Tumors in the pelvic area remain a major challenge. In the past, most
malignant tumors in the pelvis were treated with hindquarter
amputation1. With
improvements in chemotherapy, limb-preserving procedures have gradually
emerged as a viable surgical
modality2 and, in
most cases, have now replaced ablative surgery. Because the anatomy of the
pelvis is complex, extensile exposure is a surgical challenge and there is a
need to identify and preserve major neurovascular structures. Pelvic surgery
is characterized by a high number of
complications3-8.
Enneking and
Dunham3,5,
Erikson and
Hjelmstedt4, and
Steel6 were among
the first surgeons to report their results with limb-sparing surgery after
excision of a bone tumor of the pelvis. These procedures were most disabling
when the acetabulum was excised. Despite the high morbidity and mortality,
these authors demonstrated that, with an adequate margin, a number of patients
can survive. Hindquarter amputation or limb-sparing resection without
periacetabular reconstruction is nowadays less acceptable as the expected
function is poorer than that obtained with reconstructive
surgery9,10.
Johnson11
advocated reconstruction of a pelvic defect with a prosthesis and cement.
Today, a
prothesis12-17
and bone both can be used as implants in a pelvic reconstruction. The resected
bone can be
autoclaved17,18
or irradiated19
before it is reimplanted. The transfer of a bone autograft from the proximal
part of the femur is another
option20. However,
bone
allograft2,21-24
is the currently preferred material for reconstructing a pelvis with or
without the use of a prosthesis.
One of the major advantages of bone allografts is that the bone can be
shaped and customized in order to match the pelvic resection. The complex
osseous architecture can be fully restored for locomotion, avoiding a flail
hip or an arthrodesis. However, there are few available data on the outcome of
pelvic allograft procedures. Their use remains quite controversial because
they are associated with high rates of infection and mechanical
failure16,18,22.
We report our experience with twenty-four consecutive patients who
underwent reconstruction with a pelvic bone allograft following a pelvic
resection. This study was performed in order to assess the functional outcome
and the surgical complications of this mode of reconstruction.
Clinical Series
From 1985 to 2003, twenty-four consecutive patients underwent resection of
a malignant tumor of the pelvis followed by reconstruction with a massive bone
allograft that had been harvested from the pelvis of an organ donor. Patients
were informed about the planned surgery and the existing alternatives and gave
their consent. Institutional approval was obtained for this study.
There were nine women and fifteen men with a mean age (and standard
deviation) of 34 ± 23 years (range, eight to seventy-six years) at the
time of the index surgery. Table
I lists the diagnoses and the surgical staging of the tumors.
There were nineteen primary bone tumors and five metastatic lesions that were
considered to be isolated at the time of the index surgery. Ewing sarcoma was
the predominant lesion. Figure
1 shows the distribution of the pelvic tumors in the surgical
areas (zones I through IV) defined by Enneking and
Dunham5. Of the
twenty-four tumors, eighteen (thirteen primary tumors and five metastases)
involved the acetabular region (zone II). The six other tumors were located at
the ilium (zone I alone or with zone IV). The topographical distribution is
indicated in Table II.
Oncological and Surgical Procedures
Neoadjuvant and adjuvant chemotherapy was given to seventeen patients. One
patient had received preoperative radiation therapy. En bloc removal of the
tumor with wide margins was the aim of the surgery. Most patients had a
preoperative cleansing intestinal preparation. All patients had the operation
under general anesthesia. Ureteral stents were placed in selected patients.
Cephalosporin was administered prophylactically until no bacterial growth on
culture of intraoperative specimens of the allograft was reported. The
operations were performed with the patients in a three-quarter position
(semisupine and semi-lateral), maintained by bags and rolls allowing anterior
and posterior positions.
All except two procedures were performed through an extended ilioinguinal
approach, from which there was a lateral extension to the proximal part of the
femur, achieving an iliofemoral exposure. This approach allowed extensive
visualization of the inner and outer aspects of the
hemipelvis5. In two
patients with a tumor in zones I and IV, no femoral extension was necessary,
but the approach was extended at the spine by a T-shaped incision.
The surgical margins that were achieved were classified, on the basis of
the surgical and pathological reports, according to the system described by
Enneking et al.25
as radical, wide, marginal, or intralesional.
Reconstructive Procedure
All osseous resections were followed by anatomic reconstruction with a bone
(obtained from the local nationally registered bone bank) that was similar to
the one that had been resected. Pelvic allografts were sterilely procured in
an operating room from organ donors by a surgical
team26. The average
age of the donors at the time of death was 40 ± 13 years (range,
fourteen to sixty-two years; median, forty-two years). Bones were immersed in
serum with rifampicin (1.2 g/L of serum) for one hour before being frozen at
—80°C until use. The mean preservation time was 13 ± 12
months (range, 0.5 to thirty-eight months). No final irradiation for
sterilization was performed. At the time of the surgery, the allograft was
thawed for about one hour in a rifampicin solution at 37°C.
Allografts were fixed by one or preferably two molded 3.5-mm ASIF
reconstruction plates (Synthes, Brussels, Belgium) in the periacetabular area
and/or with one or two 6.5-mm lag screws in the sacrum and/or one screw in the
pubic ramus. In thirteen patients, a cemented total hip prosthesis that
included a cemented polyethylene cup and a 22.2 or 28-mm metallic head was
implanted to restore the hip joint (Figs.
2-A and 2-B). In five patients, the acetabular component of the
allograft matched the femoral head of the patient and the hip joint was
restored without a prosthesis (Figs. 3-A
and 3-B). All spared muscles were attached when possible to the
pelvic allograft. Three patients were treated with traction for three weeks,
as the hip was considered to be unstable at the time of surgery, and two
others were treated with an anti-dislocation device for eight weeks after an
episode of prosthesis dislocation. The remaining patients were not treated
with immobilization postoperatively. Weight-bearing was not allowed for two
months, to allow soft-tissue healing. After two months, patients were allowed
to walk with two crutches and progressive loading on the operatively treated
limb. Adjuvant chemotherapy, if any, was resumed at three weeks
postoperatively.
Functional Evaluation
Patient function was assessed according to the Musculoskeletal Tumor
Society (MSTS)
criteria27. Pain,
functional capacity, walking distance, use of a support, gait, and emotional
acceptance were each evaluated on a 5-point scale, with a maximum total score
of 30 points recorded as 100%. Patients who had received two successive
allografts were assigned two separate scores for function. The follow-up
period was calculated from the time of surgery to the last consultation, to
the time of allograft removal, or until the patient's death. Living patients
were analyzed at a minimum of twenty-four months postoperatively.
Statistical Analysis
Nonparametric testing was performed with the SPSS statistical package
(version 12.0; SPSS, Chicago, Illinois). The Mann-Whitney U test was used to
compare two unpaired subgroups of patients according a grouping variable such
as the location of the resection, an age of less than twenty years or an age
of twenty years or older, whether palsy had occurred, surgical revision, and
union or nonunion. These subsets were compared with regard to the duration of
surgery, the blood loss, the MSTS score, and the hospital stay. Relationships
between two sets of parameters were analyzed with use of the Spearman rank
order correlation coefficient. The level of significance was set at p <
0.05.
The duration of follow-up ranged from one to 137 months, with an average of
forty-one months. Every living patient in the series was followed for a
minimum of twenty-four months. The main data on the patients are listed in a
table in the Appendix.
Surgical Procedures
The mean duration of the complete surgical procedure (from the induction of
the anesthesia until the patient was taken from the operating room) was 10
± 2.8 hours (range, 4.5 to 16.5 hours). The average compensated blood
loss during the hospital stay was 4359 ± 2800 mL (range, 1000 to 11,300
mL). The amount of blood loss correlated with the duration of the surgery (r =
0.44; p = 0.03) but not with the duration of the hospital stay, patient age,
or the area of bone resection (zone II compared with not zone II). The mean
hospital stay was 25 ± 18 days (range, ten to ninety-four days) and
correlated strongly with patient age, with older patients having the longest
stay (r = 0.61; p = 0.001), but not with the duration of the operation. Tissue
culture of allograft specimens obtained at the time of the surgery were
negative, except for two cases of late growth of Peptostreptococcus or
Corynebacterium. Intravenous antibiotic therapy was discontinued after one
week except for the two patients with positive cultures, who were treated for
three weeks. A clinical infection did not develop in either of those two
patients. Adjuvant chemotherapy was resumed at three weeks for fourteen
patients and was delayed for three others because of skin necrosis or
infection.
The surgical resection included the periacetabular area (zone II) in
eighteen patients. Those patients did not differ significantly from the
patients without acetabular resection with regard to age, operative time,
hospital stay, or amount of blood transfused, with the numbers available
(Table II). Five patients, with
a mean age of 15 ± 5 years, had an osteoarticular allograft that
allowed sparing of the femoral head. In thirteen other patients with a
periacetabular excision (mean age, 46 ± 22 years), the hip was restored
with a cemented hip prosthesis. The sacroiliac joint was resected in four
patients and was replaced with an allograft and an arthrodesis, usually with
fixation with two screws. In two patients, the three zones (I, II, and III)
were resected and reconstructed. All except three of the resections disrupted
the pelvic arch. The surgical resection was wide in nineteen patients,
marginal in six, and intralesional in one.
Oncological Aspects
Eight (33%) of the twenty-four patients were alive and free of disease when
they were last seen, at a mean of seventy months. Thirteen patients (54%) had
died of the disease, and there were three deaths not related to the malignant
tumor. A local recurrence was observed in seven patients (29%), two of whom
underwent hindquarter amputation.
Of the nineteen patients with a primary bone tumor, six were alive without
evidence of disease at a mean of 62 ± 40 months postoperatively. In
this group with primary sarcoma, a local recurrence occurred in seven patients
after a mean time of 21 ± 19 months. Of the five patients with a
solitary bone metastasis located in or next to zone II, two survived without
evidence of disease for an average of ninety-five months. There were two
deaths unrelated to the tumor and no local tumor recurrence.
Surgical Complications and Morbidity
There was one early postoperative death from a pulmonary embolism at four
weeks. The morbidity rate was high (Table
III). One patient had a femoral artery thrombosis diagnosed in the
immediate postoperative period; this required vascular revision, which was
complicated by a compartment syndrome necessitating additional surgery. Two
patients had skin necrosis, and there were three deep infections. We found no
relationship between infection and either blood loss or the duration of the
surgery, with the numbers available.
Neurological complications and hip dislocation or subluxation were observed
only in the eighteen patients with a zone-II resection
(Table III). Sciatic nerve
palsy occurred in six patients, who had only partial recovery. One of these
patients also had a femoral nerve palsy. Patients with neurological deficits
had a significantly longer duration of the operation (p = 0.022) than did
those without a deficit. Hip complications were observed in five patients. Two
hip prostheses dislocated. Two femoral heads gradually subluxated from an
acetabulum reconstructed with an osteochondral allograft and were left
untreated. One femoral fracture occurred distal to the prosthetic stem, and it
was treated with a plate.
Of the twenty-four patients, eleven (46%) had a total of sixteen surgical
procedures to treat a complication of the surgery and nine underwent a
surgical revision related directly to the reconstruction itself. Thirteen
patients (54%), with a mean age of twenty-five years at the time of the index
operation, had no surgical revision; these patients were younger than the
eleven patients who required revision (average age, forty-four years; p =
0.031).
Evaluation of Bone Allografts
Sixteen allografts could be evaluated radiographically for healing, as they
were in patients who had been followed for at least eighteen months. An
unhealed junction after that period was considered to be a
nonunion21.
There were forty-three anastomotic junctions with host bone, and five of
them (12%) failed to unite by eighteen months postoperatively. Two nonunions
were revised with new allografts, which also failed to unite. Nonunion was not
found to be significantly associated with the age of the patient, duration of
the surgery, or amount of blood transfused, with the numbers available.
Hardware failure was observed in two patients with a nonunion and in a third
patient in the immediate postoperative period.
No resorption or lysis and no fracture of the graft were observed in this
series. Seven allografts had an uneventful course for at least three years,
and four of them had an uneventful course for five years. Of the five patients
who had had a hip joint reconstruction with an osteoarticular allograft, two
subsequently had progressive dislocation of the femoral head with wear of the
acetabular roof.
Functional Evaluation
The Musculoskeletal Tumor Society (MSTS) score could be determined for
twenty-three patients who had a total of twenty-five allografts, as one
patient had died one month postoperatively. The mean MSTS score was 21.5
± 6 points (73% of the maximal possible
score)27. The MSTS
score was found to be inversely correlated with patient age (r = —0.58;
p = 0.002) as well as with the duration of the operation (r = —55; p =
0.007). The eleven patients who were less than twenty years of age had an
average score of 24.7 ± 7 points (82% of the maximal score) whereas the
older patients had an average score of 19 ± 5 points (65% of the
maximal score). This difference was significant (p = 0.016).
All except four patients were able to walk. The patients who could walk had
an average MSTS score of 23 points (78% of the maximal score). Three patients
needed two crutches for walking, five patients occasionally used one crutch,
and two routinely used one crutch. Ten patients walked without any device, and
five of them had normal function with no or only a slight limp and an
unlimited walking capacity. The walking performance of the eleven patients who
were younger than twenty years old was excellent; eight of the children walked
without support, two of them occasionally used a crutch, and one was unable to
walk.
The six patients with a reconstruction in zone I or in zones I and IV had
an average MSTS score of 26 ± 3.8 points, whereas the seventeen with a
reconstruction in zone II alone or in combination with one or more of the
other zones had an average score of 20.4 ± 6.6 points. With the numbers
available, this difference was not significant
(Table II).
When excision of a pelvic tumor and reconstruction have been combined in
one procedure, the reported complication rate has been high, ranging from 30%
to 90% in series ranging in size from nine to ninety-six
patients8,13-17,19,22-24,28,29.
Carter et al.1
reported that a hindquarter amputation without reconstruction was associated
with a 41% complication rate in a series of thirty-four patients. The reported
durations of operations that included both pelvic resection and pelvic
reconstruction have been high, ranging from five to ten
hours8,10,14,22,24,
with blood loss ranging from 2500 to 8300 mL depending on the extent of the
reconstruction8,10,14,17,22.
The data in our study are consistent with those findings.
There were no perioperative deaths in our series, and the postoperative
mortality rate was low. The femoral artery thrombosis that occurred in a
patient with Ewing sarcoma resulted from prolonged excessive traction during a
lengthy procedure and might invoke the question of whether surgery is the best
option to control a radiosensitive tumor. Studies dealing with pelvic Ewing
sarcoma have had conflicting results; the impact of surgery on overall
survival could not be verified in some of
them30-32,
whereas others have indicated that surgery combined with chemotherapy tends to
offer better local control and survival than any other combination of
treatments without
surgery33-40.
Surgery should be considered only when the tumor can be completely removed.
The role of surgery for less radiosensitive tumors, such as osteosarcomas, and
for nonresponsive tumors, such as chondrosarcomas, is more
obvious4,5,41,42.
Neurological problems were the most frequent surgical complications, with a
prevalence of 25%, and were exclusively observed in patients with a
periacetabular resection (zone II). The reported rate has ranged from 3% to
30% in series ranging in size from nine to ninety-two
patients7,10,13,16,28,29.
Most palsies were complete and partially resolved; full recovery was not the
rule in our series. According to Capanna et
al.7,41,
neurological complications are predominant following iliac resections with
disruption of the greater sciatic notch.
Limb-sparing surgery has become accepted standard surgical treatment for
primary bone sarcoma unless the lesion cannot be surgically removed in an
appropriate manner. Wide resection and reconstruction remain a questionable
approach for a solitary bone metastasis in zone II of the pelvis, where
curettage, cementation, and the placement of a supportive ring and prosthesis
have been reported to have good
results43. In
patients with metastatic carcinoma, the risks can outweigh the expected
benefit of a wide resection. A potentially curative reconstruction appears not
to be the procedure of choice in this circumstance, except for an isolated
metastasis that is resistant to radiation therapy and
chemotherapy43.
Limb-sparing surgery exposes the patient to a tumor resection with narrower
margins than are achieved with a hindquarter amputation, which also does not
guarantee eradication of the local
disease1,23.
In studies of limb-salvage procedures in series ranging in size from
thirty-five to ninety-two patients, the local recurrence rate has ranged from
28% to
35%13,15,28,41,42;
the rate in our study is within that range.
There is no unanimous opinion regarding the most appropriate method of
reconstruction. Periacetabular resection without reconstruction will result in
pelvic
instability44. Many
authors8,10,45,46
have demonstrated that patients with a stable, reconstructed pelvis have
better MSTS functional scores than do patients without pelvic reconstruction.
Palliative surgery such as iliofemoral or ischiofemoral arthrodesis, although
better than amputation, is less well accepted nowadays.
In zone II, there are several alternatives to an allograft reconstruction.
The range of reconstruction techniques has been broadened by the development
of new prostheses, new
biomaterials9,12-14,29,47,
and new surgical
methods20. There
are three current techniques for periacetabular reconstruction: insertion of a
saddle prothesis14,
use of a computer-aided-designed prosthesis to bridge the
resection12,13,47,
and a hip arthroplasty with cement and with or
without11 the
support of an
autograft17,18,20
or an
allograft18,23,24.
The advantage of a saddle prosthesis is modularity and ease of
reconstruction14.
The other techniques are more complex as the reconstruction must fit the
resection. The use of bone, whether autograft or allograft, allows the surgeon
to implant a conventional total or bipolar prosthesis and to match a
potentially greater resection than anticipated. A total hip arthroplasty with
cement was preferred in our series because wear of the allograft was a
potential concern with the use of a bipolar
prosthesis48.
Infection is a major surgical complication of pelvic reconstruction,
regardless of the method of reconstruction. Its prevalence has ranged from 0%
to 37% in series ranging in size from nine to ninety-six
patients8,13-17,19,22-24,28,29,46.
Ozaki et al.22 and
Hillmann et al.46
reported a 37% infection rate in their respective series of twenty-two and
thirteen patients treated with a pelvic allograft, which would indicate that
this procedure is an unacceptable choice. Such a high rate was not found by
others23,24
nor by us in the present study. This discrepancy remains unclear but may be
related to the procurement of the allograft. We could not confirm the finding
by Hong Tan and
Mankin49 that blood
loss influenced the infection rate. Impregnation of the allografts with
rifampicin at the time of procurement and at implantation has decreased the
rate of contamination of bone allografts in our
experience26. Bone
has been shown to be an appropriate carrier of antibiotics that will be
released from the time of implantation and for at least twenty-one
days50-52.
Fracture and nonunion are concerns with the use of allograft or irradiated
bone and
autograft16,19,22-24.
In our study, nonunion occurred in three young adult patients with a high
activity demand. In two of them, the allograft was inadequately matched with
the host bone, leaving a large interfragmentary gap. Cutting and adjusting a
hemipelvic graft to fit the area of osseous resection was one of the most
difficult steps in the reconstruction and was most challenging in zone II.
Selection of a size-matched pelvic allograft remains a concern. Another
difficulty is related to fixation in the ilium, which can be tenuous. We
observed two nonunions in this location. Gaps and narrow surfaces at the
junctions were both concerns associated with nonunion. We believe that there
is a need for a computer-assisted osteotomy in tumor resection and in cutting
of the bone allograft in order to achieve a better match of the allograft to
the resection.
Another concern associated with bone allografts is the potential
transmission of viral
diseases53,54.
Procedures have been designed to ensure the supply of safe tissues. These
include guidelines for donor selection, tissue quarantine, and tissue
processing. The risk of transmitting a communicable disease remains remote
with the implementation of quality systems in tissue banks and specific
procedures for safety such as back-screening and nucleic acid testing of organ
and tissue donors.
We advocate the use of a pelvic allograft instead of bones from another
skeletal location. This allows true anatomical restoration of the complex
architecture of the pelvis and leaves open various options for hip
preservation, particularly in children.
Anatomical reconstruction following a pelvic resection provides the
opportunity for much better function than does palliative reconstruction. The
range of function as calculated as the percentage of the maximum MSTS score
(normal function) ranged from 55% to 70% in series ranging from thirteen to
thirty-nine
patients9,13,23,24.
In series of patients in whom the pelvis was reconstructed with either an
allograft or a
prosthesis16,46,
the allografts compared favorably with the prostheses in adults as well
children. We reported an average MSTS score of 73%, confirming the possibility
of obtaining an anatomically and biomechanically sound reconstruction with an
allograft. More than half of our twenty-four patients walked without crutches
or only occasionally used one crutch. In our series, as in a previous
one16, children and
teenagers performed substantially better than adults, with an average score of
82% compared with 65%. Ten of the eleven children walked without support or
with only occasional use of a crutch. Whether this was due to their age or to
a lesser extent of revision surgery could not be determined.
The rate of complications of limb-sparing pelvic resection is high. Today,
limb-sparing surgery with reconstruction is preferred to amputation when
possible13,23,28,41,
but it should usually be reserved for primary sarcoma. We believe that, when
such reconstruction is anticipated, a pelvic allograft should be considered,
especially in young patients, in whom very acceptable function can be
expected.
A table showing clinical details regarding all twenty-four patients is
available with the electronic versions of this article, on our web site at
(go to
the article citation and click on "Supplementary Material") and on
our quarterly CD-ROM (call our subscription department, at 781-449-9780, to
order the CD-ROM). ?