Aseventy-five-year-old man presented with multiple, extremely painful
compression fractures of the spine secondary to primary and secondary
osteoporosis. He had a four-year history of steroid immunosuppression because
of a liver transplant that had been performed as a result of cirrhosis from
long-standing hepatitis. At presentation, he was bed-ridden and able to lie
only in a supine position. He underwent kyphoplasty at four levels: T11, L2,
L4, and L5. (The details of the makeup, type, and amount of cement that had
been used were unavailable.) The intervention provided immediate
post-operative relief, and the patient was pain-free on the day after the
surgery. He died 3.5 years later from unrelated pulmonary complications. His
body was donated for medical research, and the lumbar spine was retrieved for
analysis (Fig. 1).
Tissue Preparation
The spine was cut sagittally, and all vertebrae were examined individually
(Fig. 2). The four vertebrae
that contained polymethylmethacrylate were studied after both undecalcified
and decalcified tissue preparation. For the analysis of the undecalcified
specimens, the four vertebral sagittal slabs were embedded in Epon (Ted Pella,
Redding, California). Each portion of bone containing polymethylmethacrylate
was dehydrated in a graded series of alcohols. The specimens then were
infiltrated with graded percentages of a toluene-and-Epon mixture and embedded
in 100% Epon. A portion of each vertebra with polymethylmethacrylate also was
studied after decalcification. These specimens were dehydrated in a graded
series of alcohols, decalcified with a 1% formic acid solution, and embedded
in paraffin. The vertebrae in which polymethylmethacrylate had not been
injected also were examined histologically after decalcification.
Histologic Analysis
The undecalcified sections best revealed the morphology of the bone,
marrow, and reactive tissue. Processing had dissolved some of the
polymethylmethacrylate. However, enough cement remained for us to study the
interface between the cement and the adjacent tissue. The cement showed no
laminations, fracture lines, or pores.
In all specimens, even when the cement was juxtaposed directly to bone, all
cement islands were bounded by a thin fibrous membrane
(Fig. 3). Multinucleated giant
cells often were associated with this membrane
(Fig. 4).
Cement had infiltrated between cancellous bone trabeculae. Because of the
extreme osteoporosis, there were many pathways through which the cement could
flow. Occasionally, trabeculae were cracked and impacted together
(Fig. 5). There was no evidence
that the vertebral end plates had cracked around the cement.
All of the vertebral bodies that were examined had focal zones of
osteonecrosis adjacent to the cement. The necrosis was widespread throughout
two vertebral bodies, although no vertebral end plate or intervertebral disc
was necrotic. Dead bone trabeculae were separated by fibrotic marrow typical
of necrotic lesions at least several months old. There was no evidence of
osteoclastic resorption or secondary repair of the osteonecrosis
(Fig. 6). Bone marrow adjacent
to the cement showed occasional histiocytes that were filled with barium.
There was no evidence of intravascular cement, and cement had not
penetrated into the spinal canal. At one level only (L2), a small amount of
cement was found in the intervertebral disc space (L1-L2).
The vertebrae that had not been treated with polymethylmethacrylate showed
osteoporosis with a compression fracture. There was no osteonecrosis in these
vertebrae.
The most striking histologic finding in this study was the large
amount of osteonecrosis associated with the polymethylmethacrylate cement.
This finding is in contrast with that reported by Togawa et
al.15, who noted
only scant areas of osteonecrosis. The features of the necrotic lesions were
consistent with old lesions, which matched the time sequence since the
injection of the polymethylmethacrylate. There was considerable marrow
fibrosis and marrow calcification. The necrosis was not uniformly distributed
around the cement, so thermal necrosis was a less likely cause. There is a
remote possibility that the bone infarcts were present in the vertebral bodies
before the kyphoplasty because the patient had been treated with steroids as a
result of a liver transplant. However, there had been no radiographic
documentation of osteonecrosis in this patient before he died, and the
untreated vertebrae showed no evidence of osteonecrosis. It is also possible
that the steroid therapy lowered the threshold for the development of
osteonecrosis, which may explain why the necrosis was more extensive and not
uniformly distributed. Thus, this case might suggest that vertebral
augmentation should be used with caution in patients with risk factors for
osteonecrosis.
Although some of our findings are consistent with those in a previously
published human
study15, some were
also substantially different. Similarities include the presence of a thin
fibrous membrane at the cement-bone interface as well as its association with
multinucleated giant cells that were often filled with cement or barium. Our
study failed to identify any laminations, fracture lines, or pores in the
cement. In addition, the trabeculae adjacent to the cement appeared to be
cracked, displaced, and impacted together in the kyphoplasty specimens. Togawa
et al.15 suggested
a possible autografting effect of the inflatable bone tamp used in kyphoplasty
to explain this impaction. The high-viscosity cement that is typically
injected in a
kyphoplasty16 is
another possible explanation. A recent report suggested that intravertebral
body pressure during cement injection is elevated sufficiently to cause
compaction of the
trabeculae17, but
others have refuted this
finding18. We are
unable to comment on the exact mechanism of trabecular displacement. However,
some trabeculae were probably broken by the compression fractures before the
kyphoplasty, as evidenced by the microfractures present in the nonaugmented
vertebral bodies. Phillips et
al.19 reported that
kyphoplasty is associated with a lower risk of cement leakage than is
vertebroplasty, presumably because of the creation of a void that permits
injection at lower pressure. The histologic evidence in our study suggests the
possibility of an additional mechanism. The impacted, increased-density bone
around the void created by the kyphoplasty could have closed previously open
paths for cement extravasation and therefore could have reduced the amount of
cement leakage.
While another apparent distinction between our study and the previous human
study by Togawa et
al.15 is the lack
of evidence of intravascular cement in any of our specimens, that lack is
consistent with the time between the kyphoplasty and the retrieval.
Interestingly, Togawa et al. noted that greater amounts of intravascular
cement were associated with shorter intervals between the initial procedure
and the retrieval. This observation is consistent with our findings, in that
the retrieval in our study was performed after a four-year interval, which is
several years longer than the interval in the study by Togawa et al.
The histologic changes created by vertebroplasty and kyphoplasty may vary
depending on the technique, instruments, cement preparation, host factors, and
time of retrieval. Additional histologic evaluation and correlation of the
findings with those variables will have important implications in terms of the
materials that are used for these procedures, the manner in which the
procedures are performed, and our ability to discern their advantages and
disadvantages. ?