Introduction

During the last two decades, total ankle replacement (TAR) has gained increased acceptance as a treatment option for patients with end-stage ankle osteoarthritis [10, 11, 23, 29, 49, 59]. However, ankle arthrodesis is the preferred treatment for the majority of patients with end-stage ankle osteoarthritis [1, 12, 18, 47, 64]. Different surgical techniques have been described that often result in short- and mid-term postoperative pain relief [47]. However, the complications after ankle arthrodesis include tibiotalar malunion or nonunion [15, 22, 26], functional gait impairment [17, 20, 46, 57], and premature deterioration of adjacent joints [18, 21, 45].

In patients with a painful nonunited or malunited ankle arthrodesis, the surgical treatment is challenging and may include takedown of the ankle arthrodesis and conversion to TAR (Table 1). This procedure is technically demanding [28, 52] and only a few studies have evaluated this surgical option in this patient population (Table 1). We therefore wished to evaluate our experience with conversion of painful ankle arthrodesis to TAR.

Table 1 Complications and clinical outcomes in other series of patients undergoing conversion of ankle arthrodesis to total ankle replacement
Full size table

Specifically, we asked (1) What intraoperative and perioperative complications were observed in patients who underwent conversion of an ankle arthrodesis to a TAR? (2) Was durable fixation achieved at short term, and what was the alignment of the components? (3) What subsequent surgical procedures were performed, including revisions? (4) What improvements were observed in pain, tibiotalar ROM, and quality of life?

Patients and Methods

The institutional review board (Landesärztekammer Hessen) approved this retrospective study, and informed consent was waived. Patients were identified by searching the German Foot and Ankle Society’s TAR register [40, 41] for the period from January 1, 2007, to December 31, 2014. Only procedures performed at the first author’s institution were searched.

During the period in question, indications for conversion of ankle arthrodesis to TAR were patients with tibiotalar nonunion or malunion with goals of regaining hindfoot mobility and decreasing pain after the revision surgery; these patients all had declined revision ankle arthrodesis. For this study a nonunion (n = 2) was defined as incomplete osseous healing at more than 6 months after arthrodesis based on conventional weightbearing radiographs. Malunion (n = 18) was defined as alignment deviation of more than 7° in the sagittal and/or coronal plane. The contraindications for conversion were: acute or chronic infection, neuromuscular disorders, neuroarthropathy (Charcot arthropathy), diabetic syndrome with polyneuropathy, severe osteoporosis, and immunosuppressive therapy. During that period, all patients who met inclusion criteria (18 of 42 with painful tibiotalar nonunion or malunion) underwent conversion of their ankle arthrodesis to TAR. The database search yielded 18 patients who underwent conversion of their ankle arthrodesis to TAR. The minimum followup for inclusion was 24 months (mean, 54 months). All 18 patients were accounted for (Table 2). Fourteen of the 18 patients were men and four were women with a mean age of 51 ± 7 years (Table 2). Major comorbidities included diabetes mellitus in two of 18, hemophilia A in four of 18, and tobacco use in one of 18. The initial ankle arthrodesis was performed 6 ± 3 years before conversion to TAR (Table 2). The most-common etiology for ankle osteoarthritis was posttraumatic in 13 of 18 patients (Table 2).

Table 2 Demographic data of 20 patients who underwent conversion of painful ankle arthrodesis to total ankle replacement
Full size table

TAR was performed using the cementless HINTEGRA® total ankle system (Newdeal, Lyon, France; Integra LifeSciences, Plainsboro, NJ, USA), a nonconstrained three-component prosthesis (Fig. 1) [8, 29]. All procedures were performed by the first author (MP), an experienced and fellowship-trained orthopaedic foot and ankle surgeon. Two of the authors (TB, AB), who did not operate on any of the patients, independently reviewed register records and patient radiographs. All patients had a popliteal block placed by the anesthesia team preoperatively [2]. An intraoperative “single-shot” antibiotic prophylaxis was administered using intravenous cefazolin unless the patient had an allergy. After general anesthesia was induced, the patient was placed in the supine position. All procedures were performed with application of a thigh tourniquet. In patients with retained hardware, the hardware was removed from the most-appropriate previous incision. A 10- to 12-cm straight, midline, anterior incision was made over the distal tibia to expose the extensor retinaculum. The fused ankle was then exposed between the extensor hallucis longus and the anterior tibial tendon. Attention was paid to the anterior neurovascular bundle behind the extensor hallucis longus. The original tibiotalar joint was marked with four K-wires under fluoroscopic guidance, as planned on preoperative radiographs. Care was taken to mark the joint line at the proper level. The tibial and talar cuts were performed using cutting blocks [28]. In patients with a previously osteotomized fibula, fibular reconstruction was performed through a separate lateral longitudinal incision. Other concomitant procedures performed at the time of the conversion procedure included: fibular reconstruction (n = 2), medial malleolus reconstruction (n = 1), talonavicular fusion and medializing calcaneal osteotomy (n = 1), Achilles lengthening (n = 13), and lateral ligament reconstruction (n = 2) (Table 3) [28].

Fig. 1A–F
figure 1

A 53-year-old man underwent tibiotalar arthrodesis. (A) AP and (B) lateral views of his ankle show complete osseous fusion of the tibiotalar joint. (C) The hindfoot alignment view shows varus malalignment of the hindfoot. (D) The overall alignment of the lower extremity is neutral. His overall pain was 8 on the VAS with a maximal walking distance less than 100 m. (E) Lateral and (F) AP views of the patient’s ankle show appropriate alignment and osseous integration of both prosthesis components at 5 years followup. The overall pain was 2 on VAS with a maximal walking distance of 3–5 km.

Full size image
Table 3 Surgical details of 20 patients who underwent conversion of painful ankle arthrodesis to total ankle replacement
Full size table

After the wound was closed in a layered fashion, a well-padded short leg splint was used to support the hindfoot in a neutral position. The first dressing was removed and changed at the second postoperative day. Postoperatively, the patients were mobilized with the foot placed in a stabilizing walker for 6 weeks. Partial weightbearing with 15 kg was allowed. At 6 weeks, patients were transitioned to full weightbearing using a stabilizing walker for 2 additional weeks. After the walker was removed, physiotherapy was begun, which included active and passive ankle motion, stretching and strengthening of the triceps surae, and proprioceptive exercises [8].

Affected ankles were evaluated based on weightbearing radiographs in two planes (AP and lateral ankle views). Radiographs were evaluated at 6 weeks (nonweightbearing radiographs), 3 months, 6 months, 1 year, and then annually thereafter. Degenerative changes of the subtalar and talonavicular joints were assessed preoperatively and postoperatively using the Kellgren-Lawrence scale [19, 35]: Grade 0, no radiographic findings of osteoarthritis; Grade 1, minute osteophytes of doubtful clinical importance; Grade 2, definite osteophytes with an unimpaired joint space; Grade 3, definite osteophytes with moderate joint space narrowing; and Grade 4, definitive osteophytes with narrowing of several joint spaces and subchondral sclerosis. The angular positions of the tibial and talar components were assessed from α/β/γ angles: α angle and β angle were measured between the longitudinal axis of the tibia and the articular surface of the tibial component on the AP and lateral views, respectively; γ angle was measured between a line drawn through the anterior shield and the posterior edge of the talar component and a line drawn along the center of the talar neck on the lateral view [7, 29, 43]. The intercomponent alignment was assessed by measuring valgus-varus tilt in the coronal plane. The AP offset in the sagittal plane also was measured [4]. To assess the radiographic bone-implant interface on the tibial side, weightbearing lateral radiographs were analyzed [13, 38, 39]. Four zones were defined on the tibial side. Each zone was classified as osseointegrated (visible trabeculae at the prosthesis-bone interface), or cystic loosening (cystic lesion with a diameter at least 2 mm) [60]. Loosening of the talar component, as seen on the lateral radiograph, was defined as subsidence in the talus by more 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 [29, 37]. Patients were seen preoperatively and postoperatively in the outpatient clinic by two medical assistants (SB, SF) who had not assisted in the operations. All patients were seen postoperatively in our institution. The clinical examination involved careful assessment of ankle alignment, stability, and ROM. Ankle ROM was determined clinically with a goniometer placed along the lateral border of the leg and foot with patients in a weightbearing position as described by Lindsjö et al. [44]. Subtalar ROM also was determined clinically using a goniometer [25]. The patients rated their pain on a VAS from 0 points (no pain) to 10 points (maximal pain) [34]. For functional assessment, the American Orthopaedic Foot and Ankle Society (AOFAS) hindfoot score was calculated [36]. All patients completed the entire Medical Outcomes Study Short Form Health Survey questionnaires (SF-36) on quality of life preoperatively and postoperatively [61].

Statistical Analysis

A Kolmogorov-Smirnov normality test was performed to determine if the data were normally distributed. A paired t-test and Wilcoxon signed-rank test were used for comparison of paired normally and nonnormally distributed data, respectively. A p value of 0.05 or less was considered statistically significant. Data were analyzed using IBM SPSS® Statistics, Version 22 (IBM Corporation, Armonk, NY, USA) and SigmaPlot Version 12.5 (Systat Software Inc, San Jose, CA, USA).

Results

Two patients sustained an intraoperative medial malleolar fracture, which were treated by open reduction and internal screw fixation. Both fractures healed within 6 weeks after the initial surgery. In three patients, delayed wound healing was observed. These patients underwent daily wound care by a certified wound care specialist (SB). Two of three patients had complete wound healing within 4 weeks, whereas the third patient had final healing within 6 months of the initial surgery. No patient developed a chronic or periprosthetic joint infection. One patient had complex regional pain syndrome Type II develop postoperatively, which resolved within 6 months.

At latest followup, four patients had incomplete osseointegration (posterior quarter of the bone-prosthesis interface on the tibial side). The incomplete osseointegration was identified early after surgery (at 6 weeks followup) and was not progressive. No talar component subsidence was observed in any patient. None of the 20 patients had prosthesis loosening. Both components were neutrally aligned as measured by α angle (mean, 88° ± 1°), β angle (mean, 86° ± 2°), and γ angle (mean, 15° ± 2°). The tibiotalar intercomponent alignment was neutral in the coronal plane with mean intercomponent tilt of −0.1° ± 0.6° and in the sagittal plane with mean AP offset ratio of 0.0 ± 0.0. The osteoarthritis grade in the subtalar and talonavicular joints was comparable before the surgery and at the latest followup (Table 4).

Table 4 Degenerative changes of the subtalar and talonavicular joints assessed using the Kellgren-Lawrence scale: a comparison preoperative and postoperative scores
Full size table

Two patients had painful arthrofibrosis with reduced ROM. In both patients, an open arthrolysis and exchange of the mobile-bearing inlay were performed 1.4 and 1.5 years postoperatively. One patient had substantial tibial component medial tilt with collapse of the medial arch, this was potentially related to incomplete restoration of coronal alignment at the time of the conversion procedure. At the time of this writing, revision was pending (removal of prosthesis components and conversion to tibiotalar arthrodesis) (Fig. 2).

Fig. 2A–G
figure 2

A 67-year-old woman has a painful malunited tibiotalar arthrodesis. (A) AP and (B) lateral views of her ankle show complete osseous fusion of the hindfoot with end-stage osteoarthritis of the talonavicular joint. The patient’s initial postoperative course was uneventful. (C) AP and (D) lateral views of her ankle show appropriate alignment and osseous integration of both prosthesis components and complete osseous healing of the calcaneal osteotomy and subtalar and talonavicular arthrodesis at 2 years followup. (E) AP and (F) lateral views of the ankle and (G) the hindfoot alignment view show collapse of the medial arch and valgus tilt of the replaced ankle 6 months later. Revision surgery is pending.

Full size image

At the latest followup, mean dorsiflexion and plantar flexion were 8.5° ± 3° and 15° ± 5°, respectively. There was no correlation with the numbers available, between postoperative ROM and the time between initial ankle arthrodesis and TAR (Pearson correlation r = 0.425; p = 0.077). Mean subtalar ROM increased from 14° ± 10° to 16° ± 10° (p = 0.005), although this difference is unlikely to be clinically important. Mean VAS decreased from 9 ± 1 to 2 ± 2 (p < 0.001). Three patients reported being pain-free and 16 of 18 had a VAS of 2 or less. The mean AOFAS hindfoot score increased from 23 ± 6 to 68 ± 14 (p < 0.001). The summarized components of the SF-36 physical and mental outcome scores improved from 34 ± 5 to 74 ± 11 (range, 34–86) and from 49 ± 4 to 75.5 ± 7 (p < 0.001).

Discussion

A painful malunion or nonunion is a major complication after ankle arthrodesis [15, 47]. The surgical options for a revision arthrodesis often are limited; therefore, alternative treatment options should be discussed. Conversion of painful ankle arthrodesis to TAR is a possible salvage procedure for these patients. However, few studies have addressed the complication rate and clinical outcomes in patients who have undergone conversion of a painful ankle arthrodesis to a TAR (Table 1) [24, 27, 51]. We found that the frequency of intraoperative, perioperative, and postoperative complications was fairly high, and a substantial subset (two of 18) underwent reoperation at short-term (2-year minimum) followup. We observed durable fixation of prosthesis components with a low frequency of incomplete osseointegration around the posterior aspect of the bone-prosthesis interface on the tibial side. At the latest followup, patients reported pain relief and improved ankle ROM.

Our study has numerous limitations. Although the study design was retrospective, all patient data were collected longitudinally and transferred to the German Foot and Ankle Society’s TAR register. Second, the mean duration of followup in this study was 5 years, which may not be sufficient to detect problems that may arise during longer periods of surveillance, such as component loosening. Future studies are planned to evaluate the clinical and radiographic outcomes in this patient cohort at longer-term followup (10 years). Third, the total number of patients included in this study was relatively small. Thus, any statistical comparisons in this study should be interpreted with caution, particularly the no-difference findings, which may have been the result of insufficient statistical power. Fourth, the functional outcome in this study was partially assessed using the AOFAS hindfoot score, which is a nonvalidated score [53]. This deficiency was not known in 2007 when the clinical assessment first began. Fifth, the ankle ROM was only assessed clinically using a goniometer. Flexion and extension radiographs would be useful to determine motion through the prosthesis versus combined ankle and transverse tarsal motion. Additionally, this study highlights the experience of a single fellowship-trained foot and ankle surgeon and thus may not be generalizable. Finally, numerous other procedures were performed concurrently with the TAR in this series, and it can be difficult to discern the degree to which those procedures may have contributed to the observed results. That said, the TAR certainly was the largest of the interventions performed in any of these patients, and it seems reasonable to conclude that the TAR contributed substantially to the improvements we observed.

During the last decade, TAR has been shown to be a reliable surgical option in patients with a painful malunited or nonunited ankle arthrodesis (Table 1). The frequency of intraoperative, perioperative, and postoperative complications observed in this study is comparable to those reported by others, with respect to patients undergoing a conversion procedure (Table 1) and in patients undergoing primary TAR [10, 63]. The conversion of painful ankle arthrodesis to TAR is a technically demanding procedure [3, 28, 52, 62], and we believe it should be limited to trained foot and ankle surgeons with adequate experience performing primary TAR. In three of 20 patients, fibular reconstruction was performed owing to the previous osteotomy performed at the initial ankle arthrodesis. This condition is a possible relative contraindication for this procedure [28, 52], because the anatomic reconstruction of the mortise is important to provide the stability of the replaced ankle.

Aseptic loosening of one or both prosthesis components is the most-common reason for revision of TAR [10, 11, 33, 42, 54]. In our patient cohort, no revision resulting from aseptic loosening was performed. However, in four of 18 patients, we observed incomplete osseointegration at the posterior quarter of the bone-prosthesis interface on the tibial side. No progression of radiolucent lines was observed during the time at intermediate followup. We recommend annual followup with conventional imaging to evaluate for progressive change. In ankles in which the progression of radiolucency may be seen or suspected, we recommend performing CT for exact assessment of radiolucent lines [60]. Instability and/or subluxation is another possible reason for revision surgery in up to 9% of all TARs [54, 55]. In our patient cohort, one patient had TAR dislocation owing to collapse of the medial midfoot with consecutive valgus subluxation of the ankle.

The most-common postoperative complication in our patients was painful arthrofibrosis. Two of 18 patients sustained this complication which was treated by open arthrolysis and exchange of a mobile-bearing inlay. The frequency of arthrofibrosis in our small series seemed comparable to that previously recognized with primary TAR. Other studies suggest that arthrofibrosis occurs in up to 22% of patients undergoing primary TAR [5, 6, 14, 31, 32]. Hintermann et al. [27] reported clinical and radiographic outcomes in 28 patients who underwent 30 conversion procedures. In their study, four of the 30 (13%) patients had painful arthrofibrosis postoperatively [27]. This can be related to the degeneration of periarticular soft tissues likely resulting from ankle immobilization after ankle arthrodesis. Compromised periarticular soft tissues may be a risk factor to developing painful arthrofibrosis as this complication has been seen in patients with bleeding disorders (soft tissue damage resulting from repetitive joint bleeding) [5, 9] and in patients with revision ankle arthroplasty (soft tissue damage resulting from the previous surgeries) [30, 31].

At the latest followup, the mean ankle ROM was 23° ± 7°. This is substantially lower than reported in a recent systematic review including 7942 primary ankle arthroplasties in which a ROM improvement from 23° (95% CI, 19°–26°) to 34° (95% CI, 26°–41°) was observed [63]. Longstanding ankle fusion with consecutive degeneration of periarticular structures may be critical to achieve physiologic ankle ROM after TAR. Surprisingly, we did not observe a major correlation between the duration of ankle arthrodesis and the ROM at the latest followup. All seven patients who had an equinus-fixed deformity preoperatively had at least 5° dorsiflexion at the latest followup. The regained ROM is one possible reason for substantial postoperative improved quality of life in our patient cohort. Numerous studies including patients with TAR using modern prosthesis designs have shown a substantial postoperative improvement in quality of life [16, 48–50, 56, 58]. Another reason for the improvement of quality of life after TAR is the pain relief we observed in our study cohort. In the systematic review by Zaidi et al. [63] with 7942 primary ankle arthroplasties, the mean pooled summary VAS scores decreased from 7.4 (95% CI, 6.8–7.9) to 1.6 (95% CI, 1.4–1.8) at 4- to 5-years followup. In our patient cohort, only three of 18 patients reported being pain-free, whereas 13 of 18 patients had mild pain (VAS, 1–2). Residual pain in the hindfoot is not uncommon in patients who underwent TAR, with one study reporting 60% of patients who underwent the procedure having residual pain [23].

In our small series, the conversion to TAR was a reliable option for surgical treatment of painful ankle arthrodesis. The observed postoperative complication rate including arthrofibrosis was high, whereas functional outcomes including postoperative ROM were lower than reported for primary TAR. Further clinical studies are necessary to address long-term clinical and radiographic outcomes in this patient cohort.