The World Health Organization estimates that 10% of the world’s population over the age of 60 suffers from osteoarthritis (OA) . The majority of these cases involve the hip and knee. Symptomatic osteoarthritis of the ankle with radiographic changes is seen in approximately 11% of patients seeking treatment for “arthritis” . Idiopathic arthritis, as seen in the hip and knee, is rare in the ankle joint. The majority of ankle arthritis is secondary to trauma (70%), with a smaller percentage being attributed to inflammatory arthropathies such as RA (12%) and only 7% attributed to primary OA . Ankle arthritis can be a debilitating disease. In fact, the mental and physical disability associated with end-stage arthritis of the ankle is at least as severe as that associated with end stage hip arthritis .
The ankle joint consists of the tibio-talar joint incorporating articular surfaces between the talar dome with the tibial plafond, and the medial talus with the medial malleolus. In addition, there is an articulation between the lateral talus with the lateral malleolus. It is mainly a rolling joint with congruent surfaces at high loads, allowing it to withstand large pressures . The ankle carries 5 times body weight during normal walking activities . The majority of the load (75%) is distributed across the tibio-talar joint but some force is transmitted through the medial and lateral sides . Pathological conditions such as a mal-united ankle fracture results in changes in the contact stresses and the overall contact area .
Patients with arthritic changes to the ankle have the typical “arthritic” complaints as seen in other affected joints. Patients often present with pain, swelling and stiffness. The pain may be worse with weight bearing/impact activities, with weather changes, and may wake them at night. Painful symptoms may be worse with initiation of activities (especially first thing in the morning) but may improve during the activity. The pain usually intensifies with prolonged activity and after the activity is completed. Stiffness can be present, usually worse first thing in the morning or after a prolonged rest. Swelling can be particularly bothersome and is usually intermittent. Patients complain of pain in the anterior talo-crural area and have difficulty with reciprocal stair climbing and hills. Subfibular or lateral foot pain is not common and is more typical for patients with subtalar arthritis which may be seen in combination with ankle arthritis. Patients with subtalar arthritis complain of difficulties with walking on uneven ground.
Given that most ankle arthritis is post-traumatic, a history of previous trauma should be determined including a history of severe or repetitive ankle sprains . History of previous infection is important to ascertain as it can be a contraindication to joint arthroplasty and current infection should be treated immediately with antibiotics and possibly debridement. Systemic illness such as rheumatoid arthritis, gout, and hemophilia should be noted. Patients with diabetes mellitus can present with Charcot deformity including arthritis.
1.3. Physical exam
Patients should be inspected for general lower limb alignment, including the knees, ankle, hindfoot, midfoot and forefoot, paying particular attention to varus/valgus mal-alignment of the hindfoot and dorsiflexion/plantarflexion mal-alignment of the ankle. Inspection should note any swelling about the ankle, wasting of the calf musculature and previous surgical incisions. Gait should be observed. Patients generally have a foreshortened stance phase, and may walk with their foot externally rotated. Patients with ankle arthritis walk with a shorter stride length, reduced walking speed, and a shorter stance phase of the affected limb [4, 5, 7, 8]. Active and passive range of motion should be compared to the other side for ankle dorsi/plantar flexion and subtalar eversion/inversion. In severe ankle arthritis, it is important to differentiate range of motion coming from the ankle joint as opposed to the hindfoot and midfoot joints. Heel eversion and inversion may also show restrictions of movement. Palpation should include all bone prominences and the anterior margin of the ankle joint looking for areas of tenderness or synovitis. Sensory changes or motor dysfunction should be noted for possible neurological abnormalities. Both dorsalis pedis and posterior tibial pulses should be assessed along with capillary refill and any abnormalities may require further vascular investigations with ABI and/or CT angiogram.
Standard radiographs should include WEIGHTBEARING AP/Lateral and mortise views of both ankles, and AP and lateral views of the foot. Radiographs can be reviewed for the classic signs of OA including loss of joint space, subchondral sclerosis, cyst formation and osteophyte formation. CT scan can be useful to determine arthritic changes in adjacent joints, or to look at the surrounding bone stock when considering arthroplasty but this technique is compromised for diagnosing early OA with cartilage loss because it is not done in the weightbearing position; therefore it can be misleading regarding joint space maintenance and overall joint alignment. MRI is not required when assessing ankle arthritis. If there is a past history of infection, quiescent infection should be ruled out with a bone/gallium scan combination or a white blood cell scan.
2. Conservative treatment
Non-steroidal anti-inflammatories drugs (NSAIDs) have long been the mainstay for symptomatic treatment of arthritic joint pain. These drugs control the pain of arthritis but are not disease modifying agents. Current classes of drugs include standard NSAIDs (eg. Ibuprofen, Naproxen, Indomethacin). The newer Cox 2 inhibitors [eg. Celecoxib (Celebrex) and Rofecoxib (Vioxx)] have less gastrointestinal side effects [9, 10]; however, concern about higher risks of myocardial infarction, thrombosis and stroke lead to withdrawl of Rofecoxib (Vioxx) from the market in 2004 Patients with long term NSAID use are counseled to have blood pressure and kidney function checked regularly with their family physicians as all NSAIDS still put patients at risk for renal-related complications.
Nutraceuticals such as glucosamine sulphate and chondroitin sulphate are popular non-prescription remedies. There have been a multiple studies on these products, the vast majority done in the arthritic knee population. Findings include pain modification similar to NSAIDS with a similar side effect profile, although the onset is slower [11, 12]. Other studies suggest actual disease modification with long term glucosamine use, as seen by NON progression of joint space narrowing [13, 14]. More recent authors have found no difference between glucosamine and placebo  and no evidence of disease modification or pain relief . Popular dosing for glucosamine is 1.5 grams per day.
Corticosteroids demonstrate both anti-inflammatory response and immunosuppressive effect. After intra-articular administration, there is a decrease in erythema, swelling, heat and tenderness. This may be secondary to altered movement and function of leucocytes, reduced micro vascular dilation and permeability, and reduced prostaglandin synthesis. Side effects can include septic arthritis, tendon rupture, post injection wheal and flare, facial flushing, atrophic skin changes, toxic effect on cartilage (at large doses) and possible systemic absorption. Standard limit of injections in weight bearing joints is 3-4 injections per year .
Hyaluronic acid (HA) is a major component of synovial fluid, and comprises part of the structure of hyaline cartilage. Viscosupplementation by injection of HA derivatives (as available in pharmaceutical components) has multiple proposed mechanisms of action. Biologic effects include an anti-inflammatory as well as anti-oxidant effect, and restoration of endogenous synthesis via positive feedback loop. Viscoelastic properties include a “mechanical spongy” trapping of immune complexes and inflammatory cells and restoration of synovial fluid viscosity and elasticity. Multiple versions of HA exist from multiple pharmaceutical companies. Not all are available in all countries . Some examples of currently available products include Hyalgan (sodium hyaluronate; Sanofi-Aventis, Bridgewater, NJ, USA), Synvisc (hylan GF-20; Genzyme Corportation, Cambridge, MA, USA), Supartz (sodium hyaluronate; Smith & Nephew, London, UK), Orthovisc (high molecular weight hyaluronan; Anika Therapeutics, Inc., Bedford, MA, USA), Euflexxa (1% sodium hyaluronate; Ferring Pharmaceuticals, Inc., Saint-Prex, Switzerland), and Durolane (stabilize hyaluronic acid; Smith & Nephew, York, UK). Most require multiple injections although newer ones (eg Durolane) are coming to market that require only one injection. Side effect of HA injections are rare but include anaphylaxis, pseudogout type reaction, and Severe Acute Inflammatory Reaction (SAIR) which presents like septic arthritis.
The effectiveness of hyaluronic acid derivatives in the ankle joint has not been determined due to poorly conducted studies with mixed results from the few good studies performed [19, 20] although a recent randomized control trial article comparing HA to saline in the ankle showed no therapeutic benefit from the HA injection. Both HA and Saline injection resulted in similar changes in VAS, AOS and AOFAS scores . Further high quality studies need to be completed before a definitive recommendation for HA derivatives in the ankle can be given.
There is a paucity of any literature on bracing and shoe modifications. Practically, bracing progresses from a simple lace-up ankle brace to a more elaborate rigid Ankle Foot Orthosis (AFO) (Figure 1). A rigid AFO works by restricting painful ankle range of motion.
More severely affected joints may be helped by a clamshell AFO (offloading AFO) (Figure 2). This type of orthosis works by transferring weight from the ankle to the calf using the clamshell portion that wraps around the calf (similar in principle to a patellar-tendon brace). This type of AFO can have a hinge to maintain ankle range of motion or can be rigid to restrict motion as well as off-load the joint.
Some patients may benefit from shoe modifications such as a rocker bottom shoe (Figure 3). This provides increased toe-off, thereby stressing the ankle joint less.
3. Surgical options
After non-operative treatments have been exhausted, surgical treatments can be considered. These options will depend on the location and extent of the arthritis.
3.1. Focal joint abnormalities
Localized lesions within the ankle joint can progress to generalized osteoarthritis. As documented in the knee, larger lesions have a worse prognosis than smaller lesions [22, 23]. Smaller lesions (usually considered 1cm2 or less can be asymptomatic or often are dealt with by arthroscopic debridement of loose cartilage with or without microfracture technique. If there is underlying bone necrosis, removal of this bone and overlying damaged cartilage can provide symptomatic relief. For larger lesions, the surgical options become more complicated and less predictable. There are a variety of techniques that can be used but no specific technique has been shown to superior to the rest. The most basic and economical procedure is microfacture. This technique can be used to induce local mesenchymal stem cells to influx to the defect and produce a fibrocartilage repair surface that can function well, particularly in the short to medium term [24, 25],. Although most research has been performed in the knee, this can be successful in the ankle as well [26, 27]. Of the more complicated procedures, perhaps the most publicized is autologous chondrocyte implantation (ACI) first described in 1994 in the knee . It is also performed in the ankle . Although there is a lot of literature published on this technique in the knee, the inability to reproduce the complex nature of the articular matrix has led to further investigation into variations of the technique including matrix-associated chondrocyte implantation (MACI) [30, 31] and characterized chondrocyte implantation (CCI) . Both of these techniques attempt to improve on the quality of matrix formed during the procedure. Other techniques utilize intact matrix to replace damaged matrix. This ensures the complex structure of the matrix is maintained. These techniques include osteochondral autograft transplantation (OATS) [33, 34] and mosaicplasty ) [36, 37], and bulk osteochondral allografts [38-41]. Finally, other more experimental techniques such as synthetic scaffolds and mesenchymal stem cells have been tried with variable success [42-45].
3.1.1. Cell-based techniques
Post-operatively, the patient is allowed early ROM but should be non-weight bearing for a period of time to allow some tissue formation. The length of time depends on the size and location of the lesion. Advantages of this technique include that it is one operation with low cost. Disadvantages include that fibrocartilage instead of hyaline cartilage is formed, filling of the defect is unpredictable and often incomplete . That said, if good results can be obtained, the results in the ankle may not deteriorate over time . This is considered a first line treatment by many surgeons, especially those that do not have access to the more complicated and expensive exogenous cell- or tissue-based therapies.
MACI is the next generation of ACI that incorporates the use of a scaffold [50-52]. The procedure is similar to that of ACI except that a matrix is used instead of a periosteal flap. The matrix is impregnated with the cells and shaped to the size of the defect. It is inserted and fixed by different mechanisms. Advantages of this technique include that it can be done arthroscopically in some cases and does not require the extra step of harvesting and suturing the periosteal flap. Thus it is less invasive. It has not been shown to be conclusively better than ACI. CCI (characterized chondrocyte implantation) is almost exactly the same as ACI except that the laboratory preselects specific cells that the company believes are predisposed to form articular cartilage out of all those that are harvested. These are the cells that are then expanded and re-implanted with the theory that the identified subpopulation of chondrocytes will produce a more physiologically intact matrix. The disadvantage of this technique is some patients do not have sufficient cells that meet the criteria for an enhanced population and, therefore, are not candidates for this procedure.
The results for these procedures have generally been good. Concerns with these procedures include that two separate surgical procedures are required and the cell processing can be prohibitively expensive. These major concerns limit their availability, especially when they have not been conclusively shown to be superior to other techniques . The uncertainties of these techniques have led some researchers to investigate the use of mesenchymal stem cells instead of chondrocytes. This would allow the surgery to be completed at one time and may produce young chondrocytes that could potentially produce a more robust cartilage. These techniques are in the more experimental stage and more commonly used in the knee joint.
3.1.2. Tissue-based techniques
Osteochondral autograft transplantation is a technique developed for small to medium sized defects. It consists of harvesting osteochondral plugs from less weight bearing portions of the same joint (or a different joint such as the knee when dealing with the ankle) and implanting them into the defect. This is accomplished by using 6mm diameter reamers to a depth of 10-15mm into the subchondal bone. A corresponding osteochondral core is harvested from another area and gently impacted into the newly formed defect attempting to achieve a press-fit while maintaining the contour of the articular surface. With lesions larger than 6mm diameter (ie most lesions that require treatment), multiple cores are required and press-fit with an interdigitating method and termed mosaicplasty. Once again, there are some reported good results using this technique [33, 36, 37, 54] but this is a complicated technique with multiple drawbacks. Harvesting cores from the ankle is usually not possible, therefore another joint must be violated and the harvest sites are not without possible complications such as knee pain and instability [35, 55]. Typically an osteotomy is required to gain access to the talar dome because the plugs must be inserted perpendicular to the surface of the joint. Alignment of the multiple plugs is difficult to ensure a smooth surface and the cartilage from the harvest often is of different thickness compared to the insertion site resulting in a variable underlying subchondral bone. Obtaining precise contour matching for lesions on the “shoulder” of the talus can be extremely difficult. Bone plugs of only 6mm diameter are susceptible to fracture resulting in unstable fixation and the boundary between the transplanted cartilage and native cartilage heals with fibrocartilage.
Osteochondral allografts can be done orthotopically and can be used for partial or whole joint replacement procedures. Popularized by Gross in the 1970’s, these allografts were typically fresh and used for all types of defects in many joints including those defects due to tumours and trauma. Historically, these grafts were used in defects that consisted of bone and cartilage but can be used for cartilage-only defects as well. Good to excellent medium- and long-term results have been reported [38-41, 56] but the use of this technique became restricted as concerns regarding transmission of infectious diseases became more prevalent. Considering that this is not a life threatening disorder, the risk of transmission of infectious diseases does not seem warranted (although disease transmission has been very rarely reported) and this method is rarely used. To decrease the risk by allowing more time for testing, hypothermic storage at 4°C was developed and allows the tissues to be stored for 28-42 days prior to transplantation. Unfortunately, cellular deterioration tends to begin after 7-14 days [57, 58]. Given that regulatory clearance takes about 14 days, the viability of the cells within the tissue may not be as healthy as those in fresh allograft or in autograft. Investigations have been ongoing to develop storage techniques that eliminate the possibility of infectious disease transmission and tissue deterioration over time while enabling the creation of a large tissue bank for ready access. Recently we have developed a cryopreservation technique that can store the tissue indefinitely without deterioration and this may improve access to the tissue required for this technique . Despite the limitations of hypothermic storage, there has been a resurgence of osteochondral allografting in recent years. This is the only biologic technique that can resurface whole joints [60, 61] and has been quite successful for partial joint replacements [40, 62]. Bugbee et al  have published promising results in total ankle arthroplasty using osteochondral allografts and similar procedures have been performed in Europe  but further improvements are required.
3.2. Early osteoarthritic changes
Once osteoarthritic changes become evident, cell based cartilage resurfacing techniques become less relevant due to the generalized nature of the process. There are some surgical techniques that can be used to palliate symptoms depending on the location and severity of the arthritis.
3.3. Global osteoarthritis
Once the osteoarthritis becomes global and more severe, the treatment options become more limited. The mainstay of treatment for end-stage ankle arthritis is ankle fusion although synthetic total joint arthroplasty is becoming more common as will be discussed. In addition, total ankle arthroplasty allografting is an experimental technique that has recently been developed that may be used in select cases.
Ankle fusion is the mainstay treatment for end-stage ankle arthritis. It has been performed for over 130 years  and there are research stucies that demonstrate the long-term effectiveness [68, 69]. Complications of ankle fusion include nonunion, malunion, infection, nerve injury, persisting pain, loss of ankle range of motion, limp, and arthritis of surrounding joints due to overload [70, 71]. Ankle fusion entails removing all of the remaining articular cartilage from the joint as well as the underlying subchondral bone leaving exposed cancellous bone. The surfaces are shaped so that there is excellent contact between the talar dome and the tibial plafond. Most techniques incorporate the medial malleolus as a medial buttress while some also include the lateral malleolus. Once the surfaces are prepared, internal fixation (occasionally external fixation) is applied. Postoperative splinting in either an aircast or plaster cast is employed for 6-12 weeks with non-weight bearing from 1-12 weeks depending on bone quality and rigidity of fixation and progression of fusion radiographically.
Over the years, many techniques have been described with increasing success with regards to fusion rates. Successful fusion rates can be expected to range from 86-100% depending on patient characteristics and surgical techniques [68, 72-83]. The most commonly used techniques utilize internal fixation. This can be completed using anterior, lateral or posterior approaches. Fixation can be achieved using crossed screws, plate and screws or an intramedullary nail. Another more recent method includes the joint debridement achieved arthroscopically with percutaneous screw fixation. Alternatively, external fixation using standard external fixation techniques or Ilizarov can be used after open joint debridement.
Traditionally, open techniques using an anterior approach going either medial or just lateral to the tibialis anterior tendon have been used. This approach typically provides access to the medial and superior aspects of the joint [75, 77]. This technique can be employed in ankles that do not have much structural deformity. Once the articular surfaces are properly prepared, the talus is compressed into the medial axilla and fixed by a variety of measures including crossed screws  (Figure 8), plate and screws [75, 77] or external fixation/Ilizarov [79, 80] (Figure 9). Less commonly, an intramedullary ankle arthrodesis nail can be used for fixation but this must also cross the subtalar joint [78, 83] (Figure 10).
In ankles with more significant deformation, especially with bone loss, a more extensive approach may be considered. This entails a lateral approach and employs a fibular osteotomy approximately 5-6cm above the syndesmosis [72, 82] (Figure 11). The fibula is osteotomized obliquely over a 1cm distance and displaced from the tibia using anterior dissection while leaving an intact posterior hinge. The distal fibular segment is split sagitally and the inner ½ removed so that the other ½ (still attached by the posterior soft tissue hinge) can be used as a lateral buttress plate across the joint. Through this lateral incision, the surfaces of the talus and tibial plafond can be prepared for the fusion. This can be supplemented by a small anterior medial incision over the medial gutter for that surface preparation. This approach allows contouring of the fusion surfaces to achieve good bone apposition and to correct mal-alignment. Often supplemental bone graft is required to fill eroded defects. Fixation can be carried out by screws including two screws from the fibular strut into either the tibia and the talus or the tibia alone (to allow talar compression during weightbearing postoperatively).
Some surgeons prefer the posterior approach to the ankle for fusion [84-86]. This approach can be through or beside the Achilles tendon and can provide good exposure while minimizing risk to compromised anterior soft tissues when present. This approach can be extended to incorporate the subtalar joint in the fusion mass and a blade plate is often used for fixation.
In the circumstances of infection or soft tissue compromise, alternative approaches should be considered. The ankle joint can be adequately debrided with minimal invasiveness using arthroscopy [79, 87, 88] potentially decreasing soft tissue complications. Alternative fixation such as external fixators  or the Ilzarov  can be used to limit the issues of hardware insertion during ongoing infection. These devices may also be able to correct deformity over time.
As noted, ankle fusion can provide excellent relief of pain with significant improvement of function . Surgical techniques have advanced such that it should be expected to have a 90-100% fusion rate in uncomplicated cases. Meticulous soft tissue handling and intraoperative care can mitigate complications. Despite these results, ongoing concerns of limitations of ankle motion and the increased stress on surrounding joints resulting in increased incidence of arthritis has led to the development of the total ankle arthroplasty as will be discussed in the next section.
4. Arthroplasty implants
4.1. Historical perspective
Total ankle arthroplasties were first introduced in the 1970s . The first prosthesis, created by Lord and Marotte, was designed as an inverted total hip replacement . Although Lord and Marotte’s primary results were failures, over the next 20 years, more than 23 designs were developed using that generation's technology from hip and knee replacements characterized by cemented, two component, constrained implants . Despite good initial results using this technology, there were problems with component loosening, severe osteolysis, infection, impingement and soft tissue breakdown [93, 94]. Kitaoka reviewed results of the first generation TAR that included three studies with greater than 5 year follow-up and found a 41% revision rate . The high rate of failure for these first generation implants and difficult salvage procedures tempered enthusiasm and led to questions as to whether the ankle joint could be replaced successfully [92, 95-97]. Hintermann accurately stated, “in an era of joint replacement surgery, ankle procedures have failed to achieve what has been accomplished with other joints” .
The first generation implants were designed without proper understanding of ankle biomechanics [99, 100]. The ankle was assumed to be a hinge joint and the rotational and translational motions were not recreated which lead to loosening [97, 101]. These implants were either highly constrained or unconstrained. The highly constrained implants led to large shear and rotational forces that were distributed through a small joint surface and even smaller prosthesis interface [94, 102, 103]. The unconstrained implants were unstable due to lack of attention to soft tissue balancing and failure to address soft tissue impingement [94, 104, 105].
In general, ankle arthroplasties were implanted through an anterior approach to the ankle but the importance of the soft tissue envelope was underestimated and excess traction on the skin led to wound complications [92, 93]. The guides, jigs and instruments were not well designed and sizing was inaccurate so implants were mal-positioned and malleolar fractures were common . Patients were left with mal-aligned implants either because the angular deformity was not corrected at the time of surgery  or because the ligaments were ignored leading to problems with functional mal-alignment despite good intraoperative positioning [94, 102]. Finally, most implants were cemented and required large bone resection including up to 17mm on the tibial side and 7mm on the talar side. This caused the implant to be cemented into soft cancellous bone which was unable to support the bone cement interface [92, 106]. Subsidence was further exacerbated by non-anatomically shaped, undersized implants [5, 92, 93, 107]. The aggressive bone resection of the tibia also changed the level of the ankle axis causing alterations in the ankle biomechanics . When function was assessed via gait analysis, there was not as much improvement as had been anticipated. Gait analyses of the first generation implants revealed that patients were subconsciously protecting the joint by not bearing weight normally across the ankle despite denying pain and were not using normal ankle dorsiflexion in the stance phase despite adequate passive dorsiflexion on physical exam [4, 5, 106].
4.2. Biomechanical considerations
In order to understand the reasons for failure and to design a better total ankle arthroplasty, it was important to understand the biomechanics of the foot and ankle. The ankle is a highly constrained joint that does not function as a simple hinge joint because the axis is oblique extending further plantar and posterior on the lateral side. This creates eversion with dorsiflexion and inversion with plantarflexion . Further, the axis of rotation changes throughout the range of motion so that when the ankle moves from a plantarflexed to a dorsiflexed position there is a combination of distraction, sliding and compression [108, 109]. Finally, there are rotational and translational movements that also need to be recreated with a replacement arthroplasty [94, 110].
The most important concepts in the design of an ankle arthroplasty are constraint and congruency. Constraint can be defined as the resistance of an implant to a particular degree of freedom of motion such as anterior-posterior translation or axial rotation . In highly constrained implants such as a hinged ankle replacement, all axial torque forces are translated through the bone-prosthesis interface as shear forces that can lead to component loosening [92, 111]. Congruency or conformity can be defined as the closeness of fit of the various components of the implant [92, 94]. In a fully conforming implant, the articular surface of the tibia has the same radius of curvature as the articulating surface of the talus in both anterior-posterior and medial-lateral directions. This decreases the wear rate because the polyethylene contact stress remains below the fatigue threshold for delamination and pitting . In a partially conforming or incongruent implant the round talus articulates with a flat tibia or there are differences in the radius of curvature of the 2 surfaces . The optimal design for a total ankle replacement is with bearing surfaces that are not only highly conforming to ensure contact throughout the bearing surface thus reducing wear but also sufficiently constrained to give the implant sufficient inherent stability to minimize shear stresses at the prosthesis-bone interface [94, 112].
Second generation implants used these two concepts to improve the design of total ankle arthroplasties. In general, the two primary alternatives were 1) a two component, partially conforming ankle replacement which replaced all three articulations by changing the ankle into a two bone joint by fusing the distal tibia-fibular joint [5, 92] or 2) a three component design which resurfaced the tibio-talar joint and the talar side of the lateral and medial gutters [5, 92]. Third generation implants are more anatomically designed and rely on ligament support as well as hindfoot “balancing” to allow the components to recreate the ankle biomechanics .
Gait studies of the newer implants show a satisfactory recovery of the natural overall mobility of the ankle joint in all three planes of motion . Further, in three component designs, there may be some movement of the polyethylene separate from the other components which helps to restore normal biomechanics . Functional motion and muscle-generated moments during walking improved at the ankle joint after arthroplasty – especially plantar flexor muscle function. This function is important because it restrains the forward movement of the tibia during the first half of the gait cycle and brings about plantarflexion of the ankle during the propulsive phase of gait . Brodsky has shown significant improvement in velocity, step length for operated side, total ankle ROM during gait, and maximum sagittal ankle joint power at push-off .
4.3. Current implants
Lewis has described key aspects for a successful ankle joint arthroplasty based on careful review of the first and second generation implants . It is important that implant size be designed to balance the need for minimal bone resection with the requirement for sufficient implant to distribute stresses over a large area to decrease the risk of subsidence . The anatomical axis needs to be recreated while also understanding that there is large variability in the axis between individuals [97, 116]. It is also important to recreate all the motions at the ankle joint instead of just plantar flexion and dorsiflexion . This includes talar tilt as the ankle is coupled to the subtalar joint to accommodate to uneven ground, and the rotational motion of internal rotation coupled to plantar flexion and external rotation coupled to dorsiflexion. Failure to understand and recreate these motions will place abnormal stress through the implant-bone interface . It is critical to understand the necessity of ligament support for ankle stability . An implant designed to be sufficiently stable to negate the need for ligament support will transfer abnormal stress to the bone-implant interface leading to loosening [94, 102, 103]. In contrast, an implant designed with low stability that is implanted into an ankle without sufficient ligamentous support will fail due to component mal-alignment or dislocation of the polyethylene [94, 97, 104, 105]. The requirement to withstand high loads means that ultra-high molecular weight polyethylene is generally recommended [97, 117].
Third generation ankle arthroplasties are now in use. The new designs show better understanding of ankle biomechanics and allow for sliding and rotational movements . There are many different third generation implants used across the world and the following discussion provides a representative sample of those commonly used in the Western world but is not inclusive of all implants. The implants will be discussed in relation to design rationale.
The common surgical approach, as will be discussed in further detail later, is an anterior approach in the interval between tibialis anterior tendon and extensor hallucis longus tendon. The neurovascular bundle is carefully protected under the extensor hallucis longus tendon. The major disadvantage of this approach is the high rate of wound healing complications [92, 118, 119]. The BOX ankle replacement (Finsbury Orthopaedics, Leatherhead, Surrey, UK) uses an anterolateral approach between the extensor digitorum longus tendon and the peroneal tertius tendon . The Eclipse (Integra LifeSciences, Plaisboro, NJ) is inserted through either a medial (preferred) or lateral approach . The disadvantage of this approach is the need for fixation of the disrupted medial or lateral malleolus.
The current implants are either mobile (3 component design) or fixed (2 component design) bearing . Mobile bearing designs allow for greater congruence, which can increase the risk of dislocation, but they share the stress through 2 articular surfaces, which may increase weardue to “back side wear” [101, 120]. There is also the potential for “overstuffing” the joint with three component designs in an attempt to reduce the chance of polyethylene dislocation ). Overstuffing tightens the joint, increases ligament strain, decreases range of motion and increases the shear stress at the bone-prothesis interface [5, 121]. Fixed bearing designs increase the constraint so they may have higher rates of loosening but have decreased back side wear if the locking mechanism is intact [94, 101]. Current mobile bearing designs include the STAR (Waldemar Link, Hamburg, Germany; 1978)(Figure 12), Buechel-Pappas (Endotec, South Orange, NJ; 1989), Salto (Tornier, Saint Ismier, France; 1997), Hintegra (Integra LifeSciences/Newdeal, Lyon, France; 2002)(Figure 12), Mobility (DePuy International, Leeds, UK; 2005)(Figure 12), BOX (Finsbury Orthopaedics, Leatherhead, Surrey, UK; 2005), CCI Evolution (Van Straten Medical, The Netherlands, 2007), and Zenith (Corin, Cirencester, UK; 2009). Two component designs include the Agility (DePuy/Ace, Warsaw, IN; 1985)(Figure 13), TNK (Kyocera, Kyoto, Japan; 1996), INBONE (Wright Medical Technologies, Arlington, TN; 2005)(Figure 13), Salto Talaris (Tornier, Saint Ismier, France; 2006), and Eclipse (Integra LifeSciences, Plainsboro, NJ; 2007).
Most ankle arthroplasty implants are constructed from cobalt-chromium alloys which are hard, corrosion resistant and biocompatible (Hintegra, Agility, Mobility, Inbone talus, BOX, Salto and Salto-Talaris, and Buechel-Pappas). The CCI Evolution adds molybdenum to the alloy to increase strength (Van Straten Medical, The Netherlands, 2007). The STAR and Inbone (tibia only) use titanium implants which are softer but can be more porous to allow for better ingrowth. The TNK is a ceramic implant (Kyocera, Kyoto, Japan; 1996).
Initial implants were cemented; however, due to the high rate of failure, most implants today are uncemented relying on bone ingrowth stimulated by hydroxyapatite coating, sintered metal beads, plasma spray metals or fiber metals . The different methods to encourage bony ingrowth include ensuring the metal is porous such as the Hintegra which is 20% porous . Some implants, like the STAR, use a single or dual plasma spray. Others use a thin film ceramic coating. The Buechel-Pappas and CCI Evolution use a porous coating covered by a titanium nitride thin-film ceramic  which also improves the smoothness of the surface. The Zenith uses the trademarked “BONIT” coating which is the application of a thin calcium phosphate layer to the titanium plasma spray. The last common technique is the use of hydroxyapatite incorporated in the Hintegra .
In most implants the tibial cut is made first. Unlike the knee, the ankle cannot be dislocated sufficiently to allow for intramedullary guides to make the tibial cuts so all ankle arthroplasties, with the exception of the Inbone, use extramedully tibial guides with or without a foot holder. The Inbone uses an external guide and radiographs to guide the insertion of an intramedullary reamer and combines it with a foot holder to place the tibial cutting guide . After the tibial cut has been completed and the intramedullary canal has been reamed, the tibial stem and plate components are inserted one at a time into the joint and built inside the ankle and tibia. The downside of using this intramedullary technique is that the subtalar joint is violated as the tibial canal is reamed retrograde through the calcaneus, talus and then tibia.
Other implants, such as the Buechel-Pappas, Mobility and Agility use intramedullary posts on the tibia to provide fixation by creating an anterior cortical window to insert the tibial component which is then replaced and generally heals without incident. If the tibial component does not have an intramedullary stem, the fixation can be with a fin (CCI Evolution), a fin with a cylinder (Salto), double cylinders (STAR), bars (BOX) or an anterior shield which may or may not be augmented with screws (Hintegra).
Tibial components can either rely on support from the intramedullary stem (Inbone) or cortical bone anteriorly and posteriorly (Hintegra). Some tibial components are rectangular and others are trapezoidal designed to be larger anterior than posterior (STARand CCI Evolution). Finally, while most are flat, they can be designed with a curve (BOX).
The design of the talus implant can be with a symmetrical curve (STAR, Buechel-Pappas, INBONE, Mobility) or asymmetrical so that the radius of curvature for the medial side is less than that for the lateral side (Salto, Hintegra, BOX, CCI Evolution). Most implants resurface the talus (STAR, Buechel-Pappas, Salto, Hintegra, Mobility, BOX, CCI Evolution, and Zenith), but the Inbone and Agility use a flat talar cut. Some designs replace the medial and lateral talar articulations (STAR, Buechel-Pappas, TNK, Salto, and Hintegra) while others do not (Agility, BOX, Mobility, CCI Evolution, and Zenith,). Finally, the polyethylene is ultra-high molecular weight polyethylene in the STAR, Buechel-Pappas, Salto, Hintegra and Inbone.
In summary, the third generation implants have been designed using the knowledge gained from more indepth study of the anatomy and biomechanics of the ankle joint as well as the causes of the failures of the first and second generation implants. When choosing an implant, the surgeon should consider the approach, the need for intramedullary guides and fixation of the tibial component, the relative importance of preserving talar bone stock by resurfacing the talus as opposed to a flat cut, the need for replacing the medial and lateral articulations of the talus, and the role for ultra-high molecular weight polyethylene.
5. Surgical considerations
Clinical indications for total ankle arthroplasty include end stage ankle arthritis that has exhausted all non-operative and joint-sparing operative treatments in a patient willing to undertake the risks of a major surgical procedure. Contraindications to this procedure include:
Compromised soft tissues
Inadequate blood supply
Peripheral neuropathy (including Charcot arthropathy)
Significant lower extremity neuromuscular impairment
Severe ligamentous laxity
Avascular necrosis of a significant portion of the tibia or talus
Inadequate bone stock
High demand lifestyle
Pertinent metal allergy
Sensory or motor dysfunction
Age is not explicitly mentioned as a contraindication in the literature. The safe zone appears to be somewhere after skeletal maturity and in conjunction with the assumption of a low demand lifestyle. Excessive malalignment is a nebulous term. The commonly accepted range is from 15-20 degrees of varus or valgus to not surgically correctable. It would appear that excessive malalignment is relative to the surgeon’s skill and experience. Obesity as a contraindication to total ankle arthroplasty is controversial and was not correlated to aseptic loosening in a recent article .
5.1. Surgical approach
There are several different types of ankle arthroplasty systems currently on the market as discussed. Each system has unique features best explained by the surgical technique manual provided by the company marketing that system. There are, however, certain common aspects and principles of all of the systems which we will outline in this section.
5.2. Preoperative planning
The condition of the soft tissue envelope including locations of scars and documentation of neurovascular status are important when planning a total ankle arthroplsty. Ankle arthroplasty may be contraindicated in some cases due to an impaired soft tissue envelope or inadequate blood supply to the foot.
Long leg weight-bearing views will help identify concurrent ipsilateral limb deformities which may affect ankle alignment. Anteroposterior and lateral views of the tibia are also important because intraoperative alignment is usually referenced off of the tibia. Weightbearing anteroposterior and lateral views of the ankle and foot are important to assess the joint being replaced but also to identify coexisting deformities and/or arthritis affecting the rest of the foot which may affect the surgical plan resulting in additional procedures that can be performed in a staged manner or concurrently depending on surgeon skill and preference.
The patient is positioned supine on a radiolucent table to allow fluoroscopic guidance of the procedure. The ipsilateral hip is supported to prop the foot perpendicular to the floor to allow easy access to both the medial and lateral aspects of the joint and enable evaluation of alignment intraoperatively. Rotation of the foot relative to the leg on the contralateral side should be noted before prepping and draping to determine what is normal for that patient.
5.4. Prepping and draping
A thigh tourniquet is utilized to provide hemostasis during the procedure. The leg is prepped and draped as a free extremity right up to the tourniquet to provide access to the knee. Many total ankle arthroplasty systems use the length of the tibia and an external alignment device as a guide to proper alignment.
5.5. Surgical exposure
The most common surgical approach to the ankle for insertion of a total ankle arthroplasty is the anterior approach. Some systems that involve tibiofibular fusion as part of the procedure may require and an additional lateral incision but still utilize the anterior approach as well. An incision is made over the anterior aspect of the ankle midway between the malleoli (Figure 14). The incision needs to be long enough to allow placement of the distal tibial cutting block and visualization of the talar neck. Care is taken to minimize injury to the branches of the superficial and deep peroneal nerves. The retinaculum is then split to expose the underlying tendons. The extensor hallucis longus tendon will be retracted laterally to protect the underlying neurovascular structures which should be located deep and lateral to the extensor hallucis longus tendon (Figures 14 and 15). If possible, the sheath of the tibialis anterior should be left intact to prevent undue pressure on the wound post-operatively (Figure 16). The anterior capsule is incised vertically and then peeled away in a subperiosteal plane medially and laterally to expose the joint (Figure 17). One should be able to see the medial malleolus clearly but the lateral malleolus is more posterior and often is not as easily seen.
Distal tibial osteophytes are now removed taking note of a commonly occurring large osteophyte anterior to the fibula. Care should be taken to differentiate the osteophyte from the fibula itself. There will also be osteophytes on the neck of the talus which can be removed with a curette, rongeur or burr to reestablish its normal contour. This will help with cutting block placement and ankle positioning when cuts are made (Figure 18).
The soft tissues should be handled with extreme caution. Blunt self-retaining retractors should be repositioned periodically to prevent prolonged pressure on the soft tisse. Care should be taken with the bone as well. Levering on the medial malleolus may result in fracture. Notching the fibula during the distal tibial resection can result in fracture as well. Exposure can be facilitated by instruments such as laminar spreaders, distractors and various retractors. Performing the talar cuts can also facilitate removal of posterior malleolar fragments in systems that require this.
The tibia is used to guide placement of the cutting blocks either with an external rod placed parallel to the shaft of the tibia by direct visualization or overlapped with the tibial canal fluoroscopically. The distal tibia is cut perpendicular to the anatomical axis of the tibia in the coronal plane. Most systems build in a posterior slope to the distal tibial cut in the sagittal plane and this can be adjusted in certain cases to allow more or less slope depending on the patient‘s needs. Longitudinal alignment of the talar component is usually in line with the second ray of the foot.
5.7. Ligamentous balancing
Ligamentous balancing is not yet as sophisticated as for total knee arthroplasty. The posterior capsule is resected during the removal of the posterior malleolus. Medially, the deltoid ligament can be released in a subperiosteal plane from the medial malleolus and/or talus, pie-crusted, released mid-substance or relatively lengthened through a medial malleolar osteotomy [123-126]. Laterally, the talofibular ligaments should be treated with care. Aggressive release from the talus can result in iatrogenic lateral instability. If there is a preexisting lateral instability, consent should be obtained for a lateral ligament reconstruction if required. Generally, stability is assessed after the bony cuts are made. A gastrocsoleus slide or Achilles tendon lengthening may be required if the ankle does dorsiflex past neutral after the ankle arthroplasty has been performed.
Repair of the extensor retinaculum during closure is very important. If left uncovered, the tibialis anterior may erode through the incision putting the arthroplasty and possibly even the leg at risk.
5.9. Post-op care
In the case of an uncomplicated ankle arthroplasty with no extra ligament reconstruction or fusion the patient has a short period of non-weight-bearing activity in a splint or cast followed by weight-bearing activity in a cast until the six week post-operative visit. Subsequently, physiotherapy to work on dorsiflexion and plantar flexion, gait retraining and leg strengthening is initiated.
There are no guidelines in regards to deep venous thrombosis (DVT) prophylaxis after total ankle arthroplasty but it would prudent to use propylaxis in patients with significant risk factors. Obesity, previous venous thromboembolism, and absence of full postoperative weightbearing were identified as risk factors for the development of DVT in a study on total ankle arthroplasty in the obese . The importance of these risk factors was echoed by Barg et al in their article looking at risk factors for DVT in total ankle arthroplasty patients who were give thromboprophylaxis . Other factors associated with increased incidence of DVTs include cancer, use of high dose estrogen replacement, and smoking.
5.10. Radiological follow-up
Ankle replacements should be followed radiologically as well as clinically in the post-operative period including weightbearing anteroposterior and lateral x-rays. One should look for changes in position which would indicate loosening of the components. Persistent periprosthetic lucencies should be followed for progression. There are a variety of angular measurements based on anatomical axes and component position as well as standardized distances from anatomical landmarks to component landmarks that can be used to assess technical success of prostheses implantation or subsequent subsidence .
The complications associated with total ankle arthroplasty would be familiar to any shoulder, hip or knee arthroplasty surgeon. and can be classified in different ways. Glazebrook et al recently published an evidence-based classification system that groups complications according to their correlation with failure of the prosthesis . The goal of this classification system is to provide prognostic information to help surgeons deal with complications more effectively. To develop this classification system, the current literature on second and third generation prosthetic ankle arthroplasty outcomes were reviewed. Based on this information they were able to provide a summary of total ankle arthroplasty implant survival and complication rates. Twenty studies met their inclusion criteria with follow-up periods ranging from two to twelve years. In total 2,386 ankle arthroplasties were reviewed. They reported nine main complications including:
Subsidence – 10.7%
Aseptic loosening – 8.7%
Intra-op fracture – 8.1%
Wound healing problem – 6.6%
Technical error – 6.0%
Implant failure – 5.0%
Non-union – 4.4%
Post-op bone fracture – 2.0%
Deep infection – 1.7%
These complications were compiled into three groups depending on how likely the complication would lead to failure of the arthroplasty. The three complications that lead to arthroplasty failure greater than 50% of the time were deep infection, aseptic loosening and implant failure. Other complications not included in this review that should be noted include tendon laceration, nerve injury, and instability (due to ligamentous imbalance). Impingement is another complication that is minor in terms of prosthesis survival but can have a significant effect in terms of patient satisfaction. It is usually medial and can be caused by osteophytes, scar tissue or tibialis posterior tendon degeneration.
In order to evaluate the outcomes of contemproary total ankle arthroplasty, it is instructive to look at the results published by joint registries from around the world. While not as common as registries for total hip and knee arthroplasty, there are at least four registries reported in the literature.
In 2007 the Norwegian registry reported on their experiences with total ankle arthroplasty . This registry looked at 257 joints implanted between 1994 and 2005. This included a small number of cemented implants. The average age at the time of implantation was 58 for women and 60 for men. The The 5-year survival rate was 89% and the 10-year survival rate was 76%. There was no difference in the survival rates of the cemented and cementless prostheses.
The New Zealand registry reported on 202 joints in total also in 2007 . This included both second and third generation prostheses. The average age of the patients at the time of implantation was 65. The most common diagnosis was primary osteoarthritis. The 5-year survival rate was 86%.
In 2010 the Finnish registry published their results on 573 total ankle arthroplasties . These were all third-generation mobile-bearing designs. The average age of the patients was 55. The 5-year survival was 83%.
The Swedish Registry reported on the ten year survival of 780 third-generation total ankle arthoplasties in 2011 . The registry included a variety of uncemented, three component implants. Revision rate was used as the primary outcome variable. The overall 10 year survival rate was 69% although this increased to 78% of the STAR prostheses were excluded. It is interesting to note that the other registries did include the STAR prostheses in their results.
Overall, age, sex, and diangosis did not affect survival in the Norwegian, New Zealand or Finnish registries. The Swedish registry, however, found that the risk for revision was higher in female patients under 60 years of age with a diagnosis of primary osteoarthritis or posttraumatic arthritis.
There have been a lot of outcome studies published in recent years but a disproportionate number have been published by the prostheses designers themselves according to Labek et al . Approximately 50% of the publications reporting on total ankle arthroplasty outcomes have been produced by implant developers. Furthermore, the revision rates from the national registries (Swedish, Norwegian and New Zealand) have a much higher revision rate than those published in peer-reviewed scientific journals that include the manufacturers‘ reports .
That said, it is still worthwhile to look at all of the published data. There have been three relatively recent reviews of total ankle arthroplasty outcome study data [69, 134, 135]. The most recent one was published by Gougoulias et al in 2010 . They included thirteen Level IV studies looking at 1105 total ankle arthroplasties. The most common diagnosis was posttraumatic arthritis. The overall failure rate was approximately 10% at 5 years. Range of motion did not change in one study and improved by 4-14 degrees in four others. In the studies reporting them, there were superficial wound complications in 0-14%, and deep infections in 0-4.6%. In seven studies, rates of residual pain in the hindfoot ranged from 23 to 60%. Interestingly, the most common sporting activities after total ankle arthroplasty were identified as swimming, cycling and fitness/weight training.
In an older review published in 2007 by Haddad et al, the intermediate outcomes of total ankle arthroplasty and ankle fusion were compared . They looked at ten papers which evaluated total ankle arthroplasty and thirty-nine which looked at ankle fusion. The mean American Orthopaedic Foot and Ankle Society Ankle Hindfoot score was 78.2 for the arthroplasty group and 75.6 for the fusion group. The average age of patients undergoing arthroplasty was 58 and fusion 50. The majority of the fusion patients were male, while the major of the arthroplasty patients were female. Rheumatoid arthritis was the primary indication for total ankle arthroplasty, posttraumatic arthritis for fusion. The 5-year implant survival rate for the arthroplasty group was 90%, while the 10-year survival rate was 77%. They noted that a weakness of their review was that there were no direct comparisons of total ankle arthroplasty to ankle fusion.
Stengel et al’s review in 2005 identified 10 adequate studies with a total of 497 ankle arthroplasties . The total ankle arthroplasty patients showed a mean improvement in hindfoot scores of 45.2 points. Range of motion improved an average of 6.3%. Complications ranged from a 1.6% deep infection rate to a 14.7% impingement rate. The five year prosthesis survival rate was 90.6%.
Since the publication of the most recent of the above reviews, there have been some other publications of note. The long-term results of the United States STAR trial were published in 2011 . They prospectively followed 80 patients treated with 84 STAR prostheses. The average age at implantation was 61 years. Their American Orthopaedic Foot and Ankle Society ankle-hindoot scores improved from an average of 42.7 to 81.9 points. Average range of motion was 39.5 degrees. Implant survival was 96% at 5 years and 90% at 10 years.
Bonnin et al recently published the long term follow-up of patients they had implanted with the SALTO total ankle arthroplasty and reported on previously . They analyzed 98 prostheses with and average followup of 8.9 years. The mean American Foot and Ankle Society ankle-hindfoot score was 79. Range of motion was 27 degrees. The survival rate was 85% with an end point being the revision of a component. Limitations of this study included the series of patients being operated on by the implant designers and the implants and technique changed during the course of the study.
Wood et al compared the STAR implant to the Buechel-Pappas implant in a randomised trial . The study involved 200 joint replacements. They were followed out for 6 years. The STAR survivorship was 95% and the Buechel-Pappas 79% with the end point being revision. The difference was not found to be statistically different.
Total ankle arthroplasty has been shown to improve ankle joint mechanics. Hahn et al demonstrated that both arthroplasty and fusion impoved gait function and reduced pain but that arthroplasty patients gained a more natural ankle joint function with increased range of motion . Piriou et al were also interested in how gait was affected by total ankle arthroplasty . They compared 12 arthroplasty patients to 12 fusion and 12 control patients. Neither the arthroplasty patients nor the fusion patients regained normal movement or walking speed, but the arthroplasty group had greater movement at the ankle, more symmetrical timing of gait and better restored ground reaction force patterns than the fusion group. The fusion group had a faster gait and longer step length than the arthroplasty group. Schuh et al compared 21 ankle fusion patients to 20 total ankle arthroplasty patients in regards to sport and recreational activities as well as functional outcome directly . They found no signifcant difference between the groups concerning activity levels, participation in sports activities.
Finally, despite improved motion and decreased pain, total ankle arthroplasty does not result in weight loss. Penner at al investigated weight loss after total ankle arthroplasty and found that although pain and disability were reduced in overweight and obese patients after both ankle arthroplasty and fusion, the mean body mass index (BMI) remained unchanged .
Ankle arthritis can be a debilitating disorder as disabling mentally and physically as hip arthritis. There are a myriad of nonoperative and joint sparing operative treatments that can improved symptoms but do not arrest the disease process. When end-stage osteoarthritis develops, the gold standard has been ankle fusion. Some limitations to a successful ankle fusion such as stiffness and arthritic progression of surrounding joints as well as the successful replacements of the hip and knee joint led to the development of total ankle arthroplasty. Second and third generation prostheses have markedly improved function and long-term results compared to first generation prostheses. Although the procedure can be technically demanding and is associated with numerous potential complications, patients generally experience improved function and decreased pain. Based on current literature, it seems reasonable to expect an 80% prosthesis survival after 10 years. Further investigations into ankle joint mechanics, risk factors for implant failure and methods to minimize complications should lead to additional improvements in long-term outcomes. Finally, prospective, long-term direct comparison studies of ankle fusion versus total ankle arthroplasty are required to determine which procedure is best for which patient.