The Implant
The Agility total ankle prosthesis (DePuy) consists of a titanium tibial
component with sintered titanium beads and a cobalt-chromium talar component
with sintered cobalt-chromium beads. A modular polyethylene insert locks into
the tibial component. In this series, the minimum thickness of the
polyethylene liner varied from 3.73 to 4.70 mm, depending on the implant size.
Fusion of the distal tibiofibular syndesmosis is desired to stabilize the
tibial component.
Initially, the tibial and talar components were both titanium. Beginning in
1989, with the twenty-first treated ankle, a cobalt-chromium talar component
was introduced after two talar components showed evidence of loosening. Also,
starting with the twenty-first ankle, the tibial component was made thicker,
after two components had broken. For the patients in this study, three sizes
of implants were available at the time of the operations.
Operative Technique and Patient Demographics
A single surgeon (F.G.A.) performed all of the operations in this study.
Details of the operative technique and postoperative protocol have been
described
previously22.
Between July 18, 1984, and October 10, 1994, 126 consecutive patients
received 132 total ankle replacements to treat end-stage arthritis of the
ankle joint that was refractory to nonsurgical treatment. The diagnosis in the
majority of the ankles (sixty-one; 46%) was posttraumatic arthritis. The
remaining diagnoses were primary osteoarthritis in thirty-eight (29%),
rheumatoid-inflammatory arthritis in thirty-one (23%), and postinfectious
arthritis in two (2%). The average age of the patients at the time of surgery
was sixty-one years (range, twenty-seven to eighty-three years), and the
average weight was 85.5 kg (range, 47 to 123 kg).
Of the 132 consecutive total ankle replacements, thirtysix implants, as
best as we can determine, were in thirty-three patients who died. None of the
ankles in the patients who died had been revised or had an arthrodesis. Of the
remaining ninety-six ankles, fourteen had been revised or had an arthrodesis
and one leg had been amputated because of unrelated causes, leaving a total of
eighty-one ankles in seventy-eight living patients. We were able to follow
sixty-nine replacements (85%) in sixty-seven patients. None of the sixty-seven
patients had a Workers' Compensation claim related to the ankle.
Subjective Evaluation
Our Institutional Review Board on Protection of Human Subjects approved the
study, and consent was obtained for each patient enrolled. In the present
study, we reviewed all 132 ankles and included the findings of the previous
sensory testing of the entire group. All contact with the patient and
collection and evaluation of the data were conducted independent of the
operating surgeon.
The Ankle Osteoarthritis Scale, a validated and reliable disease-specific
self-administered instrument, was used to assess the clinical outcomes. It
consists of a pain-specific subscale and a disability-specific
subscale23. We also
asked six more specific questions used previously with this patient cohort to
assess overall
satisfaction22. In
the initial follow-up of this group performed five years earlier, we reported
the results of the first 100 consecutive ankles, but, at that time, we had
collected patient-derived, clinical, and radiographic data on the first 132
ankles.
Of the eighty-one intact ankle replacements in seventy-eight living
patients, sixty-eight ankles (85%) were followed clinically with use of a
validated ankle osteoarthritis scale and a short questionnaire. Each item of
the Ankle Osteoarthritis Scale was graded from 0 (best) to 10 (worst), and
subscales for pain and disability were independently generated. These were
then combined and averaged to yield an overall ankle osteoarthritis score.
Radiographic Evaluation
We evaluated the radiographs of all patients who had adequate studies
available. Of the 132 ankles, 117 (89%) had radiographic follow-up for a
minimum of two years. This included seventy-seven of eighty-one ankles in the
living subjects, twenty-seven of thirty-six ankles in the patients who had
died, and thirteen of the fourteen ankles prior to revision or arthrodesis.
The average duration of radiographic follow-up was 4.6 years (range, two to
13.5 years) for the patients who had died and eight years (range, two to
sixteen years) for the living patients. Overall, this yielded an average of
7.2 years of radiographic follow-up. Most patients had annual sets of
radiographs available for review. When they were available, anteroposterior,
mortise, weight-bearing lateral, and maximum dorsiflexion and plantar-flexion
lateral radiographs were reviewed. Two methods of evaluation were used.
Digitized images were systematically analyzed with use of an image analysis
program written in PV-Wave 6.21 (Visual Numerics, San Ramon, California).
These data helped to determine which components showed progressive angular
changes and/or migration over time. These images were then checked manually
with use of the technique and evaluation previously
described22
(Figs. 1-A and 1-B).
Component subsidence was judged on the lateral radiographs. The position of
the horizontal surface of the component was related to the level of the tibial
or talar bone resection scar. An assessment of linear subsidence over time was
assessed on sequential postoperative radiographs. Care was taken not to
confuse osseous overgrowth with settling of the component.
A subset of patients had good-quality non-weight-bearing lateral
radiographs that had been made with the ankle in maximum dorsiflexion and
plantar flexion at the time of the final follow-up. The arc of component
motion was assessed by measuring the change in the angulation of a line drawn
along the inferior edge of the talar component in relation to the longitudinal
axis of the tibia (angle C) (Fig.
2).
Angular migration of the component was defined as a shift of at least
5° in any one of the three component angles described
previously22. This
criterion is based on previous reliability
testing22 and was
chosen to provide a reasonable and reproducible evaluative measure, taking
into account the variability of the radiographic technique, beam angles, and
rotation.
The radiographs were also assessed with respect to the time until
syndesmosis fusion; the presence and healing of postoperative fractures; the
presence and development of hindfoot arthritis, according to the radiographic
arthritis classification system of Kellgren and
Moore24; and
peri-implant bone loss. The tibial component was divided into six zones,
extending from medial to lateral on the anteroposterior and mortise
radiographs, and three zones, extending from anterior to posterior on the
lateral radiograph, for the purposes of describing the presence of
periprosthetic lucencies and
lysis22
(Figs. 3-A and 3-B). A lucency
was defined as a radiolucent line of =2 mm in width, and lysis was defined
as a radiolucent area of >2 mm.
In distinction to the previous
report22, we
further defined lysis by its stability or progression over time and by its
characteristic radiographic appearance. Hence, we refer to lysis as being
either "mechanical" or "expansile" in nature. We
believe these two variants represent very different modes of peri-implant bone
loss. What Pyevich et
al.22 previously
called "ballooning lysis" occurred reliably at the implant-bone
interface along the lateral margin of the tibial component when syndesmosis
fusion was delayed. Typically, this focal bone loss occurred in the first six
to twelve months after implantation and stabilized thereafter. Presumably,
this may represent an unfavorable microenvironment for bone ingrowth and, in
fact, may lead to focal bone resorption secondary to abnormal motion through
the interface between the tibial component and the lateral malleolus. We now
redefine this phenomenon as "mechanical" lysis
(Fig. 4). Cessation of bone
loss and stabilization of the lesion typically coincided with definitive
radiographic evidence of fusion of the tibiofibular syndesmosis.
In contrast to this mode of bone loss, we found a subset of ankles that
presented with late, expansile bone loss that occurred without a specific
anatomical location. By its appearance and later onset, we presume that this
mode of bone loss is secondary to a wear-particle inflammatory reaction
similar to that found in the peri-implant interfaces of hip and knee
prostheses, and we refer to it as "expansile" lysis
(Fig. 5).
Furthermore, we subcategorized each form of lysis as being progressive or
stable, with respect to the change in the lesion over time on sequential
radiographs. If a lesion showed continued enlargement up to the most recent
radiograph, it was deemed to be progressive. If there were no apparent changes
in a lesion over a one-year follow-up period, it was considered to be
stable.
Statistical Methods
Descriptive statistics were calculated for the various clinical outcomes.
Kaplan-Meier survivorship curves for revision and revision to an arthrodesis
were estimated. The overall ankle osteoarthritis score and its pain and
disability subscale scores for the intact total ankle replacements were
compared with those of age-matched controls with use of an extension of the
Kruskal-Wallis test on ranks accounting for strata (age-group). The
association of these scores with lysis, syndesmosis union or nonunion, and
other radiographic findings was tested with use of either the Wilcoxon
rank-sum test or the Kruskal-Wallis test. The Fisher exact test and the
chi-square test were used to test the association of the short questionnaire
with lysis, syndesmosis union or nonunion, and other radiographic findings.
This analysis was also used in the tests of the association of subsidence with
lucency, lysis, and syndesmosis union or nonunion. The log-rank test was used
in the analysis of time to syndesmosis fusion and its association with angular
change (<5° versus =5°) and subsidence (<5 versus =5 mm).
All of the statistical analyses were performed with use of SAS/STAT (SAS
version 8.2, 2001; SAS Institute, Cary, North Carolina).
Revisions, Failures, and Secondary Procedures
At an average duration of nine years of follow-up, the 132 ankles required
fourteen major revision procedures and thirty-two minor secondary operations.
Major revisions were considered any procedure requiring removal or replacement
of one or both of the metal components. Other secondary procedures included
any procedure to the foot or ankle related to the ankle replacement.
The overall rate of major revision was 11% (fourteen of 132 ankles).
Figure 6 shows the survivorship
curves, with 95% confidence intervals, for the entire cohort with revision or
ankle arthrodesis as the end point. Seven ankles had been revised to a new
total ankle replacement, and seven ankles had an arthrodesis. One patient
underwent a below-the-knee amputation 7.5 years after total ankle replacement
because of unrelated complications from a failed arterial stent placed for
vascular insufficiency in the affected leg. Five implants failed secondary to
impaction or settling, four failed because of loosening and migration, and two
failed because of tibial component fracture. One talar component was revised
because of malalignment, and one failure was due to a deep infection. One
patient who had a revision was lost to follow-up, and no information was
available regarding the revision (Table
I).
Eight ankles had a major secondary procedure during the five years since
the last follow-up study of this cohort. Six ankles had an arthrodesis, and
two had a revision. One revision involved the tibial component only, and the
other involved both components.
Three patients had postoperative wound complications that resolved with
local wound care and oral antibiotics. Three ankles had a late deep infection,
two of which were treated with débridement alone and one was treated
with component resection and arthrodesis. One patient's postoperative course
was complicated by a deep venous thrombosis with documented pulmonary embolus
without any long-term sequelae.
Many patients had secondary (nonrevision) procedures on the ipsilateral
foot or ankle. As best as we can determine, the 132 ankles had thirty-two
secondary operations, which included a total of thirty-eight procedures. The
most common procedure was syndesmosis screw removal, which was done in
seventeen ankles. Three triple arthrodeses, three subtalar arthrodeses, and
two calcaneal osteotomies were performed. In four ankles, curettage and
bone-grafting of the medial malleolus was done because of an expansile lysis.
In two ankles, the Achilles tendon was lengthened percutaneously because of
dorsiflexion deficits; one lengthening was combined with débridement of
the anterior part of the joint. Two ankles underwent polyethylene exchange at
the time of other minor procedures. The first was during a subtalar
arthrodesis, at which time the polyethylene was noted to have fractured. The
second was during a débridement and grafting of a talar lytic area, in
the presence of a well-fixed talar component. At the time of that surgery, the
polyethylene component was noted to have a small, anterior chip-defect and was
exchanged. Two ankles were débrided because of a deep infection; two,
because of superficial infection; and one, because of lateral gutter
impingement.
Clinical Outcomes
On careful sensory examination of seventy ankles and feet, we found
postoperative sensory deficits related to the incision in fifteen feet (21%).
Sensory deficits were found in the distribution of only the superficial
peroneal nerve in six patients, in the distribution of only the deep peroneal
nerve in three patients, and in the distribution of both nerves in six
patients.
Ankle Osteoarthritis Scale
The overall Ankle Osteoarthritis Scale score was calculated from the
combined results on the pain and disability subscales. The mean overall score
(and standard deviation) was 2.55 ± 2.31 in comparison with an
age-matched control score of 0.1 ±
0.1823; the
difference was significant (p < 0.0001). The mean pain score was 2.02
± 2.24 compared with 0.09 ± 0.17 for age-matched
controls23; the
difference was significant (p < 0.0001). The mean disability score was 3.36
± 2.90 compared with 0.09 ± 0.21 for age-matched
controls23. This
difference was also significant (p < 0.0001).
Additional Questions
In response to the specific questions that were asked of these patients in
the previous follow-up study five years earlier, 94% (sixty-two of sixty-six
who responded) reported decreased pain, 92% (sixty of sixty-five) were
satisfied with the outcome of the surgery, 94% (fifty-eight of sixty-two)
would have the surgery again, and 97% (sixty-two of sixty-four) would
recommend an ankle replacement to a friend. Overall, 83% (fifty-five of
sixty-six) reported continued functional improvement. Forty-six (73%) of
sixty-three reported occasional or no pain at all.
At the time of the present review, twenty-eight (41%) of sixty-eight
patients who responded to our questionnaire used assistive devices. Of the
twenty-eight patients, five (18%) used assistive devices for problems related
only to the affected ankle; ten (36%) reported needing an assistive device for
multiple reasons, which would include the total ankle arthroplasty; and nine
(32%) reported the use of an assistive device for problems entirely unrelated
to the total ankle arthroplasty. Four patients (14%) were unable to categorize
the reasons for the use of assistive devices. Of the twenty-eight patients who
used an assistive device, twelve (43%) used a cane, nine (32%) used a walker,
five (18%) used a wheelchair, and two (7%) used an ankle brace. Of the five
who used an assistive device solely for the ankle, one used a brace, three
used a cane, and one alternated between a cane and crutches.
Radiographic Findings
Hindfoot Arthritis
The subtalar and talonavicular joints were evaluated for radiographic signs
of arthritis with use of the system of Kellgren and
Moore24 (grades 1
to 5) at an average follow-up interval of 7.2 years (range, two to fourteen
years). In an effort to define the time course and progression of the disease,
we reviewed the radiographs sequentially. In this cohort, prior to ankle
replacement, 100 (85%) of 117 ankles had grade-1, 2, or 3 findings in the
subtalar joint, while seventeen (15%) had grade-4 or 5 arthritis. At the time
of the latest follow-up, seventy-eight (68%) had grade-1, 2, or 3 findings and
thirty-eight (32%) had grade-4 or 5 arthritis. Preoperatively, 102 ankles
(87%) had grade-1, 2, or 3 findings at the talonavicular joint compared with
fifteen (13%) that had grade-4 or 5 arthritis. At the time of the last
follow-up, eighty-three ankles (71%) had grade-1, 2, or 3 findings of the
talonavicular joint and thirty-three (29%) had grade-4 or 5 findings. With use
of a minimum two-grade advancement in the Kellgren-Moore scale as the
criterion for substantial arthritic progression, twenty-two (19%) of the 117
ankles with complete radiographic follow-up had arthritic progression at the
subtalar joint and seventeen (15%) had arthritic progression at the
talonavicular joint (Table
II).
Union of the Distal Tibiofibular Syndesmosis
Union of the syndesmosis was assessed on serial radiographs for 111 of the
117 ankles (no early radiographs were available for assessing union of the
syndesmosis in six patients). The time to union was determined as the point at
which there was trabecular bridging across the syndesmosis with signs of
remodeling. Fifty-six (51%) of 111 ankles had union of the syndesmosis in six
months or less, forty-six (41%) had union after six months (delayed), and nine
(8%) persisted as a nonunion (Fig.
7). The average time to union of the syndesmosis among the ankles
that eventually had a successful union was ten months (range, two to sixty
months).
Range of Motion
At the time of the final follow-up, we were able to obtain high-quality
radiographs with the ankle in maximum plantar flexion and maximum dorsiflexion
for thirty-three patients. The average arc of motion was 18° (range,
2° to 40°), with average plantar flexion of 19° (range,
—1° to +36°) and an average dorsiflexion to neutral (range,
—24° to +16°).
Lysis and Lucency
Eighty-nine (76%) of the 117 total ankle arthroplasties had radiographic
signs of lucency. Thirty-three (37%) of the eighty-nine ankles had lucency at
the tibial fin in anteroposterior zone 3, while eighteen (20%) had
circumferential lucency. Forty-nine (42%) of all 117 ankles showed some
evidence of mechanical lysis, with seven (14%) of the forty-nine showing
progressive worsening of this problem over time. Eighteen (15%) of all 117
ankles had expansile lysis. Four of them showed definite signs of progression
over time, and fourteen appeared relatively stable on serial radiographs.
Expansile lysis was a late-onset entity, appearing, on the average,
thirty-five months (range, nine to eighty-five months) after the
arthroplasty.
Subsidence
Overall, sixteen ankles had =5 mm of subsidence and/or =5° of
angular change at an average of nine years of follow-up. Five ankles had only
tibial subsidence (average, 8 mm; range, 7 to 8 mm). Nine ankles had only
talar subsidence (average, 10 mm; range, 5 to 20 mm). No ankle had concurrent
subsidence of both components. Eight ankles had angular changes of =5°
(average, 15°; range, 5° to 25°), and six of them had concomitant
subsidence of =5 mm (five on the tibial side and one on the talar side)
(Figs. 8-A and
8-B).
Associations Among Findings
Short Questionnaire
A significant and strong relationship was detected between the patients'
assessments of the ankle pain (p = 0.012), functional limitations of the ankle
(p = 0.013), and recreational limitations (p = 0.023) and the time to a
successful syndesmosis fusion. The presence of anterior tibial lysis in
lateral zone 1 was predictive of more pain as well (p = 0.016).
Pain Score
Tibial mechanical lysis in anteroposterior zone 6 was predictive of a
higher pain score (p < 0.05). Circumferential lucency had a weak
association with higher pain scores (p = 0.077). A delay in syndesmosis fusion
and syndesmosis nonunion predicted a higher pain score (p = 0.016); however,
with the numbers available, no apparent difference was detected between
nonunion and delayed union.
Disability Score
Higher disability scores were associated with the presence of tibial
mechanical lysis in anteroposterior zone 6 (p < 0.05). Disability scores
were lowest in patients with no lysis in anteroposterior zone 6 (p = 0.014).
Delayed union and nonunion of the syndesmosis were associated with higher
disability scores (p < 0.001). With the numbers available, no difference
was found between nonunion and delayed union. Evidence of lysis on the lateral
radiograph did not significantly correlate with disability scores.
Subsidence
Subsidence of the tibial component was associated with the presence of
circumferential lucency (p = 0.007). There was a trend toward an association
between the presence of lucency at the tibial fin in anteroposterior zone 3
and tibial subsidence (p = 0.057). Subsidence was also associated with
evidence of progressive tibial lysis on any anteroposterior or lateral zone (p
< 0.033 for all). Tibial lysis in anteroposterior zones 1 and 6 was
associated with the development of circumferential lucency (p < 0.0001 and
p = 0.002, respectively). Ankles with progressive tibial lysis in
anteroposterior zones 2 through 5 tended to have greater subsidence compared
with those with no lysis (p = 0.053) and those with stable lysis (p = 0.051).
At the time of the most recent follow-up, the presence of expansile lysis was
not associated with component subsidence. Three of five ankles with tibial
component subsidence of >5 mm had a tibiofibular syndesmosis nonunion. The
other two had delayed union of the syndesmosis at eleven and twenty-nine
months, respectively. No ankles with talar component subsidence of >5 mm
had a syndesmosis nonunion (average healing time, twelve months; range, three
to twenty-nine months). Four of eight ankles with an angular change of
=5° had a syndesmosis nonunion. The other four ankles with large
angular changes had an average time to syndesmosis union of twenty-one months
(range, eleven to twenty-nine months). Of the fourteen ankles that had a
revision or arthrodesis, seven (50%) had =5 mm of subsidence or =5°
of angular change.
Syndesmosis Healing
Ankles with a successful fusion of the syndesmosis showed, on the average,
<2 mm of tibial and talar subsidence, and no difference was found between
those with delayed or early union. Overall, nine ankles had nonunion of the
tibiofibular syndesmosis. Four of them had angular changes of >5°
(average, 15°; range, 7° to 25°), and three had tibial subsidence
of >5 mm (average, 8 mm; range, 7 to 8 mm). The average time to syndesmosis
fusion was 9.7 months (range, two to sixty months) for the ninety-six ankles
with <5° of angular change or <5 mm of subsidence and 13.6 months
(range, three to twenty-nine months) for the twelve ankles with =5° of
angular change or =5 mm of subsidence (p = 0.004). Ankles with a
syndesmosis nonunion had significantly more subsidence compared with those
with a successful fusion (p = 0.023). Circumferential lucencies were also
associated with delayed healing of the syndesmosis (p < 0.0001). A delay in
syndesmosis union predicted the appearance of tibial lysis in any zone (p <
0.0001) and mechanical lysis in zone 6. The most recent radiographs showed
evidence of progressive mechanical lysis in seven of the nine ankles with a
syndesmosis nonunion, in contrast to only two (4%) of forty-six ankles with a
delayed union and no ankles with successful healing of the syndesmosis within
six months.
The number of screws (one or two) placed across the syndesmosis was not
associated with the end point of fusion (p = 0.169). However, the time until
successful syndesmosis fusion when only one screw was employed (forty ankles)
was significantly longer (median, ten months; interquartile range, five to
eighteen months) than that when two screws (seventy-seven ankles) were used
(median, six months; interquartile range, four to twelve months) (p =
0.05).
This review of the first 132 Agility total ankle replacements performed in
one surgeon's practice is the continuation of the study performed by Pyevich
et al.22, with the
addition of five more years of follow-up and thirty-two additional patients.
As with the previous study, patient satisfaction with the procedure remained
high. Since the last follow-up, the revision rate increased by 5%, with seven
ankles converted to an arthrodesis and seven converted to another Agility
ankle implant.
In the previous study, questions were raised regarding the clinical
relevance of radiographic signs of lysis and migration in patients who seemed
to be doing well at the time. With five additional years of radiographic and
clinical data, it seems that certain radiographic signs are more troubling
than others. The ankles that had suggestive signs of component instability or
loosening according to our criteria (migration of =5 mm or =5°)
seemed to have a less favorable clinical outcome and more often needed
revision or arthrodesis. Other problematic radiographic markers were the
presence of progressive lysis in any zone, circumferential lucency, and
anteroposterior zone-3 lucency. These factors were also associated with higher
pain and disability scores. Other zonal lucencies, expansile lysis, and stable
mechanical lysis were not associated with subsidence. Talar component
subsidence was ultimately more common and more likely to be progressive than
was tibial component subsidence. The current survivorship curves for the talar
component are of concern, and with further time we expect to have more
failures because of subsidence. The tibial component appeared to be more
stable, especially after syndesmosis fusion, and was not tending to continue
to subside with further follow-up.
Pyevich et al. defined an entity that was typically found in
anteroposterior zone 6 as "expansile or ballooning
lysis."22 In
the present study, the appearance of the implant-bone interface in
anteroposterior zone 5 and anteroposterior zone 6 was linked to the status of
the syndesmosis fusion. Typically, serial radiographs made over the period of
syndesmosis healing prior to fusion demonstrated a slowly progressive widening
of the space between the tibial implant and the bone, with the development of
a sclerotic cortical margin along the medial aspect of the lateral malleolus.
Invariably, this phenomenon ceased to progress when union of the syndesmosis
occurred. In the cases of delayed union or nonunion of the syndesmosis, the
focally progressive lysis along the lateral bone-implant interface likely
comes from persistent micromotion between the distal aspect of the fibula and
the tibial component. Although, in our experience, the distal part of the
fibula does not grow into the tibial component, these conclusions support the
concept that the distal part of the fibula is an important lateral support
buttress for the tibial component. In contradistinction to this early
postoperative lysis, we found less distinct, expansile lytic defects appearing
at approximately three years after surgery in several ankles. The appearance
of this type of lysis was consistent with the inflammatory bone loss
associated with the generation of wear particles, as is found in hip and knee
arthroplasty. At the time of the latest follow-up, progressive expansile lysis
was an unusual finding and had no apparent clinical correlations. However, it
remains a concerning factor that may be problematic in the future, as is the
persistent talar component settling.
These patients were among the first to have an Agility ankle replacement.
It is important to note that several changes that may lead to more predictable
and improved outcomes have been implemented since this study was initiated.
There is a wider array of implants available, including six component sizes
(rather than the three sizes used in this study) and a polyethylene spacer
that is 2 mm thicker. The so-called footprints of both components have been
enlarged with the intention of increasing osseous support and reducing
subsidence, especially on the talar side. Improved techniques of addressing
the syndesmosis are now incorporated into the surgery with the more common use
of fibular plating and multiple screw fixation.
Most follow-up reports on ankle fusion have described the prevalence of
painful hindfoot arthritis after ankle fusion as >50% at an average of
eight
years25-27,
with the prevalence increasing to nearly 100% after twenty
years28. Compared
with arthrodesis, total ankle arthroplasty with the Agility design appears to
have reduced the prevalence of radiographically apparent hindfoot arthritis.
Using strict radiographic criteria in the patients with adequate radiographs
and retained implants, we found that <25% of our patients had progressive
hindfoot arthritis at an average of 7.2 years following total ankle
arthroplasty.
In conclusion, the results of ankle replacement at an average follow-up
period of nine years are encouraging, and patient satisfaction with the
procedure remains high.