Between August 1988 and February 1998, twenty-one patients (twelve
male and nine female) with a failed ankle arthrodesis underwent revision with
use of the Ilizarov apparatus. The average age at the time of the Ilizarov
procedure was forty years (range, twelve to sixty-three years). One patient
had rheumatoid arthritis, one had human immunodeficiency virus, and the other
nineteen patients had an unremarkable medical history. The mean duration of
follow-up from the time of the surgery was 83.4 months (range, twenty-six to
144 months) (see Appendix).
The original diagnosis that led to the primary fusion was posttraumatic
arthrosis in nine patients (43%), painful deformity of the hindfoot or
forefoot in seven (33%), osteonecrosis of the talus in two (10%), ankle
instability in one (59%), primary osteoarthritis in one, and a neuropathic
ankle joint in one. The primary fusion had been performed at the tibiotalar
joint in eight patients and at the tibiotalar and subtalar joints in eleven;
two patients had undergone pantalar arthrodesis.
Before the Ilizarov procedure was performed, each patient had undergone an
average of three operations (range, one to five operations) to salvage the
failed ankle fusion. The primary fusion had been attempted with various
techniques: internal fixation had been used in sixteen patients; external
fixation, in three; minimal internal and external fixation, in one; and
bone-grafting and cast immobilization, in one. Four patients had undergone at
least one unsuccessful operation (range, one to five operations) for revision
of the failed arthrodesis. The Ilizarov procedure was the first attempt at
revision in the other seventeen patients (81%). The average interval between
the primary fusion attempt and the Ilizarov reconstruction was 104 months
(range, ten to 480 months).
The indications for the Ilizarov procedure were malunion at the site of an
ankle fusion in eleven patients, aseptic nonunion in eight (six mobile and two
stiff nonunions), and infected nonunion in two. Of the eleven patients with a
malunion, seven had an equinovarus deformity, three had an equinovalgus
deformity, and one had the ankle fused in an excessive calcaneus and varus
position.
The patients had a mean of 2.2 coexisting pathologic conditions (range, one
to four conditions) (Table I).
Eight patients had been offered amputation as the only alternative before they
presented to our center.
Technique
All patients were treated with the basic principles of the Ilizarov
technique. The construction of the frame was individualized for each patient
to address the site of the failed arthrodesis and the coexisting problems.
Malunion
The eleven patients with malunion were treated with a supramalleolar
osteotomy. Eight patients underwent gradual correction, and three underwent
acute correction-compression. Eight patients underwent single-level
lengthening (by means of a distal osteotomy in eight and a proximal osteotomy
in one), and two underwent double-level lengthening. Two patients with
malunion and coexisting chronic osteomyelitis of the tibia, proximal to the
malunion, underwent radical resection of the dead bone, acute shortening, and
a distal tibial osteotomy for correction of the malunion before the
lengthening.
Nonunion
Six patients with an aseptic mobile nonunion underwent excision of the
nonunion and resection of the bone ends to achieve acute angular correction
and maximum bone contact for compression of the fusion. Two patients with a
hypertrophic stiff nonunion were treated with gradual distraction. Six
patients in this group also underwent a proximal tibial osteotomy for limb
lengthening. Two patients with an infected nonunion underwent thorough
resection of the necrotic bone, acute angular correction and shortening, and a
proximal tibial osteotomy for lengthening during the same procedure.
In addition to limb-length discrepancy and infection, eighteen other
concomitant pathologic conditions were addressed by adjusting the Ilizarov
frame as needed; this increased the total number of simultaneously treated
deformities to forty (Table
II). Furthermore, two patients underwent tarsal tunnel release,
two underwent lengthening of the Achilles tendon, and one underwent neuroma
excision during the same surgery. A supplemental bone graft was used at the
arthrodesis site in nine patients; two of the bone grafts were obtained from
the iliac crest, and seven were obtained locally from the fibula and
tibia.
The circular Ilizarov frame was used for fixation in all patients. In one
patient, a retrograde tibial nail was also used, and tibial lengthening was
performed over the nail. After application of the Ilizarov external fixator,
all patients were permitted to bear weight as tolerated while using crutches.
Although full weight-bearing was allowed, most patients limited their
weight-bearing until partway through the consolidation phase. After the
lengthening osteotomy was performed, a latency period of one week preceded the
beginning of distraction. The initial distraction rate ranged between 3/4 to 1
mm per day. At the end of the distraction phase, the timing of fixator removal
was based on the radiographic appearance of the bone at the distraction site
and the healing of the ankle arthrodesis site. Once the fixator was removed,
all patients except one wore a cast or brace for additional protection.
Assessment
Evaluation of the results consisted of a retrospective review of the
records and subjective and objective clinical evaluations, including
radiographic and gait analyses.
Record Review
The records of the twenty-one patients were reviewed to determine the
reason for the primary fusion, the diagnosis (malunion, nonunion, or infected
nonunion), concomitant problems, and the amount of motion of the subtalar and
midtarsal joints preoperatively. Details of the revision surgery (the
osteotomy site, the frame construction, and additional procedures), the
postoperative course, and the complications were also recorded.
Clinical Interview
Eighteen patients were able to return for clinical examination. Three
patients were unable to return for formal follow-up and were interviewed by
telephone. All patients were questioned regarding the severity and location of
any pain, the distance that they were able to walk, the type of shoes that
they usually wore, their ability to walk up and down stairs, their ability to
run and to participate in recreational activities, their current employment
status, and whether they were satisfied with the result.
Physical Examination
During the physical examination, special attention was paid to the
patient's gait pattern, any signs of infection, the position of the hindfoot
and forefoot, and the motion of the subtalar and midfoot joints. With the
patient standing, the angle between the long axis of the lower limb and the
axis of the sole of the foot in the sagittal plane and the angle between the
long axis of the lower limb and the axis of the heel of the foot in the
frontal plane were measured with a goniometer. The thighfoot angle was
measured with the patient prone.
Radiographic Analysis
The preoperative radiographs of twenty patients were reviewed. At the time
of final follow-up, standard standing anteroposterior long leg radiographs,
standing anteroposterior and lateral radiographs of the ankle, and
anteroposterior and lateral radiographs of the foot were made for every
patient. In addition, radiographs to assess hindfoot alignment were made with
the technique of Saltzman and
el-Khoury32 for
fifteen patients. To assess the union of the arthrodesis site, tomograms were
made for seventeen patients and oblique radiographs of the arthrodesis site
were made for the remaining four.
A series of measurements was made on the radiographs. On the lateral
radiographs of the ankle, we measured the distance from the lateral process of
the talus to the mid-diaphyseal axis of the tibia and the angle subtended by
the intersection of the longitudinal axis of the tibia and the weight-bearing
surface of the foot. On the hindfoot alignment radiographs, we measured the
angle between the longitudinal axis of the calcaneus and the mid-diaphyseal
line of the tibia. On the anteroposterior radiographs of the ankle, we
measured the angle between the longitudinal tibial axis and the lateral border
of the talus.
Evaluation
The results were divided into two categories—bone results and
functional results—according to Paley's previously published
classification33.
Bone Results
Four criteria were evaluated to assess bone results: union, infection,
deformity, and limb-length discrepancy. The criteria for solid union included
a clinically stable ankle that was painfree with manipulation and
weight-bearing as well as radiographic evidence of bridging trabeculae without
a change in the position of the ankle. The bone result was considered to be
excellent when there was solid union of the arthrodesis site and all
additional osteotomy sites; no infection; a neutral or slight calcaneus
position (=5° of dorsiflexion), neutral to 5° of valgus, external
rotation of 0° to 15°, neutral or slight posterior translation, and no
deformity; and a limb-length discrepancy of =1.5 cm. The result was
considered to be good when there was solid union as well as no infection;
slight equinus (=5° of plantar flexion), 5° to 10° of valgus or
varus, =5° of internal rotation, or =1 cm of anterior translation;
and a limb-length discrepancy of =3 cm. The result was considered to be
fair when there was union as well as no infection; 5° to 10° of
plantar flexion or dorsiflexion, 5° to 10° of varus or >10° of
valgus, internal rotation of the foot of >5°, anterior translation of
>1 cm, or moderate forefoot deformity; and limb-length discrepancy of >3
cm. If, at the end of treatment, nonunion or infection or malunion equal to or
worse than that seen preoperatively remained, the result was considered to be
poor.
Functional Results
The functional results were rated on the basis of four criteria: pain, gait
abnormality, daily activities including employment status, and walking
distance. The functional result was considered to be excellent if the person
was active, had no pain, had no or only a slight limp, had returned to his or
her previous job or its equivalent, and was able to walk for more than six
blocks. The functional result was considered to be good if the person was
active, had mild pain necessitating only occasional analgesics, had a mild
gait abnormality, had changed his or her job to a more sedentary one because
of the ankle, and was able to walk for three to six blocks. The functional
result was considered to be fair if the person had pain that necessitated
regular analgesics and prevented daily activities, was unable to work, or was
unable to walk more than three blocks because of the ankle problem. If a
person had constant pain and was unable to walk, the result was considered to
be poor.
Gait Analysis
We performed gait analysis for eighteen of the twenty-one patients at the
time of clinical review. We employed two techniques. The first involved use of
an in-shoe foot sensor manufactured by Tekscan (Boston, Massachusetts). This
technology involves use of F-Scan foot sensors, which are very thin (0.007 in
[0.178 mm]) and provide 960 individual sensing points. Each sensor is
calibrated to the patient's body weight. Sensors are connected to a personal
computer, and data are analyzed with use of software for pressure distribution
throughout the foot during walking.
For the second technique of gait analysis, we used a GAITRite mat (CIR
Systems, Clifton, New Jersey), which automates the measurement of time and
distance parameters (stance time, swing time, step length, and velocity) of
walking. The mat contains six sensor pads. The total active recording zone for
the mat is 144 in (365.8 cm) long and 24 in (61 cm) wide. As the patient walks
over the walkway, the computer connected to the mat captures the relative
geometry of the footfalls. The GAITRite software collects the raw footfall
data and computes the time and distance parameters of walking. The testretest
reliability rate of the GAITRite mat is 0.970 to
0.95034.
Union
Union was achieved in all patients, although one patient required
second-stage bone-grafting of the nonunion site. There was radiographic
evidence of consolidation at an average of 6.5 months (range, 2.5 to 10.5
months) after the operation. The eleven patients in whom malunion had been
treated with a supramalleolar osteotomy had solid union of the osteotomy site
at an average of 6.2 months (range, 2.5 to 11.5 months). The ten patients in
whom nonunion had been treated either with excision of the ununited bone and
acute compression or gradual distraction followed by compression had solid
union at an average of 7.7 months (range, 4.5 to 15.5 months). The Ilizarov
frame remained in place for an average of 7.8 months (range, 2.5 to eighteen
months). The nine patients who had undergone only distal tibial corticotomy
for correction of both the pathologic abnormality of the ankle and the
limb-length discrepancy had the Ilizarov frame in place for an average of 7.2
months, whereas the eleven patients who had undergone both distal and proximal
osteotomies (the latter for lengthening) had the frame in place for an average
of 8.2 months. The duration of the external fixation averaged 2.5 mo/cm in the
group of patients who had undergone distal osteotomy, 2.5 mo/cm in the group
who had undergone proximal osteotomy, and 2.1 mo/cm in the group who had
undergone double-level osteotomy.
Pain and Satisfaction
Sixteen patients (76%) were satisfied with the result, two were somewhat
satisfied, and three were dissatisfied. Twelve patients (57%) had no pain, and
seven (33%) had mild, occasional pain that did not necessitate analgesics. Two
of the three dissatisfied patients continued to have chronic disabling pain
and required long-term analgesic medication. The radiographic measurements of
all of the evaluated parameters for these two patients were within normal
values. Signs of tarsal tunnel syndrome developed in one of these two
patients; it was alleviated by a reduction in the rate of distraction, but it
never resolved completely. It is possible that the patient was experiencing
residual pain of neurogenic origin. The other patient with disabling pain
refused to return for reexamination, so the source of the pain remained
unknown. The third dissatisfied patient had reactivation of chronic
osteomyelitis in the midpart of the tibia, far from the ankle arthrodesis
site. One patient had no pain at the arthrodesis site but had a very sensitive
surgical scar.
According to our criteria, fifteen patients (71%) had an excellent
functional result; three, a good functional result; two, a fair functional
result; and one, a poor functional result (see Appendix). Ten patients (48%)
had an excellent bone result; nine (43%), a good bone result; one, a fair bone
result; and one, a poor bone result. With the numbers available, no
significant difference was found between the final outcome for the patients
with malunion and that for the patients with nonunion (p = 0.477).
Infection
Twenty patients showed no evidence of active infection at the time of final
follow-up. In one patient with a history of mid-tibial osteomyelitis, which
was not active at the time of the Ilizarov procedure, the infection was
reactivated at the original site.
Motion and Alignment
A plantigrade limb was achieved in each patient; the average angle formed
by the tibial axis and the weight-bearing surface of the foot was 88°
(range, 86° to 91°). Radiographically, ten patients had a neutral
hindfoot in the coronal plane, four patients had mild valgus (<5°),
three patients had mild varus (<5°), two patients had >5° of
valgus, and two patients had >5° of varus. The average tibiocalcaneal
angle, as measured on fifteen hindfoot alignment radiographs, was 0°
(range, 5° of varus to 13° of valgus). The average tibiotalar angle,
as measured on six anteroposterior radiographs of the ankle, was 2° of
valgus (range, 5° of varus to 12° of valgus).
Nineteen ankles were fused an average of 3 mm posteriorly (range, 4 mm
anteriorly to 9 mm posteriorly). Two ankles were fused in 11 mm of anterior
translation without compromising the final result. No ankle was fused so that
the foot was in internal rotation. The average amount of external rotation was
12° (range, 8° to 15°).
Only seven patients in this series had presented with satisfactory subtalar
and midfoot motion before the Ilizarov procedure, either because the
arthrodesis had been performed at more than one level or because the long
period of immobilization had led to a fibrous ankylosis. Five of these
patients retained the same motion after the Ilizarov procedure, and the other
two had almost no motion at the subtalar level and only a few degrees of
midfoot motion at the time of final follow-up. Of the fifteen patients who had
an excellent functional result, only three had normal motion of the subtalar
joint and only five had flexible midtarsal joints.
Employment Status, Shoe Wear, Walking Distance, and Recreational
Activities
One patient had retired before undergoing the Ilizarov procedure. Of the
remaining twenty patients, seventeen returned to their previous employment,
one changed his employment status to a more sedentary job, and two were not
able to work. At the time of the last follow-up, fifteen patients presented
wearing regular shoes and six wore shoes that had some modification. Fifteen
patients were able to walk more than six blocks, three were able to walk two
to three blocks, and one was not able to walk outdoors. The patient with
rheumatoid arthritis was not able to walk without the use of a cane. The
patient who had undergone amputation on the contralateral side wore a
prosthetic limb and was not able to walk more than five blocks. Twelve
patients were able to participate in recreational activities, including
limited running, aerobics, and rowing.
Limb-Length Discrepancy
The goal of treatment was for the operatively treated limb to be 1 to 1.5
cm shorter than the normal limb. This goal was achieved after the Ilizarov
procedure in sixteen patients and after a reoperation in two more patients.
Two patients had a final limb-length discrepancy of >1.5 cm (3 and 3.7 cm);
each compensated well with a shoe-lift. One patient had had an amputation on
the contralateral side as a result of childhood meningococcemia, so
limb-length discrepancy was not a treatment consideration.
Gait Analysis
Our pedobarograph (F-Scan) results showed that all eighteen patients
included in the gait analysis walked with a heel-to-toe gait. The transition,
or time taken, from heel-strike to foot-flat was notably shortened on the
fused side in all patients. Patients with a mobile midfoot tended to achieve a
smoother transition in a shorter time from foot-flat to toe-off than did
patients with a stiff midfoot. These observations were assessed qualitatively,
and the time difference was not significant (p = 0.23).
No patient had a normal transition of the center of pressure during the
stance phase (from the heel to the lateral border of the foot to the fifth
metatarsal head to the first metatarsal head to the great toe). In most
patients, the center of pressure went from the heel to the second and third
metatarsals to the great toe.
Our GAITRite examination of the time and distance parameters showed that
twelve of the eighteen patients were able to walk with normal velocity for
their age, height, and gender. All twelve patients had an excellent or good
result. The singlelimb-stance period on the fused side was within 3% of that
on the normal side for these patients, and the step length of the normal limb
was within 5 cm of that on the fused side. Of the six patients who did not
have a normal walking velocity, four had a single-limb-stance time within 6%
of normal. The walking velocity of the remaining two patients was reduced by
>30% compared with normative values, and both of those patients showed
substantial asymmetry in the single-limb-stance times. These two patients also
walked with a substantial limp. The cause of the gait dysfunction was thought
to be neuropathic pain in one of these patients and recent femoral surgery in
the other.
All of the patients used an external rotation foot-progression angle when
walking. The patients with a stiff midfoot had a significantly greater
external rotation foot-progression angle (mean, 10°) than did the patients
with a mobile midfoot (mean, 5°) (p = 0.018)
(Figs. 1-A through 1-E). Five
of the eighteen patients wore a shoe with a rocker-bottom sole on the side of
fusion to facilitate stance-phase progression. We also assessed the patients
while they walked with and without shoes on the GAITRite mat. Without shoes,
seventeen of the eighteen patients walked with significantly reduced velocity,
greater stance asymmetry, and a limp (p = 0.034).
Complications
From the day of the surgery until the removal of the Ilizarov frame,
forty-one minor complications (treated conservatively) and twenty major
complications (treated surgically) were recorded. Seventeen additional
operations were performed in twelve patients
(Table III). Nine of those
operations were performed on five patients with a history of infection.
After completion of the treatment and removal of the circular frame, seven
complications were recorded and four additional operations were performed. One
patient with preexisting malalignment of the distal part of the ipsilateral
femur and limb-length discrepancy underwent corrective osteotomy with
simultaneous acute shortening of the ipsilateral femur. One patient with a
limb-length discrepancy of 6.5 cm after the Ilizarov procedure underwent acute
shortening of the contralateral femur. One patient with distal tibial valgus
malalignment and a limb-length discrepancy of 3 cm underwent double-level
tibial osteotomies for correction and lengthening. One patient underwent
excision of a recurrent neuroma.
The high complication rate associated with primary ankle fusion
emphasizes that ankle arthrodesis often fails to achieve the patients'
requirements and may lead to or be associated with other abnormalities. Late
reconstruction following a failed ankle arthrodesis can be complicated by the
presence of infection, osteopenia, osseous defects, limb-length discrepancy,
and hindfoot and forefoot
malalignment15,19,25,30,35.
A question the surgeon often faces when he or she treats a patient with a
failed ankle arthrodesis is whether an effort to salvage the limb is indicated
or whether amputation is a more realistic or practical option. MacKenzie et
al. and the LEAP Study
Group36 reported
that factors thought to influence the decision regarding limb salvage or
amputation include injury severity, physiologic reserve, and patient
characteristics. They concluded that the severity of soft-tissue injury has
the greatest impact on the decision regarding whether to salvage the limb or
amputate. In our study, the soft tissues were intact with normal plantar
sensation. Amputation was therefore not considered to be the best option for
treatment.
The aims of revision ankle arthrodesis are to obtain solid union; to
correct any malposition of the ankle, hindfoot, and forefoot; and to address
coexisting deformities. Once these goals are achieved, a pain-free ankle, an
almost normal gait pattern, and a foot capable of wearing a regular shoe can
be expected. There is sound evidence to suggest that revision can achieve a
satisfactory long-term clinical result without deterioration of
function3,4,37-39.
The method of treatment following a failed ankle arthrodesis should be
individualized primarily on the basis of two factors: the pathologic condition
at the arthrodesis site (nonunion, malunion, and/or infection) and the
concomitant problems. The three-dimensional nature of the Ilizarov frame
renders the technique extremely versatile and capable of accommodating complex
abnormalities of the tibia, ankle, and foot. This offers the surgeon the
option of a more comprehensive approach, particularly in cases in which
alternatives are limited.
An atrophic mobile nonunion can be treated with resection of the nonunion
and acute correction with compression of the bone ends. A hypertrophic stiff
nonunion can be treated with gradual distraction osteogenesis. In this series,
union of all ununited ankle fusion sites was achieved with use of the Ilizarov
apparatus in conjunction with distraction, open reduction, and/or
bone-grafting.
Varus or valgus malunion combined with equinus, which are the most common
deformities to occur after a failed ankle arthrodesis, can be corrected by
supramalleolar osteotomy. Posterior translation of the foot can also be
achieved with the same osteotomy. By moving the foot posteriorly, the surgeon
reduces the anterior lever arm of the foot, providing a mechanical advantage
during the second part of the stance phase of walking, when the tibia must
move over the foot to allow the body to progress
forward40.
Limb-length discrepancy can be addressed with either a distal or a proximal
corticotomy, followed by distraction osteogenesis combined with compression at
the arthrodesis
site33,41.
The position of the forefoot is as important as the position of the ankle and
hindfoot after ankle arthrodesis, and an inability to obtain the appropriate
forefoot position can compromise the final result. Foot deformities should be
corrected simultaneously to allow appropriate axial load transmission through
the arthrodesis site and to provide proper foot-ground contact. Half the
patients in this study presented with major forefoot deformities. Such
deformities can be addressed with either soft-tissue distraction or midfoot
osteotomy if fixed secondary bone changes have
occurred33,41.
The optimum foot position after ankle arthrodesis has been the subject of
much debate. The majority of surgeons suggest fusion in neutral flexion, with
avoidance of extremes of plantar flexion or equinus (an optimum position of
<5° of either), but some surgeons recommend as much as 20° of
plantar flexion13.
There is less controversy regarding alignment in the coronal plane, with
almost all surgeons agreeing that varus malposition should be
avoided3,6,10,11,14,26,38,39,42-44,
recognizing that excessive valgus or varus malposition alters the biomechanics
of the foot45.
Residual motion at the subtalar and Chopart joints after ankle arthrodesis
has also been the subject of several studies, with most authors confirming a
significant decrease in
eversion-inversion4,22,38,46
and an average of 15° of motion (mainly plantar flexion) of the midtarsal
joints14,38,39,41,47,48.
Flexibility of the hind-foot and midfoot is a well-recognized compensatory
mechanism for minor malposition of the
ankle13, but its
influence on the final functional result is still unclear. Duquennoy et
al.37, Mazur et
al.4, and Ottolenghi
et al.49 stated
that the final results depend mainly on the existing tibiopedal motion and not
on anatomic restoration, whereas Buck et
al.43 and Jackson
and Glasgow18 found
no correlation between the degree of foot motion and the final result. In this
series, nearly 70% of the patients presented with a stiff hindfoot and
midfoot; after the Ilizarov procedure, only five had flexible feet.
Accurate positioning of the ankle is more critical in patients with a stiff
foot because of the inability to compensate through the subtarsal and
midtarsal joints. Any malalignment will create problems, both proximal and
distal to the fusion. It seems that fusion of the ankle in neutral or a slight
calcaneus position is advantageous for the stance phase of walking when
midfoot motion is
restricted3,5.
A neutral or slightly valgus position of the hindfoot avoids weight-bearing on
the lateral border of the foot, decreasing the risk of lateral foot pain and
calluses. In addition, external rotation of the foot allows normal knee
motion, avoiding the need for external rotation of the knee during the stance
phase, which can lead to laxity of the medial collateral ligament. Finally, a
slight amount of limb-shortening is desirable to facilitate toe clearance
during the swing phase of
gait1.
The ability to correct the position of the hindfoot and forefoot by
adjusting the frame as needed during the regeneration phase is a unique
advantage of the Ilizarov method. It allows the surgeon to address any
intraoperative errors or early postoperative loss of position.
The complications that occurred in this series were consistent with the
complications reported in other studies of patients treated with the Ilizarov
circular
frame30,31,41,
with pinsite infections occurring most frequently. Patients with a history of
infection had higher complication rates, and the frame needed to remain in
place longer in such patients than it did in patients with no history of
infection (10.9 months compared with 6.8 months). Infection, however, does not
seem to have a detrimental effect on the final result. This finding is
consistent with those of other
authors50,51,
suggesting that a salvage procedure should be undertaken even in the presence
of ongoing infection.
In conclusion, revision with use of the Ilizarov technique following a
failed ankle arthrodesis is a worthwhile limb-sparing procedure. Even though
the complexity of the problem increases the complication rate and the time
needed in the Ilizarov frame, the final outcome can be satisfactory. The final
functional result is closely associated with the hindfoot and forefoot
position and with resolution of associated abnormalities.