Pelvic discontinuity is a separation of the cephalad aspect of the pelvis
from the caudad aspect, as a result of fracture or bone loss, wherein the
anterior and posterior columns are nonsupportive. Pelvic discontinuity in a
patient undergoing revision hip arthroplasty is often a result of massive bone
loss due to osteolysis, infection, m echanical abrasion, or fracture. In this
setting, revision hip arthroplasty is a challenge. In its classification of
acetabular bone loss in patients undergoing revision hip surgery, the American
Academy of Orthopaedic Surgeons (AAOS) defined a pelvic discontinuity as a
Type-IV deficiency1.
Berry et al. subclassified the degree of bone loss associated with pelvic
discontinuity as Type IVa if the discontinuity is associated with cavitary
(Type-II) or mild segmental (Type-I) bone loss, Type IVb if the discontinuity
is associated with a large segmental (Type-I) or a combined (Type-III) defect,
and Type IVc if the pelvis had been previously irradiated regardless of the
presence of cavitary or segmental bone
loss2. Authors of
published reports have used this modified classification system to describe
the clinical results of a variety of treatment
options2,3,
including the use of an anti-protrusio cage and a cemented cup with or without
augmentation with a posterior column plate, a so-called press-fit jumbo or
oblong cup with augmentation with a posterior column plate, dual anterior and
posterior column plates with a cup inserted with or without cement, a cemented
or cementless cup without ancillary fixation of the discontinuity, dual
anterior and posterior column plates with bulk allograft and a cemented cup,
and a total acetabular allograft with a cemented cup. The clinical and
radiographic outcomes of treatment of patients with pelvic discontinuity have
been
mixed2,3.
The purpose of this review was to describe the mid-term to long-term
results of revision hip arthroplasty with use of a custom-made porous-coated
triflange acetabular prosthesis in a consecutive series of patients with
pelvic discontinuity.
Twenty-eight patients (thirty hips) with a failed total hip arthroplasty
and pelvic discontinuity were identified. From January 1992 through October
1998, all patients with a pelvic discontinuity who were seen at our
institution were treated by two surgeons (D.K.DeB. and M.J.C.) with a
custom-made triflange acetabular component. No acetabular bone defect was
considered to be so severe that a resection arthroplasty was chosen over a
reconstruction. Eight patients died and were followed for less than seven
years; thus, they were not included in the study. Two patients were lost to
follow-up after three months postoperatively. The resulting study cohort
consisted of twenty hips in eighteen patients who were followed for a mean of
123 months (range, eighty-nine to 157 months).
The eight patients who died were an average of seventysix years old (range,
fifty to eighty-eight years old) at the time of death. No death was related to
the index arthroplasty. No patient who died had had failure or impending
failure of the arthroplasty at the time of the last follow-up.
Prospectively collected clinical data on all patients, including the
presenting medical history, preoperative and post-operative Harris hip scores,
clinical and radiographic findings, surgical details, and follow-up status,
were retrospectively reviewed. Of the eighteen patients (twenty hips) who were
followed for a minimum of seven years, three were men and fifteen were women;
their mean age at the time of the surgery was 55.8 years (range, thirty to
seventy-seven years). Nine left and eleven right hips were operated on. The
underlying diagnoses included osteoarthritis (seven hips), rheumatoid
arthritis (three), posttraumatic arthritis (three), developmental dysplasia of
the hip (three), and radiation osteonecrosis, poliomyelitis, osteomyelitis,
and osteonecrosis of the femoral head (one each). The mean time since the
primary total hip arthroplasty was 16.8 years (range, three to twenty-five
years). The index arthroplasty was a first revision for seven hips and a
second revision for eight. Five hips had had three to seven revisions at the
time of the index reconstruction.
Anteroposterior and true lateral radiographs were made for all patients
preoperatively and postoperatively. The lateral radiographs were made so that
the greater and lesser trochanters were superimposed. No Judet radiographs
were made. Postoperative radiographs were made routinely at six weeks, three
months, six months, and yearly thereafter. The radiographs were reviewed by
three of us (M.J.C., D.K.DeB., and J.C.M.). The early postoperative and the
most recent radiographs were compared to detect the presence of radiolucent
lines, evidence of osseous remodeling, healing of the discontinuity, evidence
of migration, and evidence of screw motion or breakage. The ability to assess
every interface was limited because of the size and bulk of the prosthesis,
but it was possible to evaluate the discontinuity in all cases.
The preoperative bone loss was classified with the method of the
AAOS1 as modified by
Berry et al.2. If a
patient had massive bone loss or if a pelvic discontinuity was suspected, a
computed tomography scan was made of the hemipelvis and a solid one-to-one
bone model of the hemipelvis was created with use of
stereolithography4.
A pelvic discontinuity was considered to be present if there was a complete
separation of the cephalad aspect of the pelvis from the caudad aspect.
The custom triflange acetabular component was constructed with use of the
bone model of the hemipelvis. Physical models have been shown to be
substantially more accurate than radiographs for measuring acetabular defects
and for surgical
planning5. The
locations of the three flanges were identified by the surgeon on the ilium,
ischium, and pubis, and the center of rotation of the femoral head and the
orientation of the cup were determined. A clay prototype of the cup was
prepared and was examined with the pelvic bone model to evaluate these
parameters. The cup orientation was established by setting the abduction and
anteversion angles of the cup. The abduction angle was generally targeted at
approximately 40° from the horizontal, generally in plane with the
obturator foramen. The anteversion angle was established in plane with the
iliac wing6. The
three flanges were designed to facilitate initial as well as long-term
fixation. Typically, five to eight 6.5-mm titanium screws were placed in the
iliac flange and three or four 6.5-mm screws were placed in the ischium. The
pubic flange was the smallest of the three and normally did not contain screw
holes. The medial aspect of the component was either hydroxyapatite-coated
(seven hips) or had a hydroxyapatite and porous coating (thirteen hips) to
facilitate osseointegration. Once the design of the implant was finalized,
reverse engineering techniques were used to digitize the surface of the
prototype into a numeric format that allowed manufacturing of the component
from wrought titanium bar stock. The blank was prepared with use of a
hemispheric inner geometry compatible with standard, snap-in ultra-high
molecular weight polyethylene acetabular liners; thus, the locking mechanism
was identical to the mechanism in the manufacturer's standard off-the-shelf
cups. No liners were initially cemented. The components were
custom-manufactured initially by Techmedica (Camarillo, California) and then
by Biomet (Warsaw, Indiana). The custom device is currently manufactured by
Biomet; DePuy Orthopaedics, a Johnson and Johnson Company (Warsaw, Indiana);
and Stryker (Kalamazoo, Michigan). The same engineer designed most of the
implants in this study.
The surgical technique consisted of a standard posterior approach in all
patients. In two hips, an extended femoral trochanteric osteotomy was used to
aid with removal of the femoral stem and cement. The sciatic nerve was
identified and was traced proximally as it passed over the ischium into the
greater sciatic notch in order to avoid injury when the ischial flange was
being secured. The gluteus medius muscle was elevated subperiosteally off of
the ilium for placement of the iliac flange while the screws were placed
through the gluteus medius to avoid tension on the superior gluteal nerve.
Dissection was restricted into the sciatic notch, and traction on the superior
gluteal nerve was limited by abducting and translating the femur proximally. A
portion of the hamstring tendon origin was sharply released from the ischium
to facilitate placement of the ischial flange. In all patients, supplemental
particulate freeze-dried allograft bone was placed at the site of the
discontinuity. In most cases, screw fixation was initiated by placing screws
in the ischial flange, where the bone was typically the poorest and osteolysis
was common. The iliac flange was then fixed with screws, and the liner was
inserted. These iliac screws pulled the flange down into intimate contact with
the bone, which often reduced the discontinuity and rotated the caudad half of
the hemipelvis into correct orientation relative to the cephalad aspect of the
hemipelvis. In two hips, a constrained polyethylene liner was placed at the
time of the index operation to prevent dislocation. Both had an absent or
severely deficient greater trochanter and were thought to be at high risk for
dislocation. The index arthroplasty consisted of an isolated acetabular
revision with exchange of the femoral headball in three hips, acetabular
revision with retention of a monoblock femoral component in two, and
acetabular revision with concomitant revision of the femoral component in
fifteen. The new femoral headball was 28 mm in seventeen hips and 32 mm in one
hip.
The patients were restricted to partial weight-bearing with a walker or
crutches for six weeks postoperatively. No patient wore a brace or a cast in
the immediate postoperative period.
The pelvic discontinuity was considered to be healed if bridging callus or
trabecular bone was visible across the site of the discontinuity. The
discontinuity was considered to be unhealed if the fracture line was still
visible or if there was evidence of loosening of the prosthesis or broken
screws at the time of the latest follow-up. The pelvis was said to have
evidence of osseous remodeling if medially placed bone graft appeared to have
incorporated and the ilioischial and iliopectineal lines of the medial wall
had become more anatomic in their location.
This retrospective study was approved by our institutional review
board.
At the time of follow-up, at a mean of 123 months (range, eighty-nine to
157 months), no component had been removed from any of the eighteen patients
(twenty hips). The mean Harris hip score improved from 41 points
preoperatively to 80 points (range, 64 to 100 points) at the time of the
latest follow-up. Preoperatively, nine patients (ten hips) required an aid for
walking, four patients were unable to walk, and data regarding walking were
incomplete or unavailable for five patients (six hips). Preoperatively, two
patients used one cane, four patients (five hips) used two crutches, and three
patients used a walker. Postoperatively, a walker was used by three patients;
one or two crutches, by three patients; one cane, by five patients (seven
hips); and no aid, by seven patients. No patient was unable to walk
postoperatively.
Reflecting the complexity of these cases, the mean surgery time was 300
minutes (range, 140 to 575 minutes) and the mean blood loss was 2060 mL
(range, 600 to 6000 mL). A mean of 4.3 units (range, 0 to 10 units) was
transfused.
Radiographic interpretation was partially limited by the bulk of the
metallic component, and some interfaces were not visible. Eighteen (90%) of
the twenty hips had evidence of healing of the discontinuity, and seventeen
(85%) of the twenty hips demonstrated osseous remodeling of the medial pelvic
wall. Six (30%) of the twenty hips had small (<1-mm), nonprogressive
radiolucent lines at the time of the latest follow-up. These radiolucent lines
were present at the iliac flange in one hip, at the ischial flange in one hip,
at the pubic flange in two hips, at the medial cup surface in one hip, and
around an iliac screw in one hip. Although the head-liner interface was not
always easily distinguished, four hips had evidence of polyethylene wear. No
hip had evidence of osteolysis. No hip had evidence of loosening, migration,
or broken screws.
Six hips underwent another operation related to the acetabular component,
with five of the six treated for dislocation. In two hips, the acetabular
liner was exchanged for a constrained insert and no subsequent dislocation
occurred. In the other three hips, multiple surgical procedures were
undertaken to prevent recurrent dislocation. In one of them, the headball and
liner were ultimately converted to a tripolar component. A second patient had
a constrained liner cemented into the triflange prosthesis after numerous
dislocations had been treated with closed reduction. A third patient, who
remained noncompliant with treatment recommendations after numerous attempts
at constraint of the femoral head, had a partial resection with removal of the
femoral component.
In addition to the patients mentioned above, one patient in this series had
a dislocation at three years postoperatively. The hip was relocated at a local
emergency room, and no operative intervention was required. Therefore, the
overall dislocation rate in this series was 30% (six of twenty hips).
One patient had radiographic evidence of poor ischial fixation and loose
screws at six weeks postoperatively (Figs.
1-A and 1-B,
1-C and
1-D). She had symptoms of
sciatic nerve irritation at that time, and the three loose ischial screws were
removed. The cup was stable eleven years later, despite poor ischial contact,
and the patient was asymptomatic; the sciatic nerve symptoms had resolved
completely.
An additional complication occurred in a patient who sustained pelvic
trauma in a motor-vehicle accident after the index arthroplasty. The patient
underwent open reduction and internal fixation of an ipsilateral pelvic
compression fracture nonunion and a contralateral acetabular fracture three
years after the revision with the triflange prosthesis. The acetabular and
femoral components were stable and were retained. At eight years
postoperatively, the patient continued to have a stable prosthesis without
signs of loosening or failure.
Two patients had evidence of incomplete healing of the discontinuity at the
time of their latest follow-up, at forty-two and fifty months. Both of these
patients died before the time of this review, and they were excluded from the
long-term data collection. When last seen, neither patient had any evidence of
loosening, migration, or broken screws.
There have been few reports of the results of treatment of this most
complex type of acetabular reconstruction, and no method of reconstruction has
been identified as being obviously superior. Berry et al. reported on
thirty-one hips in twenty-seven patients in whom pelvic discontinuity had been
treated with a variety of
methods2. At the
time of follow-up, at a mean of three years postoperatively, nine of the
twenty-seven acetabular reconstructions had failed, requiring revision or
removal of the component. All four hips treated with a cemented cup, bulk
structural allograft, and one or two column plates failed, and all seven hips
treated with structural allograft that was not supported by an anti-protrusio
cage failed. All thirteen hips treated with an anti-protrusio cage remained
stable at the time of the latest follow-up. The authors concluded that most
Type-IVb and IVc deficiencies, as classified with their modification of the
AAOS system, should be treated with particulate or structural allograft
protected with an anti-protrusio cage.
Paprosky et al. reported the results in twelve patients one to three years
after treatment of pelvic discontinuity with a trabecular metal hemispherical
cup supplemented with trabecular metal augments that reapproximated a rim on
which to support the hemispherical
cup7. One of the
twelve patients had evidence of broken screws at the time of the latest
follow-up.
Stiehl et al. reported on a series of seventeen patients with massive
acetabular bone
loss3. Ten of these
patients had a Type-IV pelvic discontinuity, with nine of them having a
Type-IVb deficiency and one having a Type-IVc
deficiency2. Bulk
structural allograft was used in combination with posterior and/or anterior
column plates. At the time of follow-up, a mean of seven years
postoperatively, six of the ten patients had had a failure requiring
rerevision surgery. Cementless cups that rested on a bulk allograft had high
failure rates.
We studied a relatively large series of patients with pelvic discontinuity,
and all of our patients had the same prosthesis, a custom triflange cup. Our
patient population was very similar to those reported on by Berry et
al.2 and Stiehl et
al.3, with the
majority of the patients being women with Type-IVb bone loss. Our complication
rate was also very similar to those in the other studies, with the majority of
the complications being dislocations. This is not unexpected in patients with
multiple revisions of a hip arthroplasty. This phenomenon has altered our
treatment protocol for our more recently seen patients. If abductor muscle
tone is poor preoperatively, we now place a constrained acetabular liner at
the time of the index revision surgery. Careful preoperative cup design, with
particular attention paid to the anteversion and abduction angles of the cup
face, is also critical for minimizing the risk of dislocation.
Previously, we reported the results in a series of hips reconstructed with
the triflange cup, including those with pelvic discontinuity and large
acetabular defects without discontinuity (AAOS Type-III
defects)6. At a mean
of 4.4 years, no hip was deemed to be a mechanical or clinical failure. Joshi
et al. reported on a series of triflange reconstructions performed for AAOS
Type-III defects8.
At a mean of 4.8 years, one hip had been converted to a Girdlestone
pseudarthrosis because of infection and one had been rerevised with a new
triflange cup because of dislocation, but all of the other reconstructions
were stable without breakage of screws, migration, or heterotopic ossification
beyond grade 2.
Holt and Dennis reported their results at a mean of 4.5 years after the
performance of twenty-six triflange reconstructions, including three done for
AAOS Type-IV pelvic
discontinuity9. Two
of the three hips with discontinuity preoperatively demonstrated loosening of
the ischial screws with disengagement of the ischial flange. No broken screws
were observed. Although these three cases were deemed radiographic failures,
two of the three patients refused additional surgery. Poor ischial fixation is
not always associated with implant failure. In our series, one patient with a
similar disengagement of the ischial flange had a stable implant at eleven
years (Figs. 1-A and 1-B,
1-C and
1-D). Ischial bone quality is
often poor because of extensive osteolysis. To prevent loosening, the screws
should be placed first in the ischium; this allows the screws placed in the
ilium to pull the component down and into contact with host bone. Another
option to improve screw purchase is to add bone cement to the ischial defects
prior to screw insertion. Although most of the patients in this series were
treated prior to the availability of locking screws, use of such screws is
another option that may decrease the risk of this complication.
The primary disadvantage of the use of a custom-made triflange cup is the
relative complexity of the preoperative process. The custom design process
generally takes several weeks to complete and involves the surgeon working
closely with the engineer. This disadvantage is compensated for by a
relatively uncomplicated surgical technique without the need to shape, fit,
and fix structural allograft or to bend and fix an antiprotrusio cage. The
total cost of the computed tomography scan, model, and implant may be more
than the cost of other reconstruction techniques. The price of a triflange cup
averaged about $8500 (United States dollars) in 2006. This higher cost may be
offset by the improved clinical outcomes.
In conclusion, the triflange cup is a viable option for a difficult
acetabular reconstruction in a patient with pelvic discontinuity. By achieving
stable implant fixation on host bone, reapproximating anatomic load-bearing,
and providing the potential for biologic fixation, use of the triflange cup
allowed us to accomplish the major goals of reconstruction. ?