Mark Ritter, Florian Nickisch, and Christopher W. DiGiovanni
DEFINITION
The number of total ankle arthroplasties (TAA) both designed and implanted continues to grow rapidly worldwide.
The success and survivorship of any joint replacement are difficult to determine before a minimum 5-year follow-up. Because the early stage of most new arthroplasties can be marked by a steep learning curve while the later stage is often plagued by some level of polyethylene-induced osteolytic failure, foot and ankle specialists can expect to face increasing numbers of patients requiring revision or salvage surgery— even with the newest-generation TAA designs.
Failure of a total ankle replacement can broadly be defined as septic or aseptic and usually results from either clinical (recalcitrant pain, instability, or malalignment) or radiographic (progressive loosening, subsidence, or osteolysis) deterioration.
Implant failure due to septic or aseptic failure typically necessitates removal and leaves the surgeon with extensive areas of bone loss that must be addressed. Other potential problems include wound breakdown, infection, limb length discrepancies, scar formation, instability, malalignment, and, of course, choosing between complex and limited reconstructive options.
Due to the anatomy and limited bone stock inherent to the ankle and particular to the talus, revision ankle replacement is frequently not possible in these cases, and arthrodesis of the ankle or tibiotalocalcaneal region remains the only viable salvage option.
Fusion has traditionally been reported and performed through either an anterior or lateral approach. These approaches and the hardware used for them, however, are often limited by wound complications from the thin or previously operated soft tissue envelope, as well as difficulties in assessing rotatory, angular, and longitudinal alignment.
The posterior approach usually provides the healthiest and deepest soft tissue bed for any postimplant failure reconstruction (through a single incision), permits ready access to the area of greatest potential bone graft harvest (posterior superior iliac spine [PSIS]), uses the fibula to aid in healing and in determining proper alignment, allows use of large fixed-angle devices applied on the tension side to enable safe and early postoperative weight bearing, and facilitates rapid intraoperative assessment of radiographic and clinical position.
Because the fibula rarely provides enough bone to fill any remaining defect after TAA failure, it is often more useful in its native position as part of the reconstruction as opposed to being partly or completely sacrificed as part of a lateral or combined anterior operative approach.
ANATOMY
The posterior aspect of the lower leg, ankle, and hindfoot is covered by several layers of well-vascularized soft tissues.
The superficial posterior compartment contains the gastrocnemius and soleus complex, separated from the deep posterior compartment by a dense investing fascia.
The sural nerve and the lesser saphenous vein should be carefully avoided as part of the superficial midline dissection in this approach.
Once the superficial posterior compartment fascia is opened, the gastroc–soleal complex can be mobilized as a unit medially or laterally, or the Achilles tendon can be Z-lengthened to enable immediate access to the deep posterior compartment septum.
The flexor hallucis longus (FHL) muscle is considered the “lighthouse” to the back of the ankle and hindfoot. It is readily identified by its uniquely low-lying muscle belly within the deep posterior compartment and provides the landmark for the posterior tibial neurovascular bundle, which courses immediately medial to it through the lower aspect of the leg and into the foot (FIG 1).
Once identified, the FHL can be easily swept medially to protect the bundle during retraction and provide maximal exposure of the posterior aspect of the distal tibia, ankle joint, and subtalar joint (FIG 2).
Once the reconstruction has been performed and fixation is indwelling, the deep soft tissue bed present in this region enables the hardware and any bone graft augmentation to thereafter be safely covered by repositioning of the FHL and Achilles complex before untensioned subcutaneous and skin closure.
PATHOGENESIS
TAA failure can result from many causes, but the predominant mechanism, based on hip and knee replacement data, will likely become aseptic failure. This can occur from polyethylene wear over time, ballooning osteolysis, subsidence of the implants into surrounding host bone, heterotopic ossification, or frank implant fracture or dislocation. Septic failure can also occur at any point after joint implantation, and if this happens in the nonacute (beyond the first 6 to 12 weeks after surgery) setting, implant removal and some form of exchange or reconstruction as a single or staged procedure is usually required for successful salvage.
FIG 1 • Anatomy of the posterior ankle. Star, posterior tibial nerve; arrow, posterior tibialis artery. TTJ, tibiotalar joint; TCJ, tibiocalcaneal joint; FHL, flexor hallucis longus.
FIG 2 • The flexor hallucis longus is carefully retracted medially, allowing excellent visualization of the ankle (star) and subtalar joint (arrow).
NATURAL HISTORY
Although historically TAA has been in existence since the earliest hip and knee designs of the late 1960s and early 1970s, it has never enjoyed the same clinical success.
This reason is likely multifactorial but no doubt stems in part from the many unique design aspects of the ankle joint that set it apart from the many other large joints that undergo very successful replacement today. These issues include the ankle's functional dependence upon the alignment and quality of many surrounding nearby joints, its inability to be dislocated during implant insertion, its thin and unforgiving soft tissue envelope, its limited native motion, its size-to-weight bearing ratio, and our present inability to completely resurface it such that the same stress is borne by the same surface area.
Much of our understanding about ankle joint replacement stems from the adult reconstructive literature regarding hip and knee replacement. Most, if not all, earlier-generation ankle replacement designs have been considered failures by today's standards, and a great number eventuated into fusion or other forms of revision surgery.
The most recent (third-generation) designs seem to have begun capitalizing on many newer technologies and design concepts ushered in by the successes in the hip, knee, and shoulder, and while early and midterm survivorship data (5 to 10 years) appear to be promising for these modular implants, time of implantation will be the ultimate judge.
If history is any indication, it is quite possible that the next decade or two of foot and ankle specialists will have to become very familiar with reconstructive salvage of failed TAA in the form of revision replacement or fusion.
Since the survivorship of the average total hip or knee arthroplasty approaches 90% to 95% at 15 to 20 years and TAA has never approximated this success, the rapidly increasing number of ankle replacement designs being accepted by the FDA and in countries abroad, as well as the seemingly exponential rate of TAA implantation today as an alternative to primary fusion, would suggest that revision ankle replacement surgery will need to be a part of every foot and ankle specialist's armamentarium in upcoming years.
PATIENT HISTORY AND PHYSICAL FINDINGS
A thorough history and physical examination, as well as an appropriate set of weight-bearing ankle radiographs, are paramount to identifying the patient with a failed TAA.
Unremitting or new-onset pain is often the chief complaint of a poorly functioning or infected ankle replacement.
Patients should also be assessed for associated ankle swelling and warmth, which if recent are reasonable indications for a more in-depth evaluation to assess the integrity of the TAA.
Time from initial implantation as well as any history of prior surgery or implantation in this region should be noted.
Patients who are diabetic or neuropathic or who have any systemic illness that could predispose them to infection, immunologic compromise, or undetected abnormal wear (Charcot) should also be more carefully assessed.
Any history of fever, chills, sweats, or recent dental surgery without antibiotic prophylaxis should be noted, as should complaints of ankle or hindfoot instability.
The examiner should look for obvious ankle or hindfoot deformity, either new or old. Particular attention should be paid to the varus or cavus foot malalignment, which has the highest association with implant failure.
Restricted range of motion or the presence of pain, crepitance, or grinding on examination should be noted.
The examiner should look for any surrounding fluctuance, erythema, or draining sinus around the ankle.
IMAGING AND OTHER DIAGNOSTIC STUDIES
When implant failure is suspected, the initial screening blood work to rule out sepsis should include a complete blood count with differentiation, erythrocyte sedimentation rate, and a C-reactive protein level.
Weight-bearing, standing plain films of the affected ankle (anteroposterior, lateral, and oblique views) should be obtained. If necessary, particularly when a form of aseptic failure from mechanical malalignment of the foot is suspected, a routine set of plain films of the foot should also be obtained.
Radiologic signs of loosening include radiolucent lines around the components, as well as malposition and subsidence of any component (FIG 3). These are most valuable when they are indentified as acute changes from previous films or shown to be slowly progressive over time.
Ballooning osteolysis behind an implant is a poor prognosticator for impending implant subsidence and failure. In such patients, polyethylene wear (a narrowed joint space) is often identifiable on plain radiographs.
FIG 3 • AP (A) and lateral (B) radiographs of a failed total ankle arthroplasty. Note lysis around tibial implant and subsidence of the talar component.
An implant should be considered to be infected until proven otherwise when the history, physical examination, and blood work suggest such.
A bone scan can also be useful as an adjunct for diagnosing septic or aseptic loosening.
In these patients or in those with an equivocal examination, an office based or radiologically guided aspiration is indicated for routine Gram stain and culture.
Percutaneous biopsy can also be performed, although intraoperative cultures are considered most sensitive and specific for the etiology of the implant failure. These can be assessed pathologically for polymorphonucleocytes per high-power field, as well as for the presence of bacteria and poly debris.
In cases of suspected or documented infection, consultation with the infectious disease team is suggested to determine the appropriate microbiologic and chemotherapeutic aspects of the subsequent management.
Failure can be on one side or both sides of the joint, and hence can involve either the polyethylene spacer (if present), one implant, or all implants in the TAA design. This determination affects whether the TAA can be salvaged.
CT scanning can be useful to assess the degree of osteolysis and bone destruction, which is often otherwise difficult to discern behind the implants. Such information can also be very useful in planning a revision procedure, as can the integrity of any nearby joints (subtalar, Chopart).
DIFFERENTIAL DIAGNOSIS
Pain of unknown etiology (implants still well fixed): complex regional pain syndrome, stiffness, fibromyalgia, neuroma, tendon incarceration, neurovascular injury or compromise, heterotopic ossification, occult fracture, syndesmotic nonunion, arthritis or impingement of nearby joints
Septic failure (infection)
Aseptic failure (impingement, osteolysis, implant or polyethylene fracture, subsidence, circumferential loosening, malposition, malalignment, dislocation, instability, periprosthetic fracture, syndesmotic nonunion when applicable)
NONOPERATIVE MANAGEMENT
Nonoperative management is generally not indicated for a septic TAA failure. When these are very acute, these can occasionally respond to serial aspiration and antibiotic therapy, but even in these cases surgical intervention (arthroscopy or single-stage exchange) has proven most effective.
In cases of aseptic failure, treatment depends on the cause.
Gross instability, uncontrollable pain, catastrophic implant failure (fracture), periprosthetic fracture, and aggressive (ballooning) osteolysis are generally best treated surgically.
Other causes of aseptic failure can be considered for conservative management, which includes some form of bracing, mechanical offloading with assistive devices, pharmacologic pain control (or osteolytic inhibition), and a RICE protocol.
Sometimes simple tolerance is the most appropriate course when the risks of revision surgery might outweigh any of the potential benefits.
The risks of such complex surgery, as well as its limitations, must be discussed in detail with any patient in this situation, and this discussion should always include the possibility of below-knee amputation.
SURGICAL MANAGEMENT
A blade plate or fixed-angle device applied from posteriorly in the prone-positioned patient addresses all the problems associated with arthrodesis of a failed TAA, and it is our preference.
This procedure can be performed as a singleor two-stage procedure.
The technique is versatile because it can be used for both tibiotalar or tibiotalocalcaneal arthrodesis, with the only difference being the size of the fixation device.
Many different implant sizes can be used for this technique, varying from smallto large-fragment fixation, locking or nonlocking constructs, and fixed-angle or straight plates.
The prone position allows access to a deep and usually healthy, unscarred soft tissue bed capable of accessing or removing the indwelling TAA as well as covering appropriate hardware and bone graft without tension.
The prone position also allows for the easiest clinical determination of hindfoot position before fusion, and it affords access to the posterior iliac crest for maximal amounts of bone graft procurement. The opposite leg can also be prepared out, if need be, for comparison.
AP and lateral radiographic images are also easily obtainable, requiring minimal to no manipulation from the surgeon when the operative leg is elevated on two or three folded blankets.
Lastly, the fixation in this approach is placed on the tension side. This acts to compress the fusion mass under the load of weight bearing, facilitating a more rapid return to ambulation.
Preoperative Planning
All radiographs and laboratory parameters, as well as the patient's skin envelope, are reviewed before surgery.
If the patient has an infected TAA, this procedure should be performed in staged fashion and only after the decision has been made not to reimplant an ankle prosthesis at the second stage. A carefully contoured, anatomic polymethylmethacrylate antibiotic spacer impregnated with tobramycin and vancomycin can easily be inserted and removed through the same approach to maintain alignment and soft tissue tension between stages.
Symptom production from the subtalar joint must be carefully assessed preoperatively (but can also be assessed visually intraoperatively) to determine any potential need for adding subtalar fusion to an isolated tibiotalar arthrodesis. Surrounding bone quality and stock should also be a major factor in making this determination, particularly on the talar side.
Preoperatively, two-stage office-based or fluoroscopically guided diagnostic differential injections of the talocalcaneal and subtalar joints with local analgesic are well suited for this purpose.
Precontouring the blade plate (and determining its size) using an ankle sawbones model and a preoperative template saves significant tourniquet time (FIG 4).
Positioning
The patient is positioned prone on an image table with gelpads, using a few folded blankets as a “workbench” under the affected leg to elevate it sufficiently above the contralateral extremity to permit unimpeded imaging of the operated extremity in the cross-table lateral projection (FIG 5).
Gelpads should not be used in the area intended for fluoroscopy, since the material is radiopaque. The image machine should be checked for clearly visible AP and lateral views of the patient's ankle and hindfoot before preparation and draping.
FIG 4 • Precontouring the blade plate with sawbones as template.
FIG 5 • Posterior midline approach.
A tourniquet should be placed about the thigh, and the ipsilateral posterior iliac crest should be squared off with preliminary drapes in anticipation of bone graft procurement.
Approach
The entire ipsilateral leg and PSIS are then prepared and draped in the usual sterile fashion.
The ankle and hindfoot should be operated on to establish the size of the defect and amount and configuration of bone required before autologous harvest from the PSIS.
Under tourniquet control, a midline longitudinal incision 12 to 16 cm long is initially made directly posterior to the ankle and hindfoot (Fig 5). Imaging can be useful at times to establish this position ideally, although we prefer simply centering this over the ankle in the midline. If it is determined intraoperatively that more exposure is required distally to access the subtalar joint, this incision can be easily extended by curving it slightly posteromedially as it courses over the heel.
FIG 6 • A. Longitudinal split through Achilles tendon. B. Z-lengthening of Achilles tendon if additional exposure is required.
No skin retraction is used, and retractors are used only once the deeper tissues are encountered.
The paratenon of the Achilles and the superficial fascia are first carefully opened with the intention of later closure and separation from overlying skin and subcutaneous tissues in the rare event of wound breakdown.
A Z-plasty of the Achilles is performed longitudinally to allow access to the deep posterior compartment. Incising the fascia over the superficial posterior compartment can ease tension and improve retractability of the gastrocnemius and soleus during exposure. Care should be taken to maintain fullthickness flaps (“canyon walls”) (FIG 6).
The deep posterior compartment fascia is then incised, exposing the FHL and the remaining deep extrinsic musculature.
The neurovascular bundle is identified but not dissected, and then carefully retracted medially by retracting the lateral side of the FHL.
This permits unimpaired full access of the posterior tibia, the capsules overlying the ankle and subtalar joints, and the distal fibula immediately beneath a portion of the inferior peroneal musculature (FIG 7).
FIG 7 • Deep compartment open, flexor hallucis longus retracted medially. The failed ankle implant has been removed and the defect débrided.
The inferiormost edge of the peroneals can be removed subperiosteally from the distal fibula as needed to gain access to the ankle joint as well as to expose greater amounts of direct bleeding bone for surface-area healing of the fusion mass. Although not necessary, this permits incorporation of the fibula in the fusion mass when desired.
TECHNIQUES
TAA REMOVAL
Although most implants are placed through an anterior approach, it is generally not difficult to remove these current designs from a posterior approach.
Use of a femoral distractor, or, alternatively, an external fixator with medial pins in the tibia and the calcaneus will facilitate distraction of the joint for easy implant removal in the event of soft tissue contracture.
Once the ankle implants have been removed, any fibrous membrane or other debris within the joint can be excised and the remnant viable bone stock (defect void) and quality can be assessed to plan alignment, bone graft requirements, and implant size for the reconstruction.
Only healthy, bleeding bone should be left behind amidst a viable soft tissue envelope.
At this point, the subtalar joint should also be inspected—and fused, if deemed necessary by virtue of its integrity or the remaining available bone stock for fixation. In the case of TAA salvage, this technique is usually recommended.
Despite any preconceived opinions about the presence or absence of infection, under all circumstances it is advisable to obtain multiple deep tissue samples for pathology and culture. These should be taken ideally before antibiotic prophylaxis is given, and we recommend taking three samples for pathology and three for microbiology, all with separate instruments, from separate sites, labeled with separate identifiers, and placed in separate sterile containers. Under no circumstances should the skin be touched when performing this task, for fear of inadvertent contamination.
INFECTED TAA REMOVAL
If the joint is infected, or presumed infected, a radical débridement is performed at this time, taking similar cultures and pathology specimens from separate “highyield” areas.
In these cases, the ankle joint is then prepared for a second-stage procedure by thorough saline irrigation and interposition of a PMMA antibiotic-laden spacer fashioned to fit the bony defect and maintain alignment, length, and stability (TECH FIG 1).
Since cultures are often not yet indicative of an infecting organism, both vancomycin and tobramycin should be included in the spacer for both gram-negative and grampositive coverage.
TECH FIG 1 • External fixator medially, antibiotic spacer placed in void after removal of infected implant.
The soft tissues are copiously reirrigated, and hemostasis is then maintained, and they are completely closed around the spacer.
The infected patient is treated with adjunctive antibiotics for 6 to 8 weeks, and often an external fixator is added for additional support (in lieu of a splint or cast).
Before second-stage surgery, the blood work should have returned to normal and the ankle should be reaspirated to verify eradication of infection. All incisions should also have healed uneventfully and be deemed capable of tolerating further surgery.
At the time of staged reconstruction, the same posterior incision is used, and during exposure a stat Gram stain and frozen sections are taken to quantify white blood cells per high-power field. If these values are within normal limits, the procedure then continues as outlined for primary fusion in the aseptic patient as indicated below.
DECORTICATION AND BONE GRAFTING
After final takedown and decortication of the ankle (and possibly the subtalar joint) the surgeon can determine how much bone graft to harvest. Occasionally this includes preference of size or shape (eg, tricortical, trapezoidal, cancellous only).
If the subtalar joint is to be taken down, it is prepared in a similar manner. Bone (laminar) spreaders are very useful for this purpose in both joints.
Bone graft blocks are taken from the posterior iliac crest with a sagittal saw and osteotomes, and thereafter are fashioned to fit and bridge the resected ankle gap. Generally, tricortical grafts are most amenable to this construct and can be easily contoured into appropriate position to maintain alignment. Once these are taken, the cancellous graft between the remaining inner and outer table of the pelvis can also be harvested for packing the remaining joint space. In all these cases the cancellous bone graft should be mixed with tobramycin and vancomycin powder before being packed into all remaining articular interstices after hardware implantation.
BLADE PLATE APPLICATION
The foot should be placed in neutral alignment and preliminarily held in reduction with one or two large nonthreaded Steinmann pins. Typically, this alignment includes 0 degrees of ankle flexion with 5 degrees of hindfoot valgus and external rotation appropriate to the opposite side. This step is performed identically for both ankle and tibiotalocalcaneal fusion.
Hence, once proper length and position are established clinically and radiographically, one or two eighth-inch Steinmann pins are placed through the calcaneus from directly inferiorly, and run into the midtibia to maintain this alignment. Foot and ankle position is then verified in both the lateral and AP planes with imaging. The precontoured 4.5-mm 90-degree fixed-angle blade plate (recommended) is then laid next to this to assess proper contouring and positioning via imaging. Small alterations in this device are best made at this time before it is actually implanted.
Predrilling a trough for the blade of the plate is usually unnecessary when doing a tibiotalocalcaneal fusion because of the soft cancellous bone found within the calcaneus. In this case, the starting position and angle of insertion are far more important. In the less common circumstance of having enough talar bone to simply fuse only the ankle primarily, a precut trough is advisable before the blade plate is introduced into the denser talus bone. In this latter circumstance, often a smaller blade plate (3.5 mm) is more amenable to this fusion construct.
In both cases, attention must also be paid to the following:
Proper length of the blade to avoid cutout upon implant seating
Proper angle of the blade to ensure adequate positioning once seated
Number of screw holes traversing the tibia such that adequate fixation is maintained above the fusion mass
Proper rotation of the blade such that it sits centrally located along the posterior tibial metadiaphysis once fully seated (TECH FIG 2)
Serial imaging during blade plate insertion can be very helpful in making these determinations before completely bottoming out the implant.
Once the blade plate is fully seated, the position of the limb clinically and radiographically should be reassessed. After this, the plate can be locked in position by placing a single proximal and distal compressive screw, followed by Steinmann pin removal.
TECH FIG 2 • Blade plate placed posteriorly with foot in neutral position, calcaneus in 5 degrees of valgus, held by Steinmann pin.
The remaining compressive screw fixation construct for the plate can then performed in routine fashion, including the use of the articulating tensioning device where applicable.
Often, several screws can be used to cross several joints not only to enact a neutralization plate construct but also to permit some articular compression across the fusion mass as well.
TECH FIG 3 • Additional bone graft is packed around the blade, the ankle, and the subtalar joint.
TECH FIG 4 • Lateral fluoroscopic image verifying blade and screw placement and alignment.
After fixation, any residual graft can be packed in and around the plate and joints (TECH FIG 3).
Final films should be taken and saved, and a repeat clinical examination should be performed to ensure satisfactory alignment before final closure (TECH FIG 4).
The deep posterior compartment is then swung back into place to easily cover the plate (TECH FIG 5) and the Achilles can be reapproximated in neutral position. A Hemovac drain should be placed at this time before the overlying fascia, subcutaneous tissue, and skin are closed.
TECH FIG 5 • The flexor hallucis longus is replaced in its anatomic position covering most of the implant.
POSTOPERATIVE CARE
An ankle blockade with Marcaine and lidocaine eases pain in the immediate postoperative setting, but it should be placed well above the operative site to avoid wound tension.
Steri-Strips should be applied across the incision site to distribute stress optimally at this level, and away from the incision itself. To this end, these should be uncut, and benzoin should be avoided to minimize blister formation.
A meticulously padded Jones dressing, posterior splinting, and taped suction drainage help to avoid edema, hematoma, and pressure sore formation.
While the patient recovers in bed or rests in a supine position, no pressure should be permitted beneath the lower leg, ankle, and foot. Hence, this area should be “suspended in midair” by placing pillows or blankets underneath the proximal calf, knee, and thigh to avoid pressure on the incision.
The patient should remain strictly non–weight-bearing for the first 2 weeks postoperatively.
After the posterior skin wound has healed and sutures have been removed, cast immobilization with partial weight bearing is allowed until week 6. Placing the plate posteriorly across the ankle creates a tension band phenomenon during the gait cycle, helping to compress the fusion site with weight bearing. Physical therapy can also begin during this time interval. At 6 weeks, consideration can be given to transitioning the patient to boot immobilization, depending on the clinical and radiographic progress, and slow progression to full weight bearing can begin.
FIG 8 • Weight-bearing lateral radiograph 3 months postoperatively.
All forms of cast or boot immobilization are discontinued after radiographic evidence of healing, usually at about 12 weeks postoperatively (FIG 8). The patient can be advanced into a sneaker with a SACH heel or rocker sole.
OUTCOMES
In our experience, this operation has been very effective for salvaging difficult revision of failed TAA with reasonable patient satisfaction.
We have not done enough of these procedures, however, to enable us to reasonably discuss outcome. We do believe, though, that this operation will become more pertinent over time, and that it is a very easy, safe, and versatile technique to address this difficult problem.
COMPLICATIONS
In our limited experience with this approach, we have encountered no complications, although we believe the potential complication list would certainly be similar to any major revisional fusion operation dealing with intercalary defects.
REFERENCES
1. Bruggeman N, Kitaoka H. Arthrodesis after failed total ankle arthroplasty. Tech Foot Ankle Surg 2002;1:60–68.
2. Hammit MD, Hobgood ER, Tarquinio TA. Midline posterior approach to the ankle and hindfoot. Foot Ankle Int 2006;27:711–715.
3. Hansen T, Cracchiolo A. The use of a 95 degree blasé plate and posterior approach to tibiotalocalcaneal arthrodesis. Foot Ankle Int 2002;23:704–710.
4. Peyvich M, Saltzman C. Total ankle arthroplasty: a unique design. J Bone Joint Surg Am 1998;80A:1410–1420.
5. Quill G. Tibiotalocalcaneal arthrodesis with medullary rod fixation. Tech Foot Ankle Surg 2003;2:135–143.
6. Wapner K. Salvage of failed and infected total ankle replacements with fusion. AAOS Instructional Course Lectures 2002;51:153–157.