Study Population
Three hundred and eleven consecutive PCA total hip replacements
(Howmedica, Rutherford, New Jersey) were implanted at our institution
between 1985 and 1988 in 279 patients. During this time, the two
senior authors (R.B.B. and C.H.R.) performed approximately 700 total
hip replacements. A patient was more apt to have a cementless PCA
hip replacement if he or she was less than seventy years of age,
had good bone stock, had a diagnosis of osteoarthritis or avascular
necrosis, and was expected to place increased demands on the joint replacement.
The average age of the 279 patients (146 men and 133 women) at
the time of the index surgery was sixty-one years (range,
twenty to eighty-one years). There were 155 right hips and
156 left hips. The preoperative diagnosis was osteoarthritis
in 255 hips, rheumatoid arthritis in twenty-two, osteonecrosis in
sixteen, degenerative arthritis secondary to congenital dislocation
in eleven, posttraumatic arthritis in four, and other
conditions in three. All operative procedures were performed by
or under the direct supervision of the two senior authors through
a modified direct lateral surgical approach5,6.
All but two acetabular components were inserted without cement with
use of line-to-line acetabular reaming. The acetabular
components ranged in size from 40 to 64 mm (outer diameter). The
metal backing on all of the acetabular components was 5 mm thick.
All of the femoral components were inserted without cement. Two
different sizes of cobalt-chromium-alloy heads were used
in this series: a 32-mm head was used in the first eighty-two
hips, and a 26-mm head was used in the subsequent 229 hips. Postoperatively,
the patients were limited to 50% partial weight-bearing
on the involved limb for the first six weeks. Postoperative warfarin
was used routinely for prophylaxis against deep-vein thrombosis.
Sixty-four patients (seventy-six hips) died
before the time of the latest review for the present study. Eleven
of the deceased patients (thirteen hips) were alive at the time
of the ten-year follow-up. The average age of
the deceased patients at the time of the index surgery was sixty-two
years (range, twenty-one to eighty-four years).
The average duration from the operation to the time of death was
eight years (range, four days to thirteen years). Excluding the
patient who died four days after the operation, the average duration
of follow-up was six years (range, one month to thirteen
years).
Forty-five patients (forty-seven hips) were
lost to follow-up. The average duration of follow-up for
those patients was five years (range, two months to nine years).
None of these patients had a revision operation before
they were lost to follow-up.
One hundred and thirty-nine patients (154 hips) returned
for examination, and thirty-three patients (thirty-seven
hips) were assessed with use of a questionnaire administered during
a telephone interview or sent by mail. Four of these patients were
excluded from the study because of their poor medical condition.
Thus, 168 patients (187 hips) accounted for the study population.
The average age of the study population at the time of the latest
follow-up was seventy-one years (range, thirty
to eighty-seven years).
Clinical Outcome
Any complication related to the hip arthroplasty was recorded.
Patients were assessed postoperatively at six weeks, three
months, six months, one year, and annually thereafter. Any revision
of the femoral or acetabular component was recorded, and survival
statistics were calculated. Postoperative evaluation included assessment
with use of the recently validated Harris hip score7. Each patient was asked specifically
about thigh pain and was instructed to grade this pain, if present,
on a scale from 1 to 10. Scores of 1, 2, and 3 were considered to
indicate mild pain; 4, 5, and 6, moderate pain; and 7 to 10, severe
pain. Thigh pain was characterized by dividing the postoperative period
into three time-intervals: early (less than three years postoperatively),
middle (three to nine years postoperatively), and late (ten or more
years postoperatively). For each period, the worst score was chosen
and the change in the score was analyzed.
Radiographic Assessment
A standard radiographic assessment that included anteroposterior
pelvic and true lateral views was performed at the patient follow-up
intervals defined above8. The
locations of radiographic findings on the anteroposterior radiograph
were recorded with use of the three zones described by DeLee and
Charnley for the acetabulum9 and
the seven zones described by Gruen et al. for the femur10. We also assessed lateral radiographs
by dividing the proximal part of the femur into the seven zones
described by Johnston et al.11 and
the acetabulum into the three zones described by Johnston et al.
The status of biological fixation of the femoral component was
assessed with use of a modification of the criteria described by
Engh et al. for a proximally porous-coated femoral component12. Radiosclerotic lines were defined
as radiodense lines that roughly paralleled the surface contour
of the implant but were separated from it by a radiolucent zone
of varying thickness. Bone ingrowth was assessed according to the
presence or absence of radiodense lines in the zones with the porous
coating (zones 1, 7, 8, and 14). A femoral component was classified
as stable with bone ingrowth if radiosclerotic lines were observed
in fewer than two of the four zones. It was classified as stable
with fibrous ingrowth if radiosclerotic lines were observed in at
least three zones without progressive subsidence after the early
postoperative period, and it was classified as unstable if there
had been progressive subsidence of >3 mm or if the alignment
of the component had changed. Subsidence of the femoral component
was referenced from both the tip of the greater trochanter and the
midpoint of the lesser trochanter, with adjustments made for change
in radiographic magnification13.
Alignment of the femoral component was classified as valgus, neutral,
or varus, as seen on anteroposterior radiographs.
An acetabular component was considered to be stable with possible
bone ingrowth if radiosclerotic lines were observed in fewer than
two of the three zones on the anteroposterior radiograph and in
fewer than three of the six zones on the anteroposterior and lateral
radiographs combined. An acetabular component was considered to
be stable with fibrous ingrowth if there were radiosclerotic lines
in at least four zones but no migration, and it was defined as unstable
if there was a circumferential radiosclerotic line of >2
mm in width, migration of >2 mm with use of the teardrop
as a reference, or a change in inclination or anteversion of >5°.
The location of osteolysis was recorded with use of the zones mentioned
above. The extent of osteolysis was graded as small (<2
cm in any dimension) or large (2 cm in any dimension) on both the
acetabular and the femoral side.
Statistical Analysis
Comparisons were performed with use of SPSS 9.0 software (SPSS,
Chicago, Illinois). Categorical variables were analyzed with use
of the chi-square test or the Fisher exact test where appropriate.
Student two-tailed t tests were used to compare continuous
variables. Kaplan-Meier survivorship analysis was performed with
use of three end points (any revision, femoral revision, or acetabular
revision) and with use of two covariates (femoral head size and
gender). Binary logistic regression (enter method) was used to determine
the effect, if any, of multiple covariates on revision status (dependent
variable). The regression was performed twice, each time with the revision
status (revised for any reason or unrevised) as the dependent variable
but with a modification in the covariates for the two regressions.
For the first regression, the covariates were femoral head size,
acetabular component size, gender, age, time since the operation,
and diagnosis. For the second regression, femoral head size and
acetabular component size were replaced with polyethylene thickness.
Polyethylene thickness was not included in the first regression
as it is directly related to the sizes of the acetabular component
and the femoral head and would therefore have introduced a covariate
interaction error.
Clinical Outcome
Three acute dislocations of the hip occurred and were treated with
closed reduction. Four patients sustained a limited longitudinal
fracture of the femoral neck during insertion of the femoral component.
All fractures were treated with cerclage wires. Sciatic nerve palsy
developed in two patients, one of whom had only mild residual symptoms
and the other of whom had a residual neurological deficit. Femoral
nerve palsy developed in one patient and resolved spontaneously. Deep-vein
thrombosis developed in nine patients and was treated with anticoagulant
therapy. Pulmonary embolism developed in three patients but was
fatal in none. One patient died four days after the arthroplasty
because of an acute myocardial infarction.
Seventeen patients (seventeen hips, 5%) had a revision.
Ten acetabular components were revised. Eight were revised because
of aseptic loosening; one, because of advanced polyethylene wear;
and one, because of instability of the hip joint. Ten femoral components
were revised because of aseptic loosening or thigh pain. The average
time to the revision was 9.3 years for the acetabular component
and 3.7 years for the femoral component.
The Kaplan-Meier survival rate, with any revision as
the end point, was 98.2% ± 1.5% at
five years, 94.7% ± 2.8% at
ten years, and 90.0% ± 5.4% at
fourteen years (Fig. 1). With acetabular revision as the
end point, the survival rate was 99.6% ± 0.8% at
five years, 97.7% ± 2.0% at
ten years, and 92.7% ± 5.1% at
fourteen years. With femoral revision as the end point, the survival
rate was 98.2% ± 1.6% at
five years, 96.9% ± 2.1% at
ten years, and 94.9% ± 3.6% at
fourteen years.
The survival of the acetabular components differed significantly
between men and women (p = 0.032). The survival rates at
five, ten, and fourteen years were 97.9% and 99.3%, 96.0% and
94.2%, and 93.9% and 86.6% for men and
women, respectively. With both femoral and acetabular revisions
or only femoral revision as the end point, the survival rates did not
differ significantly between men and women (p = 0.124 and
0.389, respectively). The survival rates for replacements with a
32-mm head and for those with a 26-mm head were, respectively, 98.7% and
98.1% at five years, 95.9% and 94.6% at
ten years, and 89.5% and 91.0% at fourteen years. These
differences were not significant (p = 0.401).
Binary logistic regression with use of revision because of aseptic
loosening as the dependent variable and femoral head size, acetabular
component size, gender, age, time since the operation, and diagnosis
as covariates showed that, when considered collectively, acetabular
component size (p = 0.039, standard error = 0.087)
and time since the operation (p = 0.003, standard error = 0.070)
were the only significant covariates for revision status. Patients
with a small acetabular component were more likely to have a revision
than were patients with a large acetabular component. Substituting
polyethylene thickness for femoral head and acetabular component
size in the regression equation showed polyethylene thickness (p = 0.011,
standard error = 0.144) and time since the operation (p = 0.007,
standard error = 0.074) to be significant covariates.
The Harris hip scores were calculated for 172 hips. The Harris hip
scores preoperatively, at the five-year postoperative examination,
and at the latest examination (at an average of twelve years; range,
ten to fourteen years) were 43 14, 88 13, and 85 14 points, respectively.
At the latest follow-up evaluation, eighty-one
hips (47%) were graded as excellent; forty-eight (28%),
as good; eighteen (10%), as fair; and twenty-five (14%),
as poor. When the relationship between the Harris hip score and
thigh pain at the latest follow-up was analyzed, thigh
pain was found to negatively affect the Harris hip score (r = -0.453,
p < 0.01).
Complete clinical records on thigh pain at all postoperative periods
were obtained after 157 hip replacements. Twenty-four hips
(15%) were associated with thigh pain in the early period
(less than three years postoperatively), and fifty-six
(36%) were associated with it both in the middle period
(at three to nine years) and in the late period (ten or more years
postoperatively). In the late period, thirty-one hips (20%)
were associated with mild thigh pain; seventeen (11%), with
moderate thigh pain; and eight (5%), with severe thigh pain.
Sixty-eight hips (43%) were not associated with
thigh pain at any time during the follow-up period. The
nature of the thigh pain changed with time (that is, it was present
at the time of one review but not at the time of the next) for the remaining
eighty-nine hips (57%). No significant relationship was
found between thigh pain and the size of the femoral component,
gender, or age.
Radiographic Assessment
Radiographic assessment was completed for 156 hips. Fixation
of the femoral component was classified as stable with bone ingrowth
in 137 hips (88%), stable with fibrous ingrowth in four
(3%), and unstable in fifteen (10%). Alignment
of the femoral component was classified as neutral in 104 hips (67%),
valgus in fifteen (10%), and varus in thirty-seven (24%).
The femoral components that were in neutral alignment were more
likely to have stable fixation with bone ingrowth and were less
likely to subside than those aligned in a valgus or varus position
(p < 0.01). Fifteen (10%) of the femoral components
had subsided; three of the four femoral components that were classified
as stable with fibrous ingrowth had subsided in the early postoperative
period, and twelve of the fifteen femoral components that were classified as
unstable had subsided.
The average thigh pain score associated with stable components
with fibrous ingrowth (4.7 points) was higher than those associated
with stable components with bone ingrowth (1.0 point) and with unstable
components (0.5 point). Although the difference was significant
(p < 0.01), there were only four femoral components with
fibrous ingrowth.
Fixation of the acetabular component was considered to be stable
with possible bone ingrowth in 130 hips (83%), stable with
fibrous ingrowth in fourteen (9%), and unstable in twelve
(8%).
On the femoral side, radiosclerotic lines in the vicinity of
the porous coating were most commonly noted in zone one (49% prevalence
of radiosclerotic lines in zone one). Radiosclerotic lines in zone
one were usually seen in the proximal part of the zone (the shoulder
of the femoral component), and they rarely extended into the distal
half of the zone. Radiosclerotic lines were frequently noted about
the non-porous-coated distal portion of the femoral component. On
the acetabular side, radiosclerotic lines were frequently noted
in zones three and six.
Femoral osteolysis was observed in sixty-five hips (42%).
Its most common locations were zones one and seven. Small lesions
were observed in forty hips (26%), and large lesions were
observed in twenty-five hips (16%). Men had a
significantly higher prevalence of femoral osteolysis than women did
(p < 0.001). Distal osteolysis around the stem tip was noted
in seven hips (4%). All distal lesions were small. Two femoral
components loosened after the progression of proximal femoral osteolysis.
In other patients, a proximal osseous seal limited loosening of
the femoral component and progression of the distal osteolysis (Figs. 2-A and 2-B).
Acetabular osteolysis was observed in fifty-one hips
(33%). The most common locations were zones two and three.
Small lesions were observed in thirty-one hips (20%),
and large lesions were seen in twenty hips (13%).
The prevalence and size of acetabular and femoral osteolytic lesions
differed significantly between the hips with a 32-mm head and those
with a 26-mm head, even when we accounted for differences
in the time since the surgery. Acetabular osteolysis was present
in 49% (twenty-three) of the forty-seven hips
with a 32-mm head and in 26% (twenty-eight) of the
109 hips with a 26-mm head (p < 0.01). Femoral osteolysis
was present in 70% (thirty-three) of the hips
with a 32-mm head and in 30% (thirty-three) of those with
a 26-mm head (p = 0.01). Furthermore, the acetabular and
femoral osteolytic lesions in the hips with a 32-mm head were significantly larger
than the lesions in the hips with a 26-mm head (p < 0.01).
Thigh pain continued to be a problem at the late follow-up evaluation
in this series. The prevalence of thigh pain increased from 15% in
the early period (less than three years postoperatively) to 36% in
the middle period (three to nine years postoperatively), but then
it remained the same at the time of follow-up more than ten years
postoperatively. We previously reported that the prevalence of thigh
pain was 22% (thirty-three of 148 hips) at two
years and 27% (twenty-seven of 100 hips) at five
years2-4. There are two
reasons for this difference in the prevalence of thigh pain between
this study and our previous studies. First, only patients with primary
osteoarthritis were included in the previous studies, whereas all
diagnoses were included in this study. Second, the overall prevalences
of thigh pain at two and five years were reported in our previous
studies, whereas the worst score was chosen from each period in
this study. Thigh pain associated with the PCA stem has been reported
previously by others14, as has
thigh pain with the PCA midstem component (Howmedica, Rutherford,
New Jersey)15. The cause of thigh
pain is multifactorial and might include fibrous fixation, the high
modulus of elasticity of the femoral component, or endosteal irritation2.
At fourteen years, the rate of survival of the PCA total hip replacement
was comparable with that reported, after similar follow-up
periods, by Engh et al.16 and by Xenos et al.14 with
use of 32-mm heads and metal-backed acetabular components.
Owen et al. reported a lower survival rate (73% at seven
years) with use of the same PCA implant with a 32-mm head17. These
historical comparisons are of limited value, however, because of
variation in case-mix and surgeon bias, and they should
be interpreted with caution (see Appendix).
At this late follow-up interval, we found no significant
difference between the survival rates of replacements with a 32-mm head
and those with a 26-mm head when other covariates were accounted
for. However, the outer diameter of the acetabular component was
a significant factor (p = 0.039), as was the time since
surgery (p = 0.003). Acetabular diameter and femoral head
size are both related to polyethylene thickness; therefore, a separate
analysis was performed with use of polyethylene thickness as a covariate
in place of the sizes of the femoral head and the acetabular component.
On the basis of these regression analyses, it appears that the polyethylene thickness
is the most important factor in the long-term survival
of the PCA hip—more important than the sizes of the femoral
head or the acetabular component.
Osteolysis and wear of the polyethylene liner became major problems
associated with the PCA total hip replacement. The prevalence of
femoral osteolysis in our series (42%, sixty-six of
156 hips) was comparable with those reported by Engh et al. (39%,
fifty-four of 138 hips)16 and by Xenos et al. (51%, thirty-nine
of seventy-seven hips)14,
after similar follow-up periods. Both groups of authors
used cementless metal-backed acetabular components with a 32-mm
cobalt-chromium-alloy head. Hellman et al. reported an even higher
prevalence of femoral osteolysis (62%, forty-seven
of seventy-six hips) in a younger patient population18. Failure of the PCA acetabular component
has resulted from the combination of a poor polyethylene locking
mechanism, polyethylene wear, acetabular osteolysis, and migration19-21. Elfick et al. suggested that
the high volumetric wear rate for the PCA prosthesis can be attributed
to its larger head size and the younger, more active patient population
in which it is used20. Astion
et al. found that the prevalence of osteolysis around acetabular
components with an outer diameter of £55 mm (22%,
twenty-five of 116 hips) was significantly higher (p = 0.03)
than the prevalence around those with an outer diameter of 58 mm
(6%, two of thirty-one hips)19 .
This observation is in agreement with our findings. Astion et al.
concluded that the combination of the thinner polyethylene liner
and a 32-mm femoral head exacerbated the problem of wear. Because
of greater production of polyethylene wear debris and associated
osteolysis, 32-mm heads in combination with a small acetabular socket
should be used with caution22,23.
In conclusion, thigh pain, uncertainty about biological fixation
of the implants, and increasing prevalence of osteolysis with time
were problems with this early design of cementless total hip replacement.
Therefore, we abandoned the PCA total hip replacement and now use
a cementless total hip replacement with a tapered/proximal
fixation femoral component.