|Home | About | Journals | Submit | Contact Us | Français|
Total ankle arthroplasty provides an alternative to arthrodesis for management of ankle arthritis. What is the outcome of total ankle arthroplasty implants currently in use? We conducted a systematic literature search of studies reporting on the outcome of total ankle arthroplasty. We included peer-reviewed studies reporting on at least 20 total ankle arthroplasties with currently used implants, with a minimum followup of 2 years. The Coleman Methodology Score was used to evaluate the quality of the studies. Thirteen Level IV studies of overall good quality reporting on 1105 total ankle arthroplasties (234 Agility™, 344 STAR, 153 Buechel-Pappas™, 152 HINTEGRA®, 98 Salto™, 70 TNK, 54 Mobility™) were included. Residual pain was common (range, 27%–60%), superficial wound complications occurred in 0% to 14.7%, deep infections occurred in 0% to 4.6% of ankles, and ankle function improved after total ankle arthroplasty. The overall failure rate was approximately 10% at 5 years with a wide range (range, 0%–32%) between different centers. Superiority of an implant design over another cannot be supported by the available data.
Level of Evidence: Level IV, therapeutic study. See Guidelines for Authors for a complete description of levels of evidence.
Total ankle arthroplasty (TAA) has been performed in selected patients with end-stage ankle idiopathic, posttraumatic osteoarthritis, and inflammatory arthritis since the 1970s . Initial implant designs were associated with failures requiring revision or fusion as high as 72% at 10 years , raising substantial concern about the devices.
Studies of normal ankle biomechanics and review of previous implant failures have led to the development of new TAA designs [5, 8, 10]. First-generation implants were constrained, had an all-polyethylene tibial component, and used cement for implant fixation [9, 10]. Implants used today may have either two- or three-components with either fixed or mobile bearings. Cementless fixation is considered better by most implant manufacturers and surgeons [3, 5, 7, 10, 11, 14, 19, 20, 34]. The procedure is considered by some a reasonable alternative to ankle arthrodesis , and the demand by younger and more active patients likely will increase as failure rates diminish. Therefore, it is in the interest of clinicians and patients to evaluate the outcome of current TAAs for management of ankle arthritis.
We therefore systematically reviewed the literature to determine: (1) the quality of the literature reporting outcomes of TAA; (2) the indications for TAA (eg, inflammatory arthropathy, osteoarthritis) in different centers; (3) the clinical failure rate and survivorship for different implants; (4) the methods used to salvage failures; (5) the wound complication and deep infection rates; (6) the functional outcome and the ability to participate in sports after TAA; (7) the range of motion (ROM) after TAA; (8) whether pain is eliminated; and (9) the radiographic outcome.
We conducted a literature search of MEDLINE®, Cochrane, EMBASE™, and CINAHL® databases using the terms “total” and “ankle” and “arthroplasty” or “replacement”. No language restrictions or date limits were applied to the search. In addition, the Google™ search engine and the electronic contents of several key journals were searched: The Journal of Bone and Joint Surgery (American and British Volumes), Clinical Orthopaedics and Related Research, Foot and Ankle International, Foot Ankle Clinics of North America, Journal of Foot and Ankle Surgery, and Orthopäde (German). Search of the German journals databases also was performed using combinations of the following keywords: “OSG”, “Oberes Sprunggelenk”, and “Sprunggelenkprothese”. The date of the most recent search was October 24, 2008.
We excluded articles irrelevant to TAA (Fig. 1), articles not reporting outcomes (eg, reviews, letters to the editor, biomechanical and cadaveric studies), case reports, and studies reporting results of TAA implants that had been abandoned (St Georg, ICLH [Imperial College London Hospital], Irvine, Beck-Steffe, Mayo, Newton, Bath-Wessex, New Jersey prostheses) . We also excluded data for prostheses that have been replaced by new versions of similar design but with fundamental differences as the older designs would not be relevant to current practice. The excluded devices include the LCS prosthesis replaced by the Buechel-Pappas™ , the shallow sulcus design of the Buechel-Pappas™ prosthesis replaced by the deep sulcus design , the early design of the Scandinavian Total Ankle Replacement (STAR) prosthesis for cemented implantation , and the early generation implants of the TNK prosthesis . Finally we excluded data for prostheses without documented use in the last 10 years (Thompson-Richards prosthesis) .
From each article, two investigators (NG, AK) independently extracted the year of publication, type of study (randomized, controlled trial, prospective study or retrospective case series), number of patients and ankles treated, patients and ankles available for followup, length of followup, complications (superficial wound healing problems, deep infections), and prosthesis survival. Also, ankle ROM, validated outcome scores, numbers and proportions for patients’ satisfaction (when available, although validated outcomes were not often reported), and residual pain in the ankle were recorded.
To evaluate the methods of studies reporting on TAA, we modified the score of Coleman et al. (commonly known as the Coleman Methodology Score or CMS) which initially was described to assess the quality of studies reporting outcomes of tendon disorders [6, 30] (Table 1). The CMS assesses methodology using 10 criteria, giving a total score between 0 and 100. A score approaching 100 indicates the study has a robust design and largely avoids chance, various biases, or confounding factors. The subsections that compose the CMS are based on the subsections of the CONSORT statement  (for randomized, controlled trials) but are modified to allow for other trial designs. Two investigators (NG, AK) scored the quality of the studies independently. Each investigator scored the quality of the studies twice with a 3-week interval between measurements. Intraobserver and interobserver reliability were examined. Where differences were encountered, agreement was achieved by consensus, for final data presentation (Table 2). To assess the reliability of quality scoring using the CMS, we used intraclass correlations for interobserver and the Spearman-Brown coefficient for intraobserver reliability. To compare means of CMS between the two examiners we used the Wilcoxon test. There was no difference (Wilcoxon test, p = 0.066, z = −1.84) between the mean CMS of the two examiners (71 versus 69). Intraobserver Spearman-Brown coefficient was 0.98, and the intraclass correlation was 0.98 (substantial agreement) . Disagreement occurred in four studies (one parameter in each study). After disagreements were solved by consensus, the mean CMS was calculated (71; standard deviation, 11).
Confidence intervals (95% CI) were calculated where pooling of data was appropriate. The level of statistical significance was 0.05.
We identified 13 studies [1–3, 15, 18, 20, 21, 26, 27, 30–32, 34] published from 2003 to 2008 and reporting on 1105 TAAs (234 Agility™ [DePuy Orthopaedics, Inc, Warsaw, IN], 344 STAR [Waldemar Link, Hamburg, Germany], 153 Buechel-Pappas™ [Endotec, South Orange, NJ], 152 HINTEGRA® [New Deal, Lyon, France], 98 Salto™ [Tornier, Saint Ismier, France], 70 TNK [Kyocera, Kyoto, Japan], and 54 Mobility™ [DePuy International, Leeds, UK]) with a minimum of 2 years followup. There were no randomized trials. All included studies were graded as Level IV evidence . Patients’ recruitment rate in 12 of the studies was greater than 90% [1–3, 15, 18, 20, 21, 27, 30–32, 34].
With revision, arthrodesis, or amputation as an end point, we identified 108 failures of 1105 TAAs (9.8%; 95% CI, 3.1%–16.5%). The weighted followup for all prostheses was 5.2 years (95% CI, 3.9–6.5 years). Eight studies [1, 3, 15, 18, 20, 27, 31, 34] provided Kaplan-Meier survivorship analysis data  ranging from 67% at 6 years to 95.4% at 12 years (Table 4).
Failures were salvaged with revision of the TAA in the majority of ankles (62%), whereas amputations were rare (Table 5).
Superficial wound healing complications (including superficial infections, delayed healing, and skin necrosis) were documented in 66 of 827 (8%) TAAs, ranging from 0% to 14.7% in the individual studies, and deep infections in seven of 827 (0.8%), ranging from 0% to 4.6% [1–3, 15, 18, 21, 27, 30, 31].
The American Orthopaedic Foot and Ankle Society Ankle (AOFAS)-Hindfoot score  was used most commonly to assess ankle function after TAA (Table 6). Some of the designers of ankle implants have developed their own scores (Kofoed score  and New Jersey ankle score ). Ankle scores improved after TAA in all studies (Table 6).
Ankle range of motion (ROM) as an outcome measure was documented in nine studies [1–3, 18, 21, 27, 30–32] (Table 7). Several methods have been used to measure ROM (radiographic, clinical with the patient sitting or standing). Mean postoperative ROM was equal to preoperatively  or improved by approximately 4° to 14° (Table 7) [2, 3, 30, 31]. Two studies [26, 32] investigated the ability to participate in sports after TAA. In one study , 55 of 152 patients (36%) were active in sports before surgery compared with 85 of 152 after surgery (56%). The most common activities were hiking, swimming, and cycling. In another study , 62.4% of the patients were active in sports preoperatively. This was similar to the 66% who were active after surgery. The patients participated in 3.0 ± 1.8 different sports and recreational activities preoperatively and in 3.0 ± 1.6 activities after surgery. The sports frequency remained unchanged (2.0 ± 1.6 sessions per week before TAA and 2.3 ± 1.7 postoperatively). The most common disciplines after TAA were swimming, cycling, and fitness/weight training.
Patients’ satisfaction after TAA was documented in eight studies [1, 2, 21, 26, 27, 30–32]. Naal et al.  used a visual analog scale to assess satisfaction with surgery. The mean score was 8 (± 2.5) of 10. Other authors did not use rigorously validated scales to evaluate patients’ satisfaction. They stated patients were questioned regarding their satisfaction with the outcome (Table 9).
Ten studies [1–3, 18, 21, 26, 27, 30, 34] reported radiographic evaluation of TAAs. Most studies evaluated the presence of radiolucency and prosthesis subsidence or migration (Table 10), with heterogeneity in methods used and in definitions of radiographic loosening. One study evaluated alignment of the TAA . Progression of osteoarthritis in adjacent joints was examined in two studies [18, 34]. Knecht et al.  reported progression at the subtalar joint in 22 of the 117 ankles (19%) and in 17 of 117 (15%) at the talonavicular joint, whereas Wood et al.  reported “deterioration” of subtalar joint arthritis in 15% of 95 ankles without arthritis in this joint before TAA.
Early attempts of TAA with implants have been disappointing [10, 23]; however, implant designs have evolved [5, 7, 10]. What can we learn from the literature regarding the outcome of TAA with implants currently in use?
We note numerous limitations in the literature reviewed. (1) The level of surgeons’ experience and variability in patients’ selection may have influenced results in the individual studies. (2) Heterogeneity in study design and outcome measures did not allow direct comparisons of much of the data. It therefore is not possible to show superiority of certain implants or directly compare TAA with alternative management options (eg, arthrodesis). A multicenter trial comparing the outcomes of fixed- versus mobile-bearing implants, and a trial comparing TAA with arthrodesis, would be clinically relevant. However, comprehensive cohort studies reporting on the long-term effects of interventions (eg, TAA), although not providing treatment effect estimates, are useful estimates of prognosis, can detect adverse effects and complications, and are indicative of daily clinical practice achievements . (3) The length of followup varied among studies, thus reported outcomes are not directly comparable. (4) Different scales and methodologies of assessment (patient recruitment, questionnaires, independent examiner or not) were used in different studies. (5) Clinical outcome measures frequently were not validated, whereas some TAA implant designers have produced their own outcome scales (Kofoed , New Jersey ). Results reported in the individual studies therefore could be biased. (6) Patient satisfaction was not assessed using rigorous validated methods. (7) Definitions of the radiographic variables used in the assessment were not identical in different studies, and the radiographic examinations were not always standardized.
We evaluated the quality of studies using the CMS [6, 29]. The substantial interobserver and intraobserver agreement is indicative of the reliability of the CMS, although formal validation was not performed. The reader can easily compare the total score, with the maximum possible of 100 points, to get an impression of the study quality. It shares some similarities with the STROBE (Strengthening of Reporting of Observational trials in Epidemiology)  guidelines (study design, type and size, data collection, and recruitment of participants), although these are not a study quality scoring tool. They were developed to provide a checklist and recommendations that could aid authors to conduct and present observational studies .
Rheumatoid arthritis was the primary indication for TAA (reported rates of 39% and 37.5%) in two previous meta-analyses [12, 28]. Our data (Table 3) showed trauma was the leading cause (34%), with a wide range (range, 12%–73%) in reports from different centers. This may reflect extension of the indications by some surgeons.
An overall 9.8% of ankle replacements required revision or conversion to ankle fusion at 5.2 years. The wide CI (CI range, 3.1%–16.5%) shows inconsistency in the presented data from individual studies, and should be interpreted with caution. A meta-analysis comparing TAA with ankle arthrodesis  that included studies published from 1990 to 2005 reported a TAA survival rate of 78% (95% CI, 69.0%–87.6%) at 5 years and 77% (95% CI, 63.3%–90.8%) at 10 years. The data in the current investigation are not directly comparable to those by Guyer and Richardson , as more recent publications have been included and different methods for data analysis were used in the two studies. Another meta-analysis , which reviewed 18 studies on mobile-bearing prostheses published from 1997 to 2002, found the weighted survival rate was 90.6% at 5 years. This is comparable to the survival rate of mobile-bearing implants in our study (Table 4). TAA survivorship data, however, should be interpreted with caution. Results from the prosthesis’ inventors can be biased and may reflect the higher familiarity with the implant. Knecht et al.  included the surgeries performed by the designer of the Agility™ prosthesis, evaluated by independent authors, and reported a 95% survival rate at 6 years, whereas others achieved only 67% . Similarly, the designer of the STAR  reported a 95% survivorship rate at 10 years, whereas an independent high-volume surgeon  was reported to have a survivorship rate of 80% at 10 years. Others  reported a better survival rate in their latter 31 TAAs compared with the initial 20. The Swedish Joint Register , possibly representing more closely the average surgeon’s outcomes, reported a 77% survival rate. Their data  showed the 5-year survival rate increased from 70% before to 86% after the surgeon had performed 30 TAAs. The designer of the Buechel-Pappas™ prosthesis reported a 92% survivorship rate at 12 years in 75 TAAs with the newer, deep sulcus implant. These results were reproduced by an independent surgeon , however, in patients with rheumatoid arthritis (low demand). Differences therefore may be symptomatic and reflect the surgeon’s familiarity with the procedure, or selection of patients, rather than the effect of the intervention and the implant.
Comparing functional outcomes of different implants requires caution because of the different methodologies used, as described earlier. Haddad et al.  reported the mean AOFAS score was 78.2 points, which is comparable to the reported outcomes in our study (Table 6). Two studies [26, 32] suggested participation in certain sports is possible after TAA. It is not known, however, whether this is advisable and how it would affect failure rates in the long term.
The improvement in ankle ROM was relatively small (0°–14°) (Table 7). This is in agreement with the results of others [12, 28]. Our patients therefore should be informed preoperatively, improvement in ankle motion is not one of the expected benefits from TAA.
Furthermore, residual pain after TAA is relatively frequent (23%–60%) (Table 8), whereas the methodologic flaws in assessing patients’ satisfaction in the individual studies raise concerns regarding the high satisfaction rates reported (Table 9).
Current TAAs improve ankle function; however, residual pain is common and wound complications can occur. The overall failure rate is approximately 10% at 5 years with a wide range from different centers.
Each author certifies that he or she has no commercial associations (eg, consultancies, stock ownership, equity interest, patent/licensing arrangements, etc) that might pose a conflict of interest in connection with the submitted article.
An erratum to this article can be found at http://dx.doi.org/10.1007/s11999-010-1229-4