Reports from the 1980s showed discouraging long-term survival following
revision total hip arthroplasty with
cement1-3.
Hunter et al. reported a 22% rate of excision arthroplasty after 140 femoral
revision arthroplasties that had been performed with
cement4. Engelbrecht
et al. later reported that 31% of 138 cemented revision femoral components had
radiographic evidence of loosening at an average of 7.4
years5. Studies
investigating the use of modern cementation techniques in groups ranging in
size from forty-three to 399 hips have shown encouraging intermediate and
long-term
results6-8,
renewing interest in this technique of femoral revision. Several studies have
identified factors that are predictive of failure, including bone
quality7,9,
patient
age9,10,
and cement mantle
quality11,12.
The purpose of the present study was to evaluate the long-term survival of
cemented revision femoral components and to determine which clinical and
surgical factors are predictive of failure.
We identified 129 revision total hip arthroplasties (117 patients) that had
been performed at our institution by two surgeons (C.H.R. and R.B.B.) between
July 1975 and May 1996 with use of a cemented femoral component. During the
same time-period, 1871 revision total hip arthroplasties had been performed
without cement. The indication for a femoral reconstruction with cement was
poor host bone stock with an intact cortical tube (consistent with a Dorr
type-C, or so-called stovepipe,
femur13). As the
purpose of the present study was to investigate long-term outcomes, thirty-two
hips in patients who had died after less than five years of follow-up were
excluded from the analysis. Of the thirty-two hips that were not included,
three (9%) had been rerevised twenty-eight to thirty-five months
postoperatively. However, none of these three hips had been revised because of
mechanical failure of the femoral prosthesis: one had been revised because of
infection, and the other two had been revised because of acetabular
instability (one hip) or liner dislocation (one hip).
Ninety-seven cemented revision femoral stems in patients who were alive at
a minimum of five years postoperatively were analyzed. The most recent results
for patients who had died after at least five years of follow-up were included
in the analysis. The mean duration of follow-up was 10.3 years (range, five to
23.2 years). The diagnoses that had led to the ninety-seven primary total hip
arthroplasties included osteoarthritis (sixty-one hips), rheumatoid arthritis
(thirteen), fracture (nine), osteonecrosis (six), developmental dysplasia of
the hip (five), ankylosing spondylitis (one), Legg-Perthes disease (one), and
Paget disease (one). Sixty-nine hips had had one previous operation, nineteen
had had two, seven had had three, one had had four, and one had had five.
Seventy-one hips had been previously cemented, whereas twenty-six were
cementless.
The mean interval since the most recent previous operation was 8.1 years
(range, 0.1 to 22.1 years). The diagnoses that led to the ninety-seven index
procedures included aseptic loosening (sixty-four hips), infection (eleven),
periprosthetic femoral fracture (ten), femoral component fracture (six), pain
(three), greater trochanteric nonunion (two), and instability (one). The mean
age of the patients at the time of the revision was sixty-eight years (range,
thirty-nine to eighty-six years). Sixty-nine hips had revision of both the
femoral component and the acetabular component, twenty-six had revision of the
femoral stem alone, and two had revision of the femoral component along with
exchange of the acetabular liner. Fifty-five revisions were performed on the
left side, and forty-two were performed on the right.
The functional status of each patient was classified with use of the
Charnley system14.
Before the index revision, forty-one hips were in patients who had unilateral
hip disease (Charnley category A), thirty-six hips were in patients who had
bilateral hip disease (Charnley category B), and twenty hips were in patients
who had multiple-joint involvement (Charnley category C). The mean weight of
the patients at the time of the revision was 76.1 kg (range, 40 to 117 kg),
and 73% of the patients were classified as overweight on the basis of the body
mass index15.
Fifty-five hips were in men, and forty-two were in women.
Surgical Procedure
A direct lateral approach was used for all
procedures16. Bone
graft was used in thirty-four hips: twenty-four (71%) received strut grafts,
and ten (29%) received cancellous grafts. Allograft bone was used for
twenty-four hips (71%), autograft bone was used for five (15%), and a
combination of both was used for five (15%). The index femoral prostheses
included twenty-seven Mallory-Head stems (Biomet, Warsaw, Indiana), fifteen
Head and Neck Replacement (HNR) stems (Stryker/Howmedica, East Rutherford, New
Jersey), thirteen Harris Design 2 stems (Stryker/Howmedica), nine Endurance
stems (DePuy/Johnson and Johnson, Warsaw, Indiana), five Iowa stems (Zimmer,
Warsaw, Indiana), two Centralign stems (Zimmer), and twenty-six miscellaneous
stems. Forty-one stems were <210 mm in length, and fifty-six were 210
mm.
The generation of the cementing technique that was employed was based on
Harris's criteria as described by Wixson and
Lautenschlager17
(Table I). During the study
period, five hips were treated with first-generation cementing techniques
(before 1980), thirty-nine were treated with second-generation cementing
techniques (between 1980 and 1990), and fifty-three were treated with
third-generation cementing techniques (after 1990).
Clinical Evaluation
Clinical data were prospectively collected and were entered into the
computerized registry at our
institution18. In
particular, Harris hip scores were recorded preoperatively, six months
postoperatively, and then annually thereafter. If a recent report was not
available, patients (or the next of kin of deceased patients) were contacted
by telephone to obtain an updated assessment and to determine if there had
been any complications or rerevisions.
Radiographic Evaluation
A complete set of radiographs was available for ninety (93%) of the
ninety-seven hips. The average duration of radiographic follow-up was 8.3
years (range, one to 23.2 years). Standard anteroposterior and true lateral
radiographs were made preoperatively, immediately postoperatively, at six
weeks, at six months, and at each annual visit thereafter. An independent
observer (R.M.) evaluated all radiographs. Preoperative radiographs were used
to evaluate bone quality. Bone loss was classified according to the system
described by
Mallory19
(Table II). The quality of the
cement mantle on immediate postoperative or six-week radiographs was evaluated
with use of the criteria described by Barrack et
al.20
(Table III). The most recent
radiographs were compared with the earliest available radiographs to evaluate
radiographic loosening according to the criteria described by Harris et
al.21
(Table IV).
Data Analysis
Kaplan-Meier survival
curves22 with end
points defined as rerevision for any reason, rerevision for aseptic loosening
of the femoral component, and mechanical failure of the femoral component
(rerevision because of aseptic loosening, or probable or definite radiographic
loosening) were calculated for (1) all patients, (2) patients younger than
sixty years of age and patients sixty years of age and older, and (3) patients
treated with second and third-generation cementing techniques. If the
acetabulum alone was subsequently revised, the femoral component was
considered to be stable and was studied until the latest followup.
Wilcoxon-Breslow tests were performed on the results of the Kaplan-Meier
survival analysis to determine significant differences between groups.
Clinical, radiographic, and surgical factors that had a significant
association with failure were identified with use of chi-square tests, one-way
analysis of variance with Tukey's analysis, or the Student t test or
Mann-Whitney U test where appropriate. The level of significance was set at p
< 0.05. Preoperative and postoperative Harris hip scores were compared with
use of the Wilcoxon signed-rank test.
Reoperations
At the time of the latest follow-up, 85% (eighty-two) of the ninety-seven
cemented femoral components remained in situ. Fifteen of the femoral
components had been rerevised because of aseptic loosening (nine),
periprosthetic fracture (three), stem fracture (one), hip instability (one),
or infection (one). The mean time to rerevision was 8.0 years (range, 3.3 to
14.2 years). Men had a slightly higher rate of rerevision for aseptic
loosening than women did (p = 0.047).
Of the fifteen femoral components that required rerevision, twelve had been
implanted during a first revision procedure, two had been implanted during a
second revision procedure, and one had been implanted during a third revision
procedure. The diagnoses that had led to the index revisions included aseptic
loosening (nine hips), fracture of the femoral prosthesis (two),
periprosthetic femoral fracture (two), infection (one), and instability
(one).
Clinical Evaluation
The fifteen hips that had rerevision of the femoral component were excluded
from the clinical analysis in order to address the long-term clinical outcomes
for hips that had the component in situ. Thus, eighty-two hips were available
for review after a mean duration of follow-up of 10.3 years (range, five to
23.2 years). The individual Harris hip scores improved from an overall mean of
52 points (range, 19 to 85 points) preoperatively to 71 points (range, 15 to
100 points) at the time of the latest follow-up (p < 0.001). Specifically,
fifteen (18%) of the eighty-two hips were rated as excellent (90 to 100
points), nineteen (23%) were rated as good (80 to 89 points), twenty-three
(28%) were rated as fair (70 to 79 points), and twenty-five (30%) were rated
as poor (<70 points). The individual Harris hip pain scores also improved;
the mean pain score was 24.3 points preoperatively, compared with 36.7 points
at the time of the latest follow-up (p < 0.001). At the time of the most
recent follow-up, 71% of the hips had no or slight pain.
Patients who were sixty years of age or older had less pain than did
patients who were younger than sixty years of age (mean Harris hip pain score,
30 ± 13 points compared with 36 ± 10 points); however, this
difference was not significant (p = 0.064). With the numbers available, no
relationship was found between the most recent Harris hip score and patient
age, gender, weight, or body mass index.
Radiographic Analysis
The Mallory classifications of preoperative femoral bone quality for the
ninety hips with available radiographs are presented in
Figure 1. The preoperative
femoral bone quality in the subgroup of thirteen failed hips with radiographs
was graded as type II for four hips, type IIIA for eight, and type IIIB for
one. Hips with type-IIIA femoral bone quality had a significantly higher rate
of failure secondary to aseptic loosening than did hips with type-II femoral
bone quality (p = 0.021). Interestingly, none of the hips that were
categorized as type IIIB (seven) or IIIC (two) failed secondary to aseptic
loosening.
In the group of ninety hips with available radiographs, the postoperative
cement mantle was graded as category A for thirty-seven hips (41%), category B
for thirty (33%), category C for nineteen (21%), and category D for four (4%).
In the subgroup of thirteen failed hips with radiographs, the cement mantle
was graded as category A for two hips, category B for eight, and category C
for three. With the numbers available, no relationship was found between the
generation of the cementing technique and the grade of the cement mantle on
postoperative radiographs.
Analysis of femoral component fixation according to the radiographic
criteria described by Harris et
al.21 demonstrated
that forty-four (57%) of seventy-seven stems were stable, twelve (16%) were
possibly loose, sixteen (21%) were probably loose, and five (6%) were
definitely loose. Five hips (6%) had symptomatic loosening (moderate to severe
pain and radiographic evidence of loosening). Hips in which the cement mantle
was classified as category C or D were more likely to have probable or
definite loosening at the time of the latest follow-up (p = 0.017) and were
more likely to experience mechanical failure (p = 0.004) than were hips in
which the cement mantle was classified as category A or B. Hips in which the
cement mantle was classified as category A had significantly better long-term
fixation than did hips in all other categories (p < 0.001). Also, hips that
had been treated with second-generation cementing techniques had a higher rate
of probable or definite radiographic loosening compared with hips that had
been treated with third-generation techniques (p = 0.039).
A one-way analysis of variance demonstrated a significant difference
between Harris hip scores and radiographic fixation status (p = 0.013). Tukey
analysis demonstrated that the mean score for hips that were classified as
stable (73 ± 16.4 points) was significantly higher than that for hips
that were classified as definitely loose (50 ± 20.9 points) (p =
0.016). When the same analysis was applied to the most recent Harris hip pain
score, hips that were classified as definitely loose had a significantly lower
score than did hips in all other categories (p = 0.001).
Component Survival
Kaplan-Meier analysis revealed that the ten-year survival rate was 91% (95%
confidence interval, 87% to 95%) with rerevision of the femoral component
because of aseptic loosening as the end point
(Fig. 2), 87% (95% confidence
interval, 83% to 91%) with rerevision for any reason as the end point, and 71%
(95% confidence interval, 65% to 76%) with mechanical failure as the end
point.
The ten-year survival rate for patients who were sixty years of age or
older was significantly better than that for patients who were less than sixty
years when rerevision of the femoral component because of aseptic loosening
was used as the end point (96% compared with 83%; p = 0.044)
(Fig. 3) and when rerevision
for any reason was used as the end point (93% compared with 71%; p = 0.003).
However, there was no difference between the groups when mechanical failure
was used as the end point (75% compared with 65%; p = 0.11).
The ten-year survival rate for hips that had been treated with
third-generation cementing techniques was significantly better than that for
hips that had been treated with second-generation techniques when rerevision
of the femoral component because of aseptic loosening was used as the end
point (94% [95% confidence interval, 89% to 99%] compared with 85% [95%
confidence interval, 79% to 91%]; p = 0.049)
(Fig. 4). No relationship was
found between the generation of the cementing technique and rerevision for any
reason or between the generation of the cementing technique and mechanical
failure (p > 0.05). Compared with second-generation techniques,
first-generation techniques were associated with a significantly higher rate
of rerevision because of aseptic loosening and rerevision for any reason (p
< 0.01).
Complications
Perioperative
Intraoperative complications included four shaft fractures, all of which
were treated with cerclage wiring. Other complications included two shaft
perforations, one of which was treated with wiring and one of which was
treated with bone graft. One patient had an intraoperative pulmonary embolus
and died two days later.
Postoperative
Three (3%) of the ninety-seven hips had dislocated by the time of the
latest follow-up; two had a single dislocation, and one had four dislocations.
Each hip was successfully treated with closed reduction.
Two patients had a periprosthetic femoral fracture five months
postoperatively. One hip was treated with internal fixation and was rerevised
twelve years later, and the other was treated with cerclage wiring and
remained well-fixed nine years postoperatively.
Despite the results of earlier studies that discouraged the use of a
cemented femoral stem during revision total hip arthroplasty, recent
evidence7,8
has supported this technique as an acceptable mode of reconstruction in
selected patients. With rerevision of the femoral component because of aseptic
loosening and mechanical failure of the femoral component as the end points,
our study demonstrated ten-year survival rates of 91% and 71%, respectively.
We found that early-generation cementing techniques, a poor cement mantle,
poor bone quality (Mallory type III-A or higher), an age of less than sixty
years, and male gender were significant risk factors for failure (p <
0.05).
The clinical benefits of third-generation cementing techniques have
remained
uncertain11,23,24.
In the present study, we found that the ten-year survival rate associated with
third-generation cementing techniques was significantly better than that
associated with second-generation techniques when rerevision because of
aseptic loosening was used as the end point (94% compared with 85%; p <
0.05). Furthermore, we found that femoral components that had been inserted
with third-generation techniques demonstrated significantly better
radiographic signs of fixation than did those that had been inserted with
second-generation techniques (p = 0.039). Thus, both clinically and
radiographically, third-generation techniques appear to produce improved
long-term results. Although first-generation cementing techniques were used in
only a small proportion of cases, we found that this early mode of
reconstruction was associated with a very high failure rate.
There has been conflicting evidence in the literature regarding the
importance of the quality of the cement mantle as a risk factor for
failure7,9,11,12,25,26.
With use of the radiographic criteria described by Harris et
al.21, we found
that hips with a category-A cement mantle demonstrated significantly better
long-term fixation than did hips in all other categories and that hips with a
category-A or B cement mantle demonstrated better mechanical survival than did
hips with a category-C or D cement mantle (p < 0.05). This finding suggests
that, over the long term, the integrity of the initial postoperative cement
mantle appears to be predictive of future radiographic signs of fixation.
Although 27% of the hips were classified as having probable or definite
radiographic loosening, only 6% had symptomatic loosening. This finding
questions the validity of using the criteria of Harris et
al.21 to analyze
the fixation of cemented revision components, as this system was intended for
evaluating primary cement mantles. As cement interdigitation with sclerotic
bone is poor, radiolucent lines may be expected. This was reflected in the
present study, in which 36% of the hips were classified as having possible or
probable radiographic loosening. This finding emphasizes the importance of
reviewing sequential radiographs to effectively monitor the progression of
radiolucent lines, rather than just noting the presence of such lines, before
deciding on the stability of the prosthesis.
On the basis of previous
studies7,9,27,28,
we chose to analyze the impact of patient age by comparing the results for
patients who were less than sixty years old with those for patients who were
sixty years old or more. Patients who were younger than sixty years of age had
a significantly higher rate of rerevision for aseptic loosening (p = 0.044)
and demonstrated a trend toward experiencing more pain than did older patients
(p = 0.064). Although age has been identified as being predictive of failure,
activity level, which depends greatly upon age, actually may be more important
for predicting prosthetic survival.
In the present series, we found that poor bone stock was related to failure
due to aseptic loosening; specifically, we found that type-IIIA femora were
more likely to fail than type-II femora were (p = 0.021). An inadequate number
of type-IIIB and type-IIIC femora precluded meaningful conclusions regarding
more extensive bone loss as a risk factor for failure. An intact cortical tube
is essential for femoral reconstruction with cement; however, the amount of
corticocancellous deficiency that would preclude reconstruction with cement
remains unknown.
There is consistent evidence that revision of the femoral component with
use of second-generation cementing techniques demonstrates acceptable ten-year
results and fair radiographic evidence of fixation. Preoperative bone quality,
the use of third-generation cementing techniques, patient age, and the
postoperative quality of the cement mantle may be important predictors of
outcome. Revision of the femoral component with cement remains a relatively
uncommon procedure; however, understanding the long-term outcomes and specific
indications for its use will provide the surgeon with another option for
revision femoral reconstruction.