An eighty-two-year-old woman with intractable back pain was transferred to
our service from another hospital. A diagnosis of osteoporosis had been made
following multiple vertebral fractures that had occurred approximately one
year before and from which she had fully recovered after treatment with oral
analgesics, calcium, vitamin D, and use of a brace. When renewed back pain
developed, she first had received outpatient therapy with a variety of orally
administered analgesics, including opioids for about five weeks, and oral
alendronate (70 mg once per week). At the time of transfer to our department,
she was receiving high doses of narcotic analgesics both orally and
intravenously, but she was still immobilized because of pain. The maximum pain
was in an area of enhanced signal intensity on magnetic resonance imaging
consistent with bone marrow edema in the twelfth thoracic and first lumbar
vertebrae. There was no associated pain or enhanced signal intensity in the
other, also deformed, vertebrae. After failure of an intensified non-invasive
treatment regimen including intravenous patient-controlled analgesia with
piritramide (a strong synthetic opioid) and the use of a rotationally stable
custom brace for another seven days, the patient provided informed consent for
vertebroplasty. She had comorbidities, which were arterial hypertension and a
history of transient ischemic attacks, for which she was taking 100 mg of
aspirin per day. The use of aspirin was discontinued five days prior to the
procedure. The patient had no history or clinical signs of a bleeding
disorder, and the preoperative prothrombin time, partial thromboplastin time,
international normalized ratio, and platelet counts were within normal range.
A bleeding time was not performed preoperatively. Prior to the start of the
procedure, a hypertensive episode (blood pressure of 220/110 mm Hg) was
treated with sublingual nitroglycerin and nifedipine. When the blood pressure
returned to 180/90, the procedure was begun under biplanar fluoroscopic
control after the administration of local anesthesia with a total of 20 mL of
0.5% bupivacaine and under constant monitoring that included a single-channel
electrocardiogram and measurement of the heart rate, blood pressure, and
peripheral oxygen saturation. The passage of the trocar needle through the
left pedicle into L1 was not problematic, but the left pedicle of T12 could
not be entered easily and access was then obtained through the right pedicle
of T12. Cement augmentation was performed with use of standard
polymethylmethacrylate cement and an applicator kit (Biomet, Berlin, Germany).
There was no visible cement extravasation into the spinal canal, but there was
filling of a short venous segment (Fig.
1). Immediately after the procedure, the patient was
hemodynamically stable and almost completely free of pain. The postoperative
neurological examination demonstrated normal findings.
Thirty minutes later, the patient reported pain in the anterior part of the
leg bilaterally. Within fifteen minutes, the pain had increased and
paresthesias and weakness began developing first in the left and then in the
right leg. A computed tomography scan was performed immediately (Figs.
2 and
3), but only a slight amount of
cement extravasation into the epidural veins was noted. Since the intraspinal
structures could not be clearly distinguished, a magnetic resonance imaging
scan was acquired. Sagittal T2-weighted sequences showed a massive epidural
hematoma compressing the cauda equina and the conus medullaris from T11 to L2
(Fig. 4). On the axial images,
the dural sac was noted to be displaced to the right by the hematoma. The
patient was immediately returned to the operating room for an emergency spinal
cord decompression. When the spinal canal was entered less than three hours
after the initial onset of leg pain, a hematoma under pressure drained from
the intraspinal space. A left-sided hemilaminectomy from T11 to L2 was
performed, and the hematoma was evacuated. No intraspinal injury, fracture at
the pedicle root, cement extravasation, or arterial bleeding source could be
identified. Intraoperative hemostasis proved challenging because of a very
congested venous plexus and profusely bleeding cancellous bone from the
hemilaminectomies. Sealing of all bleeding bone with bone wax and hemostasis
with bipolar coagulation, gelatin sponges, and silver nitrate cellulose was
performed. Nevertheless, during the procedure, the patient lost 3 L of blood,
requiring cell-saver autotransfusion, and the administration of packed
red-blood cells and fresh-frozen plasma. Postoperatively, she immediately had
partial return of sensory and motor function. Within forty-eight hours, there
was complete recovery of all neurological function. The patient remained in
the intensive care unit for several days to monitor the cardiac and pulmonary
systems, and she made a complete recovery. Fourteen days after the
vertebroplasty, the patient was walking and did not require any pain
medications. She was transferred to a rehabilitation facility from where she
was discharged home three weeks later.
Analysis of the case of this patient does not reveal a clear single cause
for the complication, but there are several factors that may have been
responsible in combination. First, low-dose aspirin was not discontinued until
five days prior to the procedure. After that period, >50% of the platelets
in the circulation should have been functional, since it is generally accepted
that 10% of circulating platelets are replaced within twenty-four hours. There
is no definitive evidence that antiplatelet medications increase the risk of
an epidural hematoma after neuraxial anesthesia, which, on the basis of the
scale of the intervention, might be considered a procedure comparable with
uncomplicated
vertebroplasty5. A
period of five days between the discontinuation of aspirin and spinal
anesthesia procedures is recommended by some authors on empirical grounds, and
a period of between several days and one week is considered sufficient for
endoscopic spinal arthrodesis in the United
States6,7.
A recent survey among German spinal surgeons found that the average period
between the discontinuation of aspirin to the time of elective spinal surgery
was 6.9 days (range, zero to twenty-one
days)8. This
indicates that some surgeons would have considered a five-day period
insufficient. In our patient, a preoperative bleeding time was not performed,
since there had been no clinical signs of impaired platelet function and no
history of prolonged bleeding. Standard coagulation tests had been normal, and
an array of platelet function tests after the complete recovery of the patient
showed no underlying abnormality. Nevertheless, the possibility that a low
platelet turnover, which may occur in an older person, might have prolonged
the effect of platelet inhibition cannot be ruled out. Second, the
hypertensive period prior to the procedure could have caused a spontaneous
spinal epidural hematoma. Such hematomas are rare, and less than 300 cases
have been reported. Among the suspected causes are
hypertension9-12,
the use of low-molecular-weight
heparins13, and
pregnancy with and without
preeclampsia14-16.
Vascular malformation is also considered as a cause of a spontaneous epidural
hematoma17,18,
but no evidence of a vascular malformation was seen in our patient. In many
cases, no apparent reason has been
found19-21.
The blood pressure in our patient at the beginning of the procedure was
slightly elevated, but this level would not qualify as a hypertensive crisis.
In addition, the close time relationship between an invasive procedure and the
epidural hematoma makes a spontaneous hematoma unlikely. Third, the
obstruction of transosseously draining veins in two adjacent vertebrae might
have caused regional venous congestion, which is what we observed during
surgery. Neurological deficits resulting from epidural venous obstruction have
been
described22-24;
however, none of those patients presented with acute paraplegia but rather
with slowly developing symptoms as a consequence of myelopathy or nerve root
irritation. In our patient, the paraplegia was clearly caused by a hematoma,
but venous congestion might have exacerbated the bleeding caused by other
factors. Fourth, the increase in local temperature by the curing cement might
have caused thermal damage to epidural vessels. There is discussion as to
whether temperatures at the posterior border of vertebrae during
vertebroplasty reach levels that can cause tissue
damage25,26.
Since, in T12 (Fig. 2), there
was a considerable volume of cement close to the epidural space and some
cement filling the epidural veins, thermal damage and subsequent bleeding
cannot be ruled out. Fifth, angulation of the trocar needle while attempting
to enter the left pedicle at T12 might have caused a fracture at the pedicle
root, directly causing intraspinal bleeding from the cancellous bone or by
injuring the epidural venous plexus at that location. We examined the left
pedicle root of T12 during surgery and did not see or feel a fissure or
fracture, but visibility was impaired because of the bleeding. The
discontinuity of the pedicle seen in Figure
2 could also have been associated with the original vertebral
fracture. In either case, it seems unlikely that such a small injury alone
would have caused such an extensive hematoma.
An epidural hematoma is an extremely rare occurrence, but it should be
remembered that vertebroplasty, while in general a very safe procedure,
carries risks similar to those of open spinal surgery. If such emergencies
cannot be addressed in a timely fashion, permanent neurological deficits may
result. We therefore recommend that vertebroplasty and similar procedures
should be performed only in settings where emergency spinal surgery is
available. As a consequence of the case of our patient, our current policy is
to discontinue aspirin at least seven days prior to elective open spinal
surgery and minimally invasive spinal procedures. In addition, we routinely
perform preoperative bleeding times in these patients. ?