Radial head fractures account for 1.7% to 5.4% of all fractures and for 30% of fractures involving the elbow1-3. The management of unreconstructable comminuted radial head fractures remains controversial4,5, and options include radial head resection or arthroplasty. Radial head arthroplasty has been shown to restore axial and valgus stability reliably and to return elbow kinematics to nearly normal levels in biomechanical studies6-8.
The insertion of a radial head prosthesis that closely replicates the dimensions of the native radial head is of paramount importance. Insertion of an implant that is incorrectly sized has been associated with altered joint architecture and mechanics that may lead to complications9-12. Use of a prosthetic radial head that is too thick leads to lengthening of the radius, or "overstuffing" of the radiocapitellar joint, and has been reported to cause loss of elbow flexion, capitellar erosions, pain, and early-onset osteoarthritis13-16.
Despite the acknowledged importance of radial length, there are few intraoperative guidelines to aid the surgeon in the selection of an appropriately sized implant13,17. Such guidelines are particularly important when the native radial head is extensively damaged and cannot be used to estimate prosthetic size, or in revision procedures in which the radial head has been previously excised. There are also no guidelines for diagnosing overlengthening of the radius radiographically in the postoperative setting. The relationship between the lateral coronoid facet and the radial head may be a useful intraoperative landmark for selecting an implant of the correct thickness13,17; however, it is of limited utility in the postoperative diagnosis of overlengthening.
It has been proposed that widening of the lateral ulnohumeral joint space on an anteroposterior radiograph may be indicative of radial head overlengthening. Rowland et al.18 recently reported that, because some degree of widening is a normal anatomic variant, it cannot be used to diagnose overlengthening. However, those authors did find that the medial ulnohumeral joint space is normally parallel. On the basis of this observation, we theorized that implantation of a radial head prosthesis that resulted in overlengthening would lead to widening of the medial ulnohumeral joint and loss of the normal parallelism of the apposing joint surfaces.
The purpose of this study was to develop radiographic and intraoperative guidelines with which to reliably diagnose overlengthening (overstuffing) of the radius by a radial head prosthesis. We hypothesized that (1) the macroscopic presence of a gap in the lateral ulnohumeral joint correlates with insertion of a radial head prosthesis that has resulted in overlengthening, (2) a radiographic appearance of incongruity of the medial ulnohumeral joint correlates with overlengthening, and (3) progressive widening of the lateral aspect of the lateral ulnohumeral joint as seen radiographically indicates that an implant is too thick, resulting in overlengthening.
Specimen Preparation and Elbow Rig
Seven fresh-frozen human cadaveric upper extremities were thawed overnight at room temperature. Prior to their inclusion in the study, the specimens were assessed radiographically and macroscopically for pathological changes in the elbow as well as for the range of motion and stability of the elbow. The average age of the specimen donors at the time of death was sixty-nine years (range, sixty-one to eighty-one years). The arms were transected at the midpart of the humerus with the elbow, forearm, wrist, and hand left intact. The skin was removed from the specimens, and the biceps, brachialis, and triceps tendons were identified and were sutured with a number-2 ultra-high molecular weight polyethylene suture (Hi-Fi; ConMed Linvatec, Utica, New York) in preparation for loading to simulate muscle tone. The biceps and brachialis sutures were tied together in a loop to allow a shared load to be evenly distributed. Specimens were tested with the muscles loaded (to simulate resting muscle tone in an awake non-paralyzed patient) and unloaded (to simulate muscle paralysis in a patient who is under general anesthesia in an intraoperative environment). For the tests done with the muscles loaded, the sutures were passed over a pulley, a 40-N force was applied to the triceps, and a 40-N force was applied to the biceps and brachialis together16.
The anterior and posterior aspects of the elbow capsule were removed, and the anterior bundle of the medial collateral ligament and the lateral collateral ligament were left intact. Additionally, the interval between the anconeus and the extensor carpi ulnaris was incised in order to replicate a Kocher surgical exposure, which is commonly used during radial head arthroplasty. A custom radiolucent rig was developed to stabilize the specimens at reproducible angles of flexion/extension and forearm rotation (Fig. 1). Once the intact specimen (stage 1) was mounted on the rig, clinical measurements of the lateral ulnohumeral joint space and a standardized series of radiographs were made (as described under Radiographic Measurements).
The lateral collateral ligament was then released from its origin on the lateral epicondyle and repaired. The repair of the lateral collateral ligament involved creation of two divergent bone tunnels, both starting at the isometric point on the lateral epicondyle and exiting along the posterior aspect of the lateral condyle. The lateral collateral ligament was sutured with a number-2 Hi-Fi suture in a running locking fashion, with the suture ends then passed through the bone tunnels and tensioned at 20 N with the elbow in 90° of flexion and the forearm in neutral rotation. A 20-N force was obtained by applying a 20-N load to the sutures attached to the ligament and then clamping the sutures to the rig in order to maintain constant ligament length. Use of a clamp to stabilize the lateral collateral ligament repair made it possible to release it, to allow access to the joint, and then to retension it with the same 20-N force. Once the lateral collateral ligament was repaired (stage 2), the lateral ulnohumeral joint space was again measured clinically and radiographic images were again made.
Stage 3 replicated a radial head resection and lateral collateral ligament repair. This stage, however, was tested last as the in situ native radial head was required for precise templating of a correctly sized implant. Stage 4, which involved insertion of a radial head prosthesis of the correct thickness, was therefore tested after stage 2. Stage 4 was created by unclamping the lateral collateral ligament repair and dislocating the elbow joint. The maximum and minimum diameters of the native radial head were measured in situ with digital calipers (Digimatic; Mitutoyo, Aurora, Illinois) and a radial head implant (Evolve Proline; Wright Medical Technology, Arlington, Tennessee) was selected on the basis of the minimum diameter. The thickness of the implant was measured with digital calipers, and an equal thickness of the native radial head was resected with a microsagittal saw. The radial canal was reamed and prepared according to the guidelines in the implant manufacturer's technique manual. A radial head of the correct thickness was then inserted, the lateral collateral ligament was retensioned, and radiographs and joint space measurements were repeated (stage 4). Subsequently, the correct-size implant was replaced sequentially with radial head implants that were 2 mm (stage 5), 4 mm (stage 6), 6 mm (stage 7), and 8 mm (stage 8) greater in thickness. The lateral ulnohumeral joint space was measured clinically and radiographs were made after each replacement. Finally, the radial head implant was removed, replicating a radial head resection, and a final set of radiographs and clinical joint space measurements were made (stage 3).
To summarize, all measurements were performed at eight different stages: the intact specimen (stage 1); lateral collateral ligament repair (stage 2); radial head resection with lateral collateral ligament repair (stage 3); radial head arthroplasty with an implant of the correct thickness (stage 4); and radial head arthroplasty with implants that were 2 mm (stage 5), 4 mm (stage 6), 6 mm (stage 7), and 8 mm (stage 8) thicker than the correct-size implant. Each stage was tested with and without muscle loading.
Radiographic Measurements
Anteroposterior radiographs of the elbow (relative to the forearm) were made in full extension and 45° of flexion, in three positions of forearm rotation (full pronation, neutral, and full supination). The radiographic beam was oriented perpendicular to the forearm (Fig. 1). Each series of images was made with the muscles loaded and unloaded, as previously described. A minimum of ninety-six radiographs were obtained for each specimen, and a maximum of 132 were obtained for the specimens that were selected for repeatability study.
The radiographic joint-space dimensions were determined by measuring the distance between the ulna and the humerus at four locations with use of commercial viewing software (Centricity; GE Healthcare, Burlington, Vermont) at 2× magnification. The four locations of measurement (from lateral to medial) were the lateral side of the lateral ulnohumeral facet joint, the medial side of the lateral ulnohumeral facet joint, the lateral side of the medial ulnohumeral facet joint, and the medial side of the medial ulnohumeral facet joint (Fig. 2). Incongruity of the medial ulnohumeral joint was calculated by subtracting the width of the lateral aspect of the medial ulnohumeral joint from the width of the medial aspect of the medial ulnohumeral joint.
In order to assess intraobserver and interobserver reliability, each measurement was performed twice by one observer (S.G.F.) and a single time by a second observer (G.S.A.). Several radiographic series were repeated to study repeatability. For each repeatability test, the implant was removed, the elbow was repositioned, the prosthesis was reinserted, and the radiographic series was repeated.
Clinical Measurement of the Lateral Ulnohumeral Joint Space
Clinical measurements of the lateral ulnohumeral joint space were performed with the elbow in 45° of flexion; the muscles unloaded; and the forearm in full pronation, neutral, and full supination. The gap between the ulna and the humerus was measured anteriorly at the most lateral aspect of the ulnohumeral joint. A set of custom metal feeler gauges was used for gaps of <1.5 mm, whereas a pair of digital calipers (Digimatic) with an accuracy of 0.03 mm was used for larger gaps (>1.5 mm). Two examiners conducted six joint-space measurements each per stage, with one examiner repeating the measurements; thus, there was a total of 144 joint-space measurements per specimen.
Data and Statistical Analysis
To determine intrarater reliability, intraclass correlation coefficients (model 2,1) and their 95% confidence intervals were calculated for the two repeated radiographic measurements of the joint spaces and the single clinical joint-space measurement. To determine interrater reliability, similar calculations were performed to compare the first observer's values with the values obtained by the second observer. Effect significance was determined with two-way analysis of variance, with use of the post hoc Tukey test to perform multiple pairwise comparisons between groups (a = 0.05).
Source of Funding
This research was financially supported by the Alexandra Kirkley Young Investigator Award from the Canadian Orthopaedic Foundation (G.S.A.). Equipment was provided by Wright Medical Technology (Arlington, Tennessee).
Intraclass correlation coefficients demonstrated excellent intrarater and interrater reliability for all measures. The intrarater reliability was 0.996 (95% confidence interval, 0.995 to 0.996) for clinical measurement of the lateral ulnohumeral joint gap, 0.856 (95% confidence interval, 0.814 to 0.890) for radiographic measurement of incongruity of the medial ulnohumeral joint, and 0.993 (95% confidence interval, 0.991 to 0.995) for radiographic measurement of the lateral side of the lateral ulnohumeral facet joint space. The interrater reliability for these measurements was 0.998 (95% confidence interval, 0.998 to 0.999), 0.823 (95% confidence interval, 0.797 to 0.845), and 0.991 (95% confidence interval, 0.985 to 0.994), respectively.
Clinical (Visual) Measurement of the Lateral Ulnohumeral Joint Space
When measured clinically, the mean lateral ulnohumeral joint space gap was found to be negligible in stages 1 through 4 (range, 0 to 0.024 mm), but there was a significant increase in this gap with all amounts of overlengthening (2, 4, 6, and 8 mm) (Fig. 3). The mean gap (and standard deviation) for specimens with 2 mm of overlengthening was 0.9 ± 0.44 mm (range, 0.0 to 1.4 mm), which was significantly greater than the measured gaps in stages 1 through 4 (p = 0.005). The mean gap was 2.3 ± 0.9 mm (range, 0.3 to 3.0 mm) in the specimens overlengthened by 4 mm, 3.4 ± 1.0 mm (range, 0.9 to 4.6 mm) in those with 6 mm of overlengthening, and 4.7 ± 1.5 mm (range, 1.2 to 6.5 mm) in those with 8 mm of overlengthening. These gaps were also significantly greater than the clinically measured gaps in stages 1 through 4 (p < 0.001) (Fig. 4). There were no significant differences among the lateral ulnohumeral joint space gaps measured in pronation, supination, and neutral forearm rotation at any stage (p = 0.3). These data suggest that visualization of a gross gap in the lateral ulnohumeral joint in an unloaded model with an intact medial collateral ligament and a repaired lateral collateral ligament is an indication of overlengthening by the radial head prosthesis.
Radiographic Measurement of Ulnohumeral Joint Incongruity
The widths of the ulnohumeral joint space measured radiographically at the four locations (the lateral and medial sides of the lateral ulnohumeral facet joint and the lateral and medial sides of the medial ulnohumeral facet joint) did not differ among the forearm positions of pronation, supination, and neutral rotation at any stage (p = 0.3). The tests performed with muscle loading demonstrated no significant difference in medial ulnohumeral joint incongruity among stages 1 through 6 (p > 0.1), although a significant difference in this parameter was detected in the specimens with 6 mm (p = 0.003) and 8 mm of overlengthening (p < 0.001) (Fig. 5). In the tests without loading, there was no significant difference in medial ulnohumeral joint incongruity among stages 1 through 5 (p > 0.1), whereas a significant difference in medial ulnohumeral joint incongruity was found in unloaded specimens with 4 mm (p = 0.02), 6 mm (p = 0.014), and 8 mm (p < 0.001) of overlengthening (Fig. 6).
In the unloaded specimens, the mean widening of the lateral aspect of the lateral ulnohumeral facet joint in stages 1 through 4 was 3.3 mm (range, 2.4 to 4.2 mm). This distance increased significantly in all specimens that were overlengthened (p < 0.001). The mean width of the lateral side of the lateral ulnohumeral facet joint was 4.4 ± 0.6 mm (range, 3.7 to 5.7 mm) in the specimens overlengthened by 2 mm, 5.8 ± 0.6 mm (range, 5.1 to 7.1 mm) in those overlengthened by 4 mm, 7.1 ± 0.4 mm (range, 6.4 to 8.2 mm) in those overlengthened by 6 mm, and 8.4 ± 0.5 mm (range, 6.9 to 9.3 mm) in those overlengthened by 8 mm. In the tests with muscle loading, only specimens with 4, 6, or 8 mm of overlengthening showed a significant increase in the lateral side of the lateral ulnohumeral facet joint space as compared with that in stages 1 through 4 (p < 0.001).
The radial head provides an important anterior and valgus buttress to the elbow6-8,19. Adverse outcomes associated with overlengthening or "overstuffing" by a radial head arthroplasty have been reported, but specific criteria for diagnosing overlengthening of the radius have not been well defined13-18,20,21. It is difficult to determine intraoperatively if a radial head prosthesis is too thick, especially in patients with associated collateral ligament injuries, in reconstructive procedures where the native radial head is absent, or during revision radial head arthroplasties13,16,18,21.
Recent studies have identified local anatomic landmarks that facilitate intraoperative insertion of a radial head implant of correct thickness13,17. Doornberg et al.13 analyzed seventeen computed tomography scans of the elbow to determine the relationship between the radial head and the lateral aspect of the coronoid. They concluded that the most proximal osseous extent of the radial head was, on the average, 0.9 mm more proximal than the lateral osseous edge of the coronoid. A weakness of this study was that the authors did not account for the articular cartilage, which may vary in thickness22. Van Riet et al.17 conducted an anatomic study involving a similar measurement. In eight cadaver specimens, they found that the lateral edge of the coronoid, which they termed the proximal edge of the lesser sigmoid notch, was a reliable landmark for assessing the correct thickness of the radial head.
Shors et al.21 conducted a radiographic study of cadaver specimens in which radial head arthroplasties resulting in various degrees of overlengthening had been performed. They examined four scenarios: the collateral ligaments intact, the lateral collateral ligament intact with disruption of the medial collateral ligament, the medial collateral ligament intact with disruption of the lateral collateral ligament, and disruption of the medial and lateral collateral ligaments. They concluded that radiographs could not be used to diagnose incorrect lengths in the range of -2 mm to +4 mm. These results agree with our findings with regard to the limitations of postoperative radiographs in the diagnosis of overlengthening. The study by Shors et al. had some limitations. The cadaveric elbow joints were accessed through epicondylar osteotomies, which do not reproduce the pathoanatomy of ligamentous disruption seen in elbow fracture-dislocations—i.e., the clinical scenario in which radial head arthroplasties are most commonly performed. By using an epicondylar osteotomy, Shors et al. were able to maintain the correct length of the collateral ligaments after implanting a too-thick radial head prosthesis, a model that reflects the clinical scenario of an unreconstructable radial head fracture with intact collateral ligaments. The data obtained in that experimental model may not be generalizable to the more common situation of an elbow with ligamentous injury, in which implantation of an overly thick prosthesis would result in the lateral collateral ligament being repaired in the lengthened position.
The studies to date have focused on intraoperative determination of the correct thickness of a prosthetic radial head13,17,21. To our knowledge, no one has reported postoperative radiographic parameters that correlate with overlengthening. A not-uncommon scenario is one in which a patient has pain or limited flexion following radial head arthroplasty and a diagnosis of overlengthening is entertained but cannot be confirmed because there are no radiographic criteria with which to do so. It was initially believed that gapping of the lateral ulnohumeral joint seen radiographically was diagnostic of overlengthening, but Rowland et al.18 determined that lateral gapping was present in about one-third of normal elbows. They also found that the medial ulnohumeral joint was normally parallel, and therefore we hypothesized that lateral gapping of the medial ulnohumeral joint may be a useful indicator of overlengthening. The data from our tests that included muscle loading indicate that gapping, or asymmetry, of the medial ulnohumeral joint seen on anteroposterior radiographs is a reliable indicator only when the radial head is overlengthened by =6 mm; it cannot detect smaller degrees of overlengthening. Our tests without muscle loading, which mimicked the intraoperative environment with muscle paralysis, showed that asymmetry of the medial ulnohumeral joint was a reliable indicator of overlengthening of =4 mm.
In this study, visualization of a lateral ulnohumeral joint-space gap was consistently related to overlengthening of the radial head. In the unloaded specimens, overlengthening of =2 mm reliably produced a detectable gap. Sensitivity to overlengthening at the 2-mm level is particularly useful given the finding, by Van Glabbeek et al.16, of considerable alterations in elbow joint kinematics and radiocapitellar joint pressure occurring with overlengthening of 2.5 mm. To see this gap intraoperatively, a dental mirror can be used to view the lateral ulnohumeral joint after provisional tensioning of the lateral collateral ligament (Fig. 7). The gap can also be directly visualized by releasing more of the anterior aspect of the joint capsule and the extensor origin from the anterolateral aspect of the supracondylar ridge. Although further anterolateral capsular release seemed possible in our cadaver specimens, we do not recommend this for elbows with ligament injuries. Additional studies are required to define the best way to assess the gap visually.
The change in the radiographic appearance of the lateral aspect of the lateral ulnohumeral joint space can also be used as an intraoperative indicator of overlengthening by an arthroplasty. Sequential fluoroscopic anteroposterior images after insertion of trial radial head implants of various thicknesses can be used to determine the correct thickness. An arthroplasty with a trial implant that results in the same radiographically measured width of the lateral aspect of the lateral ulnohumeral joint space as seen on the images obtained after the radial head resection has not produced overlengthening.
All of the measured parameters differed between the loaded and unloaded states. A gross gap that was present in the lateral ulnohumeral joint in an unloaded specimen with an implant that was, for example, 2 mm too thick was diminished by the application of load. Likewise, the widening of the lateral aspect of the lateral ulnohumeral joint that was seen radiographically was decreased with the application of load, as was our ability to identify asymmetry of the medial ulnohumeral joint. Asymmetry of the medial ulnohumeral joint was a reliable indictor of 4 mm of overlengthening in the unloaded state, but the reliability decreased in the loaded state, with an ability to detect =6 mm of overlengthening. We theorize that this was the case because the application of load across the elbow joint compressed the joint surfaces together and subtle amounts of overlengthening were absorbed by the soft tissues, such as the interosseous membrane and the proximal and distal radioulnar joints.
The limitations of this study include the use of a cadaver model; the small sample size; and the fact that only the biceps, brachialis, and triceps muscles were loaded (i.e., the common wrist flexors and extensors were excluded). Also, as overlengthening was the specific focus of this study, we did not test the effects of underlengthening, or understuffing. Another limitation is that a gap in the lateral ulnohumeral joint, if present, can be viewed intraoperatively only with a dental mirror or with further release of the anterolateral aspect of the capsule, which we do not recommend in an already unstable elbow. The strengths of this study are that we tested the most common lateral ligament injury scenario that could theoretically lead to overlengthening, and testing was done with the muscles both loaded and unloaded. A testing jig was used to ensure that radiographs were made in a standard and reproducible manner.
One goal of this study was to identify a method for diagnosing radial head overlengthening postoperatively. Unfortunately, incongruity of the medial ulnohumeral joint was proven to be insensitive in the determination of subtle overlengthening. Because overlengthening is difficult to diagnose postoperatively, the primary goal should be intraoperative prevention of overlengthening. This study identified another method with which to assess radial head thickness intraoperatively that may be used in concert with the methods described by Doornberg et al.13 and van Riet et al.17. Several key intraoperative steps should decrease the likelihood of overlengthening. Whenever possible, the excised fragments of the radial head should be reassembled to determine the correct thickness of the implant. The criteria described by van Riet et al., which reference the length of the radial head off the lesser sigmoid notch, can also be used if the native radial head is unavailable. The lateral ulnohumeral joint should be examined during the use of trial radial head implants of various thicknesses. If a gap is present when the joint is viewed through the lateral arthrotomy site, the trial implant is too thick. Intraoperatively, the trial implants can also be examined fluoroscopically. First, before any trial implants are inserted, the width of the lateral aspect of the lateral ulnohumeral joint space should be measured, with the elbow held reduced, on an anteroposterior fluoroscopic image made with the beam directed perpendicular to the proximal part of the forearm. Then this image is obtained again after the trial implant has been inserted. If there is any increase in the lateral ulnohumeral joint space, one can be certain that the trial implant is too thick and the radius has been overlengthened. Also, any asymmetry of the medial ulnohumeral joint on an anteroposterior fluoroscopic image is diagnostic of overlengthening. Additional studies are required to identify better tests to diagnose overlengthening postoperatively. 