Ankylosis of the hip is defined as total loss of hip motion.
Whether it is a spontaneous or postoperative condition, ankylosis
provides a durable, painless, and stable hip. However, in the long
term, especially when the hip is in a poor functional position,
ankylosis of the hip can be responsible for pain and degenerative
changes in the lower back and the knee. Conversion to a total hip
arthroplasty has been recommended to relieve the overstresses in
the neighboring joints in order to stop the degenerative process1,2. However, conversion of an ankylosed
hip to a total hip replacement is a challenging problem. Because of
the initial disease, the effect of previous operations on both bone
and soft tissues, and the atrophy of the periarticular muscles,
this procedure is technically demanding. Currently, there are few
reports on total hip arthroplasty in ankylosed joints1-7. The mid-term results have been
promising, but the long-term results have been less satisfactory,
with complications and failures being more frequent than they have
been after primary hip replacement8-10.
Furthermore, postoperative motion and stability of the hip have
been poor compared with those after conventional arthroplasty.
The goal of the present retrospective study was to evaluate the
clinical and radiographic outcome of total hip arthroplasty for
the treatment of an ankylosed hip. The minimum duration of follow-up
was five years. The results in the neighboring joints were examined,
and the factors likely to influence the functional results were
analyzed.
Patients
From January 1969 to December 1993, forty-five consecutive ankylosed
hips in forty-five patients were managed with a total hip arthroplasty
performed by the senior one of us (M.K.). Hips with ankylosing spondylitis
or fibrous ankylosis were excluded from this study. No other ankylosed
hips were excluded. All hips included in our series were soundly
fused, as determined on both clinical and radiographic examination.
There were twenty-six women and nineteen men. The mean duration
of ankylosis prior to the arthroplasty was 35.7 14.2 years
(range, three to sixty-five years). The mean age of the patients
at the time of the index hip replacement was 55.8 12.3 years (range,
twenty-eight to eighty years). The mean weight and height were 61.5
10.8 kg (range, 40 to 84 kg) and 164.8 9.5 cm (range, 140 to 182
cm), respectively. Nineteen right hips and twenty-six left hips
were involved. The ankylosis had been spontaneous in twenty hips
and postoperative in twenty-five. The arthrodesis had been intra-articular
in eleven of the twenty-five hips with postoperative ankylosis,
extra-articular in ten, and a combination of both techniques in
four. The initial diagnosis was tuberculosis in twenty-six hips,
developmental hip dysplasia in nine, hematogenous bacterial infection in
five, osteoarthritis in three, Perthes disease in one, and unknown
in one. None of the hips in which ankylosis developed after infection
had biopsies performed prior to the operation. Medical histories,
clinical examination, and laboratory tests (C-reactive protein level
and erythrocyte sedimentation rate) were used to rule out the possibility
of a persistent or recurrent infection. The mean number of previous
operations on the ankylosed hip was 0.96 0.14 (range,
zero to three). The involved limb was usually shorter than the contralateral
limb. The mean limb-length discrepancy was 3.6 2.2 cm (range, 0
to 8 cm).
The correct functional position for an ankylosed hip in this
study was considered to be 10° to 20° of flexion, 0° to 10° of adduction,
and 0° to 10° of external rotation. Malposition was defined as a variation,
in any plane, of at least 10° from the correct position and was
noted, usually in a combination of planes, in thirty-six hips.
All patients had a perceptible contraction of the abductor muscles
on palpation.
The indication for conversion was isolated pain in the lower
back in twenty-two patients, isolated pain in the ipsilateral knee
in seven patients, and pain in both the back and the knee in fifteen
patients. One patient complained of a walking disability related to
excessive fixed flexion of the hip. Progressive pain in the lower
back typically was not radicular. It was associated with mild degenerative
changes of the lumbar spine in twenty patients, severe arthritis in
six patients, and scoliosis in eleven patients. Knee pain was associated
with mild-to-severe unicompartmental femorotibial arthritis in fourteen patients
and with tricompartmental arthritis in seven patients (four of whom
had severe arthritis and three of whom had mild arthritis). The
remaining patient had no radiographic signs of arthritis. Two patients
had had a successful total hip arthroplasty on the contralateral
side three and fifteen years before the conversion. Three patients
had severe symptomatic primary osteoarthritis in the contralateral
hip.
The mean duration of follow-up for the entire series was 8.5
3.4 years (range, five to 20.8 years), with a median duration of
follow-up of 7.6 years. No patient was lost to follow-up. Five patients
died of unrelated causes at a mean of 6.2 years (range, five to
eight years) postoperatively.
The implants used in the present series included a Charnley-type
prosthesis in five patients, a McKee-Farrar prosthesis in one, and
a Charnley-Marcel Kerboull prosthesis in the remaining thirty-nine patients
(all prostheses were manufactured by Benoist Girard, Howmedica,
Herouville Saint Clair, France). Both components were inserted with CMW
type-1 bone cement (DePuy 1; DePuy, Exeter, Devon, England).
Operative Technique
The operative technique has been described in detail elsewhere11. A lateral approach with a standard
trochanteric osteotomy was routinely used. The dissection of the trochanteric
area was often difficult because of dense fibrous scar tissue. Before
the osteotomy of the greater trochanter, the anterior and posterior parts
of the gluteal muscles were dissected. The status of the gluteal
muscles was systematically assessed with respect to their thickness
and color. Any internal fixation device was first removed. The anterior,
posterior, and superior parts of the articular capsule, which were
usually strongly adherent to the femoral neck and head, were completely removed.
Multiple specimens of the capsule were routinely obtained for culture
at the time of the operation. The femoral neck and the site of the fusion
between the ilium and the femoral head were clearly visualized.
The site and direction of the section were located on the femoral
neck as planned preoperatively on outlined transparent templates. All
hips were assessed to determine whether an osseous bridge was present
between the ischium and the femur. If a bridge was found, it was
osteotomized first. The osteotomy of the femoral neck was performed
with use of an oscillating saw or an osteotome. After dislocation
of the hip and resection of the inferior aspect of the capsule and
adjacent fibrous tissue, a hooked retractor was placed distally
in the obturator foramen to widely expose the acetabular cavity
filled by the femoral head fused to the ilium. The primary landmark
to identify was the inferior margin of the acetabulum. After excision
of bone or fibrous tissue, which usually covered the inferior margin
of the acetabulum, this landmark was always recognized. In hips
that had had spontaneous ankylosis or an extra-articular arthrodesis,
the transverse ligament of the acetabulum was present. A hemispherical
bone cavity was prepared with the use of curved gouges, until the preoperatively
chosen cup could be inserted. The cup was anteverted 10° to 15°,
and its inferior rim was tangential to the inferior margin of the
acetabulum. In this respect, the socket was placed 40° to 45° to
the horizontal, irrespective of the position of the pelvis. The
femur was routinely prepared. However, before the femoral component
was cemented, a trial reduction was carried out, and the possibility of
fixing the greater trochanter in its correct position was assessed.
These procedures were important, especially when limb-lengthening
was >2 cm. If reduction was impossible, or if the greater
trochanter could not be securely fixed in its correct position with <25°
of abduction, the femoral osteotomy was revised. However, in order
to retain the insertion of the iliopsoas muscle the osteotomy was never
performed distal to the lesser trochanter.
Postoperative Care
Postoperatively, patients received anticoagulation therapy, systemic
antibiotics, and nonsteroidal anti-inflammatory drugs (ketoprofen;
100 mg/day) to prevent heterotopic ossification. Immediately
after the operation, passive motion exercises of the involved joint
were begun, with the assistance of a therapist, and were continued
until active motion of the hip was possible. The patients were free
to walk with two supports after three days. Full weight-bearing
was usually allowed after six weeks. In hips with ankylosis related
to tuberculosis, specific antituberculous agents were prescribed
for six to nine months to avoid reactivation of the tuberculosis12,13.
Methods of Evaluation
Clinical and radiographic evaluations were performed at six weeks,
three months, six months, one year, and then every one or two years.
Function of the hip was rated according to the grading system of
Merle d’Aubigné14.
The hip score was then classified as excellent (18 points), very
good (17 points), good (16 points), fair (15 points), poor (14 points),
or bad (£13 points). Pain and osteoarthritis in the neighboring joints
were evaluated at the time of each follow-up examination.
Serial anteroposterior radiographs of the pelvis were analyzed.
The position of the socket relative to the horizontal and vertical
teardrop line, the inclination angle of the cup, and the presence
and progression of radiolucent lines on the pelvic side were evaluated
according to the system of DeLee and Charnley15.
Loosening of the socket was defined as >3 mm of migration
of the cup, >3° of angular rotation, or a continuous radiolucent
line >1 mm wide at the bone-cement interface. The parameters
evaluated on the femoral side included the progression of radiolucent
lines in the seven zones of the femur16,
calcar resorption, and subsidence of the stem.
Statistical analysis of the relationship between various preoperative
factors and the clinical results was performed with use of nonparametric
tests. Significance was determined with StatView statistical software
(version 5.0; SAS Institute, Cary, North Carolina) and was defined
as a p value of <0.05.
Survivorship analysis was performed to determine the overall
success of the joint replacement. Failure was defined as an implant
that had been revised at the time of follow-up. The survival curve
was derived from the cumulative survival rate over time, as calculated
from the actuarial life table17.
The standard error, given as a percentage, and the 95% confidence
intervals were calculated from the data in the life table. The log-rank
test was used for statistical comparison of the survivorship-analysis groups18.
Hips that had had spontaneous ankylosis were compared with those
that had had postoperative ankylosis. No difference was detected
except for the duration of ankylosis, which was significantly longer
(Mann-Whitney test, p = 0.004) for the hips that had had
spontaneous ankylosis (Table I).
Complications
Five patients had postoperative complications. Two patients had
deep-vein thrombosis associated with nonfatal pulmonary embolism.
One patient had a deep hematoma that needed operative débridement. One
patient had a common peroneal nerve palsy that required an operative
release; it completely resolved two weeks later. One patient had
a periprosthetic and inguinal abscess one year postoperatively.
This patient, whose underlying diagnosis of tuberculosis was not
determined until after the abscess developed, had not been treated
with specific antituberculous agents. General treatment included
repeated aspirations of the hip joint for one year; no additional
treatment was necessary. No prosthesis dislocated in this series.
Clinical Results
The mean functional hip score, according to the system of Merle
d’Aubigné, significantly improved from 11.3 0.7
points (range, 10 to 13 points) preoperatively to 16.5 1.5 points
(range, 10 to 18 points) at the time of the last follow-up (Wilcoxon test,
p = 0.0156). The final pain and mobility scores were achieved
by one year; however, the definitive score for walking ability continued
to improve for two to three years (Fig. 1). The mean range of flexion was
88° 23° (range, 30° to 130°).
The global score, according to the system of Merle d’Aubigné,
was excellent for twelve hips, very good for twenty, good for nine,
fair for two, and poor and bad for one each. Forty-three patients
had no pain related to the hip, and two patients had moderate pain.
Twenty-three patients were able to walk with a normal gait, sixteen
needed a cane to walk a long distance, five had a marked limp and often
used a cane, and one had a severe limp and always used a cane. The
range of flexion was >90° in twenty-three hips, 75° to
85° in fourteen, 55° to 70° in three, 35° to 50° in four, and <30°
in one hip.
The mean hip score was 16.4 1.6 points for the hips that had
had postoperative ankylosis compared with 16.6 1.4 points for the
hips that had had spontaneous ankylosis; the difference was not found
to be significant, with the numbers available (Mann-Whitney test,
p = 0.81). No significant correlation was found between
the duration of fusion, the age at the time of conversion, the preoperative hip
score, or the number of previous operations and the global Merle
d’Aubigné score at the time of the latest follow-up
(Spearman rank correlation coefficient, p > 0.05). There
was a trend toward a higher functional score for the hips that had
had a shorter duration of fusion, but it was not significant (p = 0.101).
The gluteal muscles were categorized as one of three types during
the conversion procedure: type 1 (poor and very thin but continuous),
type 2 (fair, pink, and quite thin), or type 3 (satisfactory, red, and
quite thin). The number of previous operations was not found to
have a significant effect on the status of the gluteal muscles (Kruskal-Wallis
test, p = 0.994). The mean functional hip scores were not significantly
affected by the category of the gluteal muscles, as demonstrated
with use of the Kruskal-Wallis test (p = 0.562). However,
the walking ability was significantly associated with the category
of the gluteal muscles (chi-square test, p < 0.02) (Fig. 2). No correlation
was found between this parameter and the number of previous operations
(Spearman rank correlation coefficient, p = 0.074).
The mean postoperative limb-length discrepancy was 1.1 1.2
cm (range, 1 to 2 cm), which was significantly smaller than the
preoperative discrepancy (Wilcoxon test, p = 0.0044).
Radiographic Results in the Involved Hip
The mean angle of inclination of the socket was 37.3° 5.1°.
The mean distance from the most medial point of the prosthetic cup
to the vertical teardrop line was 3.2 0.6 mm, whereas the distance
from the lower point of the prosthetic cup liner to the horizontal
teardrop line was 3.3 0.8 mm (Figs. 3-A and 3-B). Three hips had a nonprogressive
radiolucent line, <1 mm wide, in zone III. One patient
had asymptomatic aseptic loosening of the acetabular component,
with superior migration of 5 mm and a tilt of 5°, at the time of
the twelve-year follow-up. Three hips demonstrated radiolucent lines
in zones 2 and 3 of the femur. Two hips had extensive calcar resorption,
and in one of them the stem had subsided 5 mm.
Results in the Neighboring Joints
Pain in the lower back was reduced in twenty-two patients and
was unchanged in fourteen. One patient with scoliosis had an increase
in pain that necessitated a lumbar arthrodesis six years after the conversion.
When degenerative changes were mild, pain always decreased. Knee
pain decreased in ten patients and remained the same in five patients.
In one patient, the progression of the arthritic process led to
a total knee replacement four years after the total hip arthroplasty.
Another patient had a total knee replacement fourteen years after
the conversion at the age of fifty-five years. Moreover, because
of severe arthritis, total knee replacement was indicated in four
additional patients at the time of conversion and was performed
at a mean of 4.5 months (range, four to six months) following the hip
replacement. These four patients were followed for a mean of 7.3
years (range, five to twelve years) after the total knee replacement,
and knee function was graded as excellent. Six patients had a total
hip arthroplasty on the contralateral side at a mean of 3.8 years
after the conversion. All of them had an excellent result after
a mean duration of follow-up of 5.5 years (range, one to twelve
years).
Revisions
Two patients had a revision of the total hip arthroplasty. One
of the revisions was performed at an outside institution, for unknown
reasons, at 8.6 years postoperatively. The other one was performed at
the time of the seven-year follow-up because of major osteolysis
of the calcar and subsidence of the stem. The socket was not grossly
loose. However, as the initial diagnosis was tuberculosis, both
components were revised. One year after revision, the result in
the hip was graded as excellent.
Survivorship Analysis
The survivorship analysis, with revision as the end point, yielded
a cumulative survival rate of 96.7% (95% confidence
interval, 90.2% to 100%) at eight years and 91% (95% confidence
interval, 78.6% to 100%) at ten years (Fig. 4). The survival
rate for the hips that had had spontaneous ankylosis was 100% at
both eight and ten years, and the survival rate for the hips that
had had postoperative ankylosis was 94.6% (95% confidence
interval, 84.3% to 100%) at eight years and 83.5% (95% confidence
interval, 61.0% to 100%) at ten years. The difference
in survival rates between the hips with postoperative ankylosis
and those with spontaneous ankylosis could not be tested with the
numbers available.
As late as the 1970s, ankylosis of the hip was regarded as one
of the most reliable solutions to septic arthritis and to severe
osteoarthritis, especially in younger patients. Patients with a
successful ankylosis in a good position usually maintain excellent
function for a long period. Symptoms related to the back or to the
adjacent joints become intrusive only after a lapse of twenty years19-21. Conversely, with a malpositioned
fusion, pain related to degenerative changes in the neighboring joints
develops more rapidly22,23. Therefore,
to relieve the knee and spine from overstress, the logical solution
is to manage the ankylosed hip with a total hip arthroplasty and
not to treat the adjacent joints directly. This is particularly true
for patients with an ankylosed hip and severe degenerative changes
in the ipsilateral knee. Total knee replacement in patients with
an ipsilateral hip fusion has led to limited range of motion and
the frequent need for manipulation because of stiffness24. Moreover, as these artificial joints
function under abnormal overstress, their long-term survival is
a cause of major concern.
The effectiveness of total hip arthroplasty following hip fusion
was clearly demonstrated in our series. Ten years after the conversion,
forty-three (96%) of the forty-five patients had not had
an operation on a neighboring joint. The outcome, as noted in previous
reports5,9, was more effective
with regard to the spine than to the ipsilateral knee.
The functional results in this series were very encouraging.
Forty-one (91%) of the forty-five patients had a satisfactory
functional outcome, and forty-three (96%) were pain-free
with respect to the involved joint. However, as noted in previous reports1-10, the postoperative range of flexion
and hip stability were not as good as those following primary hip arthroplasty
in joints that had not had ankylosis. The mean range of flexion
in our series was 88° (range, 30° to 130°), which is similar to
the range of 75° to 87° reported in other studies5,8.
Thirty-nine (87%) of the forty-five patients had a normal
or subnormal gait (a mild limp requiring a cane for walking a long
distance). It is important to note that, unlike pain and mobility,
gait improved for two to three years after the operation. This finding
has been observed in other series3,5,
underlining the need for strengthening exercises of the abductor
muscles following the procedure. In this limited number of patients,
the duration of the hip fusion, the age of the patient at the time
of the hip replacement, the type of ankylosis, and the number of
previous operations did not significantly influence the final functional
result, contrary to findings in previous reports2,8,9.
The absence of a significant difference between the hips that had
had spontaneous ankylosis and those that had had postoperative ankylosis
in the present study could be related to the relatively limited
number of patients in each group. The fact that the duration of
the fusion, which tended to be negatively associated with the final
functional result, was significantly longer for the hips that had
spontaneous ankylosis (Mann-Whitney test, p = 0.004) could
also account for this observation.
The only factor that was predictive of the final score for walking
ability in the present series was the intraoperative status of the
gluteal muscles. This finding was not surprising to us. As a matter
of fact, the parameter that most limits the final functional result
of a total hip arthroplasty following hip fusion is the capacity
of the gluteal muscles to regain strength. This factor is also important
for patient satisfaction as it is directly related to the ability
to walk with no or minimal support. In this respect, when conversion
of an ankylosed hip to a total hip arthroplasty is being considered,
evaluation of the status of the gluteal muscles is of major importance.
The usefulness of electromyography has been discussed25. Most investigators have not found
any association between the intraoperative findings and the results of
electromyography1,3,5. Computed
tomography has also been proposed3,26.
We agree with Amstutz and Sakai1 that
palpation of the contracting abductor muscles is a reliable method
for assessing the ability of the abductors to regain strength. If
contractions are not felt and the hip is fused in an acceptable
position, the conversion should be reconsidered. The classification
proposed in the present study is somewhat unsophisticated, but it
is predictive of the final result. A patient with red, bleeding,
and rather satisfactory gluteal muscles (type 3) can be expected
to have a nearly normal score for walking ability. A patient with
poor, very thin, but continuous gluteal muscles (type 1) can be
expected to have a lower, but acceptable, final score for walking
ability. When the gluteal muscles are not continuous or continuity
is impossible to obtain, we do not perform the conversion if the
hip is fused in a correct position. Amstutz and Sakai1 made a similar recommendation. Moreover,
a tight reduction that imposes stretching of the capsular and ligamentous
structures around the hip in order to both restore length and obtain
a firm tension of the abductors has been recommended2. Otherwise, instability and subsequent
subluxation are to be expected. We have not used muscle transfer
to replace absent abductors in the conversion of a fused hip to
a total hip arthroplasty, as proposed by Besser27.
Other investigators have reported that the number of failures
increases with long-term follow-up8-10.
However, this observation was not supported by the findings in our
series, which had a cumulative survival rate of 96.7% at
the time of the eight-year follow-up, with a rate of 100% for
the hips that had had spontaneous ankylosis and 94.6% for
those that had had postoperative ankylosis.
Total hip arthroplasty following hip fusion has had very encouraging
long-term results, provided that the operative technique was accurate
and was performed in patients whose gluteal muscles allowed restoration
of abductor strength. Hip function was reliably maintained for the
long term, both in patients who had had spontaneous ankylosis and
in those who had had postoperative ankylosis. However, the procedure
is technically demanding.