An otherwise healthy, thirty-eight-year-old woman was involved in a
motor-vehicle accident that resulted in a right upper-extremity radiculopathy.
She was initially evaluated and treated by an orthopaedic surgeon who
recommended physical therapy and cervical epidural injections. After a short
course of physical therapy, the patient underwent a series of three
fluoroscopically guided right transforaminal epidural injections at the C7-T1
level. She noted only mild relief of symptoms after each injection.
Approximately four days after the final injection, the patient awakened
with severe upper thoracic back pain and progressive loss of sensation in the
lower extremities. By late morning, the patient had progressive neurological
symptoms and the onset of weakness, and she was promptly referred to the
emergency department for evaluation. At this point, physical examination was
notable for decreased strength, sensory loss, and diminished deep tendon
reflexes. The strength of the iliopsoas, quadriceps, extensor hallucis longus,
and gastrocnemius and soleus muscles was graded 0 of 5 on the right and 2 of 5
on the left. The deep tendon reflexes were diminished in both lower
extremities. Sensation to pinprick was diminished below the midline of the
abdomen. The rectal tone was normal.
Emergent magnetic resonance images of the spine were acquired
(Fig. 1, A and
B). A large heterogeneous mass that was compressing the
spinal cord from T1 to T5 was noted in the posterior epidural space, prompting
an urgent transfer of the patient to a level-1 orthopaedic trauma center for
surgical decompression.
Laboratory values at the time of presentation at the level-1 center
included a prothrombin time of 9.7 seconds (normal, 8.7 to 11.5 seconds), an
activated partial thromboplastin time of 29 seconds (normal, 22 to 36
seconds), and an international normalized ratio of 1.0. A complete blood-cell
count was normal, with a platelet count of 352 × 103/dL (352
× 109/L). The patient was not taking any antiinflammatory
drugs, and she had no personal or family history of coagulopathy.
Emergent posterior decompression of the spinal canal was achieved with
laminectomies from T2 to T5. The T1 lamina was left intact to preserve the
native cervicothoracic junction. On removal of the posterior elements, a thick
layer of coagulated blood was encountered superficial to the dura. It was
noted to be compressing the dura and adjacent to a normal-appearing confluence
of epidural veins. The clot was carefully separated from the dura with a curet
and sent for Gram stain and culture. In order to adequately decompress the
spinal canal at the T1 level, a long 18-gauge intravenous catheter was used to
irrigate and lavage the epidural space at T1 with normal saline solution.
After the Gram stain failed to identify any organisms, the wound was closed
over a subfascial drain. The final cultures of the wound and the hematoma were
negative.
Postoperatively, the patient began to regain function in the lower
extremities. By the third postoperative day, the strength of the iliopsoas,
quadriceps, extensor hallucis longus, and gastrocnemius and soleus muscles had
improved to grades 3 of 5 on the right and 5 of 5 on the left. A follow-up
magnetic resonance imaging scan at one month showed good re-expansion of the
spinal canal (Fig. 1,
C). By the time of the six-month follow-up, the patient
had regained full strength and sensation in both lower extremities.
Selective cervical nerve-root injections can be effective in both the
diagnosis and treatment of cervical
radiculopathy1,15-17.
However, complications from this procedure may cause devastating
morbidity2-4,8,9.
Several reports have described cerebellar infarcts and brain-stem
herniations after transforaminal epidural steroid injections, which have
resulted in permanent neurological
deficits8,9.
Inadvertent introduction of corticosteroid particulate embolus during arterial
injection may be a possible cause. Brouwers et
al.2 as well as
Ludwig and Burns4
described cases of spinal cord infarction after cervical nerve blockade.
Brouwers et al.2
suggested that perfusion to the major anterior radicular artery of the spinal
cord may become compromised, resulting in an ischemic spinal cord injury.
Their patient experienced flaccid paralysis and breathing difficulties and
required endotracheal intubation immediately following C6 selective nerve-root
injection. The postoperative magnetic resonance imaging scan of the cervical
spine revealed an extensive spinal cord infarct from C2 to T1. Unfortunately,
their patient died several weeks later from medical complications of pneumonia
and gastric perforation. Ludwig and
Burns4 also cited
vascular compromise as a mechanism for spinal cord injury following selective
cervical nerve-root injections. Their patient experienced an immediate
paralysis of the left arm and both lower extremities following a
fluoroscopically guided transforaminal injection on the left side of C6. A
magnetic resonance imaging scan once again showed diffuse spinal cord
infarction, from which the patient did not recover.
Injection-induced hematoma in a patient without underlying coagulopathy or
bleeding diathesis is unusual. Stoll and
Sanchez14 described
a patient in whom Brown-Séquard syndrome developed from an epidural
hematoma eight days after a cervical epidural injection. The injection was
described as being performed in the "C5-C6 interspace," presumably
as an interlaminar injection. Similar to our patient, this patient did not
have any known preexisting risk factors for bleeding, and he recovered full
neurological function after an urgent decompression of the hematoma. Williams
et al.11 reported
on the formation of an intracanal hematoma following an epidural injection;
however, their patient was taking indomethacin and allopurinol. The paralysis
in that patient developed within forty-five minutes of the injection, and he
fully recovered after emergent decompression.
In the case of our patient and in the case of the patient described by
Stoll and
Sanchez14, there
was a delay in presentation of the paralytic symptoms following the last
injection. We believe this may be due to venous bleeding that slowly
accumulated over several days. The patients likely became symptomatic when the
hematoma reached a critical volume and had a mass effect on the spinal cord.
Therefore, physicians should be aware and should inform their patients that
potentially devastating neurological complications can occur days after an
injection.
In a prospective study of 337 patients, Furman et
al.6 determined that
the rate of intravascular injection during a fluoroscopically guided cervical
nerve block can be as high as 19.4%. Detection of an intravascular penetration
by the spontaneous presence of blood in the needle hub (the so-called blood
flash) after an attempted aspiration by pulling back on the syringe plunger
was 97% specific but only 45.9% sensitive. Therefore, the absence of blood in
the needle hub, despite aspiration, is not a reliable indicator of a needle
penetrating a vascular
structure6.
Typically, cervical transforaminal nerve blocks are performed only after
confirmation of the needle position with a nonionic contrast
injection18. Under
real-time fluoroscopy, a contrast injection should outline the nerve if the
needle position is correct. If the contrast material is rapidly carried away
in a cephalad direction, an intra-arterial (vertebral or radicular artery)
penetration should be
suspected18.
Injection into the venous plexus can be much more difficult to visualize.
Venous plexus penetration may show a serpentine pattern of the dye around the
nerve root. However, depending on the quality of the image, intravenous
injection may appear similar to a safe injection. Multiple attempts of needle
placement may cause iatrogenic injury to the epidural vascular plexus around
the nerve root, causing a slow bleed.
Despite these rare complications, we believe that fluoroscopically guided
cervical nerve-root injections can be an effective diagnostic and treatment
modality for cervical radiculopathy. However, the possibility of catastrophic
complications should be appreciated and discussed with the patient prior to
the procedure. Physicians should also be aware that symptoms from a slowly
developing epidural hematoma can present even days following an injection.
?
Note: The authors thank Erin Oneill for her help with the
preparation of the manuscript.