subluxation of the radial head relative to the capitellum was found, compared with three of 24 patients (3 of 26 elbows) in the fixed-stem unipolar group. The loose-stem unipolar replacement was designed to realign itself with the capitellum, but this suggests that it may not always be the case. Despite its improved alignment, the fixed-stem bipolar radial head replacement had the highest incidence of capitellar erosion, but this prosthesis was used more often in a reconstruction. Biomechanical studies have shown bipolar prostheses to provide less stability at the elbow.^31,32 There was no statistical evidence of this in our study, but there was a trend towards increased stability in the fixed-stem unipolar group. Clinical signs of instability were found in five of 24 patients in the fixed-stem unipolar group versus three of 13 and two of eight in the loose-stem unipolar and fixed-stem bipolar groups, respectively.

This study has several weaknesses. It is retrospective and includes a heterogeneous group of patients with a wide range of indications for radial head replacement using different designs. However, we believe it provides the first in-depth analysis of failed radial head replacements in the literature.

We conclude that failed radial head replacements may lead to re-operation for pain, reduced range of movement, instability or infection. Radiological loosening was prevalent in this group of failed replacements. In addition to a delay in surgical intervention, technical errors leading to over-lengthening or malpositioning appear to be the cause of some failures. Further studies are required to identify the ideal design, fixation and timing to avoid such failures.

No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article.