Ankle arthritis is a prevalent condition with different treatment options available for end-stage disease, of which arthroplasty is one1,2. Total ankle arthroplasty, initially introduced in the 1970s, had frequent complications that led to the development of newer, second-generation prostheses3,4. However, even with these updated designs, several studies continued to show high rates of failure5-7. Several mechanical factors can contribute to the failure of the prosthesis, including component malpositioning and talar component subsidence. When the latter occurs, it may be associated with subtalar degeneration and arthritis. This may leave the surgeon with few options because the revision components for ankle arthroplasty are limited. As a result, the most common salvage procedure for ankle replacements that have failed because of subsidence or subtalar joint changes is tibiotalocalcaneal arthrodesis with bone-grafting8.
Unfortunately, ankle arthritis can present with advanced hindfoot arthritis and/or deformity. This condition is seen most often in posttraumatic arthritis and stage-IV flatfoot deformities9. Limited options are available for this scenario, and historically pantalar arthrodesis has been the main salvage procedure. One study has suggested that a below-the-knee amputation would be more functional than pantalar arthrodesis10. More recently, total ankle arthroplasty in combination with a subtalar or triple arthrodesis has been suggested as a possible treatment option, although we know of no published results on this technique11. In order to perform this salvage procedure, custom long-stemmed talar components, which are placed across the subtalar joint into the calcaneus, have been created. The advantage of this prosthesis is that it can be used in the setting of talar subsidence or for patients who have ankle arthrosis with associated hindfoot pathology. This allows for preservation of ankle motion with combined arthrodesis of the hindfoot and midfoot joints.
The purpose of this study was to retrospectively review the clinical and radiographic outcomes of patients who had advanced degenerative changes of the ankle and hindfoot or failed total ankle replacements with subtalar joint arthritis who underwent total ankle arthroplasty with a custom, long-stemmed talar component. Data were collected prospectively during the study.
Between November 1, 2004, and February 28, 2006, thirty-three complex and/or revision total ankle arthroplasties were performed with use of a custom Agility long-stemmed talar prosthesis (DePuy Orthopaedics, a Johnson & Johnson Company, Warsaw, Indiana) in thirty-two patients. All surgeries were performed by the senior authors (R.S. and M.M.). Inclusion criteria for the study were (1) failure of a total ankle replacement because of severe talar subsidence either into the subtalar joint or with progression of arthritis at the subtalar joint and (2) severe arthritis and deformity of the ankle and hindfoot joints due to primary osteoarthritis, posttraumatic arthritis, or stage-IV flatfoot deformity associated with a rupture of the posterior tibial tendon. After clinical examination and standard radiographs, and before a final treatment option was offered, all patients had a computed tomography (CT) scan with use of an ankle and hindfoot protocol. This allowed the surgeon to assess the extent of the arthritis of both the ankle and subtalar joints. Also, the CT scan was required by the manufacturer to create a custom stem to match the individual patient anatomy. Patients who met the indications for participation in the study were offered either a tibiotalocalcaneal or pantalar arthrodesis, or a total ankle arthroplasty with use of a long-stemmed talar component combined with an extended hindfoot as needed.
The average age (and standard deviation) of the patients was 59.1 ± 10.6 years (range, forty-one to seventy-seven years). Twelve men (twelve ankles; 36%) and twenty women (twenty-one ankles; 64%) participated in the study. Nineteen right ankles and fourteen left ankles were involved. Six ankles (18%) were in patients with stage-IV flatfoot deformity. Thirteen ankles (39%) had a failed primary total ankle replacement with severe talar component loosening and subsidence associated with subtalar joint involvement. Fourteen ankles as well as the hindfoot joints (42%) had extensive changes from primary, inflammatory, or posttraumatic arthritis.
Thirteen ankles had a revision ankle arthroplasty with a stemmed talar implant with subtalar arthrodesis (Figs. 1-A and 1-B). Twenty ankles were managed with primary ankle arthroplasty with a stemmed talar implant with distal tibiofibular arthrodesis. Of the twenty ankles, ten had previous or staged hindfoot arthrodesis procedures (seven triple arthrodeses and three subtalar arthrodeses). Ten ankles had hindfoot arthrodesis as a simultaneous procedure with ankle arthroplasty (nine subtalar arthrodeses and one triple arthrodesis). Individual patient characteristics are detailed in Table I.
Patients who elected to undergo this treatment were asked to complete the Short-Form-36 (SF-36) questionnaire to assess functional outcomes12. The SF-36 is composed of a thirty-six-question evaluation form that has a Physical Component Summary (PCS) composed of twenty-one questions and a Mental Component Summary (MCS) composed of fifteen questions. The American Orthopaedic Foot & Ankle (AOFAS) hindfoot scores and Maryland Foot Scores (MFS) were also completed preoperatively13,14. Additionally, goniometric and dynamic radiographic measurements of ankle range of motion were performed. For patients who had undergone a previous ankle arthroplasty, preoperative talar component subsidence and talar body osteolysis were measured radiographically and recorded.
All patients signed a consent form, indicating they were aware that a custom prosthesis would be prescribed and made specifically for them, on the basis of the radiographic parameters of their individual talar and hindfoot deformity. In order to accomplish this, preoperative anteroposterior, mortise, and lateral radiographs with use of a magnification marker as well as a computed tomography scan were made and sent to the manufacturer. Blueprints were then returned to the surgeon for approval. On the basis of a review of the template and radiographs, the stem length, its width, the implantation angle in two planes, and the sagittal stem position relative to the talar body were determined. In addition, the height, width, and offset of the talar body were also verified relative to the size of either (1) the previously implanted tibial component or (2) the overall size of the ankle bones, and the associated bone loss. Any changes were noted, and adjustments were made until the surgeon was satisfied that the prosthetic design was optimized for the individual patient. At that point, final approval was given and the prosthesis was manufactured. Average time from final approval to manufacture was three months.
The operative treatment depended on the joints involved and the level of deformity. For the patients requiring a triple arthrodesis and total ankle replacement, the triple arthrodesis was performed first with use of standard methods, followed by the ankle arthroplasty at three months, so that the subtalar fusion was healed and the compression screws could be removed prior to talar component stem placement. If the operation was limited to a subtalar arthrodesis and an ankle arthroplasty, these two procedures were performed as one operation. In this case, the subtalar joint was prepared through a separate incision made laterally over the sinus tarsi, was denuded of cartilage, repositioned, grafted with cancellous allograft bone chips, and pinned. After insertion of the long-stemmed component, a large cannulated lag screw was placed across the subtalar joint to augment the fixation of the arthrodesis.
The technique used for insertion of the custom talar prosthesis was developed by us. Standard anterior midline incisions were used, and either preparation for a total ankle arthroplasty or debridement and removal of the failed talar and/or tibial components was performed. After talar surface preparation, a custom cannulated talar dome trial implant without a stem, which was manufactured to match the custom prosthesis, was placed onto the cut surface of the talus, under either a tibial trial component or the existing stable tibial component. Position was verified both visually and under lateral fluoroscopic control. Correct rotation of the talar body was addressed at this point. With use of a wire driver, a long shaft, short-tipped, 3.2-mm drill-bit was drilled antegrade through the trial component’s cannulation (which corresponded to the position of the stem) into the talus and calcaneus, and exiting the skin posteriorly, just above the heel pad. If the position of this drill-bit was unsatisfactory, either repositioning of the talar component on the talus or performing a calcaneal osteotomy, or both, was required. Once the surgeon was satisfied with wire placement, the wire driver was disengaged and reattached to the tip of the protruding drill-bit distally. Drilling was then continued in an antegrade manner until the proximal portion of the drill-bit was disengaged from the trial talar component, which was then removed. The drill-bit was then advanced in a retrograde manner until it protruded slightly from the cut surface of the talus. Thereafter, rigid cannulated reamers were used to prepare the talus and calcaneus to accept the stem of the prosthesis. After the custom long-stemmed trial component was satisfactorily placed and additional bone preparation was performed as needed, the actual prosthesis was implanted. Often, because this was a nonmodular construct, a rotational maneuver was required to place the two-plane angled stem into the hindfoot. Once completed, if subtalar arthrodesis was required, the additional cannulated large-fragment lag screw was placed across the joint, in either an antegrade or retrograde manner.
Patients who underwent a primary total ankle arthroplasty (tibia and talus) required a distal tibiofibular arthrodesis, and as a result were instructed not to bear weight for a minimum of eight weeks postoperatively. During this time, ankle range of motion in a boot was permitted. If only a custom long-stemmed talar revision was performed, weight-bearing was begun four weeks after the index procedure. If a subtalar arthrodesis was part of the procedure, however, the patient was to remain non-weight-bearing for twelve weeks. The final decision to begin weight-bearing in any of these scenarios was based on clinical (lack of pain) and radiographic evidence of union (bridging trabeculation).
Postoperatively, patients underwent radiographic evaluation of the ankle at each visit. For patients who had a previous ankle arthroplasty, talar subsidence and talar body osteolysis had been initially measured and these data were compared with serial radiographs made after surgery. For patients who underwent a primary arthroplasty with use of a custom stem, talar subsidence and osteolysis were analyzed with use of sequential postoperative radiographs. Also, for patients who underwent primary custom arthroplasty, healing of the syndesmotic fusion was assessed clinically (lack of pain) and radiographically (bridging trabeculation), with delayed healing being defined as that occurring more than twelve weeks after surgery. At six months and yearly intervals following surgery, patients were again asked to complete the SF-36, the AOFAS hindfoot score, and the MFS outcome measures. In addition, goniometric range of motion and dynamic radiographic measurements were made. Results were then calculated with use of the most recent scores and range-of-motion data for each individual.
Statistical Methods
The Student t test was used to compare the clinical outcomes of the patients preoperatively and postoperatively, assuming the data were continuous and followed a normal distribution. A p value of <0.05 was considered significant.
Source of Funding
There was no external source of funding for this project.
Follow-up was available at an average of 58.6 months (range, fifty-two to sixty-six months) for thirty-one of the thirty-three ankles, including thirteen of thirteen in the revision ankle arthroplasty group, six of six in the stage-IV flatfoot group, and twelve of fourteen in the arthritis group. Of the twenty ankles that had a primary arthroplasty, nine of nine with an arthroplasty and triple arthrodesis and nine of eleven with an arthroplasty and subtalar arthrodesis were available for follow-up.
Clinical Outcomes
Twenty-two of the thirty-one ankles with complete follow-up had an arthrodesis (twenty-one subtalar and one triple procedure) at the time of implantation of the stemmed talar component. There was no clinical or radiographic evidence of delayed healing or nonunion of the arthrodesis site in any patients. The overall position and alignment of the foot was acceptable (from 5° of varus to 10° of valgus) in thirty of the thirty-one lower extremities on the basis of weight-bearing radiographs and clinical assessment. One ankle had a malunion, with >5° of varus alignment of the prosthesis, and was symptomatic, requiring a calcaneal osteotomy to resolve the problem.
Ankle Range of Motion
The mean preoperative dorsiflexion was −1.4° ± 7°, and the mean preoperative plantar flexion was 25.5° ± 12°, with an average arc of motion of 21.3° ± 14°. Following surgery, the average dorsiflexion increased to 6.4° ± 6° (p < 0.05). The average postoperative plantar flexion was 25.8° ± 10° (p = 0.26). There was an overall increase in the arc of motion following surgery to 32.2° ± 11° (p < 0.05). All ankles except one had an increased arc of motion following surgery, with the one ankle losing only 2° of motion.
Radiographic Evaluation
For the eighteen ankles that had a primary arthroplasty with use of a custom stem device, healing of the distal tibiofibular joint and syndesmotic fusion was evaluated. At ten weeks, seventeen of the eighteen ankles had a solid fusion. One ankle had delayed healing that eventually required revision with bone-grafting. Ten weeks after revision surgery, the syndesmosis was noted to have fused in that ankle.
The most recent radiographs were assessed to determine the total amount of talar component subsidence. Three (9.7%) of thirty-one ankles exhibited talar component subsidence, but in no ankle was this >3 mm. Osteolysis around the talar component was also assessed but was seen in only one ankle, which demonstrated lysis around both the component and the stem. This occurred following revision, five years after a primary total ankle replacement, in a patient with hemophilia whose prosthesis never achieved ingrowth.
Functional Outcomes
The mean PCS score of the SF-36 was 28.2 ± 5.6 (range, 16.9 to 39.3) preoperatively compared with 39.7 ± 6.5 (range, 23. 7 to 49.3) postoperatively (p < 0.05). Two of thirty-one ankles were in patients who had worse PCS scores postoperatively. A significant difference was also seen in the mean MCS values on the SF-36, which improved from 42.2 ± 13.8 (range, 10.7 to 64.1) preoperatively to 50.8 ± 12.6 (range, 20.3 to 65.4) postoperatively (p < 0.05). Similar significant increases were seen postoperatively in both the MFS and AOFAS hindfoot score. The mean MFS was 47 ± 13 (range, 18 to 78) preoperatively and 75 ± 10 (range, 45 to 89) postoperatively (p < 0.05). Every patient exhibited an increase in the MFS following surgery. The average AOFAS hindfoot score was 41 ± 16 (range, 16 to 78) preoperatively compared with 68 ± 12 (range, 20 to 88) postoperatively (p < 0.05). Only one patient had a worse AOFAS hindfoot score postoperatively.
Complications
Two surgical wounds had superficial dehiscence; both healed without intervention. One patient had a major wound dehiscence, which required vacuum-assisted closure over a seven-week course. Four patients underwent additional procedures following the arthroplasty. One had a delayed union of the distal tibiofibular and syndesmotic fusion and underwent successful revision of the arthrodesis with bone-grafting. Another patient developed anterolateral pain from synovitis and underwent arthroscopic debridement with resolution of the symptoms. A third patient had a twisting injury to the ankle and sustained a lateral ligament injury, which required lateral ligament reconstruction with use of a peroneal rerouting procedure. The injury healed without residual instability. One patient required a sliding calcaneal osteotomy to treat persistently symptomatic varus alignment of the stemmed prosthesis. There were no early cases of deep infection. Three patients had subsequent failure of the stemmed talar prosthesis, all at greater than thirty-six months after implantation. Two failures were due to late septic seeding of the prosthesis, and one was due to trauma with fracture around the talar prosthesis. All three patients were treated with removal of the prosthesis and ankle arthrodesis with use of an intramedullary device and a femoral head allograft. All three arthrodeses healed, and in each patient “spot-welding” was seen at the allograft-bone interface by the time of the twenty-four-month follow-up.
Primary or posttraumatic arthritis involving the ankle and hindfoot joints is a challenging problem. A tibiotalocalcaneal or pantalar arthrodesis continues to be the standard salvage procedure for these problems11,15,16. Both procedures result in substantial stiffness and limitation of activities of daily living for the patient, while at the same time placing high levels of stress on adjacent joints in the foot8,10,15-24. Millett et al. reviewed the results of tibiotalocalcaneal arthrodesis in fifteen patients22. Although they had good average postoperative AOFAS scores (69.2), there were a total of twelve complications. While Hammett et al.23 found that the rate of satisfaction for fifty-two patients who had a tibiotalocalcaneal arthrodesis was >80%, their postoperative AOFAS scores averaged 63 of 100. The average AOFAS score in this study was 68, which is comparable with the outcomes after tibiotalocalcaneal arthrodesis.
Although these procedures can be somewhat effective in eliminating pain, functional outcomes have varied. Acosta et al.24, in a study of thirteen patients with tibiotalocalcaneal arthrodeses and fourteen patients with pantalar arthrodeses, found a 37% rate of complications, with 33% of the patients rating their results as fair or poor. Sanders et al.25 evaluated eleven open grade-IIIB ankle or talus injuries with bone loss, all requiring either a tibiotalocalcaneal or a pantalar arthrodesis. Although fusion was achieved and infection was eradicated in this group of patients, substantial functional and psychosocial disability remained. Eight of eleven patients had considerable pain, difficulty with stairs, and limited ability to walk. All changed jobs or were unemployed. The authors concluded that these patients may have benefited from early amputation instead. The complication rate in the present study was less than or comparable with those rates reported for tibiotalocalcaneal or pantalar arthrodesis. Also, there is an added benefit in terms of retaining functional motion at the joint, as patients had an average arc of ankle motion of 32°, which would have otherwise been sacrificed with a fusion-based procedure.
A failed total ankle replacement with substantial talar component subsidence creates a dilemma for the surgeon, as there is often extensive loss of the talar body secondary to erosion, and arthritic changes have occurred at the subtalar joint. The majority of failed total ankle replacements that were revised in this series were first-generation Agility prostheses. The failure rate for total ankle replacement has been reported to be between 4% and 32%26. Two studies specifically looking at the survivorship of the Agility prosthesis have shown failure rates of 10.6% at seven to sixteen years of follow-up27 and 32.3% at two to six years of follow-up6. Outcomes of revision total ankle arthroplasty generally have been poor, in part because of the fact that surgery must be performed with standard implants, as no revision components have been available. As a result, ankle arthrodesis (with structural bone graft to fill the defect left following removal of the prosthesis) with arthrodesis of additional joints of the hindfoot has been considered the primary salvage procedure, despite its obvious limitations. Kitaoka and Romness reported on thirty-eight failed total ankle replacements in thirty-six patients treated with tibiotalocalcaneal fusion28. In that study, 80% of the patients had mild or no pain at eight years after a successful fusion. Eighty-nine percent achieved osseous union, but there was an overall complication rate of 13%. No patient in that study underwent a below-the-knee amputation. A protocol for the management of failed total ankle replacement has been described by Kotnis et al.29. In that study, nine of fourteen patients with aseptic loosening of the components underwent a hindfoot fusion to treat inadequate talar bone stock. Spirt et al. noted that component migration and failure occurred almost exclusively with the talar component30. At our institutions, we have experienced similar findings with component collapse and loosening occurring almost exclusively on the talar side. Prior to the development of a custom long-stemmed talar component, arthrodesis was our only option.
Aside from failed total ankle replacement and complex ankle and hindfoot arthrosis, which can be managed with salvage procedures, stage-IV posterior tibial tendon dysfunction has limited reconstructive options. Myerson modified the staging of this disease to describe a stage-IV flatfoot condition in which the tibiotalar joint is also affected, making this a pantalar disease31. A variety of surgical procedures have been described for the treatment of stage-IV flatfoot, with either pantalar or tibiotalocalcaneal arthrodesis being the procedures most commonly used11,32,33. Some have advocated deltoid ligament reconstruction in combination with hindfoot fusion procedures33-35. In our study, no patient underwent deltoid ligament reconstruction, although care was taken to adequately tension both the medial and lateral compartments during the arthroplasty. There have been a limited number of studies on the treatment of stage-IV posterior tibial dysfunction, with fewer addressing functional outcomes with either arthrodesis or joint-sparing procedures. The subset of patients with stage-IV posterior tibial tendon insufficiency in our study was small. However, all showed an increase in the functional outcome scores, and only one patient had evidence of clinical or functional medial instability of the ankle at greater than thirty-six months following surgery.
This study includes a wide range of pathology, including failed total ankle replacement, severe hindfoot arthritis, and stage-IV posterior tibial tendon dysfunction. While this is a limitation of the study, it is important to note that nearly all of the patients had an individual increase in the functional outcome scores. Therefore, we believe that custom total ankle arthroplasty is a reasonable alternative to traditional arthrodesis procedures in these situations. Also, the data presented, although encouraging, were obtained at a minimum of four years following arthroplasty. Continued follow-up will be required to determine if the increases in functional outcomes and clinical parameters will be maintained or will change over time.
In conclusion, our results indicate that the use of a custom long-stemmed talar component placed either primarily in patients with ankle and hindfoot arthritis or as a revision prosthesis in patients with a failed total ankle replacement is a promising option. The use of this device to maintain tibiotalar motion, coupled with hindfoot fusions to correct malalignment and relieve pain, is equal, if not superior to, the results of other procedures. It must be reiterated that the principal benefit of this technique is the restoration of a reasonable arc of ankle motion, in the face of severe disease. While our results indicate that the direction to pursue is continued development of revision prostheses, long-term clinical and radiographic follow-up will be necessary to determine the survivorship of this prosthesis and its role in treating combined complex pathology of the ankle and hindfoot.