The brachial plexus: normal

ma

,c,*

en

t of R

, Miami, FL 33136, USA

al Center, 1611 NW 12th Avenue, Miami, FL 33136, USA

y of M

ivers

A 981

Neuroimag Clin N Amdiagnostic evaluation can now be acquired routinely

because of improvements in gradient coil technology,

MR pulse sequence design, and radiofrequency (RF)

coil design. In particular, the use of phased arrays and

integrated arrays of RF coils for dedicated brachial

plexus imaging has made it possible to directly

evaluate the plexus components (ie, roots, trunks,

divisions, and cords) and, frequently, to distinguish

between intrinsic and extrinsic pathologic changes.

The traditional approach to evaluation of the plexus

focused on morphologic evidence for a mass lesion

the additional information promises to yield more

accurate diagnoses in patients who have plexopathy

[25].

Anatomy

Each component of the brachial plexus has the

basic endoneurium-perineurium-epineurium organi-

zation and fascicular structure that have been de-

scribed for isolated peripheral nerves, such as the

median and ulnar nerves. The largest components

are approximately 5 mm in mean diameter [6]. To

localize plexus lesions detected on MR imaging,* Corresponding author. Division of Neuroradiology/spatial resolution and sufficient signal-to-noise ratio

(SNR) and tissue contrast-to-noise ratio (CNR) for

have a larger database from which to formulate a dif-

ferential diagnosis. Based on published case material,MR i

Brian C. Bowen, MD, PhDa,b

Efrat Saraf-Lavi, MDb, KaDivision of Neuroradiology/MRI Center, Departmen

1115 NW 14th StreetbDepartment of Radiology, Jackson Memorial Medic

cMiami Project to Cure Paralysis, UniversitdNeuroradiology and MR Research Laboratory, Un

Seattle, W

MR imaging of peripheral nerves at high spatial

resolution, with and without fat suppression, has

been shown to detect features of intraneural anatomy

not previously seen on diagnostic imaging studies and

to localize pathologic lesions in conditions where

electrophysiologic and physical findings are nonspe-

cific or nonlocalizing [1]. The improvements in nerve

imaging have led to the increasing use of MR imaging

in the evaluation of peripheral nervous system dis-

ease, including brachial plexopathy. Images with high1052-5149/04/$ see front matter D 2004 Elsevier Inc. All right

doi:10.1016/j.nic.2003.12.002

MRI Center, Department of Radiology, University of

Miami School of Medicine, 1115 NW 14th Street, Miami,

FL 33136.

E-mail address: bbowen@med.miami.edu

(B.C. Bowen).iami School of Medicine, Miami, FL, USA

ity of Washington, 1959 NE Pacific, Box 357115,

95, USA

infiltrating perineural fat. The approach has been

broadened and now includes an assessment of the

intrinsic MR features of nerves, such as signal inten-

sity on short tau inversion recovery (STIR) or fat-

saturated T2-weighted fast-spin-echo (FSE) images,

the appearance of the intraneural fascicular pattern, or

the pattern of postcontrast enhancement on fat-satu-

rated T1-weighted images. By analyzing the signal

characteristics of the nerves (and innervated muscle)

and the regional morphologic features, radiologistsanatomy, pathology, and

ging

, Pradip M. Pattany, PhDa,c,neth R. Maravilla, MDd

adiology, University of Miami School of Medicine,

14 (2004) 5985knowledge of the anatomy of the plexus and the

relationship of the plexus to adjacent muscles, ves-

sels, and osseous landmarks is necessary [1].

Descriptions of brachial plexus anatomy (Fig. 1)

begin with the roots, which are continuous with the

s reserved.

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 598560ventral rami of the spinal nerves. In the most common

configuration of the plexus, there are five roots: C5,

C6, C7, C8, and T1. Continuing peripherally, the

roots form three trunks: upper (from the C5 and C6

roots), middle (from the C7 root), and lower (from

the C8 and T1 roots). Each trunk separates into an

anterior and posterior division, resulting in a total of

six divisions. The divisions form three cords: lateral

(from the anterior divisions of the upper and middle

trunks), posterior (from the three posterior divisions),

and medial (from the anterior division of the lower

trunk). In the lateral aspect of the axilla, each cord

divides into two terminal branches: musculocuta-

neous nerve and lateral root of median nerve (lateral

cord), axillary nerve and radial nerve (posterior cord),

and ulnar nerve and medial root of median nerve

(medial cord). The upper branches of the plexus,

which arise in the neck rather than the axilla, include

Fig. 1. The brachial plexus with selected vascular and musculoske

subclavian/axillary artery and the lower is the subclavian/axillar

intersect the following landmarks: A, interscalene triangle; B, later

trunk and its posterior division are located posterior to the subcla

corresponding approximately to planes A and B are shown in Fig.

plexus. In: Bradley W, Stark D, editors. Magnetic resonance Ima

with permission.)the dorsal scapular, suprascapular, and long thoracic

nerves, and the nerve to the subclavius.

Branches of the lateral and medial cords innervate

the anterior muscles of the upper limb, and branches

of the posterior cord innervate the posterior muscles.

This pattern also may be expressed in terms of the

plexus divisions: the anterior divisions innervate

anterior muscles and the posterior divisions inner-

vate posterior muscles. Of the upper limb peripheral

nerves, the ulnar nerve is the only one derived ex-

clusively from the medial cord, which is a favored

site for plexus involvement in metatstatic breast

carcinoma, or from the lower trunk, which is usually

involved early in superior sulcus carcinoma (Pan-

coasts tumor). Therefore, in patients who have breast

or lung carcinoma, lower trunk plexopathy or ulnar

neuropathy should raise suspicion for early tu-

mor infiltration.

letal landmarks. Of the two vessels shown, the upper is the

y vein. The three dotted lines indicate sagittal planes that

al margin of the first rib; and C, coracoid process The lower

vian artery, which appears translucent. Sagittal MR images

4. m, muscle; n, nerve. (Reprinted from Bowen B. Brachial

ging. Philadelphia: Mosby-Year Book.; 1999. p. 182132;

small field-of-view (FOV) with a high SNR. By

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 5985 61By identifying the anatomic landmarks shown in

Fig. 1, the location of the plexus components may be

estimated (Fig. 2). The first landmark is the cleft

between the anterior and middle scalene muscles

referred to as the interscalene triangle (see Fig. 1,

plane A). The anterior scalene muscle originates from

the transverse processes of C3C6 and inserts on the

first rib anteriorly. The middle scalene originates from

the transverse processes of C2C6 (FC7) and insertson the first rib posterolaterally. On the medial aspect

of the interscalene triangle, the plexus is composed

primarily of roots, and on the lateral aspect, primarily

of trunks. The trunks continue inferolaterally and

form the divisions before, or in the vicinity of, the

plane of the second landmark, the lateral border of

the first rib (see Fig. 1, plane B). As the plexus

Fig. 2. Gross anatomy of the brachial plexus (frontal view).

R, T, D, and C identify the approximate locations of the

roots, trunks, divisions, and cords. SCA, subclavian artery.

Note the sizes of the trunks relative to the subclavian artery.courses between the first rib and the clavicle to

enter the axilla, the divisions form the cords. These

relationships are variable, and for example, the lower

trunk often extends into the axilla before dividing

into a small posterior division that completes the pos-

terior cord and a large anterior division that continues

as the medial cord. On reaching the plane defined by

the medial border of the coracoid process, which is

the third landmark (see Fig. 1, plane C), the cords are

fully formed.

The subclavian artery passes posterior to the

anterior scalene muscle, whereas the subclavian vein

passes anterior to the muscle. The second part of the

subclavian artery, which is the part immediately

posterior to the muscle, accompanies the plexus

through the interscalene triangle. The C5C7 roots

are located superior to the artery, whereas the C8 and

T1 roots have a more horizontal course posterior to

the artery. The trunks and divisions are usually foundcombining the signals from the multiple coil ele-

ments, an image (Fig. 3B) can be produced that has

the high SNR of each element yet encompasses a

composite FOV similar to that of a single larger

surface coil [8,9]. For the six-element phased array

shown in Fig. 3A, three elements span the neck-to-

shoulder region on each side of the body. To image

the left plexus, for example, the three elements on

the left side and one element on the right side are

activated. The dedicated brachial plexus phased ar-

ray [9] has not been widely used for routine clinical

studies, in part because of its limited commer-

cial availability.

Alternatively, an integrated array of RF coils,

which is achieved by combining multipurpose phased-

array and other coils, can be implemented on some of

the commercially available MR scanners. For exam-

ple, in Fig. 3C, the phased-array neck coil, quadrature

head coil, phased-array spine coil, and a quadrature

multipurpose flexible coil are shown in place and cansuperior and posterior to the third part of the subcla-

vian artery. As the plexus and the subclavian artery

emerge from the interscalene triangle, they carry with

them a sleevelike diverticulum of the prevertebral

fascia, which accompanies them to the axilla where it

contributes to the axillary sheath [7].

In the axilla, each of the three cords is named

based on its anatomic relationship to the axillary

artery: lateral, posterior, or medial. In the vicinity of

plane B in Fig. 1, the divisions and cords are bordered

posteriorly by the serratus anterior muscle and ante-

riorly by the clavicle or pectoralis major muscle. In

the vicinity of plane C, the cords are bordered

posteriorly by the subscapularis muscle and anteriorly

by the pectoralis major and minor muscles. Just la-

teral to the pectoralis minor muscle, the cords divide

into the terminal branches. Axillary lymph nodes lie

in the loose connective tissue around the major neu-

rovascular structures.

MR techniques and image contrast

Radiofrequency coils

High-resolution images of the plexus can be

obtained by using a phased-array RF receiver coil

designed specifically to cover the plexus anatomy

(Fig. 3A). Phased-array RF coil technology combines

the signal data from multiple small coils. The RF

signal from each small-coil element is collected

independently of the other coil elements in the

phased array and each contributes signal from a

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 598562be combined. Fig. 3D shows an image produced by

combining specifically the elements of the anterior

neck coil, the upper two elements of the phased-

array spine coil, and the flexible coil.

With the coil configurations described previously,

the goal is usually to image either the right or left

plexus at high spatial resolution, rather than a bilat-

eral examination with lower resolution. A compre-

hensive MR study extends from the roots and trunks,

located in the supraclavicular region, to the terminal

Fig. 3. Radiofrequency receiver coils for brachial plexus imaging. (

on the right, three on the left). The MR scanner operator determine

weighted image of a normal brachial plexus obtained using the co

locations for the roots, trunks, divisions, and cords. (Reprinted f

nervous system: evaluation of peripheral neuropathy and plexo

permission.) (C) Commercially available integrated array of co

the brachial plexus. The coils shown are the anterior neck coil a

chest of the volunteer, and the head coil. Not shown is the spine c

the flexible coil array, one element of the neck coil array, and two

saturated T2-weighted image obtained using the integrated array

mate sagittal locations for the roots, trunks, divisions, and cords. N

neck on the right. This results from incomplete fat saturation due to

SCA, subclavian artery; tm, thyroid mass.branches of the cords, located in the infraclavicular

region just lateral to the pectoralis minor muscle.

For optimal results, the MR study is targeted to a

specific region (Fig. 4) of the plexus by a careful clin-

ical examination.

Pulse sequences

At field strengths of 1.0 to 1.5 T, the brachial

plexus is commonly evaluated based on its appearance

A) Dedicated phased array consists of six coil elements (three

s which of the six coil elements are active. (B) Coronal T1-

ils shown in A. The arrows indicate the approximate sagittal

rom Maravilla KR, Bowen BC. Imaging of the peripheral

pathy. AJNR Am J Neuroradiol 1998;19:101123; with

ils, which can be combined to image the entire course of

rray and flexible general-purpose coil, both resting on the

oil array located underneath the supine volunteer. Typically,

elements of the spine coil array are active. (D) Coronal fat-

of coils illustrated in C. The arrows indicate the approxi-

ote the hyperintensity of the intermuscular fat in the lower

magnetic susceptibility variation in the region being imaged.

Fig. 4. MR appearance of the normal brachial plexus. (A) Gross anatomy of the roots of the brachial plexus (sagittal section).

The C5T1 roots are indicated by five arrows. Note that the T1 root is located immediately inferior to the first rib. (Reprinted

from van Es HW, Witkamp TD, Lino MPR, Feldberg MAM, Nowicki BH, Haughton VM. MR imaging of the brachial

plexus using a T1-weighted three-dimensional volume acquisition. Int J Neuroradiol 1996;2:26473; with permission.)

(B) Sagittal T1-weighted image of the roots. The distribution is similar to the anatomic specimen in Fig. 4A. The plane of

the image corresponds approximately to plane A in Fig. 1. The roots are surrounded by fat and thus easily seen. The C5 root

is depicted with an arrow. (C) Sagittal STIR image of the trunks. The trunks are located superior to the subclavian artery. The

plane of the image is slightly lateral to that shown in B. With fat suppression, the trunks are distinguished by their mild

hyperintensity relative to muscle and fat. (D) Sagittal STIR image of the divisions. The divisions are located primarily superior

to the subclavian artery in the retroclavicular region. The plane of the image is near the lateral border of the first rib and

corresponds to plane B in Fig. 1. Fig. 4C and D are reproduced from Maravilla KR, Bowen BC. Imaging of the peripheral

nervous system: evaluation of peripheral neuropathy and plexopathy. AJNR Am J Neuroradiol 1998;19:101123; with

permission.) A, subclavian artery; a, anterior; m, middle scalene muscles; V, vein; 1r, first rib.

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 5985 63

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 598564on T1- and T2-weighted images with flow compen-

sation. Standard two-dimensional Fourier transform

(2DFT) spin-echo or FSE sequences are used to

generate the T1-weighted images, although some

investigators prefer T1-weighted three-dimensional

gradient-echo images [10,11]. The T1-weighted

images display regional anatomy, including the vari-

ous muscles, blood vessels, and nerves outlined by

tissue fat planes. The 2DFT T2-weighted images

are generated with FSE sequences and are useful to

detect pathologic changes within components of the

plexus. Because abnormal intraneural signal from one

component of the plexus, such as a root or a cord,

may be obscured by adjacent fat signal, various

methods of fat suppression are used in conjunction

with the FSE sequence. The two most common

methods are as follows: (1) STIR, with nulling of

the signal contribution from fat, and (2) frequency-

selective saturation of the fat resonance.

In general, the STIR method has proved to be

more reliable because it gives uniform and consistent

suppression of fat signal from patient to patient while

maintaining excellent T2-like contrast on long TR

images. There are, however, three disadvantages to

the STIR method. The first is relatively low SNR.

The second is greater sensitivity to blood flow

artifacts that can degrade the resulting STIR image.

Flow saturation bands are used to reduce the intensity

of these artifacts. The third disadvantage is the

inability to generate images with tissue contrast that

resemble T1-weighted spin-echo images.

The frequency-selective fat saturation method

has a higher SNR, fewer blood-flowrelated arti-

facts, and can generate T1-weighted images. The

major disadvantage of this method, however, is

the variability in fat suppression across the FOV.

The variability is primarily from inhomogeneity

in the main magnetic field (B0) caused by the

magnetic susceptibility of various body tissues. Be-

cause of these effects, T2-weighted (or T1-weighted)

images may have some areas with incomplete sup-

pression of fat signal and other areas with water

saturation rather than fat saturation, which can result

in poor discrimination of peripheral nerves and the

anatomic landmarks shown earlier. The problem may

be accentuated by nonuniformity in RF signal trans-

mission (B1 inhomogeneity).

Several approaches to improving the suppression

of fat signal, or removing it from the plexus images,

are currently under investigation. The simplest ap-

proach uses the current method of frequency-selective

fat saturation combined with the following techniques

for reducing B0 inhomogeneity: (1) bags containing

clay suspensions, such as attapulgite, placed againstthe neck and suprascapular region to reduce magnetic

susceptibility effects in the plexus region [12], and

(2) careful second-order shimming of the magnetic

field over a volume encompassing the region of

interest. A more ambitious approach uses multipoint

Dixon fat-water separation [13,14] with steady-

state free precession (SSFP) or FSE imaging methods

[15]. This rapid-acquisition approach yields water-

only or fat-only images in a relatively short scan time.

Hargreaves et al [15] have shown water-only images,

acquired with high resolution (1 mm3) and with T2-

like contrast, that clearly distinguish small peripheral

vessels (bright signal) from muscle (low signal). This

technique, which has not yet been applied to plexus

or peripheral nerve studies, also may be useful for

contrast-enhanced imaging.

T1- and T2-weighted images of the plexus are

obtained in the same plane of orientation and using

the same imaging parameters for FOV, location, and

slice thickness. Coregistration of the T1- and T2-

weighted images allows direct comparison between

them, thus facilitating the characterization of the

signal properties of anatomic or pathologic structures

of interest (Fig. 5). Coregistration is necessary be-

cause the fat-suppressed T2-weighted images often

lack delineation of normal tissue planes (usually

outlined by fat), making identification of anatomic

landmarks difficult. The radiologist must then rely on

the direct comparison between T1-weighted images

with high-signal fat and the T2-weighted images to

locate critical landmarks and correctly recognize ple-

xus components.

Images are obtained in two orientations. In-plane

imaging of the brachial plexus is done in the direct

coronal plane. Although it might seem preferable to

obtain angled coronal images for evaluation of the

plexuses, experience has shown otherwise. Long

segments of the brachial plexus have a rather shallow

obliquity to the true coronal plane of the body and

may be imaged in one or two coronal sections (see

Fig. 3B, D). Cross-sectional imaging of the brachial

plexus component nerves is accomplished using

either a true sagittal or an oblique sagittal plane for

the side of the body of clinical interest. True sagittal

images are preferred by some investigators because

anatomic landmarks are recognized more easily with

this orientation, and the images can be compared with

those typically found in atlases of sectional anatomy.

The oblique course of the plexus, however, inevitably

results in some components being imaged obliquely

in the sagittal sections (see Fig. 4). Oblique sagittal

images can provide views that better approximate a

true cross-section of the plexus, and these images

may allow better detection of nerve enlargement or

alteration in signal or fascicular pattern than conven-

tional anatomic sagittal images.

The FOV for plexus imaging is targeted to the

side of clinical interest. Image quality is improved

with the use of a dedicated phased array or an

integrated array of RF coils, as noted previously.

For direct coronal images: FOV = 17 22 cm,matrix = 512 256 or 512 512, slice thickness =3.5 4 mm, and interslice gap = 0 - 0.5 mm. For sag-ittal or oblique sagittal images: FOV = 14 17 cm,

Fig. 5. Schwannoma of the lateral cord. Coronal T1-weighted (A), STIR (B), and postcontrast, fat-saturated T1-weighted

contig

erinten

al T1

. The

scle a

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 5985 65(C) images show a lobular, sharply marginated mass (arrow)

superior and lateral to the axillary artery. The mass is hyp

contiguous cord is hyperintense without enhancement. Sagitt

superior and anterior to the flow void of the axillary artery

sagittal images. (F) The mass (arrow) abuts the subclavius mu

T1-weighted image.uous with the left brachial plexus in the region of the cords,

se in B and demonstrates enhancement in C, whereas the

-weighted (D) and STIR (E) images show the mass (arrow)

location corresponds to the distribution of the lateral cord on

nd enhances homogenously on the postcontrast, fat-saturated

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 598566matrix = 512 256 or 512 512, slice thickness =4 mm, and interslice gap = 1 2 mm. The largerinterslice gap for cross-sectional imaging permits

adequate coverage of the plexus in a reasonable

scan time. Before the conventional imaging, coronal

or axial images with a large FOV covering both right

and left plexuses may be obtained with a body coil, if

comparison between symptomatic and asymptomatic

sides is warranted.

Spatial presaturation pulses are used to saturate

the signal from tissues outside the imaging slab and

suppress blood-flow artifacts [2]. Obliquely oriented

saturation bands are placed over the heart and aortic

root to saturate arterial flow, especially for long TR

images. An oblique band also may be placed over the

lateral axilla to saturate venous flow; however, this

additional saturation band increases acquisition time

and typically is not implemented. Respiratory motion

may degrade both coronal and sagittal images of

the brachial plexus region. Although respiratory gat-

ing may be added to the imaging sequence to reduce

motion artifacts, it results in considerably increased

scan time and is therefore rarely used.

Contrast-enhanced images of the brachial plexus

are obtained routinely in patients being evaluated for

Fig. 5 (contisuspected neoplasm, radiation injury, inflammation

or abscess formation, and following peripheral nerve

surgery. In addition to these indications, contrast-

enhanced images also have proved useful in some

cases of nerve entrapment and stretch injury. In

patients who have acute severe traumatic nerve injury

and simple compressive neuropathy, a noncontrast

examination can be sufficient. For contrast-enhanced

studies, a standard dose of 0.1 mmol/kg is adminis-

tered as an intravenous bolus. T1-weighted spin-echo

images, with frequency-selective fat saturation, are

acquired immediately following injection. In some

cases, delayed imaging, 10 to 20 minutes post injec-

tion, may be useful to better define abnormal areas

of enhancement.

MR imaging: normal versus abnormal plexus

On T1-weighted coronal images, the normal bra-

chial plexus has the appearance of an elongated

bundle of fibers that are isointense to adjacent scalene

muscles and frequently interspersed with thin strips

of adipose tissue (see Fig. 3B). This bundle is

identified as the plexus by its location superior and

nued).

lished series [20,21] of surgically treated primary non-

neurogenic brachial plexus tumors collected over a

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 5985 67posterior to the curvilinear flow void of the sub-

clavian-to-axillary artery and by its mild hyper-

intensity (relative to adjacent muscles) on STIR or

fat-saturated T2-weighted FSE images (see Fig. 3D).

Differentiation of the plexus from the adjacent sub-

clavian/axillary vessels may be difficult if there is

slow flow in the artery, or flow-related enhancement

in the vein, paralleling the plexus.

On the sagittal images, the individual components

of the plexus have the following appearances and

locations: (1) the roots (see Fig. 4B) and trunks (see

Fig. 4C) are round or oval-shaped and located supe-

rior and posterior to the flow void of the subclavian

artery in the region of the interscalene triangle (see

Fig. 1, plane A); (2) the divisions appear as dotlike

structures in a triangular-shaped cluster that is located

superior to the subclavian artery flow-void and ante-

rior to the serratus anterior muscle (see Fig. 4D); and

(3) the cords have rounded or elongated shapes,

sometimes incompletely resolved from each other,

and are found adjacent to the axillary artery in the

following locations (in clockwise order): anterosupe-

rior (lateral cord), superoposterior (posterior cord),

and posteroinferior (medial cord) [2].

Abnormal brachial plexus findings include the

following: loss of fat planes around all or part of a

plexus component, diffuse or focal enlargement of a

component (especially the presence of an eccentric or

nodular mass), and marked hyperintensity on T2-

weighted images or enhancement on T1-weighted

images with fat suppression. An altered fascicular

pattern is also abnormal, although this finding may

not be apparent in the brachial plexus. Demonstration

of a fascicular pattern may be more difficult for

plexus components than for individual peripheral

nerves, such as the sciatic and tibial nerves, because

of the lower spatial resolution of plexus images and

because of the difficulty in obtaining true cross-

sectional views of most plexus components [1].

Isolated hyperintensity of plexus components on

STIR images may not be abnormal and should be

viewed cautiously when unaccompanied by morpho-

logic or other evidence of disease or nerve injury.

Recently, Chappell et al [16] provided evidence of a

magic angle effect for peripheral nerves by showing

that there is a 46% to 175% increase in signal

intensity in the median nerve as its orientation rela-

tive to the main B0 magnetic field changes from 0(parallel to B0) to 55 (the magic angle), accompaniedby an increase in mean T2 relaxation times from

47.2 ms to 65.8 ms. Images depicting the signal

intensity changes in the ulnar and sciatic nerves and

in the brachial plexus as a function of orientation

relative to B0 suggest that the effect is likely to be29-year period. Of the 48 reported tumors, 44% were

benign and included fibromatosis (most common),

lipoma, myositis ossificans, ganglioneuroma, heman-

gioma, and lymphangioma. The information from

MR imaging aids in preoperative planning and may

help to shorten the surgical procedure [3,5].

Wittenberg and Adkins [22] retrospective study

of 104 patients who had nontraumatic brachial plexo-

pathy and who underwent imaging (FOV 28 cm,

matrix 256 256 or 192, and slice thickness 7 mmgeneralized for peripheral nerves and nerve plexuses.

The components of the brachial plexus are likely to

exhibit the effect because they are generally oriented

at 55 to the body axis and B0 when a patient is in theconventional supine position during scanning. The

magic angle effect, which has been well documented

in tendons and ligaments, is a manifestation of T2

anisotropy attributed to the densely packed, hydrated

collagen in peripheral nerves.

Indications for MR imaging of the brachial plexus

The accuracy of MR imaging in detecting plexus

lesions has not been widely reported. A 1994 study

by Bilbey et al [17] found conventional spin-echo

MR imaging without gadolinium to be 63% sensitive,

100% specific, and 77% accurate compared with

clinicopathologic results in the evaluation of 43 pa-

tients with suspected brachial plexopathy. Accuracy

increased to 88% when the evaluation involved only

the subset of patients (n = 34) with neoplastic or

traumatic disorders. With current high-resolution MR

techniques and the use of gadolinium contrast agents,

accuracy is likely to be increased further [18].

Mass involving the plexus

MR imaging techniques can often determine

whether a mass is intrinsic or extrinsic to a compo-

nent nerve of the plexus and, for extrinsic masses,

determine the site of the displaced and compressed

nerve fibers before surgical intervention. This infor-

mation is valuable in the diagnosis and management

of patients who have plexopathy caused by neoplastic

processes, such as nerve sheath tumors (see Fig. 5),

metastases (Fig. 6), direct extension of non-neuro-

genic primary tumor (Fig. 7) and lymphoma, or

benign processes, such as aggressive fibromatosis

(desmoid tumor) and nodular fasciitis. Saifuddin

[19] has summarized the results of two major pub-

Fig. 6. Recurrent breast carcinoma in a 60-year-old woman with a history of right mastectomy and radiation therapy for breast

cancer. Coronal T1-weighted (A), fat-saturated T2-weighted (B), and postcontrast, fat-saturated T1-weighted (C) images

demonstrate a lobular, hyperintense, enhancing mass (arrow) diffusely involving the trunks, divisions, and proximal cords. The

distal cords and peripheral nerve branches are distinct, yet are hyperintense and minimally enhancing (arrowhead). Sagittal T1-

weighted (D), fat-saturated T2-weighted (E), and postcontrast, fat-saturated T1-weighted (F) images. The plane of the images

approximates plane C in Fig. 1. The lobular, eccentric mass (arrow) engulfs the lateral, posterior, and medial cords, and abuts the

axillary artery (arrowhead).

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 598568

Fig. 7. Pancoasts tumor (adenocarcinoma) involving the T1 root of the left brachial plexus. Coronal T1-weighted precontrast

(A) and fat-saturated postcontrast (B) images demonstrate a moderately enhancing, apical lung mass (between large and small

arrows) invading the extrapleural paraspinal space inferior to the left first rib (1r). The T1 (identified as 1) and T2 vertebra and

the left T1 neural foramen are involved. In B, susceptibility effects are responsible for the lack of fat saturation in the upper

mediastinum (*) and the inhomogeneous signal in the enhancing mass at the left lung apex. Sagittal T1-weighted precontrast

(C) and fat-saturated postcontrast (D) images. The plane of the images corresponds to the interscalene triangle and approximates

plane A in Fig. 1 and the image plane in Fig. 4B. The mass invades the periapical soft tissues inferior to the body of the first rib

(1r) and involves the second rib (2r), which is hypointense in C and enhancing in D. The arrow points to the location of the T1

root, which is involved by the mass. Immediately inferior to the T1 root, and along the inferior margin of the arrow in D, there is

an oval hyperintensity, which represents a combination of tumor enhancement and susceptibility effect at the lung apex. The

tumor involvement of the T1 root in C and D is confirmed on the sagittal STIR image (E), which shows the normal mild

hyperintensity of the C5 (arrowhead) through C8 roots in the interscalene triangle and the confluence (arrow) of the T1 root with

the markedly hyperintense invasive mass. Invasion of the second rib (2r) is also evidenced by marked hyperintensity in E.

A, subclavian artery; V, vein; 3r, third rib.

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 5985 69

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 598570with a 3-mm interslice gap) found that 75% of the

explainable causes of plexopathy were attributable to

three processes: radiation fibrosis (31%), metastatic

breast cancer (24%), and primary or metastatic lung

cancer (19%). MR imaging of the plexus in patients

who have suspected cancer is useful in the detection

of malignant primary (see Fig. 7) or recurrent tumors

(see Fig. 6) that infiltrate nerve structures. High-

resolution coronal and sagittal images are especially

beneficial in cases where clinical examination and

routine imaging studies are not able to distinguish

whether a patients symptoms are caused by recurrent

tumor, postoperative or post-treatment changes asso-

ciated with scarring, or compressive neuropathy re-

sulting from regional deformity. In patients who have

plexopathy and Horners syndrome, axial images are

useful to demonstrate paraspinal extension of tu-

mor. If a mass is contiguous with the ipsilateral lon-

gus colli muscle, the sympathetic chain usually has

been invaded.

Brachial plexopathy caused by metastatic disease

is most often seen in patients who have carcinoma of

the breast or lung. Metastases from breast carcinoma

are the most common and involve the plexus mainly

by lymphatic spread [23]. Other primary malignan-

cies, such as melanoma and gastrointestinal or geni-

tourinary carcinomas, which metastasized to lymph

nodes, soft tissue, or bone and resulted in brachial

plexopathy, have been reported [17,2325]. True

hematogenous metastases to the plexus are rare [19].

Clinically, most patients who have tumor infiltra-

tion of the brachial plexus from metastatic breast

cancer or direct extension of primary lung cancer

present with moderate to severe pain as the initial

symptom. Typically, the pain begins in the shoulder

girdle and radiates to the ulnar aspect of the forearm

and hand, primarily in a C8T1 distribution [23,26].

On neurologic examination, most patients also dem-

onstrate weakness (eg, intrinsic muscles of the hand),

atrophy, or sensory changes in a C8T1, or lower

trunk, distribution. This relationship between the dis-

tribution of plexopathy and cause has not been ob-

served by all investigators, however [27,28]. Horners

syndrome (ptosis, anhydrosis, miosis, and enophthal-

mus) occurs in at least 50% of patients who have

infiltration of the brachial plexus [23,26].

Lymphoma can involve the brachial plexus in two

ways. First, enlarged lymph nodes can compress or

infiltrate the plexus. Enlarged axillary and medias-

tinal lymph nodes may involve the lower plexus,

affecting the C7T1 distributions, whereas cervi-

cal and supraclavicular nodes may affect the upper

plexus, causing pain and neurologic dysfunction in

the C5C6 distribution. In Hodgkins disease, theprevalence of plexus involvement ranges from 5% to

15%. Perineural spread of lymphoma is a common

feature, resulting in epidural and intradural extension

of tumor with or without vertebral involvement.

Second, neurolymphomatosis, which is a rare mani-

festion of lymphoma primarily involving the pe-

ripheral nerves, can affect the brachial plexus. MR

imaging in two patients with B-cell lymphoma re-

vealed diffuse thickening of plexus components, with

hyperintensity on T2-weighted images and postcon-

trast enhancement on T1-weighted images [29,30].

Neurolymphomatosis may occur in isolation or in

association with systemic or primary central nervous

system lymphoma,

The differential diagnosis of infiltrative lesions of

the brachial plexus also includes soft tissue tumors,

such as sarcomas and fibromatosis [24,31] Aggres-

sive fibromatosis is a benign fibroblastic proliferation

that occurs in the deep soft tissues, mimics fibrosar-

coma but does not metastasize. It tends to invade or

surround muscles, tendons, and nerves and vessels,

and to recur locally following excision. The principal

location of the tumor is the musculature of the

shoulder; Chui [31] noted involvement especially of

the serratus anterior and the scalene muscles, with

infiltration of the brachial plexus. Aggressive fibro-

matosis accounted for the largest fraction, one third,

of the benign, primary non-neurogenic tumors in-

volving the plexus in the two published series of

plexal tumors [20,21] reviewed by Saifuddin [19].

OKeefe et al [32] reported that aggressive fibroma-

tosis was iso- to hypointense to muscle on T1-

weighted images and heterogeneously hyperintense

on T2-weighted images, although hypointensity on

both T1- and T2-weighted images has been observed.

Postcontrast images demonstrated diffuse tumor en-

hancement, which was found to be useful in assessing

the extent of tumor invading muscle. Usually, the

tumor appears as an irregular mass with no cystic or

necrotic component. These findings differ from those

typical of larger benign and malignant nerve sheath

tumors, which tend to be more well circumscribed

and may have cystic areas [31].

The most common neurogenic tumors of the

brachial plexus are the benign nerve sheath tumors:

neurofibroma (50%65%) and schwannoma (18%

20%) [20,33]. Malignant peripheral nerve sheath

tumors (MPNSTs) account for 14% of the neurogenic

tumors. Nerve sheath tumors may involve any com-

ponent of the plexus, although the roots are the most

frequent site [25]. The schwannoma (see Fig. 5),

which commonly occurs as a solitary lesion, is a

well-defined, encapsulated tumor that typically has an

eccentric location relative to the nerve axis. Adjacent

uninvolved fascicles are displaced rather than infil-

trated, and their identification on high-resolution MR

imaging can be helpful to the surgeon in planning the

approach to resection of the tumor.

The neurofibroma, which characteristically occurs

at multiple sites, is nonencapsulated and exhibits

infiltration of the nerve fascicles histopathologically.

One third of these lesions occur in patients who have

neurofibromatosis type 1, and two thirds occur spo-

radically. It is difficult to resect a neurofibroma

without permanent damage to the nerve. Plexiform

neurofibromas diffusely enlarge major nerve trunks.

The clinical signs and symptoms of nerve sheath

tumors include pain (which may radiate during

manipulation of the tumor and is more often associ-

ated with schwannoma than neurofibroma), paresthe-

sias, numbness, sensory and motor deficits, atrophy,

and a palpable mass.

The MR signal characteristics of schwannomas

and neurofibromas are similar. The tumors are iso-

intense or hypointense to muscle on T1-weighted

images. On T2-weighted images, the lesions are

variably hyperintense and may have a central area

of inhomogeneous hypointensity, referred to as a

target sign [34,35]. The peripheral hyperintense

signal has been attributed to myxoid tissue, and the

ght br

comp

dly de

eriorly

printe

exopa

sities

iffuse

tensit

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 5985 71Fig. 8. Traumatic pseudomeningocele and neuromas of the ri

(B) images show a pseudomeningocele (arrow) at the site of

trunks to the cords (arrowheads) is hyperintense and marke

several weeks before MR imaging and also sustained an inf

traction on and compression of the plexus. (Fig. 8A and B re

nervous system: evaluation of peripheral neuropathy and pl

mission.) (C) Coronal STIR image shows nodular hyperinten

arrows) of the trunks and divisions. More proximally there is d

right C5 nerve root. For comparison, note the minimal hyperinachial plexus. Coronal T1-weighted spin-echo (A) and STIR

lete avulsion of the right C8 nerve root. The plexus from the

formed. The patient was involved in a motorcycle accident

displaced clavicular fracture. Plexopathy was attributed to

d from Maravilla KR, Bowen BC. Imaging of the peripheral

thy. AJNR Am J Neuroradiol 1998;19:101123; with per-

representing traumatic neuromas near the junction (opposing

hyperintensity and nodularity (arrow with adjacent 5) of the

y and uniform contour of the left C6 nerve root (arrowhead).

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 598572central hypointense area to the presence of fibrocar-

tilaginous tissue [34,36]. The variable tissue compo-

sition reflects the distribution of Antoni A and B cell

types within a tumor. Tumors with cystic necrosis

do not exhibit the target sign. On postcontrast

T1-weighted images, nerve sheath tumors enhance,

usually homogeneously for small tumors and hetero-

geneously for larger tumors (see Fig. 5C, F).

MPNSTs occur less frequently than benign tumors

and are found mainly in patients who have neurofi-

bromatosis or a history of previous radiation therapy

to the brachial plexus region [33,34,37,38]. Tumor

size does not predict malignancy. In the few reported

cases of MPNST involving the brachial plexus,

conventional MR imaging findings were indistin-

guishable from those of schwannoma or neurofibro-

ma [25,28]. On dynamic contrast-enhanced MR

imaging, Van der Woude et al [39] found that the

presence of early peripheral enhancement and a type 1

pattern of progression of enhancement were highly

specific for malignant lesions. This technique, how-

ever, has not been widely used. An uncommon

vascular neoplasm that may mimic nerve sheath

tumor on conventional MR imaging is hemangioperi-

cytoma, and a case involving the C8 root of the

plexus has been reported [2].

Traumatic injury

Injury to a peripheral nerve caused by trauma can

range from disruption of axonal conduction with

preservation of anatomic continuity of the connective

tissue sheaths comprising the nerve (neurapraxic

injury) to severed nerve with complete loss of conti-

nuity of the nerve (neurotmetic injury) [40,41]. By

demonstrating the location and severity of injury and

the morphology of the injured nerve, high-resolution

MR imaging complements the electrophysiologic

studies in determining the exact site and type of

nerve injury, and the potential for surgical treatment

versus spontaneous recovery. In addition, MR imag-

ing can show the relationship of the intact nerve to

post-traumatic lesions, such as spindle, lateral, and

stump neuromas, and focal or diffuse perineural

fibrosis (Fig. 8). When MR imaging is unable to

determine which plexus component or terminal

branch has been injured, its identity can often be

deduced from the pattern of abnormal signal intensity

involving the skeletal muscle or muscles innervated

by the nerve (Fig. 9) [42].

Brachial plexopathy following trauma can result

from compression, stretching, or laceration of plexal

components; perineural fibrosis; or avulsion of nerve

roots from the spinal cord. Typically, the mechanismof injury is traction on the plexus (Fig. 10). A simple

classification that is useful for surgical manage-

ment divides lesions into supraclavicular and infra-

clavicular (includes retroclavicular), the former being

approximately 3 times more common [43]. Supra-

clavicular lesions may involve the roots or trunks of

the brachial plexus or the nerve roots extending from

the spinal cord to the neural foramen. Erb-Duchenne

palsy results from injury to the C5 and C6 roots or

upper trunk of the plexus and accounts for approxi-

mately 90% of obstetric brachial plexus injuries [44].

Much less common is Dejerine-Klumpke palsy,

which results from injury to the C8 and T1 roots or

lower trunk.

It is important to distinguish intraspinal nerve root

avulsion (preganglionic lesion) from brachial plexus

interruption (postganglionic lesion) because the sur-

gical treatment differs in each case. Nerve root

avulsion cannot be repaired directly, and neurotiza-

tion by nerve-crossing using the intercostal nerves or

spinal accessory nerve has been recommended [45].

Brachial plexus interruption can be treated by local

repair, and nerve grafting is the usual method of

plexus reconstruction. Differentiation of nerve root

avulsion from plexus injury is aided by electromyo-

graphy (EMG) studies, because abnormalities of the

paraspinal muscles indicate that an injury is proximal

to the plexal trunks. Somatosensory evoked potentials

routinely have been used to diagnose nerve root

avulsion; however, because these do not enable

physicians to discriminate between incomplete avul-

sion and intact roots, or between intraforminal root

avulsion and rootlet avulsion from the spinal cord, the

inclusion of imaging studies (myelography, CT mye-

lography, and MR myelography) in the diagnostic

evaluation has been recommended [46,47].

The two major causes of cervical nerve root

avulsion are motorcycle accidents and traumatic

delivery at birth. Sunderland [48] proposed two

mechanisms of injury: (1) a peripheral mechanism,

in which lateral traction on the plexus results in

tearing of the dural sleeve around a nerve root, with

or without complete root avulsion, and (2) a central

mechanism, in which there is root avulsion caused by

cord displacement in the spinal canal without associ-

ated dural sleeve laceration. Laceration of the dural

sleeve can result in an extradural collection of cere-

brospinal fluid (CSF), termed a traumatic meningo-

cele or pseudomeningocele (see Fig. 8A, 8B). Thus,

traumatic meningocele may occur with or without

nerve root avulsion, and avulsion may occur without

meningocele. Although there is a strong association

between the finding of root avulsion and the finding

of traumatic meningocele, their distinction is impor-

Fig. 9. Subacute stab wound to the left side of the base of the neck. Injury to the upper plexus was suspected based on the

location of the wound and moderately increased signal in the C5C7 roots or the upper trunk compared with the other plexus

components. Signal abnormalities and postcontrast enhancement in the supraspinatus, infraspinatus, and subscapularis muscles

were consistent with subacute muscle denervation in the distribution of the suprascapular and subscapular nerves. These nerves

are primarily derived from the C5 and C6 roots, confirming these two roots or the upper trunk as the site of most severe

penetrating injury to the plexus. Coronal T1-weighted (A) and fat-saturated T2-weighted (B) images. The supraspinatus (long

arrow), infraspinatus (short arrow), and subscapularis (large arrowhead) muscles are hyperintense relative to the trapezius

(small arrowhead) in B. The same muscles are isointense to trapezius in A. Sagittal T1-weighted (C) and fat-saturated

T2-weighted (D) images confirm findings (arrows, arrowheads same as in B) observed in A and B. Sagittal fat-saturated pre-

contrast (E) and postcontrast (F) T1-weighted images demonstrate minimal abnormal enhancement of the same muscles (arrows,

arrowheads same as in B) that were hyperintense in B and D.

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 5985 73

tant because it is only when the former is present that

neurotization is indicated.

In the last 10 years, studies comparing myelog-

raphy, CT myelography, or MR [46,47,49,50] for

detection of nerve root avulsion and traumatic menin-

gocele have been reported. In the detection of nerve

root avulsion, some studies [46] found that myelog-

raphy/CT myelography was the most accurate ap-

proach ( > 90%), confirming separate reports of the

reliable demonstration of root avulsion with CT

myelography [51] and a 92% accuracy of MR mye-

lography compared with CT myelography [52]. Other

studies, however, found that myelography/CT mye-

lography and MR imaging achieved similar accuracy

[49]. Recently, Doi et al [47] reported that both

approaches had the same sensitivity (92.9%) for

detecting nerve root avulsion, which was documented

by surgical exploration with or without spinal evoked

then be more accurately documented by additional

targeted MR imaging or myelography/CT myelogra-

phy [47].

On MR imaging, traumatic meningocele appears

as an area of CSF-equivalent signal intensity extend-

ing from the spinal canal into the neural foramen

[24,44,50]. Dissection of CSF along the roots of the

brachial plexus may result in a paraspinal fluid

collection [2]. Because the most common sites of

nerve root avulsion are C7 and C8 [53], traumatic me-

ningocele is likely to be found at these levels [50].

Using 2- to 3-mm-thick T2*-weighted axial and

coronal images, Miller et al [44] detected traumatic

meningoceles in four of five infants with brachial

plexopathy following traumatic delivery. The C7 site

was most commonly involved. The meningoceles

corresponded to the segmental level of neurologic

deficit, which was persistent in the four infants who

ial ple

and 4

body

d) and

disea

-resol

ever,

ch of

e, is

D) re

ated

erves

. (F, G

trate r

ions o

in F

avilla

m J

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 598574potential measurements.

In the detection of traumatic meningocele, con-

ventional spin-echo MR imaging is equivalent to CT

myelography, which is more accurate than myelog-

raphy. Volle et al [50] found that MR and CT

myelography demonstrated 16 of 16 cases of menin-

goceles, compared with 14 of 16 cases (88%) for

myelography. The rate of association between trau-

matic meningocele and completely avulsed nerve root

was 83% (15 meningocles per 18 avulsed roots at

corresponding levels), whereas 1 meningocele of the

16 (6%) was observed in the absence of root avul-

sion. The high sensitivity with which meningocele is

detected on multiplanar MR imaging and the strong

association between meningocele and nerve root

avulsion make conventional MR useful as a first test

in screening for the presence of avulsion, which could

Fig. 10. Stretch injury to axillary portion of the right brach

imaging was performed at 2 weeks (A), 4 weeks (BE),

T1-weighted image (lower resolution image obtained with the

of the coracobrachialis and biceps brachii muscles (arrowhea

These findings initially raised concern for possible metastatic

pain, and a higher-resolution study was recommended. Higher

(C) images. The nerves and muscles are isointense in A; how

arrow) are diffusely hyperintense in C. The subscapular bran

cord and superior to the medial head of the median nerv

Corresponding postcontrast fat-saturated T1-weighted image (

and median nerve (long arrow). The adjacent, posteriorly loc

arrowhead), radial (single arrowhead), and median (arrow) n

blood-nerve barrier secondary to the subacute stretch injury

compared with the corresponding images D and E, demons

except the axillary nerve (double arrowhead in G). The locat

nerve (single arrowhead in G), and median nerve (long arrow

nearly resolved at the time of this study. (Reprinted from Mar

evaluation of peripheral neuropathy and plexopathy. AJNR Ahad meningoceles and reversible in the infant who

did not have meningocele.

For overall characterization of traumatic brachial

plexopathy, MR has an advantage over CT and

myelography because MR is better able to show

plexus lesions (postganglionic), in addition to detect-

ing pseudomeningocele. Examples of post-traumatic

lesions of the plexus that have been demonstrated on

spin-echo images include neuromas (tangles of regen-

erating nerve fibers), focal or diffuse fibrosis, and

masses that compress or stretch the plexus, such as

hematoma, clavicular fracture (see Fig. 8), and hu-

meral dislocation [10,17,25,54].

Gupta et al [55] described the MR findings in

10 cases of injury (9 traction injuries and 1 knife

wound) to the roots or trunks of the brachial

plexus, with surgical confirmation. Neuroma (n = 4)

xus secondary to positioning for a surgical procedure. MR

months (FG) following surgery. (A) Axial postcontrast

RF coil as receiver) demonstrates enhancement in the region

focal enhancement posterior to the axillary vessels (arrow).

se in this patient who complained of right shoulder and arm

ution coronal T1-weighted (B) and fat-saturated T2-weighted

the posterior cord (short arrow) and the median nerve (long

the axillary artery, which is located inferior to the posterior

labeled with an A in this and all subsequent images.

veals diffuse enhancement of the posterior cord (short arrow)

image (E) demonstrates enhancement of the axillary (double

. Enhancement of the nerves likely reflects breakdown of the

) Postcontrast fat-saturated T1-weighted images F and G,

esolution of the contrast enhancement for all of the nerves

f the nonenhancing posterior cord (short arrow in F), radial

and arrow in G) are identified. The patients symptoms had

KR, Bowen BC. Imaging of the peripheral nervous system:

Neuroradiol 1998;19:101123; with permission.)

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 5985 75

Fig. 10 (continued).

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 598576

that extends from the first thoracic rib to an elon-

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 5985 77and focal or diffuse fibrosis (n = 4) were isointense

to normal neural tissue. Neuroma was identifiable as

a discrete fusiform mass either replacing (end or

stump neuroma) or projecting eccentrically from

(lateral neuroma) a component of the plexus. Focal

fibrosis appeared as thickening and raggedness of

the borders of plexus components. Diffuse involve-

ment was characterized by marked distortion and

thickening of plexus components, and severe cases

demonstrated a confluent mass without distinguish-

able components.

Other investigators have found post-traumatic

neuromas to be hyperintense on T2-weighted spin-

echo images [17] and on long TR STIR images (see

Fig. 8C) [1]. The variable signal intensity of neuro-

mas on images with T2 weighting may depend on the

time interval between injury and MR imaging. Gupta

et al [55] suggested that high signal intensity occurs

at short time intervals and is caused by neural edema,

which later resolves. Neural edema, however, as the

sole cause of T2 hyperintensity remains speculative.

Diffuse hyperintensity of plexus components has

been observed in patients who have nontraumatic,

and traumatic, plexopathy [1,25]. Diffuse postcon-

trast enhancement of plexus components and ter-

minal branches in the subacute phase following

stretch injury also has been observed [18] and seems

to be a reversible process. The cause of these find-

ings is unknown but presumably results from tran-

sient alteration in the integrity of the blood-nerve

barrier or possibly from intraneurial venous conges-

tion (Fig. 10).

Although clavicular fractures, with or without

associated hematoma, may injure the plexus because

of acute compression or stretching, subsequent callus

and scar formation may also contribute to plexopathy.

Bilbey et al [17] reported the results of a patient with

persistent plexopathy who was imaged 4 months after

clavicular fracture. A soft tissue mass extended from

the callus at the fracture site to involve the plexus.

The mass was isointense to muscle on T1- and T2-

weighted images. At surgery, the mass proved to be

scar tissue, which encased the plexus trunks. In other

cases of clavicular fracture with associated soft tissue

mass, presumed to be callus and scar tissue, involv-

ing the plexus, the mass was hyperintense to muscle

on T2-weighted images, as were the involved

nerves [54]. Although used infrequently, T1-weighted

three-dimensional gradient-echo acquisitions with

multiplanar reformatting, surface rendering, and seg-

mentation of data show promise as a means of

demonstrating the relationship of clavicular fracture

and hypointense scar to the components of the

plexus [10].gated transverse process or a rudimentary cervical

rib (Fig. 11). The C8 and T1 roots, either immedi-

ately before or after they form the lower trunk, are

stretched and angulated around this band. As shown

in Fig. 11, coronal images may demonstrate only sub-

tle displacement of the plexus and should be corre-

lated with the findings on plain radiographs. Sagittal

MR images provide additional evidence of distor-

tion of the lower plexus and interscalene triangle.

Two other subgroups of TOS involve the brachial

plexus: the disputed TOS (2b and 3b in the previous

list) and the traumatic TOS resulting from clavicular

injury, usually a midshaft fracture (3a in the previous

list). As suggested by the title, disputed TOS is

controversial, and the reader is referred to a review

by Wilbourn [57] for a full discussion of this sub-Entrapment syndromes

Guided to the location of entrapment/compression

by the clinical and neurologic examination, the MR

imaging study is used to detect objective findings of

nerve compression [56]. The brachial plexus or the

subclavian/axillary artery or vein encounter three

possible sites of compression along their course: the

interscalene triangle, the costoclavicular space be-

tween the first thoracic rib and the clavicle, and the

retropectoralis minor space posterior to the pectoralis

minor muscle near its insertion on the coracoid pro-

cess. Clinical symptoms and signs caused by entrap-

ment/compression are classified under thoracic

outlet syndromes (TOS) as follows [57]:

1. Vascular: blood vessels only are affected

a. Arterial

b. Venous

2. Neurologic: brachial plexus only is affected

a. True/classic

b. Disputed/nonspecific/symptomatic

3. Combined/neurovascular

a. Traumatic (caused by clavicle disorder)

b. Disputed

The true and/or classic neurologic TOS, which is

also referred to as cervical rib (and band) syndrome,

is almost always unilateral and predominantly affects

women. The syndrome manifests as a chronic lower

trunk (C8T1) plexopathy in which there is wasting

of the hand and sensory disturbances, typically ach-

ing or paresthesias along the medial arm and forearm,

sometimes extending into the medial hand and fin-

gers. Entrapment and/or compression of the lower

trunk usually results from a congenital fibrous band

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 598578group. The traumatic TOS is an uncommon compli-

cation of clavicular fracture and occurs when the

blood vessels and the brachial plexus located between

the midportion of the clavicle and the first thoracic rib

are injured secondarily. Although these lesions may

present immediately after fracture occurrence, they

frequently are not manifested for days, weeks, or

even years. This subgroup differs from most cases in

Fig. 11. Neurologic thoracic outlet syndrome in a patient with a l

STIR MR image (B) of the cervicothoracic region. The left cervical

displacement of the brachial plexus. Sagittal T1-weighted spin-ec

interscalene triangle in the region of the cervical rib (arrow). The r

appear as hyperintense foci located superior to the arrowhead a

hypointense band is contiguous with the cervical rib in D. (E)

portion of the partially resected cervical rib (arrow).which there is clavicular fracture and plexus injury

because, in most patients, the trauma is so severe as

to cause simultaneous fracture and plexus injury, with

the roots or trunks of the plexus sustaining traction

damage (see Fig. 8).

In the traumatic TOS subgroup, the proximal

aspects of the cords of the plexus, the terminal

portion of the subclavian artery, and the initial portion

eft cervical rib. Anteroposterior radiograph (A) and coronal

rib (arrow) is shown in B to be associated with a mild inward

ho (C) and STIR (D) images demonstrate narrowing of the

oot/trunks (arrowhead), which are isointense to muscle in C,

nd superoposterior to the subclavian artery A in D. A

Anteroposterior radiograph showing the residual posterior

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 5985 79of the subclavian vein may be damaged alone or in

various combinations. Although medial cord lesions

are the most common, medial, lateral, posterior, and

pan cord lesions have been described [57]. The

plexus injuries can result from compression by dis-

placed fracture fragments, manipulation of the frac-

ture, hematoma or pseudoaneurysm of the subclavian

artery, hypertrophic callus, and nonunion. Hyper-

trophic callus and nonunion are the two causes of

injury for most cases that present long after the frac-

ture has occurred.

There is some disagreement on the value of MR

imaging in diagnosing neurologic or combined/neu-

rovascular TOS [58,59]. Panegyres et al [58] found

relatively good correlation between distortion of the

plexus on MR imaging and clinical symptoms (sen-

sitivity, 79%; specificity, 87.5%). The authors also

noted the frequent detection on MR of a fibrous

band presumably causing the distortion; however,

the band was detected in asymptomatic controls in

addition to symptomatic subjects.

Recently, some investigators [60,61] have pro-

posed that patients who have TOS be imaged with

the arm on the side of interest in abduction and

alongside the body. The measurements by Smedby

et al [60] for assessing plexus compression were

targeted to the costoclavicular space, thus excluding

evaluation of possible entrapment/compression at

the interscalene triangle or the retropectoralis minor

space. Demondion et al [61] obtained additional

measurements to assess all three sites of possible

compression and found that only three thoracic outlet

measurements differed significantly in the patient and

control groups: patients with TOS had a smaller

costoclavicular distance after the postural maneuver

(P < 0.001), a thicker subclavius muscle in both arm

positions (P < 0.001), and a wider retropectoralis

minor space after the postural maneuver (P < 0.001)

than did volunteers. The postural maneuver consisted

of hyperaduction to 130 and external rotation ofthe arm.

A hypertrophied anterior scalene muscle, some-

times accompanied by a hypertrophied middle sca-

lene, compresses the plexus against the first rib,

producing symptoms (scalenus anticus syndrome)

similar to those of the cervical rib syndrome. When

the subclavian artery is also compressed, diminution

in the radial pulse occurs as the head is turned to the

side of the plexopathy. Using multiplanar reformat-

ting of three-dimensional gradient-echo images with

T1 weighting, Collins et al [10] have shown com-

pression of the plexus and subclavian artery at the

interscalene triangle by an enlarged anterior scalene

muscle. With the same MR technique, these authorsalso demonstrated an example of plexopathy caused

by a scalene muscle anomaly, which splayed the roots

of the plexus.

Post-treatment evaluation

Patients who have a history of cancer and clinical

evidence of plexopathy following radiation therapy

may have recurrent tumor or radiation-induced plexo-

pathy. Imaging features that favor recurrent tumor are

nonuniform, asymmetric diffuse or focal enlargement

(Fig. 12), and especially the presence of an eccentric

mass with postcontrast enhancement [22,28]. Imag-

ing features that favor postradiation injury of the

brachial plexus are diffuse, uniform, symmetric

swelling and T2 hyperintensity of the plexus nerves

within the radiation field (Fig. 13). Diffuse, uniform

postcontrast enhancement for months to years after

treatment also may result from radiation injury

[22,62]. Radiation fibrosis often has low signal in-

tensity on T1- and T2-weighted images [63], and this

finding may represent the more common appearance

for chronic radiation injury, although a correlation

between the time interval following radiation therapy

and T2 signal intensity has not been reported.

When diffuse enlargement, T2 hyperintensity, and

postcontrast enhancement of the plexus (and sur-

rounding tissues) are present on MR imaging of

patients who have a history of breast cancer and

radiation therapy, differentiation between radiation

injury and local/regional recurrent cancer with axil-

lary/supraclavicular metastases may not be possible.

Preliminary results suggest that 2-(fluorine 18)fluoro-

2-deoxy-D-glucose (FDG) positron emission tomog-

raphy (PET) helps to confirm metastases (see

Fig. 12E) in patients with indeterminate MR imag-

ing findings and is useful for depicting metastases

outside the axilla [64].

Injury to the brachial plexus following radiation

therapy is associated with three distinct clinical

syndromes: (1) classic delayed, progressive radiation

injury or radiation fibrosis; (2) reversible or transient

plexopathy; and (3) acute ischemic plexopathy. The

frequency of classic radiation-induced brachial plexo-

pathy in patients who have cancer depends on the

dosage of radiation but has been estimated generally

at 5% to 9% based on studies published since 1990

[38,65]. Kori et al [23] found that radiation fibrosis is

unlikely to occur with total doses of less than 60 Gy.

If more than 60 Gy is given and if neurologic

symptoms appear within a year, the diagnosis is

probably radiation damage. If symptoms appear after

1 year, they may be from radiation damage or tumor

recurrence. The incidence of radiation-induced plexo-

Fig. 12. Tumor infiltration of medial cord of left brachial plexus from recurrent breast carcinoma. The patient had new onset of

left arm and hand pain several years after left mastectomy and radiation therapy for breast carcinoma. Differential diagnosis

based on clinical neurologic examination was neoplastic versus radiation-induced plexopathy. Coronal T1-weighted spin-echo

(A) and STIR (B) images of the plexus show subtle, diffuse enlargement the medial cord (arrows) of the brachial plexus. Oblique

sagittal T1-weighted spin-echo precontrast (C) and fat-saturated postcontrast (D) images (at the level of the acromioclavicular

joint) demonstrate the abnormal enhancement of the enlarged medial cord (arrow in C and D). The flow void located superior

to the medial cord is the axillary artery. (E) FDG-PET image reveals uptake in the distribution of the medial cord (arrows)

consistent with diffuse tumor infiltration of the cord, which was confirmed on subsequent clinical follow-up. (Reprinted from

Maravilla KR, Bowen BC. Imaging of the peripheral nervous system: evaluation of peripheral neuropathy and plexopathy.

AJNR Am J Neuroradiol 1998;19:101123; with permission.)

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 598580

nt pre

n and

ion-in

divisi

plexu

d wit

aluati

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 5985 81pathy also depends on fractionated dose, treatment

with cytotoxic drugs, patient age, and the premorbid

state of the irradiated nerves [6567]. The average

interval between the last dose of radiation and the

development of plexopathy has been reported as

approximately 5 years (range, several months to

Fig. 13. Radiation injury of the left brachial plexus. The patie

2 years after radiation therapy to the left supraclavicular regio

clinical neurologic examination was neoplastic versus radiat

STIR (B) images show enlargement of the roots, trunks, and

involvement of all visualized components of the left brachial

brachial plexus structures are markedly hyperintense compare

Bowen BC. Imaging of the peripheral nervous system: ev

Neuroradiol 1998;19:101123; with permission.)>20 years) [23,68,69].

The clinical findings associated with radiation

fibrosis often differ from those of plexopathy caused

by tumor infiltration. In 50% to 80% of patients who

have classic radiation injury, plexopathy is relatively

painless [23,26,28]. Kori et al [23] and others [70]

found that neurologic signs were predominantly dis-

tributed in the upper trunk, involving C5, C6, or C7

roots, with weakness usually involving shoulder

abduction and arm flexors. Also, progressive lym-

phedema and swelling of the ipsilateral arm occurred

approximately 5 times more frequently in patients

with radiation-induced plexopathy. Some investiga-

tors, however, have not found a predominance of

upper trunk signs [26,65]. Needle EMG findings are

valuable, because characteristic electrical activity

so-called myokymic dischargesin limb muscles

is associated with radiation plexopathy and not tumor

infiltration [26,71].

Two other clinical syndromes that occur following

radiation therapy are as follows: (1) delayed, revers-

ible brachial plexopathy and (2) acute, ischemic

brachial plexopathy. Reversible or transient brachial

plexopathy has been observed in patients who havebreast cancer following radiation therapy to the axilla

[72,73] The clinical findings differ from those asso-

ciated with classic radiation injury or tumor infiltra-

tion and consist of paresthesias in all cases, with

weakness and pain less commonly seen. The syn-

drome occurred in 8 of 565 patients (1.4%) who

sented with signs and symptoms of a progressive plexopathy

axilla for breast carcinoma. Differential diagnosis based on

duced plexopathy. Coronal T1-weighted spin-echo (A) and

ons of the left brachial plexus. Note the uniform and diffuse

s within the radiation field. On the STIR image (B), the left

h the contralateral right side. (Reprinted from Maravilla KR,

on of peripheral neuropathy and plexopathy. AJNR Am Jreceived an average axillary dose of 50 Gy in 5 weeks

[73]. The syndrome did not conform to a specific

anatomic pattern, but more often affected the lower

plexus. The median time interval between the end

of radiation therapy and onset of symptoms was

4.5 months (range, 214 mo), with subsequent spon-

taneous resolution in all patients. Seven of the

patients received adjuvant chemotherapy, and in six

of these patients, symptoms began following drug

treatment. Although a possible relationship to adju-

vant chemotherapy exists, the cause of this plexo-

pathy remains unclear, and no characteristic MR

imaging findings have been reported.

A rare cause of radiation-induced brachial plexo-

pathy is acute ischemia from occlusion of the sub-

clavian artery following radiotherapy. In a case

reported by Gerard et al [74], painless panplexopathy

developed acutely and was nonprogressive 21 years

following radical mastectomy and radiotherapy for

carcinoma of the left breast. This presentation and

clinical course differ from the description of classic

radiation fibrosis described previously. The EMG did

not show any myokymic discharges. MR imaging of

the plexus and cervical spinal was reported as normal.

interventions for peripheral neuropathy. Experimental

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 598582Blood pressure in the left arm was markedly reduced,

however, and conventional angiography showed a

segmental occlusion of the left subclavian artery with

collateral revascularization distally. Although sub-

clavian artery occlusion has been recognized as a

complication of radiotherapy for breast carcinoma

[75], only two cases with accompanying plexopathy

have been reported [74].

Miscelleneous

When the clinical examination does not reveal a

cause of the patients neuropathy, MR imaging may

identify a focal or diffuse peripheral nerve or plexus

structural abnormality, such as occurs in chronic

inflammatory demyelinating polyneuropathy (CIDP)

[76,77], multifocal motor neuropathy (MMN) [78],

hereditary hypertrophic motor and sensory neuropa-

thies (HMSNs) [79,80], and inflammatory pseudo-

tumor [81]. Hypertrophy and diffuse hyperintensity

of the plexus components, especially the roots, have

been detected on T2-weighted images in cases of

CIDP, MMN, and HMSN. In cases of unexplained

plexopathy, demonstration of a normal-appearing

nerve is also useful because it narrows the differential

diagnosis and may eliminate the need for biopsy or

invasive therapy. Not uncommonly, patients who

undergo imaging for unexplained plexopathy will

have idiopathic, probably postviral, inflammatory

conditions, such as brachial plexus neuritis.

Patients who have idiopathic brachial plexus neu-

ritis, or plexitis present with sudden onset of severe,

constant pain in the lateral neck, shoulder, scapula, or

upper arm [82]. Involvement is bilateral in 10% to

30% of patients [67,83]. The pain is exacerbated by

arm or shoulder movement. Within a few weeks, the

pain is followed by a profound weakness and atrophy

of the painful muscles, usually those of the shoulder

girdle. The most frequently affected muscle is the

serratus anterior. The most commonly affected nerves

include the axillary, suprascapular, and long thoracic

nerves, whereas the ulnar nerve is rarely affected.

Onset of symptoms may be preceded by a viral illness

or immunization. A rare heredofamilial form exists,

and the syndrome also has been observed in patients

who have newly diagnosed lymphoma [67]. Partial or

complete resolution of symptoms usually occurs,

although recovery may not begin until 6 months after

the onset of symptoms and may require up to 3 years.

In Bilbey et als [17] study, 4 of 64 consecutive pa-

tients who underwent MR imaging for suspected

brachial plexus abnormalities had a clinical diagnosis

of idiopathic or viral plexitis. In all 4 patients, spin-

echo MR findings were normal. Posniak et al [25] re-studies using mice already have shown that the

introduction of replication-defective viral vectors

directly into dorsal root ganglia using microneurosur-

gical technique leads to long-term expression of

reporter genes along the length of the sensory neuron,

from its distal portion to the ipsilateral nucleusported a case of brachial neuritis in which the nerves

of the plexus were diffusely enlarged and hyperin-

tense on T2-weighted images. These findings were

attributed to inflammation and edema but not corrobo-

rated by subsequent imaging or other methods.

Summary

High-resolution MR imaging of peripheral nerves

and nerve plexuses is an area of rapidly growing

clinical interest and importance. The number of

studies performed is rapidly increasing in response

to the need for more detailed in vivo information

about neuropathic changes and regional neural anat-

omy before treatment planning by peripheral nerve

specialists [84]. Specific information gained from

peripheral nerve imaging studies is being used to

determine the need for biopsy or surgical treatment.

In patients who have small tumors, peripheral nerve

imaging has proved useful in planning the surgical

approach and in predicting the prognosis for preser-

vation of nerve function postoperatively. In cases of

traumatic nerve injury, MR imaging results are being

considered as part of the clinical assessment regard-

ing (1) the likelihood of spontaneous recovery versus

the need for surgical repair and (2) the progression of

nerve recovery postoperatively.

Additional applications and new developments for

the MR techniques used in peripheral nerve imaging

are on the horizon. Among these are improvements in

the homogeneity of fat saturation methods and alter-

native approaches to removing fat signal, such as

multipoint Dixon fat-water separation with SSFP or

FSE imaging methods. Rapid volumetric data acqui-

sition with high CNR may allow the presentation of

the largest peripheral nerves and their branches in a

three-dimensional display, analogous to the three-

dimensional maximum-intensity projection images

of the vasculature available with MR angiography.

This presentation will provide the referring physician

with a more complete picture of the affected periph-

eral nerves, facilitating localization and diagnosis of

peripheral neuropathy.

Improvements in imaging of peripheral nerves

coupled with the availability of MR-guided interven-

tional systems should eventually lead to therapeutic

three-dimensional volume acquisition. Int J Neurora-

diol 1996;2:26473.

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 5985 83[12] Cox IH, Dillon WP. Low-cost device for avoiding bulk

susceptibility artifacts in chemical-selective fat satura-

tion MR of the head and neck. AJNR Am J Neuro-

radiol 1995;16:13679.

[13] Dixon WT. Simple proton spectroscopic imaging. Ra-

diology 1984;153:18994.

[14] Glover GH. Multipoint Dixon technique for water and

fat proton and susceptibility imaging. J Magn Reson

Imaging 1991;1:52130.

[15] Hargreaves BA, Vasanawala SS, Nayak KS, et al.gracilis and cuneatus [85]. Dorsal root ganglia injec-

tion may be superior to other therapeutic approaches,

such as intraneural injection into more distal periph-

eral nerve segments [86]. Targeted expression of

foreign genes in the peripheral nervous system has

several potentially valuable applications, including

gene therapy of neuromuscular diseases, neuroana-

tomic studies, and the elucidation of mechanisms of

axonal flow.

References

[1] Maravilla KR, Bowen BC. Imaging of the peripheral

nervous system: evaluation of peripheral neuropathy

and plexopathy. AJNR Am J Neuroradiol 1998;19:

101123.

[2] Bowen B. Brachial plexus. In: Bradley W, Stark D,

editors. Magnetic resonance imaging. Philadelphia:

Mosby-Year Book; 1999. p. 182132.

[3] Britz G, West G, Dailey A, et al. Magnetic resonance

imaging in evaluating and treating peripheral nerve

problems. Perspect Neurol Surg 1995;6:5366.

[4] Filler A, Kliot M, Howe F, et al. Application of mag-

netic resonance neurography in the evaluation of pa-

tients with peripheral nerve pathology. J Neurosurg

1996;85:299309.

[5] Aagaard B, Maravilla K, Kliot M. MR neurography

MR imaging of peripheral nerves. Magn Reson Imag-

ing Clin N Am 1998;6:17994.

[6] Sunderland S. Nerves and nerve injuries. Balti-

more, MD: Williams and Wilkins; 1968. p. 2557,

7331112.

[7] Rosse C, Gaddum-Rosse P. Hollinsheads textbook

of anatomy. 5th edition. Philadelphia: Lippincott-Ra-

ven; 1997.

[8] Roemer P, Edelstein W, Hayes C, et al. The NMR

phased array. Magn Reson Med 1990;16:192225.

[9] Hayes C, Tsuruda J, Mathis C, et al. Brachial plexus:

MR imaging with a dedicated phased array of surface

coils. Radiology 1997;203:2869.

[10] Collins JD, Shaver ML, Disher AC, et al. Compromis-

ing abnormalities of the brachial plexus as displayed by

magnetic resonance imaging. Clin Anat 1995;8:116.

[11] van Es HW, Witkamp TD, Lino MPR, et al. MR im-

aging of the brachial plexus using a T1-weightedFat-suppressed steady-state free precession imaging

using phase detection. Magn Reson Med 2003;50:

2103.

[16] Chappell KE, Robson MD, Stonebridge-Foster A, et al.

Magic angle effects in magnetic resonance neurogra-

phy. AJNR Am J Neuroradiol 2004;27, in press.

[17] Bilbey JH, Lamond RG, Mattrey RF. MR imaging of

disorders of the brachial plexus. J Magn Reson Imag-

ing 1994;4:138.

[18] Bowen BC, Maravilla KR. Imaging the peripheral ner-

vous system. In: Latchaw RE, Kucharczyk J, Moseley

ME, editors. Advanced imaging and image-guided ther-

apy of the nervous system. Philadelphia: Mosby; 2004.

[19] Saifuddin A. Imaging tumours of the brachial plexus.

Skeletal Radiol 2003;32:37587.

[20] Lusk MD, Kline DG, Garcia CA. Tumors of the brach-

ial plexus. Neurosurgery 1987;21:43953.

[21] Ganju A, Roosen N, Kline DG, et al. Outcomes in a

consecutive series of 111 surgically treated plexal tu-

mors: a review of the experience at the Louisiana State

University Health Sciences Center. J Neurosurg 2001;

95:5160.

[22] Wittenberg KH, Adkins MC. MR imaging of nontrau-

matic brachial plexopathies: frequency and spectrum of

findings. Radiographics 2000;20:102332.

[23] Kori SH, Foley KM, Posner JB. Brachial plexus lesions

in patients with cancer: 100 cases. Neurology 1981;31:

4550.

[24] de Verdier HJ, Colletti PM, Terk MR. MRI of the

brachial plexus: a review of 51 cases. Comput Med

Imaging Graph 1993;17:4550.

[25] Posniak HV, OlsonMC, Dudiak CM, et al. MR imaging

of the brachial plexus. AJR Am J Roentgenol 1993;

161:3739.

[26] Lederman RJ,Wilbourn AJ. Brachial plexopathy: recur-

rent cancer or radiation? Neurology 1984;34:13315.

[27] Harper Jr CM, Thomas JE, Cascino TL, et al. Distinc-

tion between neoplastic and radiation-induced brachial

plexopathy with emphasis on the role of EMG. Neu-

rology 1989;39:5026.

[28] Thyagarajan D, Cascino T, Harms G. Magnetic reso-

nance imaging in brachial plexopathy of cancer. Neu-

rology 1995;45:4217.

[29] Swarnkar A, Fukui MB, Fink DJ, et al. MR imaging of

brachial plexopathy in neurolymphomatosis. AJR Am

J Roentgenol 1997;169:118990.

[30] van Es HW. MRI of the brachial plexus. Eur Radiol

2001;11:32536.

[31] Chui M. Fibromatosis of the brachial plexus and shoul-

der girdle. J Can Assoc Radiol 1989;40:2831.

[32] OKeefe F, Kim EE, Wallace S. Magnetic resonance

imaging in aggressive fibromatosis. Clin Radiol 1990;

42:1703.

[33] Dart Jr LH, MacCarty CS, Love JG, et al. Neoplasms

of the brachial plexus. Minn Med 1970;53:95964.

[34] Varma DG, Moulopoulos A, Sara AS, et al. MR imag-

ing of extracranial nerve sheath tumors. J Comput As-

sist Tomogr 1992;16:44853.

[35] Bhargava R, Parham DM, Lasater OE, et al. MR im-

B.C. Bowen et al / Neuroimag Clin N Am 14 (2004) 598584aging differentiation of benign and malignant periph-

eral nerve sheath tumors: use of the target sign. Pediatr

Radiol 1997;27:1249.

[36] Suh JS, Abenoza P, Galloway HR, et al. Periphe-

ral (extracranial) nerve tumors: correlation of MR im-

aging and histologic findings. Radiology 1992;183:

3416.

[37] Foley KM, Woodruff JM, Ellis FT, et al. Radiation-in-

duced malignant and atypical peripheral nerve sheath

tumors. Ann Neurol 1980;7:3118.

[38] Pierce SM, Recht A, Lingos TI, et al. Long-term radia-

tion complications following conservative surgery

(CS) and radiation therapy (RT) in patients with early

stage breast cancer. Int J Radiat Oncol Biol Phys 1992;

23:91523.

[3