1145

Epidural Abscess of the Cervical Spine: MR Findings in Five Cases

Robert Elliot

American Journal of Roentgenology 1992.158:1145-1149.

George Harry

Knicun1

I. Shoemaker1

I. Chovanes2 W. Stephens2

Cervical epidural abscess is an uncommon infectious process of the spinal epidural space. Although this disorder is often unsuspected clinically, the patient’s signs and symptoms may suggest other diagnoses that frequently lead to an MR examination. We retrospectively reviewed the MR examinations of five patients with surgically proved cervical epidural abscess in order to assist in the diagnosis of this clinically elusive disorder. Each epidural abscess was evaluated for MR signal intensity, location, extent, delineation, and enhancement pattern. We assessed the spinal cord for compression and signal intensity and analyzed the vertebrae, intervertebral disks, and paraspinal soft tissue. Compared with the spinal cord, the abscess was isointense or hypointense on Ti-weighted spin-echo images and hyperintense on T2-weighted images. The ab-

scess was hyperintense

or isointense

relative to the cord on T2” gradient-echo

images.

Enhancement of the abscess occurred in the two patients given an IV injection of gadopentetate dimeglumine. The epidural abscess was located anteriorly in three patients, posteriorly in one, and was circumferential in one. The abscess extended from two to nine vertebral bodies in length. In each case, the abscess caused some degree of spinal cord compression, and one patient had bright signal intensity within the cord on T2-weighted

images.

and paravertebral

MR imaging associated

paraspinal

Three

patients

had

MR changes

of accompanying

osteomyelitis

abscess.

is useful

abnormality

in diagnosing of the spinal

cervical cord,

epidural

vertebral

abscess

bodies,

and in evaluating

intervertebral

disks,

and

soft tissue.

AJR 158:1145-1149,

May

1992

Epidural abscess of the spine is an uncommon disorder estimated to occur in 0.2-i .3 cases per 1 0,000 hospital admissions [1 2]. The cervical spine is the least frequent site of spinal epidural abscess. Although early diagnosis of epidural abscess is considered crucial for successful therapy, the clinical diagnosis is frequently unsuspected. MR imaging can be used to show an epidural process, ,

suggest

the diagnosis,

patients with cervical inflammatory process. Materials

Received October 30, 1991 ; accepted after revision January 2, 1992. 1 Department of Radiology, Lehigh valley Hospital Center,

1 200 5. Cedar

Crest

Blvd.,

Allentown,

PA 18103. Address reprint requests to A. Kricun. 2

Division

of Neurosurgery,

tal Center, Allentown, 0361-803X/92/1585-1 C American

Roentgen

Lehigh

PA 18103. 145 Ray Society

Valley

Hospi-

and reveal the full extent epidural

abscess

of involvement.

and describe

the

We report

MR appearance

five

of this

and Methods

During a 14-month period, we examined the MR images of five patients with surgically proved cervical epidural abscess. The patient population consisted of three men and two women 32-65 years old (average, 55 years). Four of the patients had studies performed on a 1 .5-T MR unit, while one patient had his examination on a 0.5-T unit. Ti -weighted spinecho (SE) images, 500-700/1 2-32 (TR/TE), as well as T2” gradient-echo (GRE) images were obtained in all patients. Proton-densityand T2-weighted SE images (2000-2609/35,70) were obtained in four patients. Short inversion time recovery (STIR) images, 2300/1 65/40 (TR/Tl/ TE), were obtained in one patient. Gadopentetate dimeglumine (Magnevist, Berlex Laboratories, Wayne, NJ) was injected IV in two patients in doses of 0.1 mmol/kg body weight. Tiweighted images were obtained immediately after injection.

KRICUN

1146

TABLE

1: Clinical

Sex

No.

(yr)

1

4

62 32 65 58

M F F M

5

56

M

Epidural

Abscess

Infection

Source

Fever

Location

=

Ti -weighted

images;

PD and T2

+

Anterior

-

Anterior

=

C2-T2 C5-C6 Ci -T2 C2-C3

-

Posterior Anterior

+

Circumferenti

-

proton-density-

Relative

to Spinal

Cord

Extent

Contrast Ti

Pectoral abscess Suspected IV drug abuse Buttock abscess Postintubation retropharyngeal abscess Unknown

May 1992

in Five Patients

al Cl-Ti

and T2-weighted

PD and

T2

Enhancement

GRE

Hypointense Isointense Isointense Isointense

Hypenintense Hypenintense Hyperintense NA

lsointense Isointense Hyperintense Hyperintense

Homogeneous NA NA NA

Isointense

Hyperintense

Hyperintense

Patchy,

images;

American Journal of Roentgenology 1992.158:1145-1149.

Note.-T1

of Cervical

AJR:158,

MR Intensity

Case Ae

2 3

and MR Features

ET AL.

GRE

=

gradient-echo

images;

NA

=

inhomogeneous

not available.

Fig. 1.-62-year-old man with epidural abscess, osteomyelitis, and prevertebral abscess. He had a pectoral abscess and a 3-week history of right shoulder and neck pain. At surgery, extensive granulation tissue was found that culhired S. aureus. MR images were obtained with 1.5-T magnet. A, Sagittal TI-weighted image (SE 500/12) in midline shows epidural abscess (black arrows) and prevertebral abscess (white arrow) of lower signal intensity than spinal cord. Note low signal Intensity of vertebral bodies from C4 to C7. Osteomyelitis at C6-C7 was proved at surgery. B, Sagittal T2-weighted image (SE 2609170)

shows increased signal intensity of abscesses and spinal cord compression. Signal intensity of involved vertebral bodies Is moderately Increased. C, Sagittal STIR image (2300/165/40) shows more marked increase In signal of vertebral bodles from C4 to C7 relative to very low signal of adjacent vertebrae with this fat-suppression technique. Epidural abscess and prevertebral abscess have bright signal Intensity similar to

that of CSF. D, Sagittal Ti-weighted image (SE 600/20) after IV gadopentetate dimeglumine shows marked enhancement of epidural abscess, which extends from Cl to T2 and is causing spinal cord compression. Enhancement of prevertebral abscess and C4-C7 vertebral bodies is also seen.

All the MR images

abscess tion,

and

was evaluated enhancement

were

reviewed

retrospectively.

for signal intensity, pattern.

compression and signal intensity. disks, and paraspinal soft tissue

The

spinal

Each

location, cord

was

The vertebral bodies, were examined.

epidural

extent, delineaassessed

for

intervertebral

Results

The clinical and MR features manized epidural

of our five patients

are sum-

in Table 1 . The Ti -weighted images showed abscess to be isointense relative to the spinal

the cord

in four

patients and hypointense in one patient. The abscess hypenintense relative to CSF in all patients (Figs. 1-4). Four patients had proton-densityand T2-weighted SE im-

was

ages. The proton-density

images

showed

hyperintensity

of

the abscess compared with the cord in all patients. Compared with CSF, there was hypenintensity in three patients and isointensity in one patient. On the T2-weighted images, the epidural abscess was hypenintense relative to the spinal cord and isointense with CSF. GRE images were obtained in all patients, with the abscess showing hypenintensity relative to the cord in three patients and isointensity with the cord in two

AJR:i58,

MR

May 1992

OF

EPIDURAL

ABSCESS

OF

CERVICAL

SPINE

1147

FIg. 2.-32-year-old woman with anterior carvical epldural abscess. She had a 1-week history of severe right-sided neck and arm pain that clinically was thought to be due to disk herniation. Surgery revealed pus that grew S. aureus. A history of IV drug abuse was suspected. MR images were obtained with 1.5-T magnet. A, Sagittal TI-weighted image (SE 700/15) obtained 5 mm to right of midline shows anterior epidural abscess (arrow) of intermediate signal compressing anterior subarachnoid space and

American Journal of Roentgenology 1992.158:1145-1149.

spinal cord. B, Sagittal 12-weighted image obtained 5 mm to right of midline intense epidural abscess (arrow).

(SE 2571/70) hyper-

shows

Fig. 3.-65-year-old woman with posterior cervical epidural abscess. She had a 3-day history of radicular pain radiating into posterior left arm that clinically was thought to be due to disk hemlatlon. Surgery revealed pus that grew S. aureus. A buttock abscess from prior Intramuseuler injections was found postoperatively. MR images were obtained with 1.5-T magnet A, Sagfttal Ti-weighted image (SE 700/15) obtained near midline shows epidural abscess (arrows) of Intermediate signal intensity obscurlig visualization of spinal cord. B, Sagittal T2-weighted image (SE 2000/70) near midline shows high-signal-intensity abscess (arrow) extending posterioriy through mul-

tiple cervical levels.

patients.

One

patient

had STIR

abscess to be hyperintense, The abscess was anterior patient, and circumferential extent was from two to nine

images,

which

showed

the

similar to CSF (Fig. 1C). in three patients, posterior in one in one patient. The craniocaudal vertebral bodies, with an average

of six. The epidural abscess adjacent cord in three patients

was well delineated from the and poorly delineated in two.

The abscess was homogeneous in signal on T2-weighted images in three patients and inhomogeneous in two patients. Gadopentetate dimeglumine was injected IV in two patients. In one marked

patient, the enhancement,

poorly delineated allowing better

collection distinction

showed from the

spinal cord (Fig. 1 0). The enhancement pattern was homogeneous except for a few small unenhanced areas of low signal. The mogeneous,

second patchy

patient receiving an injection had inhoenhancement. Delineation of the abscess

from the spinal cord improved

after contrast

enhancement.

Spinal

cord compression

best demonstrated

was present

on Ti -weighted

in all cases

sagittal

images.

and was

One pa-

tient had central linear high signal within the cord on T2weighted images. There was no abnormal signal within the cord on the Ti -weighted images before or after injection of contrast material. In three of the cases the marrow of two to four vertebral bodies was involved (Figs. 1 and 4). Compared with normal marrow, these vertebrae were hypointense on Ti -weighted

images and hyperintense on T2-weighted or GRE images. One patient had STIR images, which showed hyperintensity of the abnormal vertebrae that was more marked than on the T2-weighted images. Two of the three patients with abnormal marrow intensity

signal had T2-weighted images that showed hyperof the intervertebral disk. The other patient had GRE

images that showed tissue

signal

abnormality

isointensity

of the disk. Paraspinal

was present

in all three

patients

softand

American Journal of Roentgenology 1992.158:1145-1149.

1148

KRICUN

ET

AL.

AJR:i58,

May 1992

Fig. 4-58-year-old man with anterior epidural abscess, C2-c3 osteomyelitis, and prevertebral abscess. He had severe neck pain and pain on swallowing 3 months after a traumatic intubation. A retropharyngeal mass was present on examination. MR images were obtained with 0.5-T magnet. A and B, Sagittal Ti-weighted images (SE 500/32) at midline (A) and left of midline (B) show large prevertebral abscess (straight arrow) and epidural abscess (curved arrows) of intermediate signal. Epidural abscess is compressing spinal cord. Low signal intensity of C2 and 3 vertebrae is due to osteomyelitis. C, GRE image

(400/25,

15#{176} flip

angle) obtained

left of midline shows heterogeneous

bright signal intensity of prevertebral

abscess

and epidural

abscess.

showed

hypointensity

on Ti -weighted

on T2-weighted or GRE images, receiving an injection of contrast

images,

hyperintensity

and enhancement material.

in those

Discussion Spinal epidural abscess is an infectious process epidural space that may consist of pus or granulation

of the tissue.

Sources of infection include osteomyelitis, bacteremia, and postoperative infection. Hematogenous spread is most often a result of a skin infection. Cervical epidural abscess is usually associated

with

a remote

[3].

most frequently

localized

ever,

severe

patients

[4]. A sudden

pain should

of high fever and [5]. Howor have a low-grade fever.

suggest

may be afebrile

onset

the diagnosis

Regardless of the initial course of the disease, paralysis can develop suddenly, thus making

and treatment reversing

crucial

paralysis

in preventing

of short

Only one abscess teriorly, anteriorly.

aureus

duration

weakness and early diagnosis

neurologic

damage

[2, 6]. Despite

this

or ur-

gency, spinal epidural abscess is frequently not considered in the early evaluation of these patients [1 2]. The clinical diagnosis of epidural abscess is correctly suspected initially ,

in only 20-25% of cases [1 2]. Diagnoses that may be considered include extruded disk (as in two of our patients), spinal tumor, vertebral osteomyelitis, musculoskeletal arthri,

prompt

myelitis;

these

disorders

fre-

A review of several large series reveals that the primary location of spinal epidural abscess is most commonly the thoracic spine (approximately 50%) followed by the lumbar (35%) and cervical (1 5%) spine [2]. In our series, the extent of cervical epidural abscess ranged from two to nine vertebral bodies in length with an average of six levels involved. Cervical epidural abscess often occurs anteriorly, unlike the more common thoracic abscess, which usually occurs posteriorly.

found with concomitant is the organism and was cultured in four of our

involving

patients. Gram’s stain of the fifth patient’s abscess revealed gram-positive cocci, although culture failed to document growth of pathologic organisms, possibly because of ongoing antibiotic therapy. The clinical course of epidural abscess has been described in four phases: spinal ache, root pain, weak-

ness, and paralysis

the use of MR evaluation.

rounded

of infection

Staphylococcus

implicated

and transverse

quently

the

source

arm, neck, or head and is frequently osteomyelitis

tis, meningitis,

abscess

while

three

the cord

in our series was located were

circumferentially,

The frequent

is thought

anterior. anterior

myelitis found in these patients anterior epidural abscess paravertebral soft tissue and creased signal intensity on SE

signal

being location

completely

pos-

abscess

sur-

other

more

of cervical

intensity

epidural

of osteo-

[3]. Three of the four patients had signal abnormality of the adjacent vertebrae with deTi-weighted images and in-

on T2-weighted

images

with paraspinal abscess and osteomyelitis. A formed in one patient proved osteomyelitis. The a posterior epidural abscess did not have osteomyelitis. A review of our case shows that an epidural

low or intermediate

prominent

to be due to the high frequency

with

creased

The

signal intensity

consistent biopsy perpatient with evidence of

abscess has on SE Ti -weighted images

and high signal intensity on proton-densityand T2-weighted images. GRE images show the abscess as hyperintense or isointense relative to the cord. The abscess may be well delineated or poorly delineated on T2-weighted images. Delineation of the abscess was improved in the two patients who had an injection of gadopentetate dimeglumine. At the time of surgery, the epidural abscess may consist of pus or granulation tissue. The abscess in four of our patients

AJA:i58,

MR

May 1992

OF

EPIDURAL

ABSCESS

consisted offrank pus while the fifth had extensive granulation tissue. Some investigators [1 2] have found pus in acute epidural abscesses and granulation tissue in the chronic stages. Others [7] have not found a correlation between the pathologic appearance and the duration of illness. Some authors [8] have shown that various components of the ,

inflammatory process can be distinguished on T2-weighted images. The fluid portion of the abscess tends to have markedly increased signal, and the surrounding inflammatory

edema and granulation tissue have inhomogeneous areas of mildly increased signal intensity. We, like others [9], did not find an exact correlation between the signal on MR and the stage of the epidural abscess. The high signal intensity of pus and more chronic granulation tissue may be similar. Rarely, hemorrhage ogeneous

may accompany high signal on Ti

the infection,

producing

American Journal of Roentgenology 1992.158:1145-1149.

and T2-weighted has the appearance of an epidural hematoma our cases showed a hemorrhagic component.

heter-

images that [1 0]. None of

-

Because both the axial and sagittal planes could be scanned, the extent of epidural involvement and the degree of compression of the thecal sac and spinal cord were readily appreciated with MR. We found some degree of spinal cord compression on MR studies of all five patients. One of our

OF CERVICAL

images

vertebral marrow

exudative

and ambulation

after

Hyperintensity

surgery

and

of the cord has

been described in some patients with spinal cord compression [1 0]. This is of uncertain etiology, although the increased

signal in the cord may be due to changes

of myelitis

that can

regress after evacuation of the epidural collection antibiotic therapy [9]. Serial MR scans after treatment useful in the follow-up evaluation.

and/or can be

Gadopentetate tients.

dimeglumine

In one patient,

of the abscess,

which

tissue that showed ment. In the other hancement of the Three patterns of

was

injected

this was helpful

consisted

IV in two

in delineating

of extensive

pa-

the extent

granulation

marked patient, epidural

and fairly homogeneous enhancepatchy and inhomogeneous enabscess aided in its delineation. enhancement have been described and

include dense homogeneous enhancement, inhomogeneous enhancement with scattered areas of sparse or no uptake, and thin peripheral enhancement [i i ]. Gadopentetate dimeglumine also helps define the activity of the infectious process, and can be used to help localize an area for potential biopsy as well as to determine response to antibiotic treatment. However, an acute epidural abscess composed mainly of pus with little granulation

tissue

may show

either

or minimal peripheral enhancement [1 1]. MR imaging is accurate and sensitive accompanying 1 3], as was

no enhancement

in the detection

of

osteomyelitis with disk space infection [9, 12, shown in three of our patients. Ti -weighted

from

the marrow

of two

with an indistinctness

or

of the

bodies. probably

The signal changes

of the vertebral

represent

water

process

increased

and ischemia.

bone

content

Some investigators

of the

[9] have

symptoms usually will lead to imaging evaluation. The MR features of epidural abscess can be used to suggest the

REFERENCES

in strength

bodies

stressed the importance of finding high signal intensity within the intervertebral disks on T2-weighted images, while others [1 3] have found variable signal intensity. STIR images have been recommended by some investigators for the evaluation of osteomyelitis [1 2, i 3]. With this pulse sequence, an inversion time is selected that effectively suppresses the signal from fat, making the normal bone marrow appear dark. In contrast, pathologic processes of high water content such as inflammation or neoplasm appear very bright, thus providing images with high contrast between pathologic and normal tissue. MR imaging plays an important role in the evaluation of patients with cervical epidural abscess [5, 8, 12, 13]. Although the diagnosis may not be suspected clinically, the patient’s

good

course of antibiotics.

signal

vertebral

vertebral body endplates and narrowing of the disk space [1 3]. T2-weighted images show increased signal from the

diagnosis

recovery

decreased

more adjacent

patients had bright signal intensity within the cord on T2weighted images. This patient had sudden neurologic deterioration with weakness that rapidly led to paralysis. An emergency MR examination was obtained, and surgery was performed 5 hr after the onset of weakness. The patient made a a lengthy

show

1149

SPINE

and

guide

disk space infection, uated with MR.

therapy.

Accompanying

and paravertebral

osteomyelitis,

abscess

can be eval-

AS, Ojemann AG, Swartz MN, Richardson EP Jr. Spinal epidural abscess. N EngI J Med 1975;293:463-468 Danner AL, Hartman BJ. Update of spinal epidural abscess: 35 cases and

1 . Baker 2.

review 3. Lasker 1987;37:

of the literature.

BA,

Harter

1747-i

753

Rev Infect

DH.

Dis 1987;9:265-274

Cervical

epidural

abscess.

4. Heusner AP. Nontuberculous spinal epidural infections. 1948;239:845-854 5. Emtell M, Holtas 5, Norlin K, Dahiquist E, Nilsson-EhIe I. nance imaging in the diagnosis of spinal epidural abscess. Dis 1988;20:323-327 6. Vemer EF, Musher DM. Spinal epidural abscess. Med

Neurology

N Eng!

J Med

Magnetic resoScand J Infect C/in

North

Am

1985;69:375-384

7. Kaufman DM, Kaplan JG, Litman N. Infectious agents in spinal epidural abscesses. Neurology 1980;30:844-850 8. Angtuaco EJC, McConnell JR, Chadduck WM, Flanigan S. MA imaging of spinal epidural sepsis. AJNR 1987;8:879-883, AJR 1987;i49: 1249-1 253 9. Post MJD, Quencer AM, Montalvo BM, Katz BH, Eismont FJ, Green BA. Spinal infection: evaluation with MA imaging and intra-operative us. Radiology 1988;169:765-77i 1 0. Enzmann DR. Infection and inflammation. In: Enzmann DR. DeLaPaz AL, Rubin JB, eds. Magnetic resonance of the spine. St Louis: Mosby,

1990:260-300 1 1 . Post MJD, Sze G, Quencer AM, Eismont FJ, Green BA, Gahbauer H. Gadolinium-enhanced MA in spinal infection. J Comput Assist Tomogr 1990;14:721 -729

12. Bertino RE, Porter BA, Stimac GK, Tepper SJ. Imaging spinal osteomyelitis and epidural abscess with short TI inversion recovery (STIR). AJNR 1988;9:563-564

13. Thrush

A, Enzmann

1990;ii:1171-ii8O

D. MR imaging

of infectious

spondylitis.

AJNR