Hodgkinson et al.1 established
an association between radiographic demarcation at the bone-cement
interface of cemented sockets and a surgical finding of loosening.
The importance of radiolucent demarcation around cementless sockets
remains uncertain. It has been assumed that a complete 2-mm-wide
demarcation between the bone and the metal cup or migration of the
cup of 3 mm indicates loosening of the cup2,3.
Schmalzried and Harris4 reported
an association between an initial radiolucent gap and subsequent
progressive radiolucent lines, but no socket with screw fixation
was loose in their study so the ultimate outcome was not known.
Dorr et al.3 found that postoperative
gaps did not lead to progressive radiolucent lines adjacent to cups
without screws.
The purpose of our study was to evaluate radiographic demarcations
about hemispheric porous-coated cementless sockets fixed with screws
and to compare these findings with the operative findings with regard
to the adequacy of fixation.
Intraoperative evaluation of the primary acetabular component
was performed during fifty-two revision total hip arthroplasties.
All of the original sockets had been fixed with 6.5-mm cancellous
screws. Every socket was routinely tested for loosening intraoperatively
by removing all screws and applying firm blows with a mallet and a
bone tamp to the socket edge. Any movement at the bone-socket interface
was considered evidence of loosening. The fixation status of the
socket at the operation was then compared with its preoperative
radiographic findings.
The fifty-two hips (fifty patients) had undergone the primary
total hip arthroplasty between February 1985 and September 1988.
The time from the primary operation to the revision ranged from
33.8 to 150.1 months (average [and standard deviation],
89.9 ± 31.4 months). Each of these hips had been
operated on and followed by us, and each had complete clinical and
radiographic follow-up. We continued to follow well-fixed sockets after
the revision surgery to determine whether subsequent radiographic
signs of loosening developed. The well-fixed sockets were followed
for an average of 136.8 ± 18.1 months (range, 102.3
to 165.4 months) following the primary operation.
The radiographs of an additional 100 primary total hip replacements
(in eighty-seven patients) that were functioning well after a follow-up
interval of 121.0 ± 18.1 months were measured to
corroborate the data on the hips that required a reoperation. The
100 unrevised hips had been treated, between February 1985 and November
1988, with the same socket and implantation technique as had been
used in the fifty-two hips that underwent revision. Each of the
100 hips had sequential radiographic follow-up, and the presence
of radiographic gaps or radiolucent lines was recorded in the same
manner as it was for the fifty-two revised hips.
The revision group consisted of twenty-six women and twenty-four
men. The age at the time of the primary operation averaged 53.4 ±
14.3 years (range, 22.6 to 75.6 years). The initial diagnosis was
osteoarthrosis in thirty-two hips; avascular necrosis in seven; developmental
dysplasia of the hip in seven; septic arthritis in two; and ankylosing
spondylitis, spontaneous fusion secondary to infection, slipped
capital femoral epiphysis, and rheumatoid arthritis in one hip each.
All patients had an Anatomic Porous Replacement hemispheric acetabular
component (APR; Sulzer Medica, Austin, Texas). The acetabular component
was manufactured from Ti-6Al-4V alloy. Holes allowed fixation with 6.5-mm
titanium-alloy screws. In the first thirty-two hips treated in the
study, the prosthesis had patches of porous coating. In November
1986, sockets with a circumferential porous coating were available,
and these were used in the next twenty hips (Fig. 1). There was
no significant difference in the clinical or radiographic outcomes
between the sockets with patches of porous coating and those with
a circumferential coating (Table I). The posterior approach was used
for both the primary operations and the revision operations. The
acetabulum was prepared with use of hemispherical reamers that were
an average of 1 mm smaller than the actual component.
Harris hip scores5 were obtained
preoperatively and at the two-year and final follow-up visits before
the reoperation. An anteroposterior radiograph of the pelvis centered
over the pubic symphysis and including the proximal part of the femur
as well as a modified 17-in (43.2 cm) Lowenstein lateral radiograph
of the hip were used for the radiographic examination. The modified
Lowenstein lateral radiograph is similar to an oblique radiograph
of the pelvis because the patient is turned onto the affected hip at
least 45° and as much as necessary to allow the lower limb to be
in abduction and external rotation and to be flat on the x-ray table,
which provides a lateral view of the acetabular subchondral bone
and the cup after implantation (Figs. 2-A and 2-B). The radiographs were measured
by two observers (P.U. and Z.W.) who had no knowledge of the operative findings
or the clinical care of the patients. Each radiograph was also reviewed
by one observer (P.U.) on two separate occasions, one year apart,
to determine the intraobserver reproducibility.
Interobserver reliability and intraobserver reproducibility were
determined with calculation of the kappa coefficient6. A value of <0.5 indicates
poor agreement, and a value of >0.75 indicates excellent
agreement. Measurements were performed on the preoperative, immediate
postoperative, three-month, and twelve-month radiographs and then
on the radiographs made at each return visit until the hip was revised.
The magnification of the radiographs was corrected according to
the known diameter of the femoral head. On the three-month anteroposterior radiographs,
the angle of inclination of the socket and the percentage of the
surface of the component in contact with bone were measured3,7. Evidence of acetabular migration,
either linear or rotational, was measured on serial radiographs;
a linear change of >3 mm or a rotational change of 8° was considered
to indicate migration8.
Postoperative gaps, radiolucent lines, and osteolysis were each
identified according to the three zones described by DeLee and Charnley9 on the anteroposterior pelvic and
modified Lowenstein lateral radiographs. Gaps were defined as areas
where the surface of the acetabular component was not in contact
with bone on the immediate postoperative radiographs; radiolucent
lines appear on subsequent radiographs in zones where no gaps initially
had existed or after the gaps had disappeared3.
The width of the gaps and radiolucent lines were measured with
use of a digimatic caliper (Mitutoyo, Tokyo, Japan). Progression
of a radiolucent line was defined as an increase in the number of
zones and/or an increase in the width of the line to 2
mm on sequential radiographs. Gaps or radiolucent lines were recorded
as decreased if fewer were observed. The size of an osteolytic area
was measured on the anteroposterior pelvic radiograph as the longest
diameters of the lesion in the horizontal and vertical axes. Screw
breakage was also recorded. All data were analyzed with use of SPSS software
(SPSS, Chicago, Illinois).
Continuous variables including age and Harris hip scores were
compared with use of the independent Student t test. The minimum
level of significance was p < 0.05.
Findings in Fifty-two Hips with Revision
Seventeen sockets were found to be loose and thirty-five were
found to be well fixed at the revision. The preoperative radiographs
of ten of the loose sockets were interpreted as showing migration
(seven), a continuous radiolucent line in all three zones (two),
or a discontinuous radiolucent line in three zones (one). In addition, one
of the sockets that migrated had a three-zone radiolucent line prior
to migration, and one of the seven sockets that was not diagnosed
as loose because of a missed diagnosis by the physician at the time
of the clinical evaluation had a discontinuous three-zone radiolucent
line. Also, three hips had a 2-mm-wide radiolucent line in a single
zone. The patients who had a loose cup at the time of the reoperation
were younger on the average and had a lower average Harris hip score
than those with a well-fixed socket.
The immediate postoperative radiograph of fourteen of the thirty-five
hips with a well-fixed socket and seven of the seventeen with a
loose socket showed gaps, all measuring £0.5 mm, at the
socket-bone interface; no hip had gaps in all three zones. At the
end of the study, the gaps were unchanged in four of the hips with
a well-fixed socket and were not visible in ten; the gaps were unchanged
in two of the hips with a loose socket, were decreased in one, and
were not visible in four. Both in the hips with a well-fixed socket
and in those with a loose socket, the gaps that disappeared did
so at an average of 3.4 ± 2.7 years (range, one
to eight years) after the surgery.
Eleven of the thirty-five hips with a well-fixed socket and fifteen
of the seventeen hips with a loose socket had radiolucent lines
identified in the first two years after the operation. These lines
initially measured <1 mm and averaged 0.4 mm (range, 0.3
to 0.7 mm) in the hips with a well-fixed socket, whereas they initially
averaged 0.9 mm (range, 0.4 to 2.0 mm) in the hips with a loose socket.
None of the hips with a well-fixed socket had a radiolucent line
in all three zones, whereas five hips with a loose socket had such
a line. At the end of the study, the radiolucent lines had progressed
in one hip with a well-fixed socket (only in the first two years),
were unchanged in three, and were not visible in ten, whereas all
zone-1 radiolucent lines in the hips with a loose socket persisted
until the time of the reoperation. No well-fixed socket had subsequently
loosened at the latest follow-up evaluation after the revision.
At the time of the reoperation, the radiolucent lines associated
with the fifteen loose sockets that had such a line were most common
in zone 3; they were progressive in nine of these hips, unchanged
in three, decreased in one, and not visible in two. In the two hips
in which the lines were not visible, the lines had been progressive until
they were obliterated by migration of the socket. Seven of the seventeen
loose sockets migrated, at an average of five years (range, two
to eight years) postoperatively. Four were associated with progressive
radiolucent lines after two years, which preceded the migration.
In the hips with a loose socket, all of the radiolucent lines
that progressed did so only after two years postoperatively (Figs. 3-A and 3-B). Progression of
a radiolucent line after two years was suggested by the width of
the line on the initial postoperative radiograph. The initial width
of the radiolucent lines that later progressed averaged 1.0 mm (range,
0.4 to 2.0 mm) compared with 0.6 mm (range, 0.3 to 1.6 mm) for those
that did not later progress (p = 0.04). Five hips had radiolucent
lines only after two years postoperatively (average, five years;
range, three to seven years); three of these five had additional
progression (Table II).
At the time of the reoperation, four loose sockets did not have
radiolucent lines. Two of the four had migrated, obliterating the
previously observed radiolucent lines. The other two also had migrated;
one of them had had a gap on the immediate postoperative radiograph,
but this gap was not evident during the follow-up period (perhaps
because of the migration). The fourth cup never had any radiolucency
before the revision.
Findings in One Hundred Hips without a Reoperation
Initial postoperative gaps were seen on the radiographs of thirty
of the 100 hips. All gaps measured £0.5 mm. Gaps were seen
in one zone in twenty-two hips, in two zones in seven, and in all
three zones (discontinuous) in one. At an average of 121.0 ±
18.1 months, the gaps were unchanged in six hips, decreased in four,
and not visible in twenty. Gaps most commonly persisted in zone
1 on the anteroposterior radiograph. Radiolucent lines were observed
in the first two years in twenty-two of these hips. At the latest
follow-up evaluation, these radiolucent lines were unchanged in
three hips, decreased in two, and not visible in seventeen. The
radiolucent lines initially measured an average of 0.4 mm (range,
0.3 to 0.6 mm). They were most commonly seen in zone 1 on the anteroposterior
radiograph. No socket had a radiolucent line in all three zones,
and no lines were progressive.
At the final follow-up evaluation, radiolucent lines were seen
around fifteen sockets. Ten of these lines were persistent gaps
that had been seen on the initial postoperative radiograph, and
five were persistent, nonprogressive radiolucent lines that had
been first seen by two years postoperatively. All involved only
one zone, and the average width was 0.4 mm (range, 0.3 to 0.6 mm).
No socket migration or screw breakage was detected.
The two observers agreed on the radiographic appearance at the
bone-socket interface in 142 (93%) of the 152 hips in the
entire series (kappa coefficient, 0.76). One observer’s
first and second reviews, separated by a one-year interval, agreed
with regard to the appearance in 147 hips (97%) (kappa
coefficient, 0.89). These values indicated excellent interobserver
and intraobserver agreement on the measurements. The criteria had
a positive predictive value of 100% and a negative predictive value
of 97%.
Surgeons need criteria with which to make the diagnosis of cup-loosening
in a patient with a painful total hip replacement. In this study,
we found five radiographic criteria that suggested loosening of
the Anatomic Porous Replacement, a cementless hemispheric titanium-alloy
porous-coated socket with screws. These criteria, which were best
judged on sequential radiographs, were (1) the occurrence of radiolucent lines
after two years (2) progression of radiolucent lines after two years
(3) radiolucent lines in all three zones (even if they are not continuous),
(4) radiolucent lines 2 mm or wider in any zone, and (5) migration.
Our study showed that postoperative gaps were not associated
with the subsequent presence of radiolucent lines, progressive radiolucent
lines, or socket-loosening. We underreamed the bone by 1 mm compared
with the size of the metal shell in these hips. The vast majority
of these gaps adjacent to both the well-fixed and the loose sockets
as well as in the hips that did not undergo a reoperation disappeared
by an average of 3.3 years postoperatively. In a ten-year follow-up
study of 188 Harris-Galante hemispheric porous-coated cups (Zimmer,
Warsaw, Indiana) that had been implanted by Harris, some association
between postoperative gaps and radiolucent lines was reported10. At ten years, radiolucent lines
were identified around forty-seven (46%) of 102 sockets
that had not been associated with postoperative gaps compared with
twenty-five (29%) of eighty-six sockets that had been associated with
gaps.
Ranawat et al.11 and Ritter
et al.12 both demonstrated that
postoperative demarcation of a cemented cup portends revision by
ten years. In contrast, we found that immediate postoperative gaps
had no association with cup-loosening. Hodgkinson et al.1 found that the pattern of radiolucent
lines around cemented cups at one year postoperatively was predictive
of loosening. In our study, the findings on radiographs of hemispheric
porous-coated cups were not predictive until after two years postoperatively.
The importance of demarcation (gaps and radiolucent lines) of a
cemented cup in the first year after implantation appears to be
different from that of such demarcation of a hemispheric porous-coated
cup.
The most important radiographic determinants of loosening were
the measurements of radiolucent lines after two years postoperatively.
Radiolucent lines that progressed after two years
postoperatively and radiolucent lines of 1 mm in thickness that
appeared after the second year were both 100% predictive
of loosening.
A 2-mm-wide radiolucent line has traditionally been thought to
indicate loosening. A 2-mm-wide radiolucent line in one zone was
predictive of loosening both in our study and in that by Kobayashi
et al.13. Hodgkinson et al.1 found that a 1-mm wide radiolucent
line in one zone had predicted the loosening of five of twenty-seven
loose cemented cups, and we found that it had predicted the loosening
of five of seventeen loose hemispheric titanium-alloy porous-coated
cups. A 1-mm radiolucent line in two zones that appeared after two
years postoperatively was progressive in five of six hips in our
study compared with twenty of twenty-five in the series of Hodgkinson
et al.1. Schmalzried and Harris4 found an association between gaps
of 1 mm in thickness and the development of progressive radiolucent
lines in hips with a Harris-Galante socket fixed with screws. These
findings suggest that a 1-mm demarcation between the cup and bone,
especially if it is in more than one zone, is predictive of loosening.
The findings in the current study provide radiographic criteria
with which to diagnose loosening of one cup design fixed with screws.
Because this cup is a 3-mm porous-coated titanium-alloy shell, as
are many designs, these criteria may be useful for judging loosening
of other cup designs fixed with screws. Implant retrieval studies
have confirmed that the pattern of loosening does not differ among
different cup designs14-16. It
is important to make sequential radiographs because progressive
radiolucent lines and radiolucent lines of 1 mm in thickness after
two years postoperatively indicate that the socket is likely to
loosen. Hopefully, if sequential radiographs are made, this socket
can be identified and revised before extensive migration occurs,
so that acetabular bone stock can be preserved.