Participants
A retrospective review of medical records and corresponding radiographs
revealed that twenty-two patients (thirteen men and nine women) had undergone
a multisegmental anterior or posterior spinal arthrodesis with a pedicled
vascularized rib graft at our institution between March 1994 and May 2004.
Four patients were excluded from the analysis because of incomplete follow-up,
leaving eighteen patients in the final cohort. Three of the four excluded
patients had died of metastatic disease (colonic adenocarcinoma, osteogenic
sarcoma, and non-small-cell lung carcinoma), and one had had recurrent
episodes of sepsis. All four of these terminally ill patients chose hospice
care within three months after the spinal operation and subsequently could not
return for follow-up at our institution or obtain the spinal imaging studies
necessary to be included in this analysis. The remaining eighteen patients
included nine men and nine women with a mean age of 45.3 years (range,
eighteen to seventy-four years) at the time of the index procedure. A summary
of the characteristics of this group is presented in the Appendix.
Decompression and spinal arthrodesis with or without vertebral body
excision were required to treat a variety of disorders, including metastatic
or primary tumor (thirteen patients) and progressive kyphosis secondary to
chronic osteomyelitis (two), injury (one), congenital anomalies (one), or
implant failure (one). Seven of the patients with a tumor underwent
preoperative radiation, one underwent preoperative chemotherapy, and three
underwent both preoperative radiation and chemotherapy. Thus, ten of the
thirteen patients with a tumor underwent radiation treatment to the spine
prior to the surgery. Twelve patients had an average of 2.25 prior operations
(range, one to six procedures), and six had not had previous spinal surgery.
Prior surgical procedures were performed for tumor resection, spinal
stabilization, treatment of kyphosis with neurological sequelae, treatment of
infection, treatment of trauma, and hardware removal or revision. Presenting
symptoms related to spinal disease included increasingly severe back pain
(eleven patients), right groin pain (two), shoulder pain (three), neck pain
(one), lower-limb pain (one), paresthesias in the lower extremities (three),
progressive weakness in the lower extremities (three), and bowel and bladder
incontinence (one).
Outcome Measures
The primary outcome measure was the time to union between the vascularized
rib graft and the surrounding bone, as determined with spinal imaging,
including posteroanterior and lateral radiographs, flexion-extension
radiographs, and a computerized tomography scan centered at the area of
interest. Imaging was generally performed at six weeks, three months, six
months, and nine months postoperatively and annually thereafter. This schedule
varied somewhat depending on each patient's ability to schedule follow-up
appointments. In two patients, osseous union was additionally confirmed by
direct intraoperative inspection of the vascularized rib during a subsequent
procedure for implant revision. Other factors that were evaluated included the
preoperative diagnosis, the number and type of prior operative procedures, the
surgical approach, the number of vertebral levels that were spanned, the
number of corpectomies performed, the length and location of the rib graft,
the presence of graft hypertrophy at the time of follow-up, preoperative and
postoperative symptoms, perioperative complications, and the number of
subsequent operations related to the initial procedure.
Surgical Procedure
The surgical approach was chosen by the spine surgeon, with reconstructions
following tumor resection usually done through both anterior and posterior
approaches. The typical approach for the rib harvest involved exposure through
a previous thoracotomy incision. The patient was placed in the lateral
decubitus or prone position, depending on the surgical approach. The anterior
approach involved a thoracotomy through the interspace closest to the diseased
spinal segment, and the posterior approach involved a midline incision over
the relevant vertebral bodies (Fig.
1). All patients in our series had an anterior procedure, either
alone or in combination with a posterior procedure, which facilitated the rib
harvest. The spine was stabilized with posterior instrumentation, and the
affected vertebral bodies were resected. Nonvascularized tibial allograft or a
titanium cage was used to fill the anterior defect. The rib chosen for the
pedicled graft was typically one rib cephalad to the most inferior rib exposed
in the surgical field. A Doppler probe was used to identify the location of
the vascular pedicle at the costovertebral junction. Once it was located,
isolation of the rib was initiated. The intercostal muscles were divided
inferiorly and superiorly to the chosen rib. Extraperiosteal dissection was
performed to maintain blood flow to the rib; this dissection was facilitated
by leaving a small cuff of muscle attached to the rib.
Dissection was continued anteriorly, freeing the chosen rib from the
remainder of the rib cage, with care taken to identify and protect the
vascular pedicle running on the inferior margin of the rib. The rib was then
osteotomized proximally at the costovertebral junction, with protection of the
vascular pedicle. The length of rib that was harvested was determined by the
defect size, and the distal osteotomy was performed accordingly.
The procedure for anterior placement of a vascularized pedicled rib graft
has been described
previously7,8,
and our technique did not differ substantially. For posterior placement of the
graft, the entire rib was rotated 90° on its pedicle and moved into the
defect in the posterior column, with the surgeon making sure that the vascular
pedicle did not become twisted, kinked, or otherwise compromised
(Fig. 2). Shortening of the
stump of the rib at the costovertebral margin assisted in the rotation of the
rib. If the rib did not transpose to the posterior position because of its
length, a partial osteotomy was performed on the concave (pleural) surface of
the rib, and the rib was cracked and folded over itself and passed
posteriorly. The periosteal sleeve was carefully preserved on the anterior
surface of the rib if an osteotomy was performed. Once the rib was posterior,
the periosteum was elevated on the concave side of the rib, with preservation
of the pedicle, and a unicortical osteotomy was performed. The rib was then
cracked in multiple locations, in order to straighten it
(Fig. 3). The ends of the rib
were stripped of periosteum to facilitate contact with the vertebral bone, and
the rib was placed in its final resting position and secured to the posterior
elements of the spine with wire and/or sutures
(Figs. 4-A and 4-B).
Supplemental bone graft was placed around the rib and the posterior elements
of the spine to assist in the fusion. The mean time (and standard deviation)
needed for the harvest, rib preparation, and insetting was 58 ± 13
minutes. There was little mean blood loss during the rib harvest.
The average duration of follow-up was 31.8 months (range, fourteen to
ninety months). Two patients who had died secondary to metastatic tumor had a
healed rib graft at the time of their death, at sixteen and fifteen months
after the surgery.
An average of 4.4 levels (range, one to ten levels) were fused during the
spinal reconstruction. The average number of vertebral bodies resected was 1.9
(range, one to five). The mean length of the rib graft was 16.1 cm (range, 4.5
to 30 cm). A right rib was used in nine patients, and a left rib was used in
nine. A rib between the fifth and eleventh ribs, inclusive, was used,
depending on the location of the spinal reconstruction. The rib was placed
anteriorly in seven patients and posteriorly in eleven. Additional graft
material was used in all patients: nonvascularized rib autograft was utilized
in four; demineralized bone matrix, in three; iliac crest autograft, in
twelve; fibular allograft, in four; nonvascularized fibular autograft, in one;
tibial allograft, in two; and bone morphogenetic protein, in one. Fourteen
patients had segmental instrumentation, which was supplemented with an
anterior plate in one, an interbody cage or a structural graft in eight, and
interbody spacers in three. The four patients who did not require
instrumentation were treated through an isolated anterior approach, with
supplementation of the posteriorly placed vascularized rib graft with
anteriorly placed nonvascularized rib autograft, nonvascularized fibular
autograft, fibular allograft, and/or demineralized bone matrix.
The average time to union of the vascularized rib grafts was recorded as
6.8 months (range, three to fourteen months)
(Figs. 5-A and 5-B), but this
is probably an overestimate of the time to union as radiographs and computed
tomography scans were made when the patients returned for reevaluation rather
than at preset times to determine the earliest time to union. Seven of the
eleven patients who had had back pain preoperatively, both patients who had
had groin pain, one of the three patients who had had shoulder pain, the
patient who had had lower-limb pain, two of the three patients who had had
paresthesias, all three patients who had had lower-extremity weakness, and the
patient who had had a bladder and bowel deficit had resolution or a marked
decrease in the symptom postoperatively. Rib hypertrophy was seen in three
patients, and notable consolidation of the graft into the surrounding tissues
was observed in all patients.
Six patients required a total of ten subsequent surgical procedures. Four
patients required one additional operation, and two required two or more.
These procedures were performed for hardware revision (three), treatment of
deep wound infection (two), halo placement and removal (two), and treatment of
spinal instability (three). One vascularized rib that had fused both
proximally and distally by nine months sustained a midsubstance fracture (a
fracture of the rib and the fusion mass) associated with a rod failure. This
necessitated a reoperation to replace the posterior instrumentation and the
vascularized graft with another pedicled vascularized rib from the
contralateral side. Twelve perioperative complications were encountered,
including, in one patient each, seizures and dysphagia, deep vein thrombosis,
pulmonary embolism, temporary respiratory difficulty, vocal cord dysfunction
associated with excision of the tumor mass, postoperative anemia, and an
incisional hernia as well as a wound infection in three patients. These
complications were related to the extent and/or the location of the index
spinal procedure. None of the complications was directly related to the rib
harvest. Persistent postoperative symptoms that were new in onset included
abdominal pain, back pain, incisional thoracotomy pain, and interscapular
discomfort, in one patient each. During the postoperative follow-up period,
two patients died, secondary to progression of cancer, at sixteen and fifteen
months after the index surgery. There were no surgery-related deaths.
Compared with nonvascular bone-grafting, vascularized bone-grafting of
osseous defects offers numerous benefits, including the preservation of
osteocytes14, that
result in accelerated graft consolidation without the need for creeping
substitution into necrotic
areas1-5,15.
Additionally, vascularized bone-grafting obviates the need for extensive
remodeling of the graft, which involves revascularization, osteoclast
resorption, and production of reactive bone, resulting in a weakened matrix
that may fracture under
stress16.
Vascularized bone grafts also have been noted to have greater bone mass with
diminished osteopenia after transfer compared with nonvascularized
grafts17. Multiple
studies have demonstrated that the biological and mechanical properties of
vascular bone grafts are superior to those of nonvascular bone
grafts18-21.
In 1975, Rose et al. described the successful fusion of an anteriorly
placed pedicled vascularized rib graft for anterior arthrodesis of a severely
kyphotic spine6.
Subsequently, Bradford and Daher further popularized the anterior technique
for the treatment of progressive and/or painful spinal deformities due to a
variety of
conditions8,22.
Bradford and Daher implied that vascularized grafting should be avoided in
patients with spinal defects secondary to
tumor8,22
because they theorized that the graft's blood flow would be consumed by, and
contribute to the growth of, recurrent tumor. This concept was based on an
expectation of local recurrence following intralesional tumor resection. En
bloc resection reduces the risk of local recurrence. It has thus been our
general practice to perform a combined anterior-posterior resection,
stabilization, and arthrodesis with a posteriorly placed vascularized rib
graft following en bloc resection of a spine lesion. We performed this
procedure in ten of thirteen patients with a tumor, and none had local tumor
recurrence or iatrogenic metastasis. The graft's blood flow through the tumor
bed could theoretically facilitate dissemination of tumor cells into other,
remote tissues; however, this did not occur during the period of follow-up of
our small number of patients.
Despite multiple reports of positive outcomes, some authors have maintained
that the rib, because of its diminished strength and curved
body10,23,
is less suitable than a long, straight fibular graft, especially when anterior
decompressive corpectomies are performed and the apex of the kyphosis does not
come into direct contact with the
graft11. However,
the transfer of free vascularized fibular grafts may be associated with
donor-site
morbidity24 and may
involve substantial technical challenges. In one study, the procedure had to
be abandoned in three of sixteen patients because of difficulties with
microsurgical
anastomosis11.
Alternative ways to use vascularized rib graft have been reported. In one
study, the vascularized rib graft was paired with a titanium mesh cage to
reconstruct large deficits in the anterior and middle columns of the thoracic
spine, and solid fusion was attained in three
patients12. Others
have used an anteriorly placed avascular fibular graft as a strut with a
vascular rib graft placed directly behind
it25. Another group
achieved solid fusions in all twenty-three patients in whom they performed a
corpectomy using a vascularized rib graft that was folded several times to
increase its
diameter13.
Most of the literature promotes anterior placement of the vascularized rib
graft because of its curvature and the presumed need for compression to
achieve graft integration and healing. We demonstrated that pedicled
vascularized rib grafts as long as 30 cm can be harvested, can span up to ten
levels, and can be placed either anteriorly or posteriorly, with an average
time to union of 6.8 months. Additional nonvascularized graft and
instrumentation were used to stabilize the spine, and we believe that this
contributed to osseous union. However, it is our opinion that a vascularized
rib graft contributes substantially to the success and speed of spinal fusion,
particularly in patients who are predisposed to poor healing, which in our
series included individuals who were being treated for pseudarthrosis after a
nonvascularized fusion, those with tumor recurrence following irradiation, and
those with a previous perioperative spinal infection. Adverse events
associated with rib-harvesting were rare. Although our cohort did have
perioperative complications, they were not related to the use of a
vascularized graft but probably reflected the poor baseline health status of
the patients and the overall complexity of the surgical procedures. The curved
rib can be straightened easily and placed posteriorly without injury to the
vascular pedicle. We did not observe hypertrophy of the rib consistently, but
it may be difficult to detect on
radiographs7,9.
Also, hypertrophy is generally a response to the compressive forces on the
graft, and there may be much less such force on posteriorly positioned grafts.
Nonetheless, we observed hypertrophy of the vascularized rib graft in three of
our patients.
The limitations of this study include its retrospective nature, the lack of
a control group, restricted power of analysis due to the small sample size,
and the absence of uniformity regarding preoperative diagnoses. These
limitations notwithstanding, we have described a technique that requires
little operative time but adds the substantial benefit of a viable bone graft
to complex spinal reconstruction.
A table showing specific characteristics of the study 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).