How Does a Novel In Situ Fixed-bearing Implant Design Perform in Revision Ankle Arthroplasty in the Short Term? A Survival, Clinical, and Radiologic Analysis

Abstract

Background

Given the growing number of primary total ankle replacements (TAR), an increase in the number of patients undergoing subsequent revisions might be expected. Achieving a stable and balanced ankle while preserving the remaining bone stock as much as possible is crucial for success in revision TAR. Most reported techniques rely on bulky implants with extended fixation features. Since 2018, we have used a novel, three-component ankle prosthesis for revision that is converted in situ to a fixed-bearing, two-component ankle prosthesis once the components have found their position according to an individual’s anatomy. The results of this novel concept (fixation, revision, pain, or function) have not, to our knowledge, been reported.

Questions/purposes

What are the short-term results with this new revision TAR design, in terms of (1) repeat revision surgery, (2) patient-reported outcomes on the American Orthopaedic Foot and Ankle Society (AOFAS) ankle-hindfoot score, (3) pain according to the VAS, and (4) radiographic signs of fixation?

Methods

Between February 2018 and February 2020, we performed 230 TAR surgeries (in 206 patients) for any indication in our clinic. The novel semiconstrained, uncemented Hintermann Series H2^© implant was used in 96% (220 of 230) of procedures (201 patients). Fifty-four percent (119 of 220) of these were converted from an existing TAR to H2, which was the focus of the present study. However, only 45% (54 of 119) of these conversions to H2 were eligible for analysis. These patients had a mean age of 63 ± 12 years, and 43% (23 of 54) were women. The median (range) follow-up time was 3.2 years (2.0 to 4.3). The H2 design allows in situ conversion to a fixed-bearing system, with minimal bone resection. It achieves translational and rotational stability while preserving function and supporting the periarticular soft tissues. We defined repeat revision as exchange of one or both metal components, ankle fusion, or amputation and assessed it using a cumulative incidence survivorship estimator. Factors potentially associated with revision were assessed using Cox regression analyses. Clinical and radiologic outcomes were assessed preoperatively and at the most recent follow-up interval. Clinical outcomes included pain on the VAS (average pain during normal daily activity during the past seven days) and AOFAS score. Radiologic outcomes were the tibial articular surface angle, tibiotalar surface angle, talar tilt angle in the coronal plane, and AP offset ratio in the sagittal plane, as well as radiolucent lines and radiographic signs of loosening, defined as change in position greater than 2° of the flat base of the tibia component in relation to the long axis of the tibia, subsidence of the talar component into the talus greater than 5 mm, or change in position greater than 5° relative to a line drawn from the top of the talonavicular joint to the tuberosity of the calcaneus, as seen on plain weightbearing radiographs.

Results

The cumulative incidence of repeat revision after 1 and 2 years was 5.6% (95% CI 0% to 11%) and 7.4% (95% CI 0% to 14%), respectively. With the numbers available, no clinical factors we analyzed were associated with the risk of repeat revision. The median values of all assessed clinical outcomes improved; however, not all patients improved by clinically important margins. The median (range) AOFAS ankle-hindfoot score increased (from 50 [16 to 94] to 78 [19 to 100], difference of medians 28; p < 0.01), and the median pain on the VAS decreased (from 5 [0 to 9] to 2 [0 to 9], difference of medians 3; p < 0.01) from before surgery to follow-up at a minimum of 2 years. Radiographically, lucency was seen in 12% (6 of 49 patients) and loosening was seen in 8% (4 of 49). One of these patients showed symptomatic loosening and was among the four patients overall who underwent revision. We could not assess risk factors for repeat revision because of the low number of events (four).

Conclusion

The investigated new in situ fixed-bearing ankle design achieved overall better short-term results than those reported in previous research. Destabilization of the ankle joint complex, soft tissue insufficiency, and possible changes of the joint configuration need an optimal solution in revision arthroplasty. The studied implant might be the answer to this complex issue and help surgeons in the perioperative decision-making process. However, a relatively high percentage of patients did not achieve a clinically important difference. Observational studies are needed to understand long-term implant behavior and possibly to identify ankles benefiting the most from revision.

Level of Evidence

Level IV, therapeutic study.

Introduction

Advances in joint arthroplasty technology have made total ankle replacement (TAR) a viable treatment for endstage arthritis and have led to a resurgent interest in this procedure [39]. With an increase in the number of primary procedures, surgeons will also be faced with an increase of ankles undergoing revisions. Component loosening and cyst formation were reported to be the most common reason for revision, independent of ankle designs [1, 7, 12, 15, 16, 18, 25, 40, 44, 45]. Progressive destabilization of the ankle complex, including the distal tibiofibular and peritalar structures, is another frequent reason for revision, particularly in mobile-bearing ankle designs, because of their unconstrained biomechanical characteristics [33, 36].

Therefore, in addition to the loss of bone stock, instability of the ankle might be one of the most challenging problems in revision TAR. Most surgeons prefer fixed-bearing components to intrinsically stabilize the restored ankle, as well as bulky metallic components to achieve greater primary stability of the remaining bone stock [5, 8, 10, 26, 27, 34]. However, this strategy has drawbacks. It may further compromise the remaining bone stock, making further revision almost impossible and, if necessary, conversion into tibiotalar fusion more difficult. To date, bone-saving procedures have been almost exclusively associated with mobile-bearing ankle systems that accommodate rotational and translational forces in the restored ankle, thus reducing potential shear forces at the implant-bone interface, which, in turn, avoids the use of bulky implants [20, 21, 29]. In 2013, in a series of 117 revision TARs, at 6.2 years we found comparable results to those after primary TAR [20]. This indicates the components were firmly anchored to the primary bone stock, despite bone loss because of the previous TAR.

Another conflicting problem for revision arthroplasty, especially for the tibial side, is changes in the ankle configuration because of bone loss. This makes it almost impossible for the surgeon to predict the axis of rotation of the newly revised ankle and then position the component accordingly in the remaining bone. In reality, the surgeon attempts to position the tibial component to best support the remaining bone. With the use of a mobile-bearing ankle, the difference in the axis or rotation of the created ankle is compensated for by the mobile insert, which does not provide intrinsic stability. In contrast, with a conventional fixed-bearing ankle, the ankle will be forced to rotate about the axis, determined by the position of the tibial component. A potential solution could be a three-component ankle that can be converted into a fixed-bearing ankle once the polyethylene insert has found its position, which is determined by an individual’s anatomy. The Hintermann Series H2® (DT MedTech LLC) ankle, which we have used since 2018, fulfills these characteristics. To our knowledge, this novel implant has not yet been studied in a similar setting.

We therefore asked: What are the short-term results with this new revision TAR design, in terms of (1) repeat revision surgery, (2) patient-reported outcomes on the American Orthopaedic Foot and Ankle Society (AOFAS) ankle-hindfoot score, (3) pain according to the VAS, and (4) radiographic signs of fixation?

Patients and Methods

Study Design and Setting

This was a retrospective study using data from an FDA-approved, longitudinally maintained institutional registry that is managed by research associates not otherwise involved in the care of the patients. All procedures were performed the senior author (BH) or surgeons under his supervision in a public hospital covering adult orthopaedics and trauma with approximately 5000 surgeries and more than 30,000 consultations annually.

Participants

Between February 2018 to February 2020, overall, 230 TARs in 206 patients were performed. The novel semiconstrained, uncemented H2 implant was used in 96% (220 of 230) of procedures (in 201 patients). A revision TAR using H2 was performed in 54% (119 of 220 procedures in 108 patients). However, 55% (65 of 119) of H2 revisions were excluded: 34% (22 of 65) had multiple prior revisions, 32% (21 of 65) had prior ankle fusion, 18% (12 of 65) had an infection before conversion to an H2, 5% (3 of 65) had a periprosthetic fracture, 5% (3 of 65) had a neuromotor disorder, 2% (1 of 65) had a missing fibula, and 5% (3 of 65) were lost before the minimum study follow-up of 2 years. This left 54 procedures (54 patients) for survival analysis.

Clinical data were available for 85% (46 of 54) of patients (three patients had no clinical evaluation, five were lost to follow-up before the minimum follow-up of 2 years), and radiologic data were available for 91% (49 of 54; three patients had no radiologic evaluation, and two were lost to follow-up before the minimum follow-up period) (Fig. 1). The median (range) follow-up duration was 3.2 years (2 to 4.3).

Fig. 1.

This flowchart shows the patients who were included in our study; H2 = Hintermann Series H2^©.

Patients’ Baseline Data

Patients had a mean age of 63 ± 12 years, and 43% (23 of 54) were women. The right ankle was affected in 44% (24 of 54) of patients. Trauma was the most common etiology of ankle osteoarthritis (Table 1).

Table 1.

Baseline characteristics (n = 54 ankles)

Characteristic	Value
Age in years	63 ± 12
Gender  Women	43 (23)
Side  Right	44 (24)
BMI in kg/m2	28 ± 4
VAS score preoperative	5 (0-9)
AOFAS preoperative	50 (16-94)
Etiology of ankle OA
Primary  Posttraumatic  Secondarya	11 (6) 63 (34) 26 (14)

Data presented as mean ± SD, % (n), or median (range).

^aHistory of rheumatoid arthritis, hemochromatosis, hemophilia, gout, diabetes mellitus, polyneuropathy, or soft tissue rheumatism; AOFAS = American Orthopedic Foot and Ankle Hindfoot Score; OA = osteoarthritis.

Prosthesis Design and Surgical Technique

The H2 TAR implant is a semiconstrained, uncemented, fixed-bearing FDA-approved system. A unique feature of this system is that it works similar to a mobile-bearing ankle in which the polyethylene insert is fixed to the tibial component once it has taken its position as dictated by the patient’s bony geometry and ligament tension, thus respecting the axis of rotation of the ankle being replaced. It is converted in situ to a fixed-bearing ankle and provides high intrinsic stability. Sagittal plane motion and function are free, and the anatomically shaped design allows minimal bone resection and an optimal contact area at the bone-implant interface. The flat-shaped, titanium alloy tibial assembly component needs a flat resection that can be as high as 10 mm above the original apex of the tibial pilon when used with the thickest polyethylene insert (9 mm). It contains an anterior shield to stabilize against translational and rotational forces. Osteointegration is intended in three planes: along the flat surface in the horizontal plane, along the medial malleolus in the sagittal plane, and along the shield in the coronal plane. Furthermore, the tibial assembly component includes a tibial slide and tray and a locking device with a screw mechanism. The standard conically shaped, cobalt-chromium alloy talar component needs a 2-mm bone resection in three planes and is inserted in a press-fit manner. It contains two pegs to enhance primary stability. If there is bone loss, which often occurs in revision arthroplasty, the flat undersurface of the talar component is used, with an identical articulating surface and a 2.5-mm-high rim on the medial and lateral side. The anterior shield increases the contact area with the supporting bone in the sagittal plane. The ultra–high molecular weight polyethylene inlay assembly component has a mushroom-like feature on its superior surface to glide into the tibial component. It is available in three configurations, with the mushroom being in the center (neutral position) or 5-mm anterior or posterior, allowing the rotational axis to be adapted in the sagittal plane to the given anatomy.

A standard surgical technique was applied [4, 19]. After resection cuts were made on the tibial and talar side using templates, remaining bone defects were filled with an allograft (such as the Tutoplast Allograft, Tutogen Medical GmbH). After the talar and tibial components were inserted, trial inserts were used to determine the appropriate polyethylene thickness to balance the ligaments and determine the AP position of the talus regarding the tibial component. Thereafter, the selected polyethylene insert was inserted. After the ankle was moved several times into forced dorsiflexion and some plantar flexion, the locking device was inserted and the screw was tightened to fix the polyethylene insert into the tibial component while the foot was held in the neutral position [9, 14, 19, 43].

Primary and Secondary Study Outcomes

Our primary study goal was to determine the survival rate and assess factors associated with repeat revision. We defined this endpoint as repeat revision involving exchange of one or both metallic components, arthrodesis, or amputation, and assessed it using a cumulative incidence survivorship estimator. Factors associated with revision were assessed using Cox regression. Reasons for revision were defined based on existing studies [17, 33, 37]; that is, the exchange of at least one metallic component. Additional procedures at the time of the index surgery were documented and categorized based on the type and anatomic location.

Our secondary study goal was to assess clinical and radiologic outcomes. Clinical and radiologic outcomes were evaluated preoperatively and at the last follow-up interval. Clinical outcomes included pain on the VAS (average value over the past 7 days on a scale of 0 [no pain] to 10 [severe pain]) and AOFAS ankle-hindfoot score, and they were assessed by independent research staff members who were not involved in any part of clinic visits or surgery. The minimum AOFAS score is 0 (severe impairment) and the maximum is 100 (no impairment) [23, 24]. Radiologic outcomes were chosen according to previous work [2, 3, 19, 38], and they were the tibial articular surface angle, tibiotalar surface angle, talar tilt angle in the coronal plane, and AP offset ratio in the sagittal plane. We also considered radiolucent lines and radiographic signs of loosening, defined as change in position greater than 2° of the flat base of the tibial component in relation to the long axis of the tibia and/or as a progressive radiolucency of more than 2 mm and as subsidence into the talar bone greater than 5 mm or a change in position of greater than 5° of the talar component relative to a line drawn from the top of the talonavicular joint to the tuberosity of the calcaneus, as seen on plain weightbearing radiographs. All radiographic parameters were assessed by a board-certified orthopaedic surgeon (PK) who specializes in foot and ankle surgery and a fifth-year orthopaedic resident (LT).

Ethical Approval

This retrospective, single-center study was approved by our local ethics committee and was conducted in accordance with the principles of the Declaration of Helsinki and the Guidelines for Good Clinical Practice.

Statistical Analysis

We used a Shapiro-Wilk test to assess whether continuous variables followed a Gaussian distribution. The cumulative incidence of repeat revision after 1 and 2 years was calculated using R-function cuminc from R-package cmprsk (RStudio). We could not assess risk factors for repeat revision because of the low number of events (four).

Preoperative and postoperative continuous variables were compared using linear mixed-effects models using patient identification as a random effect, and gender, age, and etiology as potential covariates. Of all possible variable combinations (R-function dredge from R-package MuMIn), the model with the lowest Akaike information criterion was selected for inference. Differences in categorical variables before conversion to the H2 implant and at the last follow-up interval were assessed using Fisher exact tests. In addition, for pairwise comparisons, each category was tested individually against all other categories combined.

Minimum clinically important differences (MCIDs) for pain on the VAS and AOFAS score were obtained from existing research. For the AOFAS score, Chan et al. [6] reported an MCID ranging from 7.9 to 30.2 in hallux valgus surgery; therefore, we adopted the approximate average MCID of 20 for our analysis. For pain on the VAS, Tashjian et al. [41] reported an MCID of 1.4, which we rounded to 2 for our analysis. The level of significance was set at p ≤ 0.05. The statistical analysis was performed using R Project for Statistical Computing (version 3.4.3) via RStudio (version 1.1.423, RStudio) [32, 35].

Results

Cumulative Incidence of Repeat Revision

The cumulative incidence of repeat revision after 1 and 2 years was 5.6% (95% confidence interval [CI] 0% to 11%) and 7.4% (95% CI 0% to 14%), respectively (Fig. 2). The most common reason for conversion to an H2 prosthesis was syndesmotic insufficiency (26% [14 of 54]), followed by arthrofibrosis or pain (22% [12 of 54]) and instability (other than syndesmotic insufficiency) (19% [10 of 54]) (Table 2). A total of 7% (4 of 54) of repeat revisions were performed after a median (range) of 0.6 years (0.3 to 1.9) after the index revision surgery (Table 3). The low number of repeat revisions precluded us from performing an analysis of risk factors for revision. Overall, 89 additional intraoperative procedures concomitant to the index revision surgery were performed in 87% (47 of 54) of patients. The most common procedure was grafting (20% [18 of 89]) (Table 4). In four patients, complications occurred intraoperatively or perioperatively. In one patient, a superficial surgical site infection was treated successfully with antibiotics, and in three, an intraoperative fracture of the medial malleolus was treated with screw fixation.

Fig. 2.

This graph shows the cumulative incidence of repeat revision of revised ankles, with death as a competing risk, using a Kaplan-Meier survival curve.

Table 2.

Reason for conversion to H2 (n = 54)

Indication for conversion to H2	Value
Instability  Syndesmotic insufficiency Arthrofibrosis or pain Loosening Cyst formation Inlay fracture Technical failure	44 (24) 26 (14) 22 (12) 17 (9) 13 (7) 2 (1) 2 (1)

Data presented as % (n); H2 = Hintermann Series H2.

Table 3.

Characteristics of patients with secondary revision after conversion to H2

Number	Age in years	Gender	Side	Primary implant	Reason for primary revision	Secondary implant	Reason for secondary revision	Survival time in years	Surgical solution	Tertiary implant
1	60	Man	Left	H3	Pain or arthrofibrosis	H2	Pain or arthrofibrosis	0.4	Interpositional TTC-fusion	Pantanail
2	72	Man	Left	H3	Aseptic loosening	H2	Aseptic loosening	1.9	Exchange of tibial component	H2
3	58	Man	Left	H3	Pain or arthrofibrosis	H2	Pain or arthrofibrosis	0.8	Exchange of tibial component	H2
4	61	Woman	Right	H3	Pain or arthrofibrosis	H2	Cyst formation talar	0.3	Exchange of talar component, cyst grafting, cotton osteotomy	H2

H2 = Hintermann Series H2; TTC = tibiotalocalcaneal.

Table 4.

Additional procedures at the index revision surgery (n = 89)

Type or location of procedure

Value

Removal of osteosynthesis material Grafting  Tibia  Talus  Fibula Tibia  Medial malleolar osteotomy  Medial malleolar osteosynthesis  Medial malleolar resection Fibula  Shortening osteotomy  Osteosynthesis Medial sliding calcaneus osteotomy Hindfoot or midfoot arthrodesis  Tibiofibular  Subtalar  Talonavicular  Calcaneocuboid Lateral ligament reconstruction Tendons  Achilles tendon lengthening  PLPB transfer Tibialis posterior release or transfer Arthrolysis Debridement Tight rope removal Cyst filling

17 (15) 20 (18) 11 6 1 8 (7) 2 4 1 3 (3) 2 1 7 (6) 15 (13) 5 4 3 1 1 (1) 10 (9) 8 1 2 (2) 11 (10) 3 (3) 1 (1) 1 (1)

Data presented as % (n); PLPB = peroneus longus to peroneus brevis.

AOFAS Ankle-hindfoot Score and VAS Pain

The median values of all assessed clinical outcomes improved; however, not all patients improved by clinically important margins. The median (range) AOFAS ankle-hindfoot score increased from 50 (16 to 94) to 78 (19 to 100) (difference of medians 28; p < 0.01). The MCID for the AOFAS score was 20, which was reached in 48% (22 of 46) of patients, and an MCID of 30 was reached in 33% (15 of 46) of patients. The median (range) VAS score decreased from 5 (0 to 9) to 2 (0 to 9) (difference of medians 3; p < 0.01) when comparing before surgery to 2 years of follow-up (Fig. 3). The MCID of 2 for VAS pain was reached in 59% (27 of 46) of patients undergoing revision TAR. Altogether, 57% (26 of 46) of patients did not improve by a clinically important amount on the VAS and AOFAS ankle-hindfoot score.

Fig. 3.

These box plots represent (A) VAS for pain, (B) AOFAS ankle-hindfoot score, and (C) patient satisfaction. A color image accompanies the online version of this article.

Radiographic Findings

The cumulative incidence of loosening—defined as change in position greater than 2° of the flat base of the tibial component in relation to the long axis of the tibia, a progressive radiolucency of more than 2 mm and subsidence into the talar bone greater than 5 mm, or a change in position of greater than 5° of the talar component relative to the line drawn from the top of the talonavicular joint to the tuberosity of the calcaneus as seen on plain weightbearing radiographs—was 4.6% (95% CI 0% to 9%) at 2 years, and the cumulative incidence of lucency was 11% (95% CI 2% to 20%) (Table 5). Lucency was evident exclusively on the tibial side. Loosening of the tibial component alone was seen in 4% (2 of 49) patients, loosening of the talus component alone was seen in 2% (1 of 49), and loosening of both components occurred in 2% (1 of 49) of patients. Among the four patients with repeat revision, loosening or lucency was found in only one patient each. Radiographic measurements showed an increase in the median (range) AP offset ratio from 0.7 (-3.9 to 19.8) preoperatively to 2.9 (-0.8 to 10.3) postoperatively (p = 0.003) (Fig. 4). Inlay thickness or position after conversion to the H2 prosthesis (neutral: 51 patients, anterior: two patients, posterior: one patient) was not associated with survival. The preoperative AP offset ratio was higher in H2 ankles with an anterior inlay position (13.8 mm, range 7.9 to 19.8 mm) than in ankles with either a neutral or posterior inlay position (0.8 mm, range -3.9 to 7.7 mm; p ≤ 0.001). Although the AP offset ratio decreased by 8.1 ± 2.5 mm in ankles with an anterior inlay position (p = 0.01), it increased by 1.9 ± 0.5 mm in ankles with a neutral or posterior inlay position (p = 0.004). No other major changes in radiographic outcomes were detected.

Table 5.

Clinical and radiologic outcomes

Variable	Preoperative	Postoperative	p value
AOFAS	50 (16 to 94)	78 (19 to 100)	< 0.01
VAS score	5 (0 to 9)	2 (0 to 9)	< 0.01
TAS in °	90 (71 to 101)	90 (84 to 98)	0.58
TTS in °	89 (78 to 102)	90 (83 to 100)	0.36
TT in °	0.2 (-8 to 10)	0.4 (-4.8 to 6)	0.91
AP offset	0.7 (-3.9 to 20)	2.9 (-0.8 to 10)	< 0.01
Lucency		12 (6 of 49)
Loosening		8 (4 of 49)

Data presented as median (range) or % (n); AOFAS = American Orthopaedic Foot and Ankle Society score; TAS = tibial articular surface angle; TTS = tibiotalar surface angle; TT = talar tilt angle.

Fig. 4.

This graph shows the Gaussian distribution of the change in the AP offset ratio after revision; AOR = adjusted odds ratio.

Discussion

An increasing number of ankles undergoing revision TAR is expected because of the high frequency of primary implantations, extended indications, and expanded target population. Progressive destabilization of the ankle joint complex, insufficient periarticular soft tissues, and loss of bone stock require targeted solutions. Outcomes can be suboptimal with available fixed-bearing ankles, particularly in revision arthroplasty, in which the remaining bone stock may require the surgeon to position the tibial component optimally to achieve maximal bone support. Thus, the axis of rotation of the ankle can be predicted based on the final position of the tibial component. On the other hand, the currently investigated, novel, in situ, fixed-bearing implant allows the surgeon to determine the axis of rotation of the created ankle as given by the individual anatomy and ligament tension as it is converted from a mobile-bearing to a fixed-bearing ankle after the components are inserted. We found a survival rate of 95% and 93% after 1 and 2 years, respectively. An improvement in pain and function was evident in 59% and 48%, respectively, and the radiologic assessment showed stable implant fixation and bone-implant interface. The four perioperative complications were treated accordingly, without further surgery. This permits the revised ankle to adapt to the axis of rotation intraoperatively, as provided by an individual’s anatomy, and provides intrinsic instability because of the fixed-bearing system (Fig. 5).

Fig. 5.

A 63-year-old woman underwent revision ankle arthroplasty after primary TAR with the two-component Cadence prosthesis (Smith and Nephew) and syndesmotic fusion. (A) A plain AP-view radiograph taken preoperatively shows loosening of both components with substantial lateralization of the tibial component and accompanying bone loss. (B) A plain lateral-view radiograph shows loosening of both components with peri-implant cyst formation and bone loss. (C) Intraoperative photographs were taken after the failed implants were removed. There was substantial tibial bone loss, and the bony configuration was changed. The cysts were cleaned and filled with an allograft structural bone graft. (D) An intraoperative photograph shows the revised ankle with an H2 implant. (E) One year postoperatively, a plain AP-view radiograph of the revised ankle shows sufficient osteointegration of the new components and no signs of perioperative lucency or implant loosening. (F) One year postoperatively, a plain lateral-view radiograph of the revised ankle showed the ankle was stable and both components were integrated, without signs of lucency or cyst formation.

Limitations

The current study has several limitations. The retrospective design could have resulted in bias. First, during the study period, not all patients who underwent revision TAR received the investigated implant. We followed an intraoperative decision-making algorithm, and if substantial coronal plane instability was evident and support of the periarticular soft tissues was necessary, we decided to use the H2 prosthesis. If these signs were not evident, a mobile-bearing implant was implanted. In our clinic, we follow patients after TAR regularly. If signs of impending loosening or other implant-associated complications are seen, we advise the patient to undergo revision to prevent a catastrophic event. This explains why some patients received revision surgery, even though they had high preoperative functional scores. The risk of selection bias may be elevated. Second, the investigated follow-up time is short and may not capture all adverse events; therefore, there may be some transfer bias. Longitudinal, observational, mid-term and long-term studies are necessary to understand the behavior and performance of this implant in the revision setting. Third, patients were clinically evaluated by independent research staff who are not involved in any aspect of a patient’s treatment, including surgery. On the other hand, the radiologic assessment was performed by a surgeon and a resident who were involved in the daily clinic and assisted in some of the surgeries; thus, there may have been an assessment bias. Radiographic measurements, particularly evaluation of periprosthetic cyst formation, might be somewhat limited because parts of the implant may have been covered. In our study, two authors (PK, TH-L) made these measurements twice. If there was disagreement, a third author (BH) evaluated the radiographs.

In addition, all surgical procedures were performed by the senior author (BH) or surgeons under his direct supervision. The senior author was involved in the development of the device. To maintain independence, he was not involved in data collection or analysis. Social desirability bias should be considered because clinical outcomes were assessed by research staff members. Furthermore, surgeons with developed surgical skills and increased experience may have influenced the results. We are aware that the AOFAS ankle-hindfoot score is not a validated score and has many disadvantages, including a lack of information on the magnitude of change that would show a clinically important change and missing information about scores translated to functional status. Certain items are overweighed compared with others, and the AOFAS research committee has recommended against its use [30]. These factors were not known at the initiation of our registry approximately 20 years ago [28]. In our clinic, we are transitioning to other patient-reported outcome measures.

Cumulative Incidence of Repeat Revision

The cumulative incidence of all-cause revision was less than 10% at 2 years. This is better than the best-reported results with other revision TAR devices with this follow-up period. In contrast to primary TAR, outcome data after revision TAR are still scarce. Most reported techniques rely on larger implants and extended fixation features. This is particularly true for intramedullary fixation in the distal tibia combined with a fixed-bearing design to overcome a bone defect and instability. Using the INBONE Total Ankle System (Wright Medical Group) with intramedullary fixation of the tibia, Behrens et al. [5] reported a survival rate of 77.8% at a mean follow-up of 47.5 months after revision TAR. Loosening of the tibial and talar components occurred in 38.9% and 44.4% of their cohort, respectively. Subsidence of the revision tibial and talar components was observed in 38.9% and 55.6% of patients, respectively. Lachman et al. [26] investigated outcomes after revision TAR in 29 patients in whom five different implants were used, including four fixed-bearing designs and three with additional intramedullary fixation after a mean of 3.3 ± 1.7 years. The authors reported repeat revision of the revised TAR in 10.3% of patients. Ellington et al. [10] noted retained components in 64% of their cohort after revision TAR using the two-component Agility system after a mean (range) of 49.1 months (25.9 to 77.8 months). Although the used techniques and ankle systems yielded acceptable results, they were not performed without awareness of concern about additional bone resection and changes in the force transmission pattern. Subsequent nonphysiologic load of the ligaments because of the component’s position while seeking congruity under weightbearing also raises concerns about the components’ longevity.

AOFAS Ankle-hindfoot Score and VAS Pain

Although the median scores improved on both outcomes tools we used, the magnitudes of these improvements were often small, with 57% (26 of 46) of our patients not achieving the MCID on the AOFAS ankle-hindfoot score and the VAS pain score. It is debatable whether functional results after revision will reach the level of outcomes after primary TAR [5, 26]. Future studies should focus on identifying patients who clinically benefit the most from a revision TAR and on distinguishing ankles that might require another surgical solution.

Radiographic Findings

The cumulative incidence of radiographic loosening was 4.6% (95% CI 0% to 9%) at 2 years. The radiologic measurements did not show differences in the coronal plane before and after revision TAR. The range of the coronal plane position was smaller postoperatively than preoperatively, indicating a more precise intraoperative adjustment of the implants in revision procedures or that fewer positional changes occurred during the initial setting process. There were radiographic signs of loosening in 8% (4 of 49) of patients and in one of the four patients undergoing a second revision, suggesting there was a stable bone-implant interface in more than 90% of patients. Our data indicate that neither intramedullary fixation devices nor bulky components are necessary to stabilize an ankle prosthesis. With the in situ fixation of components that have adapted to a patient’s anatomy, we hope to minimize soft tissue load and decrease unnecessary subsequent forces between the bone-implant interface. Our results, which showed there were minimal signs of radiolucency and no cyst formation, are in accordance with the findings of Quevedo González et al. [31], who found four to five times larger values for the maximum peak micromotion for the stem design than for designs with spikes or keel. Furthermore, in a finite-element model, no significant differences in tibial bone strain were found between mobile-bearing and fixed-bearing devices [42]. Our data indicate that the resurfacing concept also works for the in situ fixed-bearing H2 ankle with osteointegration surfaces in three planes: at the flat surface horizontal to the tibia, along its medial border to the medial malleolus, and along its anterior shield to the tibia.

Defining the center or axis of rotation of the ankle during a replacement is a demanding task. Even in healthy ankles, the center or axis of rotation changes during motion. Osteoarthritic ankles commonly show some deformity and corresponding changes of the center and axis of rotation. In primary procedures, it is possibly somewhat easier to re-create the joint configuration and correctly position the implant components. In revision procedures, bone loss, cyst formation, instability, and a changed joint configuration create unpredictable anatomic and biomechanical features, making it difficult to define the center or axis of rotation. Our results indicate a slight malposition of the talar component in the sagittal plane; however, in contrast to Barg et al. [2], regarding primary TAR, we did not find an association between the AP offset change and clinical outcomes or repeat revision. However, we still believe that implant positioning is crucial to re-create physiologic biomechanics and for satisfactory long-term results. Cyst formation may occur without destabilization of the component, or with loosening and/or subsidence of the component. High success was reported after bone graft filling [13], even in larger lesions [11, 22], thus questioning the necessity of intramedullary fixation of the tibial component. This is in accordance with the results of the current study, in which stable implant osteointegration was achieved in 20% (11 of 54) of patients after tibial bone grafting.

Conclusion

The investigated implant design and novel concept in revision TAR showed acceptable repeat revision rates, with improved function and pain in approximately half of patients. Although no additional features are used in the flat contact area, the rate of lucency and loosening was lower than reported with fixed-bearing ankles, indicating that respecting the individual axis of rotation may have resulted in reduced shear forces at the implant-bone interface and nonphysiologic load of the ligaments. These preliminary findings suggest a reliable solution and allow the maintenance of function and desired quality of life. However, a high percentage of patients did not achieve a clinically important difference. This reflects the complexity of ankles treated with revision. With this evidence, foot and ankle surgeons can possibly offer better education to the patient, as well as more efficient care. However, the longevity of novel two-component implants and other implant designs used in revision TAR should be further investigated to understand their long-term characteristics and provide better, evidence-based practice. Additionally, observational investigations will help us to determine which ankles are expected to benefit the most from revision arthroplasty.