Twenty-six subjects (thirteen men and thirteen women) were recruited
from the private practices of the two senior authors (P.M.P. and
E.A.S.). The project was approved by the institutional review board,
and informed consent was obtained from all volunteers. Subjects
underwent unilateral primary total hip arthroplasty with a hybrid
construct (a cementless acetabular component and a cemented stem).
The mean age of the subjects at the time of surgery was 67.6 years
(range, forty-five to seventy-nine years). The underlying diagnosis
was osteoarthritis in each patient. The same femoral component design
(Omnifit; Osteonics, Allendale, New Jersey) and the same hemispherical
titanium-alloy acetabular component design (Trilogy; Zimmer, Warsaw,
Indiana) were implanted in all patients. All femoral components
were secured with cement with use of a distal plug, canal pressurization,
and vacuum mixing for porosity reduction. Press-fit fixation of
all acetabular components was achieved by underreaming the socket
by 2 mm. No supplemental screw fixation was used. Patients who required
supplemental screw fixation were excluded from the study because
the presence of acetabular screws precluded accurate measurement of
retroacetabular bone-mineral density. Patients with bilateral hip
involvement were also excluded so that the contralateral (nondiseased)
hip could serve as a valid internal control for each patient. Because
of the exclusion criteria and the voluntary basis for participation,
the study cohort was a nonconsecutive, selected series of patients.
Bone-mineral density was measured with quantitative computed
tomography36-38. Subjects were
positioned supine on a High Speed Advantage Gantry CT2 unit (General
Electric Medical Systems, Milwaukee, Wisconsin; manufactured in
1995). Frontal scout images were made to confirm proper pelvic position
and true axial orientation of the imaging planes (Fig. 1). A standard-density
calibration phantom was positioned in the field directly anterior
to the patient, and the scanner was recalibrated prior to analysis
of each hip. True three-dimensional volumetric density calculations
were performed with standard vertebral densitometry software (General
Electric, 1995).
Bone-mineral density was assessed within a specific region of
interest that was defined by a cylinder of cancellous bone centered
within the ilium directly cephalad to the acetabular dome (Fig. 2). Density was
calculated at five separate levels within the region of interest.
These five circular axial windows were separated by 2.5-mm increments.
The cross-sectional area of the region of interest was 100 mm2 (Fig. 3). The level
of the most caudal axial window was positioned to be tangential
to the most cephalad point of the acetabular implant.
Bone-mineral density was quantified at two time-points. Baseline
bone-mineral density was measured within the first five days following
the total hip arthroplasty. Follow-up bone-mineral density was measured
after an average (and standard deviation) of 1.28 ± 0.2
years (range, 1.00 to 1.85 years). Bone-mineral density was measured
at corresponding levels within the contralateral (nondiseased) ilium
at both time-points to serve as an internal control. A mean bone-mineral-density
value was calculated for each of the five levels on both the treated
and the untreated side at the two time-points. Statistical comparisons
between the treated and untreated sides were performed with paired
t tests. The same form of statistical analysis was employed for
comparison between baseline measurements and those made at the time
of follow-up.
At the time of the follow-up density measurements, all patients
were queried regarding their satisfaction with the result of the
hip surgery, their resultant activity level, and the presence of
any residual hip symptoms.
Patient Positioning and Measurement Reproducibility
The validity of side-to-side density comparisons was predicated
upon the fact that the imaging planes intersected both sides of
the pelvis at corresponding levels. This was ensured by examination of
the frontal computed tomography scout images (Fig. 1). The goal in
positioning the patient was to establish parallelism between (1)
the five horizontal scout-film lines (representing the five axial
imaging planes) and (2) a reference line tangential to both teardrops
(modified Hilgenreiner line). The vertical positions of the imaging
planes were adjusted so that the line corresponding to level 1 was
tangential to the most cephalad portion of the acetabular component
(Fig. 1).
Pelvic rotation within the axial plane was assessed on the frontal
scout images by noting (1) the morphology of the obturator foramen
and (2) the alignment between the symphysis pubis and the lumbar spinous
processes. Axial rotation, even if present, was not a potential
source of error because it does not alter the imaging plane of an
axial tomogram.
At the time of positioning of the patient for follow-up quantitative
computed tomography, the initial and follow-up frontal scout images
were examined simultaneously to confirm duplication of the orientation
of the axial imaging planes.
The amount of pelvic tilt (rotation in the sagittal plane) was
evaluated by observing (1) the profile of the obturator foramen
on the frontal scout films and (2) the contour of the cross sections
of the ilium on the tomographic images. Variability between the amount
of pelvic tilt on the baseline scans and that on the follow-up scans
is a potential source of error. However, we detected negligible
variability between the amounts of pelvic tilt that individuals exhibited
at the times of their baseline and follow-up scans.
A formal reproducibility study was conducted to assess the validity
of quantitative computed tomography as a means of consistently measuring
pelvic bone-mineral density. Six healthy volunteers underwent quantitative
computed tomography of the hips according to the protocol specified
above. Because none of these subjects had hip prostheses, the position
of the most caudal imaging plane (level 1) was set to be tangential
to the most cephalad aspect of the acetabular subchondral plate.
The subjects were placed in the scanner a second time, and repeat
density measurements were obtained. Pelvic position and the levels
of the five axial imaging planes were duplicated at the time of
reimaging with the methods described above.
High correlation was found between the initial and repeated bone-mineral-density
values at levels 2 through 5 (correlation coefficients 0.89). The reproducibility
at level 1 (correlation coefficient = 0.80) was inferior
to that at the more cephalad levels, but it was still satisfactory.
In fact, the difference between repeated measurements at each of
the five levels was always <10% of the average
of the two measurements.
The diminished reproducibility at level 1 was attributed to the
abrupt transition between cancellous and cortical bone that occurs
along the subchondral plate of the acetabulum. The bone-mineral-density gradient
within the ilium is more gradual at the more cephalad levels.
At both time-points, bone-mineral density within the region of
interest was greatest immediately adjacent to the implant (level
1), and it progressively decreased with further distance cephalad
to the acetabulum. This trend was consistent on both the treated
and the untreated side (Figs. 4, 5, and 6).
The mean baseline bone-mineral-density values were greater on
the treated side than on the untreated (control) side at all five
levels (Fig. 4).
The differences in baseline values were significant (p < 0.05)
at levels 1 through 4 but not at level 5 (p = 0.2).
At the time of follow-up, all patients were satisfied with the
outcome of the arthroplasty. All hips were asymptomatic, and all
patients had returned to unrestricted walking about the community.
Plain radiographs demonstrated no component migration and no radiolucent
lines.
The mean bone-mineral-density values at the time of follow-up
were greater on the untreated (control) side than on the treated
side at all five levels (Fig. 5). The differences were significant
at levels 2 and 3 (p = 0.008 and 0.04, respectively), and
they approached significance at levels 1 and 4 (p = 0.08 and
0.06, respectively). The difference at level 5 lacked significance
(p = 0.2).
On the treated side, bone-mineral density decreased between the
two time-points at all five levels (Fig. 6). This decrease was significant
at each level (p £ 0.001). The largest absolute decreases
in bone-mineral density occurred at levels 1 and 2 (75 and 76 mg/cc,
respectively). These values represented relative declines of 26% and
33%, respectively. The absolute (and relative) magnitudes
of the interval reductions in bone-mineral density ranged from 35 mg/cc
(20%) to 76 mg/cc (33%). The magnitude
of the interval reduction diminished with increasing distance cephalad
to the acetabulum. However, even at level 5 (10 mm cephalad to the
acetabular dome), the decrease (35 mg/cc [20%])
remained significant (p = 0.001). The percentage decrease
at each level far exceeded the resolution limit of the quantitative
computed tomography technique that had been established in the reproducibility
analysis. Thus, the measured declines were considered to be valid.
The subject cohort manifested considerable heterogeneity in terms
of their baseline and follow-up bone-mineral-density values. However,
despite the relatively large standard deviations at each level, the
temporal bone-mineral-density trends were extremely consistent.
Hence, the significance of the decline in bone-mineral density in
the treated hip was unequivocally established.
Evaluation of each of the tomographic images revealed a confluent
pattern of bone-mineral-density distribution—that is, there
was no evidence of focal bone resorption (osteolysis).
On the untreated side, bone-mineral density increased slightly
between the two time-points at all five levels. This increase was
not significant at any level (p 0.07).
The lack of significant alteration in bone-mineral density between
the two time-points on the untreated side suggests that the untreated
side served as a reliable internal control. Theoretically, the slight
interval increase in bone-mineral density on the untreated side
may represent an adaptive mechanical response to a general increase
in weight-bearing activity afforded by a successful contralateral
total hip arthroplasty34,35,39-41.
The discrepancy between baseline bone-mineral density on the
treated (diseased) side and that on the untreated (control) side
(Fig. 4) is
consistent with the fact that subchondral sclerosis is a fundamental
characteristic of osteoarthritis. However, the interval reduction
in bone-mineral density on the treated side (Fig. 6) cannot simply
be attributed to a return to "normal" bone-mineral
density because, at the time of follow-up, bone-mineral density
on the treated side had not equilibrated with that of the contralateral (nondiseased)
hip. Rather, the bone-mineral density on the treated side had decreased
to a value that was lower than that on the contralateral (nondiseased) side
(Fig. 5).
The confluent pattern of osseous resorption on the treated side
excludes osteolysis as an explanation for the reduction in bone-mineral
density. Both varieties of osteolysis (cavitary and linear) produce discrete
areas of bone resorption4,42,43.
We propose a mechanical explanation for the observed trend of
bone-mineral resorption on the treated side. Specifically, we believe
that the decline in retroacetabular bone-mineral density reflects
a remodeling response to decreased stress on the regional pelvic
bone. The retroacetabular cancellous bone of the central part of
the ilium is mechanically shielded because the metal-backed acetabular
implant is stiffer than the host bone.
When the acetabulum is resurfaced with a construct that is stiffer
than the native subchondral plate, the elasticity mismatch focuses
contact stress at the peripheral zones of the host-implant interface44-46. Hence, a greater portion of
the weight-bearing load is transmitted to the peripheral cortex
of the ilium, and, consequently, the cancellous bone of the central
part of the ilium is subjected to less force than under ordinary
circumstances (Fig. 7)8,47-51.
Finite element analyses have suggested that press-fit fixation of
hemispherical acetabular components further heightens rim stress
and intensifies transmission of weight-bearing force to the peripheral cortex
of the ilium as a result of generation of hoop stresses at the acetabular
orifice and decreased host-implant contact area (gap formation)
at the dome region32,33,52-54.
The trabecular bone of the central part of the ilium is shielded
as the press-fit implant transfers force to the cortical shell of
the ilium (Fig. 7)32,33,55-59. Applications of computer-simulated
remodeling theory predicted resultant attenuation of bone density
behind the acetabular dome33-35.
The findings of the present study provide clinical corroboration
of the predictions generated by these finite element analyses and
computer simulations.
As with femoral stress-shielding, the most definitive clinical
consequence of retroacetabular stress-shielding is its potential
to compromise component revision. Supplemental screw fixation is
often required to stabilize uncemented implants at the time of acetabular
revision. The strength of dome-screw purchase depends on bone quality
within the central part of the ilium. Thus, a decrease in bone density
in the retroacetabular region decreases the fixation strength of
revision acetabular components with dome screws.
Retroacetabular stress-shielding may simply reflect stable osseous
integration of the acetabular component and successful adaptation
of local bone architecture to its new functional requirements. However,
theoretical concern remains that periprosthetic stress-shielding
could eventually lead to component-loosening if the stress-shielding
proves to be progressive. As the regional bone-mineral density adjusts
to its new mechanical environment, incremental attenuation of bone-mineral
density could eventually reach a level at which the bone-implant
interface becomes critically compromised. Although studies of proximal
femoral stress-shielding by Engh et al. showed that a new steady
state is characteristically achieved within the first two years1,6, other studies have suggested that
this process may not be a self-limiting phenomenon13,60,61. Intermediate-term clinical
follow-up of press-fit acetabular components continues to demonstrate favorably
low loosening rates. However, we must bear in mind that alarmingly
high rates of loosening of cemented acetabular implants did not
become evident until long-term follow-up had been carried out. If
acetabular stress-shielding proves to be a progressive phenomenon,
then it could be a contributing factor to late failures of press-fit
components. The long-term natural history of retroacetabular stress-shielding
and its effect upon the longevity of implant fixation remain to
be defined.
It is uncertain whether this putative stress-shielding effect
would be encountered to the same extent with other types of acetabular
components. It is conceivable that more bone stock within the central
part of the ilium could be preserved by selecting acetabular component
design features and fixation methods that more effectively transmit
forces to the central part of the ilium (rather than principally
to the acetabular rim). Such features and methods have yet to be
defined, but perhaps they are a worthy subject for future research.
A component design that transmitted load more diffusely to the acetabulum-pelvis interface
could conceivably promote a more symmetric pattern of bone ingrowth.