Over the past decade, a resurgence of interest in total ankle replacement (TAR) has emerged, largely as a consequence of an improved understanding of ankle kinematics, better implant designs, and advances in surgical techniques producing better results and long-term outcomes. The longevity of these TARs, regardless of design, is unpredictable, with many short- and long-term survivorship papers reported by authors involved in the surgical design of the prosthesis. Quite apart from the short-term failures, which mostly are related to wound healing, it is the longer-term potential for complications that is of greatest concern. In view of the potential for failure of a primary joint replacement related to poor bone quality (as a result of either osteopenia or avascular necrosis), as well as the intermediate and longer-term complications such as osteolysis, subsidence, and loosening, careful consideration of the available salvage options is essential.
Key WordsRevision, ankle replacement, ankle arthroplasty, bone block, tibiotalar calcaneal arthrodesis, deformity
Failure of ankle replacements may be the result of patient, implant, or surgical factors, or a combination thereof. Poor selection of patients stemming from lack of attention to physiologic (comorbid conditions, obesity), psychologic, and lifestyle (occupational and recreational) factors can jeopardize the outcome. Outcomes can be affected by surgical decisions or technique dictating implant choice, sizing, placement, and alignment; balancing of varus and valgus deformity; and adequately addressing coexisting hindfoot arthritis and deformities.
Over the past decade, a resurgence of interest in total ankle replacement (TAR) has emerged, largely as a consequence of an improved understanding of ankle kinematics, better implant designs, and advances in surgical techniques producing better results and long-term outcomes. As a result, the implants in current use have experienced greater popularity, and the numbers of ankle replacements performed worldwide is increasing relative to ankle arthrodesis. The longevity of these TARs, regardless of design, is unpredictable, with many short- and long-term survivorship papers reported by authors involved in the surgical design of the prosthesis. Quite apart from the short-term failures, which mostly are related to wound healing, it is the longer-term potential for complications that is of greatest concern. In view of the potential for failure of a primary joint replacement related to poor bone quality (as a result of either osteopenia or avascular necrosis [AVN]), as well as the intermediate and longer-term complications such as osteolysis, subsidence, and loosening, careful consideration of the available salvage options is essential.
Failure of ankle replacements may be the result of patient, implant, or surgical factors, or a combination thereof. Poor selection of patients stemming from lack of attention to physiologic (comorbid conditions, obesity), psychologic, and lifestyle factors (occupational and recreational) can jeopardize the outcome. Outcomes can be affected by surgical decisions or technique dictating implant choice, sizing, placement, and alignment; balancing of varus and valgus deformity; and adequately addressing coexisting hindfoot arthritis and deformities. Other factors such as soft tissue complications, wound breakdown, deep infections, intraoperative technical failures, and fractures of bone or components can also cause failure.
Avoiding complications to begin with is always preferable rather than having to manage them after failed surgery. Optimizing outcomes begins with patient selection. The factors that we take into consideration are the patient’s activity level, lifestyle, and interest in exercise. The limb alignment, the quality of the bone, and the presence of adjacent joint arthritis should be taken into consideration. We will generally exclude patients who smoke and those in whom the surgery poses too high a risk, including those with peripheral vascular disease, diabetes, peripheral neuropathy, and skin conditions that preclude use of the appropriate surgical approach. Problems with wound healing may occur with increased frequency in patients who smoke and older patients with peripheral vascular disease. Bone quality is important as well, particularly in those patients with periarticular osteopenia, who have a higher risk for subsidence. Partial AVN of the talus does not preclude TAR, particularly with partial involvement of the joint. The larger or heavier patient should be very carefully selected, even if sedentary, and if osteopenia is present, we do not perform arthroplasty. To date, however, there is no evidence-based standard or proven optimal approach to managing revision of failed replacement, and these have unpredictable outcomes and high risks of failure and complications. Design and the properties of various implants are highly relevant to the success of revision total ankle arthroplasty (TAA). We have recognized that replacing the failed device with a similar prosthesis can be far easier than converting it to other systems; however, our experience has demonstrated that the number of these cases is limited, because most of the failed TAA end up with large segmental bone loss that is worsened after removal of the implants, debridement of dead bone, and resection of tibia and talus to a functional level. Very few of the available prosthesis systems are suitable for revision purposes. The ideal revision components need to have sufficient surface area, cortical contact, stability, and various options with the tibia, talus, and polyethylene inserts. Failure does not always require revision of the prosthesis. A principle of ankle replacement is that there must be a plantigrade foot distal to the ankle. For example, a flatfoot will produce unnecessary stress on the component and must be corrected either before or simultaneously with the replacement. Leaving the hindfoot in valgus or the talonavicular joint subluxated will inevitably lead to failure ( Fig. 19.1 ).
Management of Fractures
The medial malleolus is at risk for fracture regardless of the type of implant used, particularly in any patient with osteopenia, and in these patients, we protect the malleolus with prophylactic pinning. We will insert two Kirschner wires (K-wires) or guide pins for a cannulated screw, and if there appears to be higher risk for malleolar fracture, we use screws over the guide pins. The medial malleolus is at particular risk for fracture in the patient who has sustained a previous fracture, and in whom the screws are still present and are in the path of the prosthesis. If the revision implant can be inserted with the screws left in, this is preferable, but if hardware removal is necessary, then multiple guide pins or K-wires should be inserted to protect the malleolus. It is a very good idea to have a Midas Rex burr system available if hardware has been used previously. We have encountered many cases where the screw heads strip and the screw is in the path of the revision prosthesis. This is particularly the case in the cases for an Agility replacement where screws were used for the syndesmosis arthrodesis.
The largest-sized prosthesis available that will fit adequately should always be used, but care should be taken not to encroach on the medial malleolus. All tibial components are designed as a press fit—they should be reasonably tight-fitting, but not at the expense of risking fracture. The problem with revision surgery is that it is very difficult to obtain a press fit with either the tibial or talar component when there has been substantial bone loss.
If fractured, the medial malleolus must be fixed using small cannulated screws or a tension band during the same procedure. The replacement is at tremendous risk for subsequent disastrous failure unless the malleolar fracture is treated. It is essential to make sure that there is no deformity of the foot causing subsequent stress of the malleolus, which is more common with a valgus foot deformity, with stress on the deltoid and medial malleolus. Intraoperative fractures of the fibula also occur, and must be stabilized and the anatomy restored. A large cannulated intramedullary screw in the fibula is usually sufficient, and preferable to another incision laterally where an open reduction with a plate is used.
Management of Wound Healing Problems
The soft tissues surrounding the ankle joint are not well vascularized and are relatively thin and tenuous. No deep subcutaneous tissue or muscle is present surrounding the ankle, and there is little to locally cover the components in the event of wound dehiscence. The blood supply to the extremity must be monitored with measurement of the ankle-brachial index in the elderly if pulses are questionable. Patients with rheumatoid arthritis, particularly those on immunosuppressant medications or prednisone, lose subcutaneous fat, and their thin, fragile skin is more prone to wound complications.
Perhaps the most important preventive measure is to avoid any retraction and tension on the skin margins during surgery. The incision should be long enough to allow for optimal exposure without excessive retraction. We do not retract the skin edges until the periosteum is reflected, and we do not perform any simultaneous medial and lateral retraction at the same skin level. In fact, if visualization is inadequate with simple finger retraction, a slightly longer incision is probably necessary. Exactly what constitutes too much retraction of the skin to cause wound breakdown is not known, but blood flow must be maintained. The anterior tibial tendon should not be exposed, and if retinaculum over the anterior tibial tendon is torn during the exposure, it must be repaired.
A minor dehiscence of the incision is not rare postoperatively. The very best treatment of this complication is use of simple topical antibiotic agents, dressing changes, and topical care of the skin. If a superficial dehiscence occurs, it is treated with daily changes of wet-to-dry saline dressings. We like to use a wound vacuum-assisted closure (VAC) device whenever there is drainage with breakdown of the incision. Deeper wound breakdowns are treated with silver sulfadiazine (Silvadene) dressings, but these should be treated promptly, generally with a free flap, once recognized to prevent infection from developing. If an eschar develops over an area of the incision, it is best left alone. The wound is not debrided and is kept as dry as possible. The size of the eschar does not seem to be of any consequence, and over time and in the absence of infection, these dry eschars demonstrate quite a remarkable potential for healing. However, once there is exposure of the anterior tibial tendon, one has to be more careful. One can still use a wound VAC provided granulation tissue rapidly forms around it. If this is the case, we would not recommend formal debridement, because the VAC device may promote granulation and complete coverage of the tendon, without increased risk of infection. If deeper breakdown does occur, and the component is potentially exposed, coverage with a free flap is required to prevent infection and failure of the joint replacement. Removal of the prosthesis should not be necessary if coverage is adequate.
More advanced infection associated with wound problems should be treated with removal of the components, insertion of an antibiotic-impregnated cement spacer, administration of intravenous antibiotics, and coverage with a free flap as needed. Although antibiotic beads can be made to fill the gap, these are not helpful in the long run, and a solid cement spacer is preferable to keep the joint space open, distracted at the correct soft tissue tension, and maintain stability (see discussion on the use of cement later). We permit patients to walk on the ankle as soon as the skin incision has healed after treatment of infection, without concern for any bone loss or subsidence. After 6 weeks, an aspiration of the joint is performed, and if results of aspirate culture are negative, 4 weeks later the prosthesis is reimplanted. Before the surgery, we take a biopsy specimen of the synovium, and if the cell count is greater than six white blood cells per high-power field, the joint is debrided and another antibiotic cement spacer is inserted until the joint is proved to be infection free.
The early detection of loosening and subsidence is not easy, requiring a high index of suspicion, thorough history and examination, and various investigations to confirm the type of loosening. The typical history in patients with loosening of prosthetic components is one of pain on starting up with activities, which is relieved shortly thereafter once “warm-up” has been achieved. If the patient presents with pain and difficulty with ambulation, the decision to investigate the cause further, usually with computed tomography (CT) scan, may be more readily made. The situation is very different, however, if the patient presents with an asymptomatic cyst. These cysts can be massive and presumably arise in response to irritation from polyethylene wear debris, secondary to abnormal joint mechanics, and typically worsen over time. If the patient presents with an asymptomatic cyst on the plain radiograph, obtaining a CT scan is essential, but the surgeon must be prepared to perform a revision if the size of the cyst is progressively enlarging. The CT findings are always far worse than expected, and as noted, these cysts can be quite massive.
Expansile osteolysis resulting from polyethylene wear debris tends to progress, with consequent destabilization of the implant, leading to loosening and, ultimately, periprosthetic fracture. Therefore despite the total lack of symptoms in some patients, we try to encourage them to understand the cause, with its implications for the need for a timely revision to prevent catastrophic mechanical failure and to preserve and supplement the remaining bone support. A real therapeutic dilemma arises if the patient is asymptomatic, with good range of motion, and plain radiographs and CT scans show an apparently stable prosthesis but large osteolytic cysts. It is hard to convince a patient to undergo a revision when the joint is entirely asymptomatic. Serial investigations and close follow-up examination may be acceptable to help the patient decide when a revision or bone grafting is needed. These large asymptomatic cysts are worrisome: we do not know what the natural history is of these cysts, but it is reasonable to assume that they are progressive and will ultimately perforate the tibial or talar cortex, making revision more complicated. No studies quantifying the rate and exact likelihood of bone defect progression have been performed. Some prosthesis designs are difficult to evaluate radiologically, particularly on the talar side, because it is not possible to see if there is any structural loss of bone support underneath. For example, in the Salto Talaris, STAR, and Hintegra systems, the talar component covers the body of the talus, which can hide what is going on underneath it, thereby masking bone necrosis, so CT scans are essential to evaluate for cystic defects or loosening.
Some cysts do not require revision of the tibial or talar component, but exchange of the polyethylene and bone grafting. This is particularly relevant if there is bone ingrowth into the prosthesis, and one is able to bone graft around the components with exchange of the polyethylene. One must be certain that there are no mechanical causes for polyethylene wear as a result of a malpositioned prosthesis. If one addresses only the problem with bone grafting and exchange of the polyethylene, this is going to fail again. It is useful to observe the pattern of wear on the polyethylene, and if ridging is present, one can assume that there is a malposition of the components or hindfoot deformity causing the abnormal polyethylene wear. In general, there is some minor abnormality with alignment causing the abnormal polyethylene insert wear. The cysts are often far deeper than is apparent on plain radiographs, and a large curette is used to remove the lining of the cyst trying to avoid perforation of the bone. We pack the cyst with a mixture of cancellous graft and an iliac crest aspirate concentrate. The cancellous graft is packed in firmly, particularly when the bone loss is at the margin of the tibial component. It is probably not enough to simply fill the cyst with bone graft unless, as stated, there is good bone ingrowth, no obvious abnormal wear of the polyethylene, and no hindfoot deformity.
Component Subsidence and Revision
One particular problem with all of the available prostheses and implants is the potential for subsidence of either the talar or the tibial component. Talar component subsidence is far more of a problem, in particular in the United States where the Agility (DepuySynthes, Westchester, USA) prosthesis was frequently used. This was particularly common in patients who were not suitable candidates for the procedure because of poor bone quality, an increased body mass index, or a high activity level. The major problem was the design of the prosthesis, with only 38% of the surface of the talus covered by the implant. The tibial component of the Agility system was more stable and with more bone ingrowth in most revisions performed; however, removal of a third of the fibula and medial malleolus and arthrodesis of the syndesmosis, which was part of the procedure, is not appealing either.
Pattern of Talar Subsidence
With talar subsidence, the talar body is crushed, and the component is pushed inferiorly. When subsidence occurs, regardless of the prosthesis, the previous supportive rim of cortical bone also becomes the source of impingement, limiting range of motion. Although the overhanging bone in the medial and lateral gutter can be debrided and decompressed, the structurally compromised base is vulnerable to further subsidence and eventually fracture of the remaining talus. There are multiple patterns of subsidence, but one that is most common is a posterior tilt to the component as it sinks into the body of the talus. This pattern of subsidence is particularly difficult to treat, because one will have difficulty positioning the revision talar component in a plantigrade position. If there is sufficient talar body to work with, then a revision can be performed without an arthrodesis of the subtalar joint. Other cases have complete implosion of the component through the body of the talus and rests on the posterior facet of the calcaneus. This pattern of failure (i.e., posterior subsidence) is most difficult to treat, since it is very difficult to recut the talus into a plantigrade position.
Depending on the pathology, the implants may be already loose, or they may still be well fixed. It is rare that we have ever found subsidence of the talar component with adequate bone ingrowth present, and by definition, if subsidence of either component occurs, it is loose. If the one component has subsided and the other is well fixed, this can present considerable technical difficulties—specifically, fractures and bone loss that occur with attempts to pry the component loose. Removal of well-fixed components must be done meticulously, working at the interface with small, thin osteotomes to preserve as much bone as possible. Any overlying exostosis, debris, or granulation and scarred synovial tissue must be removed to aid visualization. The polyethylene insert is generally easiest to remove once the talar component has been removed. With the additional available room under the tibial component, an osteotome can be inserted under the polyethylene insert to lever it out. If the joint still seems too tight for removal of the polyethylene insert, then a laminar spreader can be inserted under the medial or lateral column of the tibial component, gaining more room to lever it out. Where possible, we therefore first try to remove the polyethylene implant to give more working space. With respect to the Agility replacement system, removal of the older versions of the polyethylene insert is difficult because it is bottom loaded and will not slide out anteriorly until the edge rails are completely disengaged. If scarring is significant and the joint cannot be distracted, the polyethylene insert needs to be cut in segments with a reciprocating saw and removed piecemeal in large sections. This technique is the most efficient way of disrupting the edge-locking section of the older Agility polyethylene component.
An option for treating subsidence used to be with a custom-designed prosthesis, but at the present time, these are not approved by the US Food and Drug Administration, and no stemmed talar component is permitted. Our experience using cement to stabilize subsidence resulting in large defects of the tibial component has not been good. For smaller defects, this is a plausible alternative, particularly in the patient who has osteopenia, is frail, and requires more immediate mobility and ambulation. However, there is always the risk for cement failure with later bone resorption, setting the stage for an even more difficult second revision procedure. Another alternative is to remove any necrotic or avascular talus and then seat the revised implant directly on the posterior facet of the calcaneus, while simultaneously performing a subtalar arthrodesis of the middle and anterior portion of the talus. This, however, is more difficult because of the plane and slope of the posterior facet. Cement is commonly used in and around the calcaneus to support the talar component and is discussed later in more detail.
If a talonavicular or subtalar arthrodesis is to be performed simultaneously, then the incisions are planned accordingly, to allow an adequate skin bridge on the lateral foot. Typically, the anterior central incision is used, and then as wide a skin bridge as possible is planned to include exposure of the subtalar joint. A short incision over the posterior facet of the joint is sufficient immediately below the tip of the fibula to expose only the posterior facet of the subtalar joint. This will maintain as much vascularity of the talus as possible, and an extensile exposure on the joint with extension into the sinus tarsi should be avoided if possible. Debridement of the cartilage is performed with a curved osteotome, but the remaining talar and calcaneal subchondral bone must be preserved and is perforated with a 2-mm drill. Once the arthroplasty components are inserted, screws are inserted to stabilize the subtalar joint from the neck of the talus, just distal to the talar component, aiming inferiorly into the calcaneus. We use two 5.5-mm screws for fixation. We do not like the idea of inserting the screw from “bottom to top.” This method of placement is not precise, and repeated insertion of guide pins or drilling may compromise the talus further. It is far safer and more accurate to insert from the dorsal surface of the neck of the talus, making sure that the screw head is not impinged against the anterior aspect of the talar component.
Sequential Steps of Revision
Access to the joint should be through the previous longitudinal midline incision. If an alternative incision has been used that is not anatomically correct, care must be taken to decide on the correct course of the revision incision. Careful dissection is needed dorsally to minimize tissue trauma, as this incision is prone to wound healing complications. It can be quite difficult to separate the deep neurovascular bundle owing to fibrosis and scarring, and the patient must be warned about the likelihood of postoperative dorsal foot numbness if the deep or superficial peroneal nerve cannot be preserved. There are occasionally problems with the original incision, typically one that is made too far medially. This creates tremendous difficulty with visualization of the lateral gutter and the fibula, however the same incision should be used. If a very nonstandard incision had been previously used, the proximal part of the incision can be reused, and then curved distally to a more standard dorsocentral position on the foot. Very quickly, one recognizes the difficulty in identifying and protecting the neurovascular bundle, which is commonly wrapped up in scar tissue. The latter is always difficult, and patients should be warned to expect numbness over the dorsum of the foot. We generally use longer incisions than used for the primary replacement; always try to minimize wound complications by decreasing retraction.
Intraoperative prophylactic antibiotic are only given once cultures have been obtained and a tissue sample taken from the synovium and after deep specimens have been taken. We recommend a standard Gram stain and frozen section of the tissue specimens in any case where there is a suspicion of infection. We would wait for the results before proceeding with revision, and will not implant if the tissue specimens demonstrate >5 cells per high-power field. Regardless of the type of prosthesis removed and replaced, prophylactic K-wires or guide pins for a cannulated screw should be placed in the medial and lateral malleolus before implant removal to protect against intraoperative fracture. This was a particular problem with the Agility revisions where even with the K-wire in place, a fracture still could occur. Using these K-wires or guide pins have the added benefit of treating a fracture with a cannulated screw that can be passed over the K-wire.
Perhaps an important and overlooked part of the procedure is to perform an extensive medial and lateral gutter release and debridement. This can be attempted before removal of the components, but it is far easier once both components have been removed. The gutters are often filled with debris in particular heterotopic bone, which not only causes pain but also limits range of motion of the ankle ( ). This step cannot be overemphasized, since bone overgrowth in the gutters may completely limit motion initially. We recommend using a rongeur alternating with a reciprocating saw to clean the gutters, such that there is a clear space in each gutter of at least 4 mm. Sometimes it means shaving off 1–2 mm of the margins of the malleoli if there is excessive heterotopic bone.
Removal of the original prosthesis commences with removal of polyethylene liner with a thin osteotome. Depending on the type of implant, this can be simple or complicated, particularly if an original Agility implant was used that has flanges in the polyethylene medially and laterally, which need to be cut with a reciprocating saw. We generally start with the loose component. If it is the tibial component, this must be removed with care to prevent fractures, which are very common. The other problem with tibial component removal is that even if one removes minimal bone, there is a space of varying size between the dorsal surface of the component and the distal tibia, a problem with the STAR, Salto, or the Agility prosthesis. One has to anticipate this, since either cancellous compression bone grafting or cement can be used above the tibial component. We use a fine chisel and pass it repeatedly until the prosthesis seems to be loose. If the tibial component is loose from the beginning, then this step is easy regardless of the type of prosthesis, but considerably more difficult if bone ingrowth is present when some bone loss is inevitable. Gentle insertion of a fine chisel is done on either side of the tibial keel if present, with care taken not to lever on bone and to carefully loosen up all bony ingrowth. Once the component is loose, one does not want to simply pull it out, since this will compress more bone anteriorly. Plantarflex the foot and then plantarflex the tibial component so that the fin of the Agility and Salto or the barrels of the STAR prosthesis do not cause any more bone loss. The majority of talar implants are easier to remove, since most of them are loose. Once the implants are removed, further debridement of the ankle is carried out focusing on any cysts. Regardless of size, the margins of these cysts must be curettaged. Once both components are removed, the most important step it to continue the soft tissue release posteriorly. This is difficult, and one has to peel off the very thick posterior capsule, which will significantly limit range of motion. We start by inserting a laminar spreader into the joint, and once on tension, with a periosteal elevator and working behind the posterolateral tibia, we peel it off the tibia and then grab the thick, hard fibrous tissue with a rongeur. Importantly, as one moves more medially, one has to be more careful with the flexor hallucis longus and the neurovascular bundle, which can be grabbed with the rongeur.
Bone loss is addressed in several ways depending on the extent and location. Impaction bone grafting with demineralized bone matrix mixed with allograft bone chips and autograft can be used in both the tibia and talus ( Fig. 19.2 and ). In smaller defects, we do use cement, for example around the tibial component, but only if the tibia is stable, not loose, and the cement is used only to fill a void. Obviously, if the prosthesis is stable, and one has a press fit and cement is used, then early weight bearing can be followed. This is not the case, however, if cancellous bone grafting is required to support the prosthesis and will limit the ability to bear weight. In these cases, whether in the distal tibia or the undersurface of the talus, cement can be used, but very cautiously and only for small areas around the tibial component that is stable, and where one wants to fill a slight defect between the component and the distal tibia. How does one address the bone loss with the anterior erosion and the extension deformity of the tibia? There are a few options available to us. Either the bone can be recut so that it is parallel with the floor, or one places the tibial component in the correct position, leaving a large gap anteriorly, and then fills the defect with bone graft, or uses cement. Impaction bone grafting can be considered, but if cancellous grafts are used, one must delay weight bearing for 3 months in large, structurally compromised areas such as the anterior tibia. The latter is not our preference (i.e., to use cancellous bone graft where there is substantial bone loss on the anterior cortical rim), since the graft may resorb slightly and the tibial component falls back into a dorsiflexed position. Structural bone graft can be considered, but one must find the ideal bone that fits the defect, and ensure bone healing.