Introduction

One third of fractures involving the elbow joint affect the radial head (RH) [1]. However, the treatment of Mason III fractures remains controversial [2,3,4]. Arthroplasty or simple RH resection is an alternative in cases of non-reconstructable RH fractures [2,3,4,5,6,7].

Patients who undergo excision of non-reconstructable RH fractures develop progressive valgus instability, potential radial ascent, and secondary ulnocarpal symptoms with alteration of the kinematics about the elbow and forearm creating a self-perpetuating cycle of degenerative change [6,7,8]. In an effort to avoid these complications and others, the use of RH replacement has been popularized in the literature since the beginning of the twentieth century (Fig. 1). Since that time, the surgical indications as well as the design of RH prostheses (RHP) have continuously evolved in an attempt to resolve the primary problems encountered during and subsequent to RH arthroplasty (RHA). Older designs are no longer used due to reports of poor results and numerous complications [9]. Heijink et al. [10] found that the mid-term functional results following RHA performed between 1993 and 2015 were good to excellent in 85% of patients using the Mayo Elbow Performance Score [11]. The long-term results of RHA, however, are unknown. Complication rates in the recent literature are variable, with reoperation rates ranging from 0 to 45% after RHA [12].

Fig. 1
figure 1

History of radial head arthroplasty: innovation, revolution, evolution

Full size image

Biomechanical and anatomic research aiming to reduce complications and improve outcomes of RHA has allowed for the development of many different models of RHP. However, current literature has yet to discern which design or material is superior to others. Enhanced awareness of the history of RHP would allow for better understanding of the state of the art and would faciliate innovation in prosthesis design.

The aims of this study are to summarize (1) the history of RH prosthesis use in traumatology and (2) the clinical and biomechanical data that engendered its evolution.

Materials and methods

A literature search was performed using Ovid Medline, Ovid Embase, Scopus and Cochrane Library, and the Medical Subject Headings vocabulary. The search was limited to English and French language literatures. The following terms were combined with “AND” and “OR”: “radial head,” “ arthroplasty,” “ prosthesis,” “ radial head prosthesis,” and “radial head arthroplasty.” Due to the limited historical timeframe that can be searched via these search engines, references from the existing literature were also searched. Results are discussed as a chronologic review of the relevant literature between January 1920 and Januray 2018 (Fig. 1).

Review of the literature

Innovation

Treatment of comminuted RH fractures at the beginning of the twentieth century was limited to conservative measures [13], or resection for cases where open reduction and internal fixation was not possible [14]. In 1924, Speed [14] even stated that: “In adults, unless the lesion is only a mere crack, there is no doubt that removal of the head is primarily indicated.” Heterotopic ossification at the proximal radius was the most commonly reported short-term complication [15] for which soft tissue interposition [16, 17] and bone grafting [15] were suggested treatments. In 1941, Speed [18] was the first to describe the use of a ferrule (Vitalium) cap (Fig. 2) that could be placed over the radial neck, in order to prevent heterotopic bone formation. The same retrospective study [18] involving three patients also led to the observation that the ferrul caps prevented shortening of the proximal radius while simultaneously allowing for complete resumption of functional elbow articulation.

Fig. 2
figure 2

Different sizes of ferrule caps for the radial head, designed by Speed [30]

Full size image

In 1951, Essex-Lopresti et al. [19] suggested the use of a RH implant until the forearm had healed and become stable in cases of distal radio-ulnar dislocation associated with RH fracture. During the same year, Carr and Howard [20] (1951) demonstrated that a metallic cap increased elbow stability, when compared to RH resection. Similarly, Cherry et al. [21] proposed a second RHP design in 1953, composed of acrylic resin, to prevent proximal translation of the radius and its related consequences; its use was, however, quite limited at that time. Twenty years after Speed’s first caps (1953), Taylor and O’Connor [22] reported that half of patients treated for RH fractures with excision presented with distal radio-ulnar joint (DRUJ) symptoms. Subsequent to this report [22], RHPs became the treatment of choice to avoid distal radio-ulnar joint subluxation related to RH resection. The first long-term results of Vitallium caps were published in 1964 and resulted in similar clinical outcomes as patients treated with RH resection, in addition to decreased prono-supination among patients receiving caps [23]. As the use of RHPs was becoming more widespread, Creyssel and De Morgues [24] changed the material of the ferrule to nylon in order to increase elasticity and lessen stress on the humeral condyle; its use remained however quite rare.

Begining in 1968, Swanson developed silastic implant (the Swanson implant, Dow Corning Corporation, USA) (Fig. 3). The implant’s low elastic modulus was intended to allow for easy implantation and provides a smooth surface for radiocapitellar articulation [25]. Swanson et al. [25] initially reported excellent short- to mid-term outcomes and advocated for more widespread use of the device. However, since 1979, criticisms of the prosthesis arose because of silastic’s propensity to create wear particles which led to inflammatory arthritis, reactive giant cell synovitis [26] and mecanical failure [27]. Silastic was too easily deformed and did not provide sufficient resistance to axial load (Fig. 3) [28]. After assessing the long-term outcomes of 17 Swanson implants, Morrey et al. [29] concluded that the indications for silastic RHPs after fracture were extremely limited, and its routine use could not be justified. Given the fact that silastic’s poor reputation had been earned on the basis of limited data with questionable surgical indications and techniques (e.g., no medial collateral ligament repair), Magen et al. [30] suggested that the use of silastic implants might be still reasonable for stable elbows or those rendered stable by ligament repair.

Fig. 3
figure 3

a The “Swanson” silastic prosthesis. b Highly deformable head. c High flexibility stem. Reprinted with permission from Morrey [14]

Full size image

However, due to the abundance of data demonstrating sub-optimal results with the Swanson implant, more rigid materials (e.g., ceramic, cobalt-chrome, or titanium) have been used manufacture RHA to since the early 1980s.

Revolution

As a result of the inability to restore valgus statibility with a RH implant, there has been increasing interest in elbow biomechanics since the 1980s; data resulting from these biomechanical studies revolutionized the approach to RHA.

In the early to mid-1980s, Morrey [29] and Carn et al. [31] demonstrated that silicone rubber deformed easily under physiologic loads and transferred minimal force to the capitellum. During the same period, Harrington and Tountas [32] demonstrated that stiffer implants provided improved resistance to valgus stress compared to softer silicon rubber. Further evidence to support the use of stiff materials came from Knight et al. [33] in 1993 when they reported restored normal axial forearm stiffness with the use of vitallium RHPs.

Metallic implants have a high modulus of elasticity, ranging from seven to 15 times greater than that of bone. This property confers improved resistance to deformation; however, it also increases the risk of capitellar wear, periprosthetic osteolysis, and stress shielding [33]. In an effort to reduce stress shielding and more closely replicate the anatomy of the proximal radius, modular and bipolar implant systems were designed and promoted (Fig. 4). These devices had the added benefit of easing implantation and compensating for potential technical errors. The first bipolar RHP was promoted by Judet et al. in 1988; this early version, made from titanium, was replaced by a cobalt-chrome design in 1994 [34, 35]. The “floating” RH prosthesis (Wright Medical, Arlington, TN, USA) had a collared stem with a 15° neck-shaft angle, a floating articulation, and concave head to allow for continuous contact with the convex humeral condyle during elbow range of motion (ROM) (Fig. 5). Short-term results of the floating RH prosthesis were also promising (83.3% excellent or good outcomes according MEP score) and the prosthesis was found to restore joint stability [34,35,36]. However, degenerative changes of the elbow after bipolar RHA were found in approximately 50% (n = 8) of patient at short-term follow-up in a multicentre  study carried out by Smets et al. [36].

Fig. 4
figure 4

Four examples of modular bipolar (a, c) and monopolar radial head prostheses (b, d): a Evolutive (cemented, Aston Medical, Saint-Etienne, France), b MoPyc (auto-expanding, Wright Medical, Arlington, TN, USA), c rHead Recon (bipolar, press fit, Stryker-Small Bone Innovation, Morrisville, PN, USA), d Evolve (Wright Medical Technology, Arlington, TN, USA)

Full size image
Fig. 5
figure 5

Pictured here is the first bipolar radial head prosthesis. The Judet RHP (Wright Medical, Arlington, TN, USA) has a collared stem, a 15° neck-shaft angle, and a floating articulation; head and stem sizes are interchangeable

Full size image

At the beginning of the 2000s, further anatomic studies demonstrated that the relationship between the RH and neck was quite variable [37, 39]. Beredjiklian et al. [37] reported that even the smallest prosthetic stem available would not fit into the intramedullary canal of the radius in 39% (n = 18) of patients. The anatomy of RH was also found not to be circular, but rather consistently elliptical in shape [39]. Biomechanical analyses demonstrated altered ulnohumeral kinematics and radiocapitellar stability with the use of nonanatomic prostheses due to their shallower articular surface and the fact that they do not replicate the elliptical shape of the native RH [37,38,39,40,41]. This data created a significant need for more anatomic and modular implant systems which led to diversification of device designs and fixation modes including the following: anatomic or nonanatomic, monoblock or bipolar prostheses, with short or long stems anchored via cement, and press fit, loose fit or auto-expanding stem systems (Fig. 4, Table 1) [10]. The first prosthesis designed to address the aforementioned requirements was the Anatomic Radial Head System (ACUMED, Hillsboro, Or, USA) which was released in the early 2000s (Fig. 6).

Table 1 Design and biomaterial composition of contemporary radial head prostheses
Full size table
Fig. 6
figure 6

The Acumed RHP (ACUMED, Hillsboro, Or, USA) is the first anatomic implant designed to replicate native radial head geometry

Full size image

Indications for RHA were also redefined as the understanding of elbow biomechanics improved. The ligamentous contribution to elbow stability has been studied exhaustively. Morrey et al. [42] confirmed, in 1983, that the medial collateral ligament (MCL) is the primary stabilizer of the elbow. The RH was therefore considered a secondary restraint to valgus and rotatory instability, significant only if the anterior bundle of the MCL was absent [32, 43]. Less than ten years later, Morrey et al. [44] published data to suggest that simultaneous rupture of the anterior band of the MCL should be a new indication for RHA and the primary contraindication to RH resection. In 1996, Olsen et al. [45] showed that the lateral collateral ligament was an equally important stabilizer of the elbow joint during varus and external rotation loads. Three years later, the same group [46] demonstrated that excision of the RH reduces tension in the lateral collateral ligament and induces varus and external rotation laxity. More recently, Beignessner et al. [47] (2004) demonstrated that RHA alone does not adequately restore stability to elbows having a ligamentous injury and recommended concomitant ligament repair at the time of RHA. In addition to ligamentous stabilizers, the coronoid also plays a central role in elbow stability by providing an attachment site for the anterior bundle of the medial collateral ligament [42, 48,49,50]. Biomechanical studies have demonstrated that deficits comprising approximately 50% of the coronoid are grossly unstable, even if the remaining stabilizers, the RH and collateral ligaments are intact or have been restored [48]. O’Driscoll et al. [49] recommended that coronoid fractures must also be addressed, especially when they occur with RH fractures as part of a terrible triad injury.

With the above data in mind, RHA became the treatment of choice for non-reconstructable RH fractures. Recognition of “complex instability” led to more global surgical treatment, inclunding concomitant repair of ligaments and coronoid fractures when present.

Evolution

After 2000, multiple RHP designs were available for the treatment of similar injuries (Fig. 4, Table 1); with no significant evidence to support the use of one implant over another [10, 12, 51, 52]. More recently, a classification system was devised to describe the four main reasons for re-operation after RHA: painful loosening, stiffness, humero-radial conflict, and instability. The study of these primary reasons for failure triggered rapid evolution in the indications for RHA with each individual prosthesis.

Painful loosening, as defined by O’Driscoll and Herald [53], is the main indication for re-operation after RHA [12, 51,52,53,54]. Despite our limited understanding of the biomechanics, fixation method seems to play a pivotal role in RHA survivorship [10, 52, 55]. Press fit implants may be the most prone to painful loosening [10, 52, 55], especially those prostheses with shorter stems of sub-maximal diameter (approximately 1 mm less than maximum diameter of the radial neck medullary canal) (Fig. 7) [55,56,57]. Difficulties in obtaining satisfactory stability when using short-stemmed and/or bipolar implants may also predispose the surgeon to favour stability over implant positioning [12, 55].

Fig. 7
figure 7

AP (a) and lateral (b) radiographs depicting loosening of a press fit, short-stemmed radial head implant

Full size image

A recent review of the literature found that radiocapitellar instability was the main reason for re-operation with implant retention (11.25% of failures) [12]. In 2009, Moon et al. [58] demonstrated that bipolar implants were more prone to radiocapitellar subluxation. O’Driscoll and colleagues [59] showed that a bipolar prosthesis depended more heavily on the integrity of the surrounding soft tissues for restoration of stability than did a monopolar prosthesis. This data would suggest that monopolar prostheses provide significantly greater radiocapitellar stability when used in the treatment of terrible triad injuries, than bipolar implants. These findings have been replicated in multiple clinical studies, though without statistically significant evidence, and the monopolar implant is currently the device of choice in cases of associated soft tissue injury [12, 58,59,60,61,62,63].

The long-term outcomes of RHA are largely unknown and concerns about capitellar wear remain (Fig. 8). The use of metal radial heads has been demonstrated to lead to severe capitellar cartilage wear [51]. Biomechanical studies demonstrated that the geometry and design of RHPs influence their contact characteritstics and can contribute significantly to changes in the articular cartilage [58, 59, 61]. Sahu et al. [62] showed that anatomic RHP with articular surfaces that match the radius of curvature of the capitellum have increased radiocapitellar contact areas and lower peak pressures compared to mono- and bipolar implants. Although comparative clinical studies could not reproduce these results, the use of the anatomic RH is recommended to avoid long-term cartilage damage; the polarity of the design does not appear to affect this endpoint [12, 51, 59,60,61,62,63].

Fig. 8
figure 8

AP (a) and lateral (b) radiographs demonstrating capitellar wear caused by an overstuffed bipolar radial head implant

Full size image

Although current studies report satisfactory mid- to long-term outcomes after RHA [10, 64], some still advocate for RH resection as an alternative to RHA for isolated, non-reconstructable radial had fractures [3]. Several studies demonstrate excellent results of RH excision for isolated RH fractures at mid- and long-term follow-up [5]; however, there are also reports of unsatisfactory results with high complication rates [7]. When magnetic resonance imaging (MRI) is performed in the setting of RH fracture, studies found even higher incidences of concomitant ligamentous injury than with physical or clinical examination alone [65, 66]. Itamura et al. [65] found a 92% incidence of associated injuries in Mason type II and type III RH fractures; Kaas et al. [66] found that 100% of Mason type III RH fractures had associated injuries. This might suggest that RHA is indicated for all non-reconstructable RH factures whether isolated or not. Radial head resection should be considered only in cases where RHA is contraindicated. Ligamentous injuries are quite common in the setting of non-reconstructible radial head fractures. Given this, we submit that monopolar modular implants with loose fitting stems (Table 1) should be used preferentially in this situation due to their improved stability (vs. bipolar implants), satifactory restoration of proximal radius anatomy, and low rate of painful loosening.

Future directions

Orthopaedic surgeons have been searching for the ideal RH prosthesis since its initial development for more than 80 years. Meticulous biomechanical studies have stimulated a revolution in the approach to RHA and complex instability of the elbow. In contrast, the clinical and radiographic literature regarding outcomes of RHA have led to inconsistent conclusions and have been largely unable to reproduce in vitro findings. We speculate that it may be due to the quasi-exclusively retrospective monocentric nature of the majority of studies and the inherent bias associated with that study design. Furthermore, the small sample size, the lack of a standardized classification system of the reasons for failure, and the plurality of methodologies used have prevented reproducible studies of RHA. A recent meta-analysis [12] estimates that 90% of current literature has insufficient follow-up and underestimates the rate of failure of RHA. Similar to what is being done in the prospective, randomized, multicentre  trial “RAMBO trial” [67], the challenge in the coming years will be to perform and publish high level of clinical studies in order to obtain reliable results and provide clear recommendations for surgeons. This will require strict adherence to the quality standards developed for observational studies [68]: clear definitions of outcomes (1) and the assessment of outcomes (2), an independent assessment of the outcome parameters (3), sufficient follow-up (4), no significant loss to follow-up (5), the identification of important confounders, and prognostic factors (6). A task of this magnitude calls for collective responsibility among authors and journal editors for transparent, comprehensive, and standardized reporting of all outcomes and study characteristics.

To conclude, despite a growing body of outcomes data and improvements in implant design and rationale, prosthetic RH replacement has yet to become the gold standard for treatment of non-reconstructible RH fractures. We suggest that RHA will be considered the treatment of choice for these injuries when a study with a high level of clinical evidence provides more definitive evidence to support its widespread use.