Abstract
Background: Progressive periacetabular osteolysis following total
hip arthroplasty may require revision surgery. The purpose of this study was
to use computed tomography scans of hemipelves retrieved at autopsy from
patients who had had a total hip arthroplasty, to define the radiographic
characteristics that differentiate clinically important osteolytic lesions
from osteoarthritic bone cysts.
Methods: We analyzed forty-four hemipelves that had been retrieved
at autopsy at a mean of eight years after a total hip arthroplasty with an
uncemented acetabular component. Computed tomography images were analyzed to
identify the location, volume, and presence of cortical erosion and/or
communication pathways with the joint space for all periacetabular bone
defects. Lesions that were not present on preoperative or immediate
postoperative plain radiographs were defined as new lesions. These new lesions
were compared with those that were present on preoperative or immediate
postoperative plain radiographs, which were defined as preexisting
lesions.
Results: Forty-six lesions were identified on computed tomography,
and sixteen of them were preexisting lesions. The mean volume of the
preexisting lesions was 1.5 ± 1.5 cm3, which was
significantly smaller than the mean volume of 5.6 ± 11.4 cm3
of the thirty new lesions (p = 0.034). Twenty-eight of the thirty new lesions
had a clear communication pathway with the joint space, while thirteen of the
sixteen preexisting lesions demonstrated no communication pathway. New lesions
were significantly more likely to communicate with the joint space than were
preexisting lesions (p < 0.001). Cortical erosion was seen in sixteen of
the thirty new lesions; none of the sixteen preexisting lesions exhibited
cortical erosion (p < 0.001).
Conclusions: The most important difference between osteolytic
lesions and preexisting bone defects was the presence of a communication
pathway to the joint space. Lesions that did not have an identifiable
communication to the joint space were smaller and were not associated with
cortical erosion. Lesions with communication to the joint through multiple
pathways or through a central dome hole were larger and more likely to be
associated with cortical erosion.
Clinical Relevance: Periacetabular lesions that are not present on
perioperative plain radiographs and that have a communication pathway with the
joint space and associated cortical erosions as seen on computed tomography
are likely to be osteolytic lesions.
Uncemented cup fixation is commonly used in primary total hip
arthroplasty and has shown excellent mid-term and long-term
results1-5,
but pelvic osteolysis is one of the major obstacles threatening the long-term
success of these uncemented acetabular
cups6-9.
Pelvic osteolysis has been observed with use of computed tomography in
association with as many as 52% of total hip
replacements10,11.
We have observed, in our clinical practice, that it is difficult to determine
on computed tomography whether a lesion represents total hip
arthroplasty-induced osteolysis or the remnants of a preoperative
osteoarthritic cyst, especially if the preoperative and immediate
postoperative radiographs are not available for comparison with the computed
tomography scan.
The purpose of this study was to use computed tomography scans to define
the characteristics that could be used to differentiate between clinically
relevant osteolysis and other bone defects, such as osteoarthritic bone cysts,
that existed prior to a total hip arthroplasty. We hypothesized that new
lesions would have larger volumes, would more frequently communicate with the
joint space, and would be more likely to be associated with cortical bone
erosion.
We examined forty-four hemipelves containing uncemented titanium acetabular
components with a modular polyethylene liner. All of the acetabular components
had been implanted at our institution with the same press-fit
method12. All of
the hemipelves were retrieved from individuals who had provided appropriate
consent for the use of their autopsy material in research studies. The index
arthroplasties had been performed because of primary osteoarthritis in
thirty-nine hips, rheumatoid arthritis in three hips, osteonecrosis in one
hip, and posttraumatic arthrosis in one hip. None of the hips had a prior
infection or had undergone structural bone-grafting.
The implanted acetabular components were of six different designs: eighteen
were Duraloc 100 cups (DePuy, Warsaw, Indiana), five were Duraloc 1200 cups
(DePuy), eleven were Triloc cups (DePuy), eight were Arthropor cups (Joint
Medical Products, Stamford, Connecticut), one was a PSL cup (Osteonics,
Allendale, New Jersey), and one was a Harris-Galante II cup (Zimmer, Warsaw,
Indiana). Screws were used with the Arthropor cup design to secure the
polyethylene liner, but they were not used with any of the other designs. The
implanted femoral components were all cast chromium-cobalt anatomic medullary
locking stems (AML;
DePuy)13. The
femoral head was chromium-cobalt in thirty-five retrieved specimens and
ceramic in nine. Twenty-four femoral heads were 32 mm in diameter, and twenty
femoral heads were 28 mm in diameter.
The forty-four hemipelves were from twenty-two men and twenty-two women
with a mean age at the time of surgery of 69.9 years (range, forty-two to
eighty-seven years). The mean age at the time of death was 77.9 years (range,
forty-seven to ninety-five years). The acetabular components were in situ for
a mean of 8.1 years (range, 1.7 to 15.9 years). All acetabular and femoral
components had stable fixation and had been functioning well at the time of
the patient's death.
Plain Radiographs
One author (C.A.E.), who was blinded to the results of the computed
tomography scans, examined the preoperative and immediate postoperative
radiographs for preexisting bone defects. The anteroposterior pelvic and iliac
oblique radiographs for each hip were placed in random order, and the reviewer
assessed each of the radiographs independently of one another. The sclerotic
border of all cystic bone defects was outlined for later comparison with the
three-dimensional computed tomography reconstruction derived from the
postmortem computed tomography images.
All clinical follow-up radiographs were also reviewed. Although we had
excellent-quality preoperative and immediate postoperative radiographs of the
hips in this study, we did not have an adequate number of annual radiographs
to perform a serial analysis of lesion growth. Only twenty-eight hips had one
or more radiographs that had been made more than three years postoperatively,
and only three hips with new lesions had clinical radiographs with visible
lesions at more than two time-points. Additionally, the time-intervals between
the last clinical radiograph and the postmortem analysis (specimen radiographs
or computed tomography images) were too variable to allow a study of the rate
of lesion growth. The time-interval between the last clinical radiographs and
the deaths of the patients ranged from 0.3 to 9.4 years (mean [and standard
deviation], 3.0 ± 2.4 years).
Computed Tomography
The use of retrieved specimens enabled us to optimize the image quality of
the computed tomography scans since we were able to image each hemipelvis
separately after we removed the femoral head and stem, which often generate
most of the image artifacts associated with clinical computed tomography
scans.
Each retrieved specimen was scanned in 1-mm axial slices in standard mode
(GE HiSpeed Advantage; General Electric, Waukesha, Wisconsin; and Somotom
Sensation 4; Siemens, Munich, Germany) at 120 kV and 220 mA. Coronal and
sagittal images were reconstructed from the axial images
(Fig. 1). The original data
were automatically segmented on the basis of statistical properties, and an
experienced orthopaedic surgeon reviewed the raw and segmented data to
identify all periacetabular bone defects. For this study, we defined
osteoarthritic cysts and other bone defects that existed prior to total hip
arthroplasty as "preexisting lesions." Expansile regions devoid of
bone that were not visible on preoperative and/or immediate postoperative
radiographs were subsequently referred to as "new lesions."
A three-dimensional computed tomography model was made of each specimen
that had one or more bone defects. This model included the acetabular
component, the bone defects, and the periacetabular bone, which was rendered
translucent, allowing us to identify the defect location through the bone. The
three-dimensional model was oriented to match the contour and size of the bone
as seen in the radiographs (Fig.
2). We then defined a bone defect as a "preexisting
lesion" if it had the same appearance and location on both the
postmortem computed tomography and the preoperative or immediate postoperative
plain radiographs. The three-dimensional volume of each lesion was calculated
with use of a computer-aided imaging program (VirtualScopics, Rochester, New
York) (Fig. 3). The
three-dimensional location of each lesion was determined by its presence in
one or more of the five zones that comprise the cup surface: anterior,
posterior, superior, inferior, and central
(Fig. 4). The diameter of the
central zone (Zone 5) was defined as one-half of the cup diameter on a
two-dimensional projection of the cup surface. The peripheral zones (Zones 1
through 4) were equally divided by four lines drawn ±45° from the
cup center with respect to the superior-inferior central axis through the most
proximal and distal points of the intersection of the coronal plane and the
edge of the cup. A lesion was associated with a particular zone if it occupied
>40% of the cup surface within that zone or if half of the total area of
involvement of the lesion on the cup surface fell within that zone. If a
lesion spanned more than three zones and did not occupy >40% of the cup
surface within any zone or if half of the total area of involvement of the
lesion on the cup surface did not fall within any zone, it was assigned to a
zone with the most cup surface area involvement.
The computed tomography images were examined to determine the presence and
pattern of communication between the bone defect and the joint space. We also
evaluated whether the lesions exhibited cortical thinning or perforation and
measured the maximum width of the cortical perforation. A cortical perforation
was defined as a defect wider than 2 mm. All other instances of cortical
erosion were classified as cortical thinning.
Data Analysis
A Mann-Whitney U test for two independent samples was used to compare the
volume of preexisting and new lesions. Chi-square analysis was used to
determine whether the prevalence of preexisting and new lesions differed among
locations. Chi-square analysis was also used to determine whether the
prevalence of preexisting and new lesions differed on the basis of
communication pathways to the joint space and the presence of cortical
erosion. A Kruskal-Wallis nonparametric test for multiple independent samples
was used to assess whether the lesion volume differed among cases depending on
the communication pathway. Statistical analysis was performed with use of SPSS
statistical software (version 8.0; SPSS, Chicago, Illinois). Probability
values of <0.05 were considered to be indicative of significance.
Periacetabular Bone Defects Identified on Computed Tomography
Atotal of forty-six lesions were identified in twenty-eight of the
forty-four hemipelves (Fig. 5).
The mean volume (and standard deviation) of the bone defects was 4.2 ±
9.4 cm3 (range, 0.3 to 54.6 cm3). The lesion volume was
<1 cm3 in twenty lesions (44%) and was >5 cm3 in
five lesions (11%).
By matching the three-dimensional computed tomography models with the
perioperative clinical plain radiographs, we identified sixteen lesions that
had existed prior to total hip arthroplasty (preexisting lesions). The mean
volume of the preexisting lesions was 1.5 ± 1.5 cm3 (range,
0.3 to 4.9 cm3). The mean volume of all other defects (new lesions)
was 5.6 ± 11.4 cm3 (range, 0.3 to 54.6 cm3). The
difference in size between the preexisting lesions and the new lesions was
significant (p = 0.034).
Lesion Location
Thirteen of the sixteen preexisting lesions appeared in the superior zone,
while only eight (27%) of the thirty new lesions were present in the superior
zone (Fig. 6). Preexisting
lesions were more likely to be identified in the superior zone (p <
0.001).
Communication Pathways Between the Lesions and the Joint Space
We were able to observe a clear communication pathway between the lesions
and the joint space in thirty-one (67%) of the forty-six lesions. Twenty-eight
of the thirty-one lesions were new lesions. We identified four different
communication pathways between the lesions and the joint space: around the rim
(nine lesions), through a central hole (six lesions), in association with a
screw or screw-hole (eight lesions), and through multiple pathways (eight
lesions) (Fig. 7). Of the six
lesions that definitely communicated through the central hole of the
acetabular shell, three lesions also had potential communication through a gap
between the bone and implant surface (Fig.
7). We could not unambiguously identify a communication pathway
for fifteen lesions. The mean volume of lesions that had a definite
communication pathway was 5.6 ± 11.2 cm3 (range, 0.3 to 54.6
cm3), which was significantly larger than the mean volume of 1.3
± 1.4 cm3 (range, 0.3 to 4.9 cm3) for the lesions
that did not have a clearly identifiable communication pathway (p =
0.012).
Thirteen of the sixteen preexisting lesions did not have a communication
pathway to the joint space. Among the three preexisting lesions that did
communicate with the joint space, two of the lesions had openings located at
the rim of the acetabulum and peripheral to the rim of the acetabular
component. The other lesion was located at the site of a screw-hole in the
acetabular cup. Twenty-eight (93%) of the thirty new lesions had at least one
apparent communication with the joint space. The two new lesions (7%) without
communication were 0.4 and 1.2 cm3 in size. Compared with
preexisting lesions, the new lesions were significantly more likely to
communicate with the joint space (p < 0.001).
Lesion Volume Compared with the Type of Communication Pathways
For the twenty-eight new lesions that had communication with the joint
space, a significant difference was detected in lesion volume depending on the
type of communication pathway (p = 0.019). The eight lesions that communicated
through multiple pathways (mean lesion volume, 14.3 ± 19.6
cm3) and the six lesions that communicated only with the central
dome hole (mean lesion volume, 5.6 ± 6.0 cm3) tended to be
larger than the seven lesions that communicated only through screw-holes (mean
lesion volume, 1.2 ± 1.3 cm3) and the seven lesions that
communicated only around the rim (mean lesion volume, 1.7 ± 1.6
cm3).
Cortical Erosion Around the Lesions
None of the preexisting lesions demonstrated cortical perforation or
cortical thinning. Sixteen of the thirty new lesions were associated with
cortical perforation. Twelve of the sixteen new lesions perforated the
acetabular rim, while four others perforated the medial wall. Three other
lesions had cortical thinning. All of the new lesions communicating through
multiple pathways were associated with cortical perforation. Four of the six
lesions communicating through the central dome hole were associated with
cortical perforation or thinning of the medial wall. Two of the seven lesions
communicating around the screw-hole had cortical thinning. Five of the seven
lesions communicating around the rim had cortical perforation adjacent to the
cup. The prevalence of cortical erosion was higher for new lesions than for
preexisting lesions (p < 0.001).
For more than ten years, we have retrieved postmortem specimens
containing primary uncemented titanium acetabular shells and modular
polyethylene liners. We recently began to analyze these specimens to better
understand the patterns of osteolysis associated with uncemented implants.
Removal of the femoral head and stem allowed us to eliminate the primary
sources of beam scatter and computed tomography image artifact. This, in turn,
allowed us to examine more carefully the characteristics of different bone
defects and to determine how osteolytic defects differ from preexisting
osteoarthritic cysts. From this analysis, we found that the most important
difference between new lesions and preexisting ones was the presence of a
communication pathway to the joint space. We identified four communication
pathways, including the rim, central dome hole, peripheral screw-holes, and
implant-bone gaps. Our results led us to conclude that, if a communication
pathway cannot be found when cystic bone lesions are examined, there is a high
probability that the bone lesion is not an osteolytic one resulting from the
implantation of the total hip replacement. Rather, lesions with no
communication pathways likely existed prior to the total hip arthroplasty. In
this study, thirteen of the fifteen lesions without communication pathways on
computed tomography were visible on the preoperative or immediate
postoperative radiographs. Moreover, our results suggest that detecting the
type of communication pathway may also be an important factor in predicting
future lesion size. We found that lesions with multiple communication pathways
or communication through a central dome hole were larger than those with
communication through a screw-hole or around the rim.
It has been postulated that a communication pathway to the joint space is
important to the pathogenesis of both osteoarthritic cysts and osteolytic
lesions14-16.
In this study, all lesions that were >5 cm3 had a communication
pathway to the joint space. The average volume of preexisting lesions was 1.5
cm3, and the largest preexisting lesion was 4.9 cm3. In
view of the lesion volumes illustrated in
Figure 5 and the absence of
reports in the literature describing complications associated with
osteoarthritic cysts, we regard bone defects with a volume of =5
cm3 and communication with the joint space as probably osteolytic
and, therefore, clinically important. We believe that when small lesions are
observed on computed tomography, one must be particularly careful to determine
whether they are truly osteolytic lesions or preexisting bone defects. Puri et
al. reported a 52% prevalence of osteolysis in fifty hips at a mean of 7.6
years with use of computed
tomography10.
However, only 14% of the cups in their study had osteolytic lesions of >5
cm3. On the basis of the findings in the current study, if the
lesions are <5 cm3, do not erode cortical bone, and do not have
a communication pathway to the joint, there is a high probability that they do
not represent osteolysis and thus are not of clinical concern.
We acknowledge several constraints and limitations of the present study.
First, we were restricted to a relatively small number of specimens because we
analyzed donated postmortem retrieved specimens from a select, rather elderly
population. While we are certain that the bone defects identified on plain
radiographs made prior to the total hip arthroplasty do not represent
osteolysis, we cannot be certain that all defects appearing in the postmortem
computed tomography images represent total hip replacement-induced osteolytic
lesions. It remains possible that some of the new lesions identified in this
study represent preexisting lesions that we were unable to identify on plain
radiographs. We suspect that this is the case for the two "new"
lesions without communication pathways. Theoretically, an immediate
postoperative computed tomography image would definitively identify any
preexisting bone defects, but such was not available for these cases.
Second, we recognize that there are many factors that may contribute to the
development of periacetabular osteolysis, such as the type of polyethylene,
wear rates, sterilization techniques, cup orientation, time in situ, and cup
design. However, we did not analyze this information for two reasons: (1) the
limited number of hips with osteolytic lesions in this study precluded a
meaningful statistical analysis with regard to these variables, and (2) the
purpose of this study was not to correlate wear with osteolysis or to clarify
the mechanism of the osteolysis but, rather, to demonstrate that not all
localized areas of periacetabular bone loss represent total hip
replacement-induced osteolysis. The correlation of osteolysis with these and
other variables remains an important area for future study.
Third, we caution that this postmortem study does not demonstrate that
routine clinical computed tomography scans can accurately differentiate
between osteolytic lesions and preexisting bone defects. Metal artifact from
the femoral components in a clinical scan could hinder the ability to
visualize osteolytic lesions. However, we believe that the advent of new
technologies and the use of optimum settings will likely mitigate future
problems associated with metal artifact.
Finally, we acknowledge that, while the determination of the pattern and
rate of the radiographic growth of periacetabular bone defects is very
important and of interest to clinicians, the lack of annual follow-up
radiographs precluded a serial analysis of the rate of lesion growth in these
hips.
Clinicians should be aware that computed tomography may lead to an
overdiagnosis of osteolysis, particularly in the absence of perioperative
imaging studies that can document the existence of preexisting bone defects.
Osteolysis is currently defined as a periprosthetic region devoid of bone with
a well-defined sclerotic border. In view of our findings, we propose that the
definition of osteolysis on computed tomography be expanded to include the
presence of a communication pathway to the joint space. ?
In support of their research or preparation of this manuscript, one or more
of the authors received funding (a general institutional grant) from Inova
Health System, Alexandria, Virginia. None of the authors received payments or
other benefits or a commitment or agreement to provide such benefits from a
commercial entity. A commercial entity (Inova Health System) paid or directed,
or agreed to pay or direct, benefits to a research fund, foundation,
educational institution, or other charitable or nonprofit organization with
which the authors are affiliated or associated.
Archibeck MJ, Berger RA, Jacobs JJ,
Quigley LR, Gitelis S, Rosenberg AG, Galante JO. Second-generation cementless
total hip arthroplasty. Eight to eleven-year results. J Bone Joint Surg
Am.2001;83:
1666-73.831666
2001
[PubMed]
Clohisy JC, Harris WH. The
Harris-Galante porous-coated acetabular component with screw fixation. An
average ten-year follow-up study. J Bone Joint Surg Am.1999;81:
66-73.8166
1999
[PubMed]
Della Valle CJ, Berger RA, Shott S,
Rosenberg AG, Jacobs JJ, Quigley L, Galante JO. Primary total hip arthroplasty
with a porous-coated acetabular component. A concise follow-up of a previous
report. J Bone Joint Surg Am.2004;86:
1217-22.861217
2004
[PubMed]
Kim YH, Kim JS, Cho SH. Primary total
hip arthroplasty with a cementless porous-coated anatomic total hip
prosthesis: 10- to 12-year results of prospective and consecutive series.
J Arthroplasty.1999;14:
538-48.14538
1999
[PubMed][CrossRef]
Udomkiat P, Dorr LD, Wan Z. Cementless
hemispheric porous-coated sockets implanted with press-fit technique without
screws: average ten-year follow-up. J Bone Joint Surg Am.2002;84:
1195-200.841195
2002
[PubMed]
Harris WH. Wear and periprosthetic
osteolysis: the problem. Clin Orthop Relat Res.2001;393:
66-70.39366
2001
[PubMed][CrossRef]
Maloney WJ, Herzwurm P, Paprosky W,
Rubash HE, Engh CA. Treatment of pelvic osteolysis associated with a stable
acetabular component inserted without cement as part of a total hip
replacement. J Bone Joint Surg Am.1997;79:
1628-34.791628
1997
[PubMed]
Maloney WJ, Galante JO, Anderson M,
Goldberg V, Harris WH, Jacobs J, Kraay M, Lachiewicz P, Rubash HE, Schutzer S,
Woolson ST. Fixation, polyethylene wear, and pelvic osteolysis in primary
total hip replacement. Clin Orthop Relat Res.1999;369:
157-64.369157
1999
[PubMed][CrossRef]
Manley MT, D'Antonio JA, Capello WN,
Edidin AA. Osteolysis: a disease of access to fixation interfaces. Clin
Orthop Relat Res.2002;405:
129-37.405129
2002
[CrossRef]
Puri L, Wixson RL, Stern SH, Kohli J,
Hendrix RW, Stulberg SD. Use of helical computed tomography for the assessment
of acetabular osteolysis after total hip arthroplasty. J Bone Joint
Surg Am.2002;84:
609-14.84609
2002
Stulberg SD, Wixson RL, Adams AD,
Hendrix RW, Bernfield JB. Monitoring pelvic osteolysis following total hip
replacement surgery: an algorithm for surveillance. J Bone Joint Surg
Am.2002;84 Suppl 2:
116-22.84116
2002
[CrossRef]
Engh CA, Hopper RH Jr, Engh CA Jr.
Long-term porous-coated cup survivorship using spikes, screws, and
press-fitting for initial fixation. J Arthroplasty.2004;19(7 Suppl 2):
54-60.1954
2004
[PubMed][CrossRef]
Engh CA Jr, Culpepper WJ 2nd, Engh CA.
Long-term results of use of the anatomic medullary locking prosthesis in total
hip arthroplasty. J Bone Joint Surg Am.1997;79:
177-84.79177
1997
[PubMed][CrossRef]
Freund E. The pathological significance
of intra-articular pressure. Edinburgh Med J.1940;47:
192-203.47192
1940
Landells JW. The bone cysts of
osteoarthritis. J Bone Joint Surg Br.1953;35:
643-9.35643
1953
[PubMed]
Schmalzried TP, Jasty M, Harris WH.
Periprosthetic bone loss in total hip arthroplasty. Polyethylene wear debris
and the concept of the effective joint space. J Bone Joint Surg
Am.1992;74:
849-63.74849
1992