At present, there is considerable conflicting evidence in the literature regarding the treatment of comminuted and displaced fractures of the radial head [2, 3, 6, 19]. Although RHR as early treatment has been reported with satisfactory to poor outcomes, the radiocapitellar contact is seriously damaged, leading to the instability of valgus or posterolateral rotation, decreased strength, and degenerative changes in elbow and wrist . Consequently, RHA and ORIF are preferable in restoring the radiocapitellar relationship. Several excellent short to midterm results have been reported in comminuted radial head fractures treated with RHA, which provides an alternative surgical approach but have inherent drawbacks, such as heterotopic ossification, joint stiffness, hardware loosening, and radiocapitellar osteoarthritis, particularly when the prosthesis overstuffs the radiocapitellar joint and further alters the biomechanics of the radial head [21–23].
Biomechanical researches have proved the critical role of the radial head as a secondary stabilizer against both valgus and posterolateral instability of the elbow, especially in those cases with the medial collateral ligament or interosseous membrane of the forearm injury [4, 24]. Thus, being aware of the importance of the radial head for elbow function, ORIF should be the initial surgical solution to retain the integrity of the fractured radial head, so as to maintain normal articulation and bone stock. If ORIF fails early, it can still be converted into RHA. Compared with RHR in the present study, seven patients except one with PLRI in the present study obtained good ROM of elbow and functional outcomes by ORIF. The results were accordant with previous views that ORIF was an excellent option for Mason type III fractures with a low level of associated injury . However, following initial reconstruction of fractured radial head and avulsed ligament, the ROM of elbow and functional score of one patient with PLRI were inferior to the remaining cases, which mainly attributed to the nature of these complicated injuries.
A majority of previous studies have reported that comminuted radial head fractures with operative fixation have a high failure rate, nonunion, and osteonecrosis . Moreover, severe Mason type-III radial head fractures are usually defined as “unsalvageable” or “unrepairable” type, which appears impossible to be treated with ORIF . Nevertheless, the progress in implant technology and new fixation techniques have gradually overturned the established views. Particularly, on-table reconstruction technique has been reported and verified clinically. Several studies that exclusively focus on such technique for the complex radial head fractures have revealed satisfactory functional outcomes [7–9]. This technique is not limited to the narrowing anatomical joint space, and allows the reconstructed radial head to replant with the use of plate, which should be worthy of supporting and popularizing. However, in view of the different characteristics of comminuted radial head fractures, difficult extent to fix fragments and a variety of internal implants, the stability of fixation in the reconstructed radial head tends to be discrepant and remains to be assessed in different cases.
Koslowsky et al  have compared four different fixation techniques for the reconstruction of Mason type III fractures in the sawbones model, and find that displacement of the radial head from the neck often occur during plate fixation. Similarly, this issue was observed in the present study after on-table reconstruction of the radial head, leading to a poor radiocapitellar contact and incongruity between the radial head and neck. It is well known that treatment of radial head fractures mainly aims to maximize elbow movement and stability to maintain function . Correct alignment of the radial neck to the reconstructed radial head is crucial for normal rotation in the proximal radioulnar joint. Therefore, only reconstruction of the radial head fails to meet the treatment criterion. Furthermore, many biomechanical studies have shown that the stability in the coronal and sagittal plane largely depends on the anatomical restoration of the radiocapitellar relationship .
We analyzed that the unstable displacement might be correlated to the insufficient axial support provided by plate fixation. Both unfirm anchoring force and weak supporting strength from proximal screws to the reconstructed radial head were critical factors. Primarily, the reconstructed radial head with Kirschner wires might block inserting of the proximal screws, and the plate must be placed at “safe-zone,” which mainly limited the amounts, location, and direction of the screws. Moreover, the shape of fracture fragments was not uniform in different cases, so that the expected effects of manipulating on-table reconstruction remained uncertainties. The power drill, Kirschner wires, and proximal screws all probably would break the fragments and made reduction and fixation more complicated. Additionally, differences in plate type and operator’s surgical experience could affect the definitive outcomes, which might be inconsistent with previously reported results. Conclusively, all above-mentioned factors might influence the anchoring force and supporting strength of screws, causing instability between the radial head and neck during plate fixation.
If plate fixation fails to offer reliable axial support, a hybrid technique combining with intramedullary pinning seems useful to compensate for that deficiency. The intramedullary pinning technique is always manipulated in children radial neck fractures, which allows minimally invasive and is easily performed, representing a favorable result with few complications . Sandmann et al  first report that intramedullary pinning not only suits to radial neck fractures in children but satisfactory in adults. A series of superior results have confirmed the feasibility of adult radial neck fractures fixed with intramedullary pinning. Gao et al  has conceived a hypothesis that radial neck component could be stabilized by intramedullary pinning fixation following open reduction and plates or screws fixation in radial head fractures. This hybrid technique was manipulated and verified in our study, by which an additional axial support was obtained in the medullary cavity to prevent the displacement of reconstructed radial head from neck, and maintain restoration of radiocapitellar contact. Also, the use of elastic nail was modified in the present study. The proximal part of the elastic nail was bent, and formed three-point supporting consisting of the nail tip, bending point and nail bottom, which further enhance the axial support to avoid longitudinal displacement of the radial head.
Early failure in plate fixation is not infrequent, particularly in severe comminuted fractures. Ring et al  have shown that 3 of 14 fractures (21.4%) with more than three articular fragments have failure in plate fixation within the first month. Therefore, if there is lack of stability and loss of normal alignment between the reconstructed radial head and neck during plate fixation, incorporating the use of intramedullary pining appears applicable. Moreover, stable fixation in fracture site allows patient to functional exercise early and promotes bone union. Kiran Kumar et al  have reported that only 3 of the 6 patients treated with on-table reconstruction achieve fracture healing after a mean follow-up of 25 months, and one case occurs avascular necrosis of radial head. There is a strong possibility that these complications are attributed to the unstable plate fixation. Comparatively, in the present study, seven patients following additional intramedullary pinning fixation had obtained bone union without avascular necrosis of radial head observed during follow-up. This hybrid technique preserved the integrity of radial head and provided satisfactory elbow function compared with RHR, which was available for selection following unstable plate fixation.
Objectively speaking, there are still some risks and limitations to this technique. Firstly, after plate fixation, distal screws may block insertion of elastic nail. Actually, the axial direction of nail tip is designed to have an angle of approximately 30 degrees with nail shaft. During elastic nail insertion, we suggest that the direction in which nail tip point on the coronal plane should try to keep parallel with the distal screws, and in this way that nail tip can pass through distal screws with the minimum contact area. Additionally, the width of the proximal radial medullary cavity is about 6–7 mm, and the diameter of distal screws we used is 1.5 mm, thus there is a sufficient interspace between screws and medial cortical bone to allow 2.0 mm-sized elastic nail insertion. Secondly, inserting the elastic nail may penetrate the articular surface through the inter-fragmental gap, and cause failure in intramedullary fixation. In addition, the nail tip might potentially lift the fracture fragments, with the articular surface secondary displaced. Therefore, this technique needs the accumulation of operating experiences and learning curves. We suggest using the intraoperative fluoroscopy to control the inserting depth of nail, till the nail tip has advanced to the subchondral bone. Due to a small number of cases, the value of our study is limited. Moreover, the axial support provided by the elastic nail requires more quantitative analysis, patient accumulation, and further biomechanical exploration.
In conclusion, on-table technique offers an optional solution in the surgical treatment of comminuted radial head fractures. However, if there is unstable displacement from the reconstructed radial head to the neck during plate fixation, intramedullary pinning technique can be selected and applied as a remedial option. Indeed, this hybrid approach availably decrease the risks of failure in fixation and provide a good clinical result.