Is End-Stage Ankle Arthrosis Best Managed with Total Ankle Replacement or Arthrodesis? A Systematic Review

Introduction. End-stage ankle osteoarthritis is a debilitating condition. Traditionally, ankle arthrodesis (AA) has been the surgical intervention of choice but the emergence of total ankle replacement (TAR) has challenged this concept. This systematic review aims to address whether TAR or AA is optimal in terms of functional outcomes. Methods. We conducted a systematic review according to PRISMA checklist using the online databases Medline and EMBASE after January 1, 2005. Participants must be skeletally mature and suffering from ankle arthrosis of any cause. The intervention had to be an uncemented TAR comprising two or three modular components. The comparative group could include any type of ankle arthrodesis, either open or arthroscopic, using any implant for fixation. The study must have reported at least one functional outcome measure. Results. Of the four studies included, two reported some significant improvement in functional outcome in favour of TAR. The complication rate was higher in the TAR group. However, the quality of studies reviewed was poor and the methodological weaknesses limited any definitive conclusions being drawn. Conclusion. The available literature is insufficient to conclude which treatment is superior. Further research is indicated and should be in the form of an adequately powered randomised controlled trial.


Introduction
End-stage ankle osteoarthritis is a debilitating condition that results in functional limitations and a poor quality of life [1]. The incidence of symptomatic osteoarthritis of the ankle has been estimated to be 47.7/100,000 in the United Kingdom [2]. More than 70% is related to posttraumatic osteoarthritis [1,3] with the majority of the remainder being primary arthritis, inflammatory arthritis, or secondary to osteonecrosis. Mild to moderate ankle arthritis can often be managed with ankle foot orthoses and a rocker-bottom shoe [4]. When surgical intervention is indicated, ankle arthrodesis (AA) has traditionally been used. However, the development of total ankle replacement (TAR) has challenged arthrodesis as the treatment of choice for ankle arthrosis [5][6][7].
AA has been shown to give good results in multiple papers [8][9][10]. Various techniques have been used which include cannulated screws, plate fixation, retrograde nail, and external fixation. The main drawback of the procedure is the high rate of later arthrosis in adjacent joints, reported between 10 and 60% [11][12][13]. This results in a subsequent subtalar fusion rate at 5 years of 2.8% which is higher than the 0.7% after TAR [14]. Additional complications associated with AA include wound infection (3-25%), nonunion (10-20%), and malalignment [15,16].
TAR is still evolving [17] but has the potential advantage to preserve range of motion, restore gait, and, thus, protect adjacent articulations [18]. The results from first generation ankle prosthesis were disappointing with stiffness, wound complications, loosening, malalignment, impingement, and instability cited as causes of failure [7,[19][20][21]. Newer prosthesis is uncemented, comprises two or three modular components, and has improved outcomes. A recent systematic review reported 89% survival of 7942 TARs at 10 years and 2 Advances in Orthopedics  an improvement in American Orthopaedic Foot and Ankle Society score (AOFAS) from 40 to 80 at a mean of 8.2 years of follow-up [22]. However, a further review from Gougoulias et al. of 1105 total ankle arthroplasties reported a 10% failure rate at 5 years with residual pain in 27%-60% of cases [23]. Haddad et al. performed a systematic review of the literature up until 2005 comparing outcomes following second generation TAR and AA. However, the majority of the 49 studies reviewed were single centre case series and none directly compared the two treatments. They concluded that both techniques gave comparable intermediate and longterm outcomes, reporting good or excellent outcomes in 68.5% of TAR and 67% of AA patients and mean AOFAS scores of 78.2 following TAR and 75.6 following AA. However, variability of reporting outcomes and lack of controlled studies restricted direct comparisons of the two groups [24]. The aim of this paper is to perform an up-to-date systematic review of the literature including only comparative studies addressing whether TAR or AA is optimal in the treatment of end-stage ankle arthrosis in terms of functional outcomes.

Methods
We conducted a systematic review of the literature using the online databases Medline and EMBASE. The review was performed and reported according to the PRISMA checklist. The search strategy used for the Medline search is shown in Table 1 and this was modified for searching EMBASE. The searches were carried out on January 15, 2014, and limited to papers available in English. The search was limited to papers published after January 1, 2005, as the previous systematic review published in 2007 included papers prior to this date [24].
Inclusion criteria were applied. Participants must be skeletally mature and suffering from ankle arthrosis of any cause deemed severe enough to warrant either TAR or AA by the treating surgeon. The intervention had to be an uncemented TAR comprising two or three modular components. The comparative group could include any type of ankle arthrodesis, either open or arthroscopic, using any implant for fixation. The study must have reported at least one functional outcome measure. Studies were excluded if no comparative group was analysed or if the study participants were divided into more than two groups. In addition, only primary research was considered for review with any abstracts, comments, review articles, and technique articles excluded. Eligibility of studies was assessed independently by two authors (R.J. and G.C.) who also appraised the included studies against the STROBE statement [25]. If there was any disagreement between the authors in assigning a score to each paper appraised, a third independent reviewer (A.C.) made the final decision.

Results
The Medline search revealed 88 and the EMBASE search 126 results. Figure 1 shows a flow diagram of the review process including the reasons for exclusion at different stages of the process. Concise details of the included studies are given in Table 2 and the appraisal against STROBE statement [25] is given in Table 3. The reasons for exclusion of the 10 articles at full paper review stage are given in Table 4.
Of the four studies included, three were level III retrospective comparative studies and the other was a level II prospective comparative study. The optimal study design when comparing two treatment modalities is a randomised controlled trial (RCT) as this allows for minimisation of bias. The main limitation that is common to all four studies reviewed is the lack of randomisation; this risks selection bias   n/a n/a n/a n/a 13 (a) Report numbers of individuals at each stage of study-for example, numbers of potentially eligible, examined for eligibility, confirmed eligible, included in the study, completing follow-up, and analysed. Main results 16 (a) Give unadjusted estimates and, if applicable, confounder-adjusted estimates and their precision (e.g., 95% confidence interval). Make clear which confounders were adjusted for and why they were included. n/a n/a n/a n/a (c) If relevant, consider translating estimates of relative risk into absolute risk for a meaningful time period.
n/a n/a n/a n/a Other analyses 17 Report other analyses done-for example, analyses of subgroups and interactions and sensitivity analyses.
n/a n/a n/a n/a 6 Advances in Orthopedics Discussion Key results

18
Summarise key results with reference to study objectives.

Study
Reason for exclusion [32] Expert opinion Conley et al., 2012 [33] No functional outcome recorded Flavin et al., 2013 [34] No functional outcome recorded Hahn et al., 2012 [35] No functional outcome recorded Krause and Schmid 2012 [36] No functional outcome recorded Piriou et al., 2008 [18] No functional outcome recorded Rouhani et at., 2011 [37] Study participants divided into more than two groups Rouhani et al., 2012 [38] Study participants divided into more than two groups Slobogean et al., 2010 [39] No functional outcome recorded SooHoo et al., 2007 [14] No functional outcome recorded with the uneven allocation of confounding factors between the groups. In addition, as none of the studies reviewed defined a primary outcome measure or included a power calculation, uncertainty is present as to whether any of the studies was sufficiently powered to show a significant difference in any recorded outcome measure.

Schuh et al., 2012.
Schuh et al. [26] performed a retrospective review of 63 patients that underwent either TAR, using a HINTEGRA prosthesis (Newdeal SA, Lyon, France), or AA, using 3 cannulated screws. No significant difference was found between the two groups in terms of activity level, participation in sport, or American Orthopaedic Foot and Ankle Society (AOFAS) hindfoot score. The surgery was performed by a single, fellowship trained surgeon between 1998 and 2006. Of the 63 patients assessed, only 41 were included for analysis. The 16% loss to follow-up forms a high proportion and the majority (70%) of these were in the AA group. The reason these patients refused follow-up is not known and their exclusion may have skewed results. Minimal details of the inclusion criteria are stated and whether patients with arthrosis of any cause were included is not clear. Clarity is required to decide whether results are applicable to other populations. In addition, no mention is given as to whether assessments were performed by an independent observer or the postoperative physiotherapy regimen was consistent between the two groups. . This is a pragmatic approach but the outcomes of different prostheses may vary and this is not accounted for in the study. Comparison between the two groups is limited by their differences at baseline and complexity of surgery performed. The fusion group was younger, had a lower proportion of rheumatoid arthritis, and had a much higher proportion of low complexity surgery (87% versus 32%); any positive results may result from these differences rather than the type of surgery received.

Discussion
Four studies were identified and reviewed which addressed our research question. Two of the four studies reported statistically significant improvements in functional outcomes following TAR [27,29]; the other two studies showed no differences between the two groups [26,28]. However, the methodological flaws present stop definitive conclusions being drawn. The main limitation in design common to all studies was the lack of randomisation. This risks differences in the study groups being present at the point of treatment allocation which has the potential to affect results. Only one study described the indications for the two procedures [28], using severe deformity or instability, poor ankle motion, no or mild adjacent joint arthritis, and younger age as indications of arthrodesis. Therefore, the arthrodesis group in this study will have had a higher proportion of patients with these factors than the TAR group, which all may have influenced outcomes. The other three studies [26,27,29] do not describe the indications used for allocation but it is likely that baseline characteristics also differed in these studies, limiting the ability to directly compare outcomes between the two groups. A further limitation of the systematic review is the use of pragmatic entry criteria for the intervention and comparator groups, with all arthrodesis procedures and any uncemented ankle replacement included. The four studies included used five different uncemented ankle prostheses and four fixation methods for arthrodesis including open and arthroscopic techniques. The inclusion of numerous surgical techniques restricts the generalisability of the results to any specific technique as it is likely that the outcomes following each method differ.
Although at least one functional outcome was measured in each study, the evidence supporting the validity, reliability, and responsiveness of the available measures following foot and ankle surgery is limited [30,31]. The AOFAS is the most commonly used outcome measure following foot and ankle surgery [30,31] but it relies on the observer to measure both range of motion and malalignment, risking observer bias. The AOS is patient reported which reduces the risk of observer bias and has been validated in ankle osteoarthritis, but it is not validated in the measurement of outcome following AA or TAR. Therefore, neither of the two functional measures provides an ideal measurement tool and may have contributed to inaccurate recording of outcomes. Future research should utilise an outcome tool that both is patient reported and is validated for use in this postoperative population.
At present, the evidence is insufficient to change clinical practice. Arthrodesis has been the traditional treatment of choice and will continue to be so until the literature definitively demonstrates one modality to be superior to the other. The increased rate of complications reported following TAR further supports this approach [28]. Ideally, future research will be of RCT design so that selection bias is limited and effects measured can be attributed to treatment allocation. The population treated needs to be clearly defined so that results can be applied to the readers' practice. They should have a patient reported outcome as the defined primary outcome and the study should be adequately powered to show a statistically significant difference between the groups.

Conclusion
Although half of the reviewed studies report some functional improvement following total ankle replacement, the lack of high quality evidence limits a definitive conclusion being drawn. Insufficient evidence is available to decide whether total ankle replacement or ankle arthrodesis improves functional outcomes and further research in the form of robust RCTs is indicated.