Recombinant human bone morphogenetic protein-2 (rhBMP-2) has been
tested as a bone-graft substitute in randomized, prospective clinical trials
since
19971-4.
In certain procedures, its efficacy with regard to both arthrodesis rates and
clinical outcome has equaled that obtained with iliac crest bone graft. The
United States Food and Drug Administration recently approved the use of
rhBMP-2 as a bone-graft replacement during anterior lumbar interbody
arthrodesis in conjunction with a tapered, threaded intervertebral cage
(LT-CAGE; Medtronic Sofamor Danek, Minneapolis, Minnesota).
Although the Food and Drug Administration has approved the use of rhBMP-2
only in conjunction with the LT-CAGE device, an off-label use of rhBMP-2 with
structural allograft bone in anterior lumbar interbody arthrodesis has been
reported. A pilot clinical trial in which allograft bone dowels and an
rhBMP-2-soaked absorbable collagen sponge were used for lumbar interbody
arthrodesis demonstrated favorable outcomes and good arthrodesis
rates3. There have
been some concerns about the potential up-regulation of osteoclastic activity
leading to rapid resorption of the allograft when the allograft is used with
rhBMP-2. Some of this concern originates from the fact that rhBMP-2 can
stimulate differentiation and enhance survival of some populations of
osteoclasts5.
However, thus far, this concern has not been clearly validated in basic
science or clinical investigations. In fact, rhBMP-2 has been used to enhance
healing of allograft fractures in
rats6 and to enhance
the biomechanical properties of allograft-native bone interface in
dogs7.
We report a case in which a femoral ring spacer filled with an
rhBMP-2-soaked sponge was retrieved from a woman who had undergone a combined
anterior and posterior spinal reconstruction but had early dislodgment of the
interbody graft. To our knowledge, this is the first report of retrieval and
analysis of an rhBMP-2-soaked collagen sponge in contact with allograft bone.
Our patient was informed that data concerning the case would be submitted for
publication.
Asixty-one-year-old woman presented to our institution chiefly
because of incapacitating back pain and worsening bilateral thigh and calf
pain while walking. Three years earlier, she had undergone lumbar laminectomy
without arthrodesis as treatment for similar symptoms. A diagnosis of
neurogenic claudication, postlaminectomy instability, and progressive
scoliosis was made on the basis of the clinical and radiographic findings.
The patient underwent a combined anterior and posterior reconstruction of
the thoracolumbar spine and a revision lumbar decompression. The first stage
of the procedure included an anterior interbody release and arthrodesis from
L2 to L5. A thorough discectomy was done at each of these levels, with the end
plates of the vertebral bodies prepared to bleeding bone. INFUSE bone graft
(1.5 mg of rhBMP-2/mL; Medtronic Sofamor Danek) was reconstituted according to
the manufacturer's instructions and placed onto a collagen sponge sheet (3
× 4 in [7.62 × 10.16 cm]). The sheet was then divided into three
sections and packed into three femoral ring allografts. The allograft-rhBMP-2
spacers were inserted into the disc spaces of L2-L3, L3-L4, and L4-L5 to
promote arthrodesis as well as to support the anterior column. The second
stage of the procedure included an L2 to L5 revision laminectomy and
foraminotomies as well as posterior spinal arthrodesis with instrumentation
from T6 to L5 with iliac crest bone graft. The patient recovered uneventfully
postoperatively and was discharged from the hospital in stable condition.
Three months after the procedure, the patient fell and presented with
new-onset low-back pain. Lumbar radiographs revealed that the anterior half of
the femoral ring allograft was protruding from the L4-L5 interspace. At six
months, the follow-up radiographs revealed that the graft had completely
extruded from the L4-L5 disc space (Fig.
1). The decision was made to perform a revision surgical
reconstruction involving removal of the posterior instrumentation, retrieval
of the extruded femoral ring allograft, revision of the anterior lumbar
interbody arthrodesis with extension of the anterior arthrodesis to the L5-S1
level, and revision of the posterior instrumentation with extension of the
arthrodesis to the sacrum and pelvis. During that operation, the implant was
found to have migrated anterior to the vertebral column without any evidence
of incorporation into the vertebral bodies. The intervertebral space had
collapsed, no woven bone was noted within the intervertebral space, and there
was no evidence of ossification of the adjacent tissue.
Gross analysis of the retrieved specimen showed that the absorbable
collagen sponge, which had been packed into the center portion of the femoral
ring, could be separated easily from the allograft
(Fig. 2). The specimen was firm
but compressible, similar to the consistency of cancellous bone. There were no
erosions at the interface between the allograft bone and the specimen.
The retrieved specimen was taken to the pathology department for further
histological examination. One half of this specimen was fixed in formalin,
decalcified, and embedded in paraffin, and sections were stained with
hematoxylin and eosin. The other half was frozen, and frozen sections were
analyzed, after von Kossa staining, to identify mineralized tissue. The
decalcified sections showed mostly woven bone, some lamellar bone, rare
cartilaginous tissue, and no evidence of a residual collagen sponge
(Fig. 3-A). There was no
evidence of infiltration of the new bone by fibrous tissue. Cytologically,
there were rare neutrophils and macrophages at the margins of the specimen,
which was consistent with a mild inflammatory reaction to the extruded
allograft-collagen sponge complex. Tissue samples were negative for infection
on culture. A very small number of lymphocytes were present within the
substance of the new bone, but no neutrophils or macrophages were observed.
Numerous lacunae with osteocytes were present, and cement lines were visible
in areas of lamellar bone (Fig.
3-B). There was no evidence of an increased number of osteoclasts
or an abnormal number of Howship lacunae.
The cartilaginous areas were rare, but present, suggesting that bone
formation may have occurred through a chondrogenic stage. Abundant viable
cells were seen within the cartilaginous tissue
(Fig. 3-C). Von Kossa staining
of the undecalcified specimen demonstrated abundant mineralized tissue.
The interface of the allograft and the new bone could not be examined
histologically because the two specimens were separated from each other.
However, grossly, there was no evidence of resorption or rapid remodeling of
the allograft at the interface of the allograft and the collagen sponge.
To our knowledge, no prior retrieval study has demonstrated the
formation of bone within an rhBMP-2-soaked absorbable collagen sponge used in
any type of spinal arthrodesis. On the basis of the histological findings, we
can conclude that trabecular-type bone replaced the sponge within six months
after the implantation. We did not detect any residual collagen-sponge
material in the specimen.
This case may illustrate the importance of obtaining stable fixation when a
femoral ring allograft is used with rhBMP-2. Our clinical experience suggests
that, when rhBMP-2 is used with a femoral ring allograft, bone formation is
first seen traversing through the ring between the vertebral bodies. The
femoral ring allograft then remodels into the adjacent vertebral bodies beyond
one year after the procedure. Earlier studies have shown that, without
rhBMP-2, femoral ring allografts in the anterior aspect of the lumbar spine
can take greater than one year to incorporate with the adjacent vertebral
bodies8. Even with
rhBMP-2, the incorporation of the allograft into the vertebral bodies may
still be dependent on the biomechanical stability of the arthrodesis
construct. Sufficient biomechanical stability may also be critical to allow
the new bone in the center of the ring to integrate into both the allograft
and the adjacent vertebrae.
The newly formed bone easily separated from the femoral ring allograft,
indicating that the new trabecular bone was not integrated with the allograft
spacer at the time of retrieval. The fact that the new bone did not integrate
into the allograft is both interesting and a matter of concern. One reason for
nonintegration could be that the amount of time from implantation to retrieval
of the specimen was not sufficient for incorporation to occur. Considering the
fact that femoral ring allografts normally take greater than one year to
integrate with adjacent vertebral
bodies8, it may take
the same amount of time for the newly formed bone to integrate into the
adjacent allografts. In addition, when the allograft-collagen sponge is
partially or fully extruded from the intervertebral space, it is not subjected
to the same biomechanical forces that may promote continued osseous
integration. Another possibility is that the incorporation was hindered by
increased osteoclastic activity. Unfortunately, the allograft-collagen sponge
interface in our patient could not be examined histologically for evidence of
increased allograft resorption. However, gross examination of the femoral ring
allograft did not reveal any evidence of robust remodeling or resorption at
the interface.
In summary, our case provides histological evidence of the transformation
of an rhBMP-2-soaked collagen sponge into trabecular bone in humans. It also
provides histological evidence of osteoinductivity of rhBMP-2 within the
allograft bone environment without excessive resorption of the allograft. Our
case may suggest that osteoinductivity of rhBMP-2 alone is not sufficient to
achieve early incorporation of the allograft into adjacent vertebrae.
Maintaining the biomechanical stability of the arthrodesis construct may be
crucial for promoting incorporation of the new bone and allograft into the
adjacent vertebral bodies. ?