Aseptic loosening continues to be the primary mode of long-term
failure after total hip arthroplasty1.
Particulate debris generated from the wear of the prosthetic components
and the ensuing inflammatory response resulting in osteolysis have
been implicated as the mechanism by which aseptic loosening occurs.
Although wear debris from any of the components (metal, cement,
or polyethylene) can stimulate the inflammatory response, wear of
the polyethylene bearing surface has been recognized as the primary
source of this debris2. Efforts
to limit the generation of polyethylene wear debris have focused
both on improving polyethylene bearing surfaces (decreasing polyethylene
oxidation and increasing polyethylene cross-linking) and on exploring
alternative bearing surfaces (ceramic-on-polyethylene, ceramic-on-ceramic,
and metal-on-metal).
Alumina ceramic was introduced as a bearing surface in the 1970s
as an alternative to the metal-on-polyethylene couplings because
of its inertness, coefficient of friction, wettability, and hardness.
Subsequent hip-simulator trials of ceramic-on-polyethylene
couplings demonstrated up to twenty times less wear compared with
that of cobalt-chromium-molybdenum-polyethylene couplings. Ceramic-on-ceramic
coup-lings showed even less wear—up to 100 times less
wear—than their metal-on-polyethylene counterparts3.
The purpose of this investigation was to evaluate the seventeen
to twenty-one-year results of total hip arthroplasties, performed
by one surgeon, with a Charnley-Müller femoral prosthesis
and an all-polyethylene cup, both cemented with use of first-generation
cementing techniques. To our knowledge, we performed the longest
follow-up to date of -ceramic-on-polyethylene total hip
prostheses.
From April 1978 to February 1981, the senior author (R.K.M.)
performed sixty-four (sixty-one primary and three revision) nonconsecutive
total hip arthroplasties using -ceramic-on-polyethylene
bearing surfaces in fifty-six patients. (Eight procedures were bilateral.)
The availability of the ceramic head was the primary variable determining
which femoral bearing surface was implanted. These sixty-four arthroplasties
comprise the focus of this study.
There were twenty-five men (twenty-eight hips) and thirty-one
women (thirty-six hips). The average age at the time of the index
arthroplasty was sixty-nine years (range, fifty-one to eighty-four
years). The average weight of the patients was 166 lb (75.3 kg),
with a range of 100 to 286 lb (45.4 to 129.7 kg). The preoperative
diagnosis was osteoarthritis in fifty-five hips (86%),
avascular necrosis in four (6%), a failed previous prosthesis
in three (5%), and rheumatoid arthritis in two (3%). The
arthroplasties were equally distributed between the left and right
hips, with thirty-two performed on each side. The average preoperative
Harris hip score was 55 points (range, 16 to 78 points).
The femoral component implanted in each patient (Fig. 1) consisted of
a stainless-steel Charnley-Müller prosthesis (DePuy, Warsaw,
Indiana) with a modular 32-mm alumina ceramic head (Feldmuhle, Plochingen,
Germany) affixed to the trunion of the femoral component at the
time of surgery. The nonmodular acetabular component (DePuy, Warsaw, Indiana)
consisted of ultra-high molecular weight polyethylene that was sterilized
by gamma irradiation in air. In this study, acetabular components
of three different diameters (44, 50, or 54 mm) were inserted on
the basis of intraoperative determinations of the diameter of the
osseous acetabulum. Both the femoral and the acetabular components
were inserted with use of Simplex P cement. The cement for the femoral
component was inserted by finger-packing, and no femoral canal plug
was used (first-generation cementing technique).
For the latest follow-up evaluation, efforts were made to contact
each patient or his or her family. Surviving patients were asked
to return for clinical and radiographic evaluation. Those who had
relocated and were unable to return to our institution were assessed,
with use of the standard system of terminology for the reporting
of results described by Johnston et al.4,
by a local orthopaedist or physical therapist who was instructed
to forward the clinical results and radiographs to us. The families
of the patients who had died were interviewed to determine whether
the patient had been satisfied with the result of the arthroplasty
and whether he or she had undergone revision prior to death.
At the time of the latest follow-up, nineteen patients (twenty-two
hips) were still alive and thirty-seven patients (forty-two hips)
had died. No patient was lost to follow-up. Four of the twenty-two
total hip prostheses in the surviving patients and one of the forty-two
prostheses in the deceased patients had been revised. Therefore,
eighteen hips had retained the index prosthesis and are the focus
of this seventeen to twenty-one-year investigation.
Clinical evaluation consisted of quantifying, with use of Harris
hip scores5, the functional level,
pain, gait, limp, muscle strength, and range of motion of each surviving
patient.
Standard anteroposterior radiographs of the pelvis and ipsilateral
hip were made at the time of follow-up (Fig. 2) and were compared with the same
views made within three months after the index arthroplasty. Each
patient’s radiographs were assessed for evidence of loosening,
osteolysis, radiolucent lines, calcar resorption, heterotopic ossification, and
wear.
Loosening of the acetabular and femoral components was categorized
according to previously accepted criteria6-9.
Definite loosening was defined as the presence of a radiolucent
line at the interface between the acetabular or femoral component
and the surrounding cement that had not been seen on the immediate
postoperative radiograph, subsidence or migration of either component,
or a fracture of the cement or component. Subsidence of the femoral
component was determined with the method of Loudon and Charnley10, and migration of the acetabular
component was determined with the criteria of Massin et al.11. Probable loosening was defined
as the presence of a continuous radiolucent line involving 100% of
the cement-bone interface. Possible loosening was defined as a radiolucent
line involving 50% to 99% of the -cement-bone
interface that had not been present on previous radiographs.
Osteolysis was defined as any area of progressive nonlinear radiolucency
surrounding the prosthesis and measuring 2 mm in its greatest diameter9. Radiolucent lines were defined as
progressive, linear, lucent areas surrounding the prosthesis, and
generally parallel to it, measuring £2 mm in their greatest
thickness9. The locations of these
areas on anteroposterior radiographs were recorded with use of the
femoral zones described by Gruen et al.12 and
the acetabular zones described by DeLee and Charnley13. Calcar resorption was considered
meaningful if the difference between the measurement between the
undersurface of the collar of the prosthesis and the stump of the
neck on the early postoperative radiograph and the same measurement
on the latest radiograph was >3 mm14.
Heterotopic bone, when present, was graded according to the classification
system of Brooker et al.15.
Wear was determined with the technique described by Livermore
et al.16. One independent observer
with experience with the technique performed the measurements. The
shortest distance between the center of the femoral head and the
periphery of the acetabular component was measured. The difference between
this length on the immediate postoperative radiograph and that on
the most recent radiograph was divided by the time that had elapsed
between the two radiographs to determine the linear wear rate. The
volumetric wear rate was calculated by multiplying pr2 by
the linear wear rate. Measurements were made with a digitizing stylus
and tablet (SigmaScan; Jandel Scientific, La Jolla, California)
with a reported accuracy of 0.025 mm. Magnification was standardized
against the known circumference of the femoral head.
The prevalence of revision was determined for all sixty-four hips
initially entered into the study and for the twenty-two hips in
the survivors at the time of the latest follow-up. The reasons for
each revision were also documented. Survivorship analysis with use
of the Kaplan-Meier method was performed with failure defined as
revision of any component of the arthroplasty for any reason. Statistical
analysis with use of the Pearson correlation coefficient, Spearman
correlation coefficient, t test, and Fisher exact test was carried
out to determine the correlation of linear wear rate with numerous
variables. These variables included the age, gender, and weight
of the patient; osteolysis; loosening; preoperative diagnosis; thickness
of the acetabular component; radiolucent lines; heterotopic ossification;
and calcar resorption.
Patient Demographics
The average age of the nineteen patients (twenty-two hips) who
were still alive at the time of the latest follow-up was eighty-two
years (range, seventy-two to ninety-two years). The average duration
of follow-up of these nineteen patients was 18.2 years (range, 17.2
to 21.3 years). The average age at time of death of the thirty-seven
patients (forty-two hips) who had died was seventy-nine years (range,
sixty-seven to ninety-three years). Ten patients (eleven hips) died
within five years after the index arthroplasty; thirteen (fourteen
hips), between five and ten years after it; six (eight hips), between
ten and fifteen years after it; and eight (nine hips), between fifteen
and twenty years after it.
Hip Scores
Preoperatively, the Harris hip scores for all fifty-six patients averaged
55 points (range, 16 to 78 points). The sixteen patients (eighteen
hips) who were alive and had retained the index prosthesis at the
time of the latest follow-up had had an average preoperative Harris
hip score of 55 points (range, 35 to 78 points). At the time of
the latest follow-up, the average Harris hip score was 88 points
(range, 72 to 99 points). Of the sixteen patients available for
this follow-up, seven had an excellent result, seven (nine hips)
had a good result, and two had a fair result. The two patients with
a fair result were nursing-home residents with advanced senile dementia
and other medical comorbidities. These two patients had a limited
range of motion and required assistance for activities of daily
living but were pain-free and able to walk with the aid of assistive -devices.
Satisfaction
Of the eighteen hips in the patients who were alive and had retained
the prosthesis at the time of the most recent follow-up, sixteen
were considered by the patient to have improved function, to be
less painful, and to have a satisfactory result. The remaining two
patients were unable to provide accurate subjective information.
One of the deceased patients had undergone a revision and, therefore,
no efforts were made to contact his family.
Information was obtained from a family member for thirty-two
of the thirty-seven patients who died before the time of the latest
follow-up. All thirty-two patients were thought to have been satisfied
with the overall result of the total hip arthroplasty before their
death. The relatives of four patients could not be reached.
Aseptic Loosening and Radiolucent Lines
None of the eighteen unrevised hips in patients who were alive
at the time of the latest follow-up had evidence of definite, probable,
or possible loosening of the acetabular component. One of the femoral
components was definitely loose according to the criteria of Harris
et al.6. There was no probable
or possible loosening of the femoral component.
Two hips had progressive radiolucent lines at the cement-bone interface
of the acetabular component. These radiolucent lines were seen in
zone III in one hip and in zones I and III in the other. In the
hip with radiolucent lines in zones I and III, the lines did not
involve 50% of the cement-bone interface and, therefore,
were not considered evidence of pos-sible loosening.
Progressive radiolucent lines at the cement-bone interface of the
femoral component were encountered in three hips. The radiolucent
lines were in zones 1 and 7 in one hip and in zone 7 in the other
two. These radiolucent lines did not encompass 50% of the
cement-bone interface, and therefore were not considered possibly
loose.
Resorption of the calcar, as defined by Blacker and Charn-ley14, was encountered in two hips. Two
other hips had 1.5 to 2.0 mm of calcar resorption. The remaining
fourteen hips had no evidence of calcar resorption.
Wear and Osteolysis
The average rate of linear wear of the all-polyethylene acetabular
component was 0.034 mm/yr (range, 0.00016 to 0.077 mm/yr)
in the eighteen hips. The average volumetric wear was 28.012 mm3/yr
(range, 0.13 to 62.09 mm3/yr).
Osteolysis was not encountered around either the femoral or the
acetabular component in any of the eighteen hips at the time of
the latest follow-up.
Heterotopic Ossification
Heterotopic ossification had developed in five of the eighteen hips
at the time of the latest follow-up. According to the criteria of
Brooker et al.15, three hips had
grade-I ossification and two had grade-II ossification.
Revision Rates
Fifty-nine (92%) of the sixty-four original prostheses
were functioning or had been in place when the patient died. Of
the twenty-two hips in the patients who were still alive at the
time of the latest follow-up, four (18%) had been revised
and eighteen (82%) had retained the original prosthesis.
The remaining revision was in a patient who had died before the
time of the latest follow-up. Revision of the index prosthesis was
performed because of loosening of the acetabular component alone
in one hip (at fourteen years), loosening of the femoral component
alone in two hips (at four years and four and one-half years), and
loosening of both the acetabular and the femoral component in two
hips (at three and eighteen years).
No femoral stem or ceramic head fractured.
Statistical and Survivorship Analysis
The Kaplan-Meier survivorship analysis with 95% confidence intervals
demonstrated that the probability of retention of the index prosthesis
at five, ten, fifteen, and twenty years was 95% (89% to
100%), 95% (85% to 100%), 89% (81% to 100%),
and 79% (57% to 100%), respectively (Fig. 3).
With the numbers available, no correlation was found between
linear wear rates and age, gender, weight, osteolysis, loosening,
preoperative diagnosis, thickness of the acetabular component, radiolucent
lines, heterotopic ossification, or calcar resorption.
Ceramics were introduced in the 1970s as an alternative bearing
surface with superior wear characteristics compared with those of
the established metal-on-polyethylene couplings used at that time
for total hip arthroplasty. Nearly thirty years have passed, and
metal-on-polyethylene coup-lings continue to be the most
commonly used bearing surfaces. Although much knowledge has been
gained and improvements have been made in the metals and polyethylene
that make up the standard bearing surfaces used today, the failure
of ceramic to supplant metal as the preferred femoral head material
is principally due to the fact that fracture of the ceramic head
has been reported in both ceramic-on-ceramic and -ceramic-on-polyethylene
couplings17. The rate of this
deva-stating complication has been as high as 13.4%18. In the last decade, however, there
has been renewed interest in ceramic bearing surfaces in total hip
arthroplasty. It is now believed that many of the failures of ceramic
femoral heads resulted from correctable manufacturing and material
flaws. A lack of material standardization for the early-generation ceramics
resulted in inferior material properties such as large grain sizes
and increased porosity. This inferior quality of the ceramic coupled
with imprecise taper mating surfaces (between the ceramic head and
the trunion of the femoral stem) is believed to have created stress
risers within the ceramic, leading to an increased susceptibility
to fracture. Current-generation ceramics are manufactured with stricter standardization,
increased strength of the ceramic (zirconium oxide), and a more
precise Morse-type taper. These modifications have led to a substantial
decrease in the prevalence of fractures of ceramic components (0% to
2.0%)17.
In this study, although early-generation alumina ceramic femoral
heads were implanted in sixty-four hips, no femoral head fractured.
We attribute the absence of ceramic fracture in our series to careful
intraoperative handling of the ceramic component and the surgeon
ensuring both a precise Morse-taper mating and that the taper mating
surfaces of the femoral head and stem remained debris-free
during impaction.
There have been numerous studies on the in vivo wear
of metal-on-polyethylene total hip replacements. A wide range of
linear wear rates has been reported, with the majority ranging from
0.08 to 0.10 mm/yr (Table I). In contrast, there have been
only a few studies on the in vivo wear of ceram-ic-on-polyethylene
total hip replacements. These too have demonstrated a relatively
wide range of wear rates (Table II), with several as low as 0.03 mm/yr19-21. The average linear wear rate
of 0.034 mm/yr in our study is consistent with the lowest -previously
reported ceramic-on-polyethylene wear rates and, correspondingly,
compares favorably with the wear rates in historical studies of
metal-on-polyethylene replacements.
A correlation between increased wear rates and calcar resorption
was reported by Saito et al.22.
Sugano et al.23 reported similar
findings with longer follow-up of the same group of patients. In
our study, no correlation between increased wear and calcar resorption
was identified, with the numbers available.
The 8% revision rate at an average of eighteen years
in the present study compares favorably with the revision rates
in previously reported long-term studies16,24-26,
despite the use in our study of what is now considered an inferior
stem design (Charnley-Müller), first-generation cementing
techniques, and a head size (32 mm) associated with greater volumetric
wear.
The absence of osteolysis in association with low wear rates in
our study support the finding of a direct correlation between wear
rates and osteolysis in the study by Dowd et al.27.
In that study, hips with a wear rate of <0.1 mm/yr
were found to have no osteolysis after ten years of follow-up. In contrast,
osteolysis was encountered in 43% of hips demonstrating
a wear rate between 0.1 and 0.2 mm/yr, in 80% of hips
with a wear rate between 0.2 and 0.3 mm/yr, and in 100% of
hips with a wear rate of >0.3 mm/yr.
Shortcomings of our study include the small number of patients
available at the time of the latest follow-up and the inherent problems
with the Livermore technique16.
While the Livermore technique is commonly employed to measure polyethylene
wear in total hip replacements, it only measures wear, or linear
penetration, that occurs in the plane of the radiograph. Furthermore,
the interobserver and intraobserver reliability of this technique
has been reported to be poor28.
In an effort to minimize this potential source of error, one individual
at an outside university who was experienced with the technique
performed all of the wear measurements. Another shortcoming of our
study is the nonconsecutive nature of the series of patients, which
potentially introduces a selection bias. However, all patients presenting
to the senior surgeon for total hip arthroplasty between 1978 and
1981 received a ceramic femoral head as long as one was available. The
demographics of the nineteen hips that instead received a metal
head during the same time frame did not differ with respect to age
or diagnosis at the time of the index arthroplasty. The strengths
of this study are that it represents one surgeon’s experience,
no patients were lost to follow-up, and to our knowledge it is the
longest follow-up study of ceramic-on-polyethylene couplings in
total hip arthroplasty.
In summary, this study demonstrates the value of ceramics as a
femoral bearing surface in total hip arthroplasty. As the indications
for total hip arthroplasty continue to expand to include younger,
more active patients, the superior wear characteristics of ceramics
provide an attractive alternative to contemporary metal-on-polyethylene
designs.