There is a consensus that severe proximal femoral bone loss
is a formidable problem in reconstructive hip surgery1. The fixation of a cemented revision
femoral component is poor compared with that of a primary component2,3. Cemented components used in revisions
for femoral loosening have a high prevalence of radiographic loosening
when the revision did not include biological reconstruction of the
deficient bone stock2,4,5. Therefore,
restoration of bone stock is thought to be necessary for long-lasting
results. There are various techniques for biological reconstruction
of the proximal part of the femur. As the amount of autogenous bone
graft is limited, allograft is widely used. When the proximal part
of the femoral shaft has sufficient stability, the so-called Exeter
technique (impaction grafting) can be employed6.
Other authors prefer massive allografts combined with a long-stem
prosthesis7. In 1987, Wagner presented
a technique in which a cementless long-stem prosthesis was fixed
in the diaphysis, and he reported excellent spontaneous osseous
regeneration8.
In October 1988, we started to use the Wagner SL revision stem
(Sulzer Orthopedics, Baar, Switzerland) for the reconstruction of
severe defects of the proximal part of the femur in revision hip
arthroplasty. We wanted to determine whether the principle of cementless
diaphyseal fixation is successful in femoral revision after total
hip replacement and whether there is spontaneous regeneration of
the bone stock of the proximal part of the femur.
The Wagner revision stem is made of a titanium-aluminum-niobium
alloy with a rough-blasted surface. The shaft of the prosthesis
has a conus angle of 2° and eight longitudinal ridges arranged in
a circle around the stem3. The
stem is available in lengths of 190 to 385 mm. Cementless anchoring
of the stem is achieved after implantation in a conically reamed
femoral shaft. The longitudinal ridges make a large amount of rotational
stability possible. If there are larger defects in the proximal
part of the femur, stable fixation of the stem can be achieved only
distally in the diaphyseal part of the femur.
Between October 1988 and September 1997, 129 Wagner SL revision
stems were implanted without cement in 123 patients. No patient
was excluded from the study. Eighty-four of the revisions involved
replacement of a cemented stem. The stems had been in situ for
a mean of 9.9 years (range, 0.13 to 27.4 years). The 129 operations
were performed by seven surgeons, and seventy-one operations were
performed by one of us (P.B.). Sixty-five patients also had a total
hip replacement on the contralateral side. Seventy-six revisions
were performed in women, and fifty-three were done in men. At the
time of the revision, the mean age of the patients was 64.9 years
(range, 36.7 to 86.3 years) and the mean weight was 73.0 kg (range,
50 to 110 kg). Nine patients had a low weight (body mass index, <20 kg/m2), forty-six had a normal weight (body
mass index, 20 to 25 kg/m2),
and seventy-four were overweight (body mass index, >25
kg/m2). Both the femoral
and the acetabular components were revised in eighty-seven hips.
In forty-two hips, only the stem was revised, and in eighteen of those
hips, the liner of a modular acetabular component was also exchanged.
The surgeon implanted the shortest stem that ensured sufficient
biomechanical stability. The conically reamed osseous bed in the
medullary cavity should ideally be 100 mm long, with a minimum length
of 70 mm9. The diameter of the
stems ranged from 14 to 22 mm (mean, 16.6 mm). The indication for
revision was painful aseptic loosening in ninety-seven hips, periprosthetic
fracture (Figs. 1-A, 1-B, and 1-C) in thirteen hips
(one of which also had an infection), and septic loosening with
a positive intraoperative culture in sixteen hips (Figs. 2-A, 2-B, and 2-C). In the three
remaining hips (one of which still had a positive intraoperative
culture), a Wagner revision stem was inserted during a second-stage
reimplantation after the performance of a Girdlestone resection
to treat chronic deep infection. The revision of the stem was the
first revision in eighty-eight hips, the second revision in thirty-four,
the third revision in six, and the fourth revision in one. The initial diagnosis
had been primary osteoarthrosis in seventy-one hips, posttraumatic
osteoarthrosis in thirteen, osteoarthrosis subsequent to bacterial infection
in one, osteonecrosis of the femoral head in ten, developmental
dysplasia of the hip in twenty-five, and rheumatoid arthritis in
nine.
Perioperative prophylactic antibiotics were administered in 118
of the 129 cases. A transfemoral approach10 was
used in sixty operations; a transgluteal approach11,
in forty-eight; a dorsal approach11,
in seventeen; a transtrochanteric approach, in three; and an anterior
approach11, in one. In six hips,
an osteotomy of the greater trochanter was performed; the approach
was transgluteal in five of these hips and transfemoral in one. Allogeneic
cancellous bone graft was used in thirty-seven hips; autogenous
bone graft, in eleven; and mixed grafts, in two. To prevent heterotopic ossification,
nonsteroidal anti-inflammatory agents were given in eighty-seven
cases, postoperative low-dose radiation therapy (10 Gy) was administered
in four, and both were given in seven. All patients were evaluated
radiographically and clinically by an independent investigator (O.B.)
who was not a member of the staff of the department. The patients
were followed for a mean of 4.8 years (range, two months to 11.1
years), until the most recent follow-up evaluation, repeat revision,
or death. Seven patients were followed for less than two years.
Three of them died, at ten, eleven, and seventeen months postoperatively,
and the other four had a revision, at seven, ten, eleven, and twelve weeks
after the index operation. The mean duration of follow-up of the
living patients without repeat revision of the stem was 5.0 years
(range, 2.2 to 11.1 years).
The prerevision femoral defects were classified with use of the
system described by Pak et al.12.
When performing a femoral revision, the surgeon must know the most
proximal part of the femur at which stable fixation of the prosthesis
is possible. In order to address this issue, we developed a classification
system based on the preoperative radiographic appearance (Fig. 3). The femur
was divided into five parts: 0 indicated the femoral head and neck;
1, the proximal quarter of the femur; 2, the second quarter of the
femur; 3, the third quarter of the femur; and 4, the distal quarter
of the femur. This classification allowed preoperative identification
of the most proximal area for fixation of the prosthesis without
a description of the bone defects.
Bone quality and restoration of the proximal part of the femur
were assessed quantitatively on follow-up anteroposterior and lateral
radiographs by measuring the width of the cortical and cancellous
bone as well as the outside diameter of the femoral shaft at a point
1 cm distal to the inferior margin of the lesser trochanter. The
entire bone mass was calculated, as an assessment of bone quality,
as the ratio of the width of the bone to the outside diameters13. Additionally, the femoral score
was calculated according to the system of Barnett and Nordin14. The radiographs made immediately
after the index operation were compared with those made at the latest
follow-up examination in order to classify the restoration of the
proximal part of the femur as A (increasing defects), B (constant
defects), or C (osseous restoration). The relative area of direct bone-implant
contact with optimal appearance of the interface was measured on
anteroposterior and lateral radiographs.
Migration was analyzed by measuring the vertical subsidence of
the femoral component according to the method of Callaghan et al.13. Allografts were assessed for incorporation
into host bone as evidenced by trabecular bridging of the host-graft
interface. A clear reduction of density or breakdown of the transplanted
bone was defined as bone resorption.
The Merle d’Aubigné score15 was
used for the functional assessment of the hip. The patient’s
general condition and activity were evaluated with use of the Karnofsky
index16. The preoperative and
postoperative satisfaction of the patient was subjectively estimated
as being very content, content, intermediate, discontent but tolerable,
and unbearable.
We performed Kaplan-Meier survival analysis (JMP software, version
3.1.6.2, 1996; SAS Institute, Cary, North Carolina) to November
1999, with removal of the stem because of aseptic loosening, removal
of the stem for any cause, and worst case (removal of the stem for
any cause and/or lost to follow-up) as the failure criteria.
All survivorship data are reported with a 95% confidence
interval. The p value for noncrossing survival curves was calculated
with the log-rank test, and that for crossing curves was calculated
with the Wilcoxon test. Correlation and significance between variables were
calculated with use of the Fisher test or the t test.
One patient, who emigrated to another country 3.5 years postoperatively,
was lost to follow-up. At the latest follow-up examination, this
patient had excellent clinical and radiographic results. Twelve patients
died before November 1999, at a mean of 4.2 years (range, 0.8 to
8.1 years) after the operation. The data on these patients were
included in the analysis of the follow-up results.
Complications
There were no deaths during or in the immediate postoperative
period after the index revision. However, one patient who had been
operated on because of septic loosening died of a pulmonary embolism during
repeat revision to treat recurrent infection 3.9 years after the
index operation. The other eleven patients who died did so of unrelated
causes.
Postoperative reinfection occurred in two of the eighteen patients
who had had a positive intraoperative culture at the time of the
revision. There was one late infection 4.6 years after revision
in an aseptic hip, but no infection developed after second-stage
reimplantation following a Girdlestone resection.
One periprosthetic fracture and one fissure of the distal part
of the femur occurred during implantation of the stem. The periprosthetic
fracture led to a repeat revision, with a longer Wagner revision stem,
ten weeks postoperatively because the fracture had been insufficiently
stabilized by means of cerclage wires. The fissure consolidated
after nonoperative treatment. Fracture of the thin, ventral, osteotomized
portion of the proximal part of the femur (the ventral lid) or of
the dorsal portion of the proximal part of the femur occurred in
thirty-six of the sixty hips treated through a transfemoral approach10, but these fractures did not affect
stability. One femoral shaft was perforated during reaming.
In seven patients, a palsy of the femoral nerve (two patients)
or the peroneal nerve (five patients) occurred. Five of these seven
patients had been operated on through a transgluteal approach, and the
other two patients had had a transtrochanteric or dorsal approach,
indicating a lower risk of nerve palsy with the transfemoral approach
(Fisher test, p = 0.0015). The two palsies of the femoral
nerve and one of the palsies of the peroneal nerve resolved within
a few weeks, but the remaining four palsies of the peroneal nerve
did not resolve. These four patients had a mean of 14 mm of limb-lengthening (range,
10 mm of shortening to 45 mm of lengthening).
Seven hips sustained a postoperative dislocation. Four of these
hips remained stable after one closed reduction. Of the three hips
that remained unstable, two had a revision of the cup and one had
a revision of both the stem and the cup because of a high angle
of anteversion of both components. In two patients, class-IV17 heterotopic ossification with ankylosis
of the hip developed. Neither patient had had postoperative prophylaxis
against heterotopic ossification.
Reoperations
Six revision stems had to be revised again during the follow-up
period. In one hip, discussed above, a high angle of anteversion
of both the stem and the cup was corrected after a second dislocation
(at twelve weeks). In another hip, with an extensive bone defect
extending to the third quarter of the femur, too thin of a stem
had been implanted and the stem subsided 23 mm within a few weeks.
During the reoperation at seven weeks postoperatively, the stem
was replaced with one with a larger diameter. There was no subsidence
at the time of the latest follow-up examination, seven years after
the repeat revision. As mentioned above, another stem was replaced
ten weeks after the operation because of insufficient stabilization
of an intraoperative periprosthetic fracture. In another three hips,
both the stem and the cup had to be removed at 0.2, 3.8, and 4.6
years postoperatively because of deep infection. There were no repeat
revisions due to aseptic loosening.
During the follow-up period, there were ten reoperations without
removal of the stem. Revision of the acetabular component was necessary
in four hips (because of aseptic loosening in two and because of recurrent
dislocation due to a malpositioned cup in two). Two periprosthetic
fractures that occurred at five weeks and 4.9 years postoperatively
required osteosynthesis. Three other periprosthetic fractures healed
with nonoperative treatment. In two hips, secondary dislocation
of the cut portion of the bone (the ventral lid) after a transfemoral
approach occurred and refixation was necessary. In one hip, refixation
of the dislocated greater trochanter was performed. In another hip,
irritation of the soft tissue by the cerclage wire around the greater
trochanter was treated with wire removal after consolidation of
the trochanter.
Functional Evaluation
Two days postoperatively, most patients started walking with
crutches, bearing partial weight (about 15 kg) for six weeks, after
which gradual progression of weight-bearing of about 10 kg/wk was
allowed. Deviations due to the extent of the osseous defect or the
ability of the patient to use crutches were necessary in a number
of cases. The mean period between the operation and full weight-bearing
was 123 days (range, two to 374 days).
At the latest follow-up examination, the mean limb-length discrepancy
was 15 mm of shortening of the revised limb compared with 19 mm
preoperatively.
The mean Merle d’Aubigné15 score
improved from 7.7 points (range, 0 to 16 points) preoperatively
to 14.8 points (range, 0 to 18 points) at the latest follow-up examination.
Preoperatively, seventy-nine hips (61%) were rated poor; forty-eight
(37%), fair; one (1%), average; one, good; and
none, very good or excellent. At the latest follow-up examination,
eight hips (6%) were rated poor; nineteen (15%),
fair; sixteen (12%), average; forty-three (33%),
good; fourteen (11%), very good; and twenty-nine (22%),
excellent. Most impressive was the improvement in the mean pain score
from 1.0 to 5.4 points, whereas the improvement in the mean mobility
score was only moderate (from 4.3 to 5.3 points). Walking ability
also clearly improved, from a mean of 2.4 points preoperatively to
a mean of 4.1 points at the latest follow-up examination. In the
sixty hips treated through a transfemoral approach, the mean Merle
d’Aubigné score improved from 7.7 points (range,
0 to 12 points) preoperatively to 14.2 points (range, 6 to 18 points) at
the latest follow-up examination.
The mean Karnofsky index improved from 62.0% preoperatively
to 78.6% at the latest follow-up evaluation. Whereas preoperatively
only five patients (4% of the hips) had a Karnofsky index
of between 80% and 100%—that is, they
were capable of normal activity—at the latest follow-up
examination eighty-three patients (64% of the hips) had
an index in that range. Preoperatively only one patient (1% of
the hips) was subjectively content with her situation, whereas at
the latest follow-up evaluation 111 patients (86% of the
hips) were content or very content.
Radiographic Evaluation
Bone Defects
On the most recent follow-up radiographs, 113 hips (88%)
had the appearance of at least some degree of restoration of the
proximal part of the femur. Increasing bone defects were seen in
only four hips (3%), and we saw no change with regard to
bone defects in twelve hips (9%). With the numbers available,
age, gender, body weight, number of previous operations, diabetes,
and use of cortisone, nonsteroidal anti-inflammatory agents, alcohol,
or cigarettes had no influence on osseous restoration. Only two
of ten patients with rheumatoid arthritis had osseous restoration
(Fisher test, p = 0.048). A transfemoral approach was positively
related to the degree of bone restoration (Fisher test, p = 0.029). The
mean relative bone mass of the proximal part of the femur increased
from 24.9% preoperatively to 46.6% at the latest
follow-up examination. Patients with a preoperative relative bone
mass of <15% (25% quartile) had poorer
osseous restoration than did patients with a greater bone mass (Fisher
test, p = 0.017). The mean preoperative femoral score14 was 47.5%, and in 44% (fifty-seven)
of the 129 hips it was <45%. A higher proportion
of patients with a femoral score of >45% had excellent
osseous restoration (Fisher test, p = 0.003). The mean
percent area of intramedullary contact between the cortical bone
and the prosthetic surface with an optimal interface appearance
increased from 56.3% (range, 16.7% to 100%)
in the immediate postoperative period to 81.5% (range,
33.3% to 100%) at the latest follow-up examination.
There was complete incorporation of the bone graft in twenty-five
(50%) of the fifty hips in which it had been used and partial
incorporation in sixteen (32%). Nine (18%) of
the bone grafts showed no evidence of incorporation. There was no
difference between the hips with and without bone graft with regard
to osseous restoration of the proximal part of the femur.
Migration
A mean subsidence of 5.9 mm (range, 0 to 45 mm) was measured
at the latest follow-up evaluation. Forty-four hips (34%)
had subsidence of >5 mm, and twenty-six (20%)
had subsidence of >10 mm. The mean subsidence in the hips
treated through a transfemoral approach was 7.7 mm (range, 0 to
38 mm). Female patients had a lower risk of subsidence than did
male patients (Fisher test, p = 0.013). Femoral defects
rated as 1B or higher according to our classification system (t
test, p = 0.0038) and osteoporosis of <45% as
measured with the Barnett-Nordin14 index
(t test, p = 0.048) were both positively associated with
the amount of subsidence.
Survival Analysis
Six of the 129 Wagner revision stems had to be revised again
during the follow-up period. The indication for three of the revisions
was instability, and the indication for the other three was deep
infection. The cumulative survival rate with failure defined as
removal of the stem for any cause was 93.9% (95% confidence
interval, 88.8% to 99.0%) at 11.1 years (Fig. 4). With use of
the worst-case criterion, the survival rate was 92.8% (95% confidence
interval, 87.3% to 98.3%) at 11.1 years. There
was no worsening of the survival rate after the fourth year. With
the end point defined as removal due to deep infection in the group
of eighteen hips with a positive intraoperative culture and the
two hips with a negative intraoperative culture and a previous Girdlestone resection,
the cumulative survival rate was 84.4% (95% confidence
interval, 63.0 to 100%) at 7.7 years compared with 98.2% (95% confidence
interval, 94.7 to 100%) at 11.1 years in the group of 109 revisions
in aseptic hips (log-rank test, p = 0.0066).
High failure rates after revisions with cement have led to the
promotion of uncemented long-stem femoral prostheses18. Uncemented femoral components have
several advantages: the difficulties and complications associated
with cement removal are eliminated, bone loss may be reduced, and
implant removal is frequently easier. A review of the literature
has shown lower rates of repeat revision after revision arthroplasties
with an uncemented femoral component19,
and most such repeat revisions have been performed within the first
few postoperative months and have been necessary because too thin
of a stem had been implanted. The results in the present series
seem encouraging because the survival rate stabilized after 4.6
years. We believe that these results are comparable with those of
arthroplasties with an extensively porous-coated chromium-cobalt
stem20-22.
In other series of patients treated with the Wagner stem, a mean
subsidence of between 3.2 mm23 and
6.1 mm24, values that are similar
to our mean of 5.9 mm, was reported. In our series, subsidence appeared
to stop at a mean of thirteen months (range, one to sixty months),
although nine stems (7%) continued to subside even after
twenty-four months. In the series of Grünig et al., most
Wagner stems ceased to subside after three months24.
Similar to our findings, the reported prevalence of subsidence of >10
mm after revision has ranged from 15% (six of forty24) to 19% (six of thirty-one23). In about 8% of the cases
(two of thirty-one cases23 and
four of forty cases24), repeat
revision with a larger Wagner stem was performed because of subsidence.
In their series of uncemented proximally porous-coated stems, Berry et
al. measured a mean of 5 mm (range, 0 to 40 mm) of femoral component
subsidence25. Hussamy and Lachiewicz
reported a mean subsidence of 8 mm after implantation of the BIAS
femoral component26; eleven of
thirteen stems with >2 mm of subsidence had stabilized
during the follow-up period. Despite a 96% cumulative survival
rate at seventy-two months in their series of curved, long-stem, titanium-alloy,
noncircumferentially porous-coated femoral components, Peters et
al. reported only a 37% chance of survival of the stem
when revision or progressive subsidence was the end point27. Krishnamurthy et al., in a series
of 297 extensively coated chromium-cobalt stems, reported a mechanical
failure rate of only 2.4% at a mean of 8.3 years20. None of the unrevised stems subsided >2
mm. The mean subsidence during the first postoperative year has
been reported to be 1.5 mm after revision with cement2 and 1.9 mm after revision with impaction
grafting28. Subsidence of the
stem within the cement as well as migration of the stem and the
cement mantle within the allograft (4 to 31 mm) has been reported after
impaction bone-grafting29.
New-bone formation has been seen to occur regularly after femoral
revision with use of the Wagner revision stem8,23,24,30-32.
However, it is necessary to be aware of the limitations of qualitative
assessment of bone formation on plain radiographs. In our experience,
mechanical stability as well as careful removal of cement and scar
and granulation tissue are essential preconditions for spontaneous
restoration of bone stock of the proximal part of the femur. In
difficult cases, a transfemoral approach is helpful, but when this approach
is used the blood supply of the osseous lid must be preserved and
detachment of muscles must be avoided. We agree with Grünig
et al. that allogeneic bone-grafting is necessary in only a small number
of patients24. Femoral bone restoration
associated with the Wagner SL revision stem may be due to proximal
transmission of force because of the conical shape of the prosthesis,
the higher elasticity of the titanium alloy, and the good histocompatibility
of the rough-blasted surface9.
In our study, the risk of postoperative nerve palsy was related
to the approach. If extensive lengthening is necessary or if the
sciatic nerve may be scarred as a result of a previous operation,
it is wise to expose the nerve. Preoperative planning of the approach,
of the length of the prosthesis, and of the extent of a femoral
osteotomy is mandatory in the use of the Wagner revision stem3. Some cases of subsidence may be
a consequence of insufficient planning of the size of the prosthesis. There
is no question that classification of defects is essential so that
the surgeon can assess the severity of the defect preoperatively
and plan the operation (the approach, necessity for bone grafts,
and suitable implant) appropriately. At the beginning of our experience
with the Wagner revision stem, we used the implant only when a patient
had a periprosthetic fracture or extensive loss of bone from the
proximal part of the femur. However, because of the encouraging
results, recently we have also preferred to use the Wagner revision
stem in patients with limited bone loss. When, in our estimation,
a cementless standard stem will provide enough primary stability,
we prefer to use it. It seems to us very important to use the shortest
stem that ensures sufficient mechanical stability because removal
of a well fixed implant in the case of late infection may be a major
problem and probably will require a transfemoral approach.
Note: The authors thank Professor Dr. Klaus Dietz, Institute
for Medical Information Processing, Eberhard-Karls-Universität
Tübingen, for statistical advice.