|Home | About | Journals | Submit | Contact Us | Français|
The differential diagnosis between preganglionic and postganglionic lesions of the brachial plexus at the level of individual roots is critical. Preoperative neuroradiologic and electrophysiological studies are considered useful in detecting of root avulsions of the brachial plexus. In this study, the predictive value of plain cervical myelography, following computed tomography myelography (CTM) and preoperative electrodiagnostic evaluation, in detecting nerve root avulsions in cases of obstetrical brachial plexus paralysis was determined. The charts of 321 patients who had sustained brachial plexus paralysis and were operated on in our center from 1982 to 2003 were reviewed and analyzed. This study includes 70 cases of obstetrical brachial plexus palsy. Preoperative cervical myelography and CTM, as well as electrophysiological studies, were compared with final intraoperative diagnosis. A total of 420 spinal nerves were examined in 70 patients. Fifty-two patients (74.3%) had avulsion injury of brachial plexus. Intraoperatively, 135 roots (32.1%) were found to be avulsed. Accuracy, sensitivity, and specificity of preoperative plain myelography were 85.3%, 71.0%, and 92.3%; of CTM they were 89.4%, 83.2%, and 92.1%, respectively; and of electrodiagnostic studies they were 76.2%, 39.5%, and 93.2%, respectively. CTM and plain myelography were significantly more accurate and sensitive than electrophysiological studies (p<0.05). There was no statistically significant difference in accuracy and sensitivity between plain myelography and CTM.
Obstetrical trauma is one of the most common causes of traction injuries of the brachial plexus, with incidence varying between one and two per 1000 live births. This incidence remains stable despite improvement in obstetrical care because of increasing mean birth weight.
The majority of the patients with obstetrical brachial plexus palsy recover spontaneously after conservative treatment, without any or with minor functional deficit. The rest of the children require microsurgical treatment, which recently, with the introduction of interposition nerve grafting and neurotization techniques, has significantly improved the prognosis of these patients. Outcomes of surgical treatment depend on the severity and localization of the brachial plexus lesion and also on the timing of surgery. The results of early microsurgical brachial plexus interventions are superior to late brachial plexus reconstructions or following secondary procedures dealing with established joint contractures and bone deformities.
Root avulsion is the most severe type of obstetrical brachial plexus lesion, usually found following use of extreme force during a complicated delivery. Presence of an avulsion lesion implies devastating paralysis without possibility for spontaneous recovery, as well as associated spread of the traction force to neighboring spinal nerves or trunks. Therefore, clinical signs of root avulsion such as Horner sign and phrenic nerve palsy are considered strong indicators for brachial plexus exploration and reconstruction. However, the absence of these signs does not rule out root avulsion. Thus, poor spontaneous recovery necessitates accurate diagnostic between preganglionic and postganglionic lesions of the brachial plexus.
Postganglionic lesions occur distally to the dorsal root ganglion at the level of the spinal nerve, trunks, or cords. Continuity usually can be restored with nerve repair or interposition nerve grafting. Preganglionic lesions occur proximal to the dorsal root ganglion, with irreversible loss of the original source of nerve fibers. Preganglionic lesions were once considered inoperable, but the use of neurotization procedures from other plexus donors or the use of extraplexus donor nerves has allowed the return of rewarding degree of function.1,2,3,4
Radiological studies such as cervical myelography, computed tomography myelography (CTM) and preoperative electrophysiological testing are commonly used in an effort to correctly pinpoint the brachial plexus lesion. However, there are some controversies regarding the place of all these methods in diagnostic of obstetrical brachial plexus lesions. Cervical myelography can indicate a sign that has a strong association with root avulsions; however, the signs can exist without root avulsion. Furthermore, large, contrast-enhancing pseudomeningoceles expanding along the cervical spine can overlap adjacent normal roots. The CTM allows more accurate assessment of the root and rootlets, but delineation of a level of avulsion on axial slides can be difficult because of discrepancy between the location of root exit zones from spinal cord segments and that of respective intervertebral foramina. The electrophysiology in obstetric brachial plexus lesions differs essentially from one in older patients. The electromyelography (EMG) in obstetric palsy is far too optimistic compared with the clinical picture.5 Others consider EMG unreliable and excluded it from their protocol for examination of the children with obstetrical brachial plexus paralysis, but some still use it with the condition that studies must be done by experienced electromyographer, and results must be interpreted expertly.6
This study was undertaken to determine the diagnostic value of plain myelography, CTM, and preoperative electrophysiology in patients with obstetrical brachial plexus paralysis. Data obtained from these studies will be correlated with intraoperative diagnosis on the basis of operative findings following exploration of these lesions, intraoperative electrostimulation, and histological studies of nerve biopsies.
The understanding of normal anatomy and physiology of the cervical spinal nerves and their relations with surrounding tissues is essential for correct interpretation of radiological and electrophysiological data. The brachial plexus is formed by the anterior rami of the lower four cervical nerves and the first thoracic nerve, while C4 also can contribute its axons to the upper trunk of the brachial plexus. The segmental ventral and dorsal nerve roots are formed from several motor and sensory rootlets (Fig. 1). Ventral roots are formed by a group of rootlets that exit the spinal cord 1–3 mm from the midline. Their diameter ranges from 1.0 to 2 mm, and the distance from the spinal cord to the intervertebral foramen varies from 5 to 10 mm. The dorsal roots are larger and have shorter than the ventral roots. Upon their exit from the dorsal horn, the dorsal roots form the spinal ganglion in the intervertebral foramen, beyond which both ventral and dorsal roots unite to form the spinal nerve (Fig. 2). The roots join the spinal cord almost one intervertebral disc above their intervertebral foramen. In the upper cervical region the course of the rootlets is directed horizontally as they converge laterally, but below the fifth cervical root the rootlets are directed obliquely caudally.
The fine connective tissue of the pia mater extends along each emerging nerve fiber to invest it with an endoneurial sheath (Figs. 2, ,3).3). The connective tissue also forms a loose endoneurial framework for the rootlets and, subsequently, the nerve roots. The nerve fibers at the level of the roots are arranged in parallel bundles, which are loosely held together. The nerve roots lack the epi- and perineurial sheaths of peripheral nerve trunks, which play a major role in maintaining the integrity of the nerve fibers when a nerve trunk is stretched.6,7 Furthermore, the collagen fibers covering the nerve fibers of the nerve roots are fewer and finer that those surrounding the nerve fibers of a peripheral nerve trunk.
The intervertebral foramina of the cervical spine are short tunnels, a few millimeters in length (Fig. 3). Each foramen receives a ventral and a dorsal root that fuse immediately beyond the dorsal root ganglion to form a spinal nerve. The nerve root–ganglion–spinal nerve junction is located at the outer end of the foramen.
The manner in which the connective tissue covering of these neural structures is formed, and their relationship to the wall of the foramen, are of particular interest and significance to the problem of nerve root avulsion. Opposite the intervertebral foramen, each pair of ventral and dorsal nerve roots invaginate the arachnoid and the dura to form a funnel-shaped depression in the wall of the dural sac.
At the bottom of this funnel, each root perforates the meninges independently, carrying with it, as it does so, a tubular dural-arachnoidal sleeve that is separated from the nerve fibers of the root by an extension of the subarachnoid space. Some nerve roots first descend intradurally to a level that may be as much as 8 mm below the center of the foramen, which they are to enter. At this level, they perforate the arachnoid and dura in the usual way. They then ascend acutely, enclosed in their sleeves, to enter the foramen by passing over its lower margin. In this way, the nerve roots are angulated at the site where they pass through the dura8,9,10 (Fig. 4). Ascending or angulated nerve roots are common in the lower cervical and upper thoracic regions and have the incidence, which is related to age.11 This angulation could be the site of injury when the root is exposed to sudden traction.12
At the inner pole of the ganglion, the dura becomes adherent to the ganglion, continues over it as a connective tissue sheath, and then goes beyond it, becoming the tough fibrous perineurium of the spinal nerve. The extension of the subarachnoid space terminates in a cul de sac, where the dura and arachnoid become adherent to the ganglion. In the case of the ventral root, the subarachnoid extension is obliterated further medially as the meninges close around the nerve to form a definitive sheath for this structure.
There are two features of particular note about these meningeal–nerve root relations, which are constant and relate to the problem of root avulsion: medially, in the intervertebral foramen the nerve roots lie freely within their dural-arachnoid sleeves, while laterally, the dural tissue forming the sheath of the ganglion, and the ventral nerve root applied against its surface, are firmly adherent to these structures.6
The neural structures and their surrounding connective tissue occupy 35 to 50% of the cross-sectional area of the intervertebral foramen, so there is a free space between the nerve and the wall of the foramen. The neural structures and their sheath are not adherent to the wall of foramen except at their entrance, where there is some exchange of fibers at the capsule of the intervertebral joint. This permits some sliding of the nerve complex inward and outward through the foramen. So when sudden traction is applied to a peripheral nerve, some movement of the complex spinal nerve-nerve root is possible, which can affect the nerve roots and the spinal cord. Normally this movement does not put undue tension on the nerve roots. There are two structural features that protect the system against traction deformation. The manner of which dura attaches to the ganglion and the spinal nerve is such that lateral traction on the nerve is transmitted centrally along the dural sleeves of the nerve roots to the dural funnel and the dural sac. When the cone-shaped dural funnel is pulled laterally into the foramen, it becomes plugged and resists further dislocation of the system laterally. Being attached to the dura, the denticulate ligament is also drawn outward. This results in some movement of the spinal cord laterally, which reduces the tension that has developed in the nerve roots following displacement of the entire system outward. This movement of the cord would, however, put tension on the corresponding contralateral nerve roots. Thus, the strength and integrity of the system are not a result of the strong attachment of the dura to the intervertebral foramen but are caused by the continuity of the spinal nerve sheath with the dural sac.7
At the level of the transverse processes of C4–C6, corresponding spinal nerves are strongly bound to periosteum and prevertebral fascia by the musculotendinous attachments and by fibrous slips, which descend from the transverse process above to blend with the sheath of the spinal nerve below. Several authors reported on the difficulty of avulsing the roots of the brachial plexus by manual traction on the exposed plexus in the cadavers.9 In the case of the C7 root, there is no groove in the corresponding transverse process, and the attachments of this nerve to the bone are only moderately strong. The C8 and T1 nerves lack any significant attachments to the neighboring structures.8
Sunderland described two mechanisms of root avulsion.15 The first, peripheral mechanism is more well known and is caused by traction of the brachial plexus, which generates the tensile stress and is transmitted centrally with subsequent stretching of the nerve roots. The second, central mechanism has its origin in the external force that deforms structures and creates stretch of the nerve roots between the spinal cord and intervertebral foramen (Fig. 5).
In the case of lateral traction, when the forces reach abnormal level, spinal nerve displacement occurs, resulting in nerve lesion. The severity and spread of the lesion are influenced by many factors, such as direction, duration, and strength of the deforming force. Nerves tolerate greater degree of stretch when this force is applied slowly rather then in situations when the nerve is stretched abruptly and violently. The amount of the nerve elongation before structural failure depends on its preliminary length. Thus, a short nerve root will suffer earlier than long one. Structurally, the weakest part of the nerve root–spinal nerve system is where the nerve fibers are attached to the surface of the spinal cord.
The first stage of avulsion is weakening and tearing of the fibrous attachments binding the spinal nerve to the transverse process (Fig. 5). The possibility for avulsion is increased if the transverse process is fractured. Then, because of the lack of the protective connections with the spinal column, the entire spinal nerve–nerve root complex is pulled outward through the foramen until the apex of the dural funnel is wedged there. By this time the nerve roots are under considerable tension. Traction of dura is transmitted to the spinal cord, which is pulled laterally toward the foramen. This movement of the spinal cord only partly compensates for the tension on the nerve roots. At this point rupture of the nerve roots may occur without tearing of the meningeal tissue. If the deforming force is still being applied, the final stage will be the rupture of the dura. At this moment the roots lose their main support, and they can be promptly and rapidly stretched to the point at which disruption of their connection to the spinal cord occurs.
The central mechanism of cervical roots avulsion is questionable. Sunderland suggested that it is very unlikely that nerve roots can be avulsed by lateral traction without damage to the distal levels of the plexus. Flexion of the head and neck has been shown to stretch the spinal cord and nerve roots on the opposite side to which the head is turned. An extreme lateral flexion and rotation of the cervical spine can be often complicated by spinal cord and brachial plexus lesions.16 Thus, the cause of “pure” preganglionic injuries may be abnormal movements of the spinal cord. This type of injury can occur without any tearing of the dura. The ventral nerve roots are more susceptible to traction injury than the dorsal roots. Because the ventral nerve roots are thinner than their corresponding dorsal roots, and the dural sheath of the dorsal nerve roots is thicker than that of the ventral roots, anterior roots have a lower tensile strength than the dorsal roots.6,7
Murphey, Hartung, and Kirklin first recognized the usefulness of myelography in assessing brachial plexus injuries in 1947.17 They introduced the term “traumatic meningoceles” to describe a bulge of arachnoid membrane through a dural tear, with leakage of contrast medium beyond the spinal foramen. Their findings were consistent with the clinical picture of the brachial plexus injury. The authors considered this as an indicator of avulsion of the corresponding nerve root. Further studies using lipid-soluble contrast media were attempted but were limited by poor opacification of the subarachnoid space. These studies were focused on looking for traumatic meningoceles rather than an abnormality of the roots themselves.18,19,20,21,22,23
An introduction of water-soluble contrast agents improved the quality of evaluation of the roots and the spinal cord.24 Such investigation is better tolerated by the patients, can be performed earlier, and can be useful even in the presence of blood within the subarachnoid space. Water-soluble contrast medium can be directly injected under fluoroscopic control into the subarachnoidal space through a lateral puncture between C1 and C2 vertebra or via lumbar puncture with slow pooling contrast medium in the cervical region.
At myelography, the cord is seen as a midline-filling defect, with a slight fusiform expansion from the third cervical to the second thoracic vertebral level corresponding to the enlargement for the upper limb. On the anterior–posterior view, five or more separate rootlets can be seen as thin, 0.5–1-mm-diameter filling defects that converge to enter the nerve root sheath, an extension of the subarachnoid space past the lateral border of the theca (Fig. 6).
The myelographic findings in cases of intradural damage may include meningocele, obliteration of a root pouch, an absent or smaller root shadow, and cystic accumulation of cerebrospinal fluid in the spinal canal (Figs. 7 and and88).
A meningocele usually means that a root is avulsed. However, it was shown that traumatic meningocele is not a pathognomonic sign of nerve root avulsion. The presence of a meningocele does suggest a force great enough to cause an arachnoidal tear, but an absence of a meningocele does not exclude a preganglionic injury. Regardless of the presence or absence of meningoceles at other levels, one has to assume that proximal injury to the adjacent nerve roots is a real possibility. Other significant myelographic findings include extravasation of dye into the subdural or epidural compartments, evidence of acute cord swelling, or after some time, evidence of cord atrophy.
Marshal and De Silva25 first reported results of computerized axial tomography of the cervical spine in comparison with conventional myelography and surgical exploration of the brachial plexus. They found that CT scan following cervical myelography with enhancement greatly improved the diagnostic accuracy.
CTM enables us to see the intradural spinal nerve and visualize separately the ventral and dorsal roots with the corresponding segment of the spinal cord in the axial plane. The sign of root avulsion is disintegration of the root with the spinal cord. In addition, traumatic extra- and intraspinal meningoceles can be clearly demonstrated by CTM.
In the ideal situation, an experienced electromyographer carries out electrodiagnostic tests. The appropriate use of these tests involves an understanding of their neurophysiologic basis, drawbacks, and limitations. EMG can be helpful in the evaluation of root avulsion by providing evidence of denervation changes in the paraspinal musculature, rhomboids, and serratus anterior muscles. This indicates a very proximal injury because the paraspinal muscles are innervated by the proximally located dorsal rami of the spinal nerves. These nerves have their origin close to the intervertebral foramina.
Bonney and Gilliat26 have shown that sensory nerve action potentials (SNAP) are important in the evaluation of root avulsion in brachial plexus injuries. Preganglionic injury produces a complete distal sensory loss but preserves distal sensory conduction. This occurs because the dorsal root ganglion in the case of dorsal root avulsion is still in continuity with peripheral sensory nerve fibers. The study is performed by stimulating the hand in the C6 (thumb and index finger), C6–C7 (index and long finger), and C8–T1 (little and ring finger) dermatomes and recording from median, or from the ulnar nerves more proximally. If the stimulated area is anesthetic to touch, recording of a SNAP indicates a preganglionic injury in the particular root. Somatosensory studies, performed by stimulating Erb's point and recording over the spinal cord and contralateral cortex, can also be used in the evaluation of plexus injuries.
The charts of 321 patients who sustained brachial plexus paralysis and were operated on in our center from 1982 to 2003 were reviewed and analyzed. This is limited to 70 patients with obstetrical brachial plexus palsy (36 males and 34 females) who underwent supraclavicular plexus exploration as a part of their surgical treatment. Data obtained from operative drawings, operative notes, and histologic reports for each of the patients was reviewed. All surgeries were performed by the senior author (J.K.T.). The age of the patients ranged from 2 months to 15 years (Fig. 9).
Cervical myelography was performed by means of a lumbar puncture under general anesthesia. The procedures were done with injection of 3–5 mL of water-soluble contrast media (Iohexol) with 240 or 300 mg/mL iodine concentration in the subarachnoid space, with subsequent fluoroscopic guidance of a more concentrated dose of contrast medium into the cervical region by gravity. The lumbar injection was performed slowly over a period of 1–2 minutes to avoid excessive mixing with cerebrospinal fluid and subsequent loss of contrast medium, as well as premature cephalad dispersion. Anterior–posterior and lateral views were taken.
The myelographic findings were judged according to the classification proposed by Nagano et al,27 who divided myelographic abnormalities into six types: N, normal; A1, slightly abnormal; A2, A3, D, more distinct abnormalities; and M, traumatic meningocele (Fig. 8). In our study, the types N and A1 were considered to indicate an absence of roots avulsion.
After myelography, 1.5-mm axial slice CT myelograms were obtained from C2 to T2. When in doubt or when evident disease was detected on the initial scan, additional 1-mm axial slices were obtained. Scan was obtained in the plane oriented parallel to the cervical discs. A CTM diagnosis of root avulsion was based on the absence of either one or both ventral and dorsal roots with or without meningocele. When axial scans revealed both ventral and dorsal roots from the spinal cord to the intervertebral foramen, the root was considered to be intact. Determining the presence of an avulsed or intact root was further aided by comparison with the contralateral intact root. Although the exact pathological anatomy of each category is different, partial lesions and complete lesions in both myelography and CTM were all considered to be avulsions.
Electrophysiological evaluation included needle and percutaneous EMG, motor and sensory nerve conduction velocities, and SNAP.
Data of plain myelography, CTM, and electrophysiological evaluation were compared with intraoperative diagnosis, which was based on observation under magnification of the injured brachial plexus, the results of electrostimulation of the spinal nerves, and intraoperative frozen sections and histochemistry.
To determine the diagnostic accuracy of the predictors examined in this study, the sensitivity, specificity, positive predictive value, and negative predictive value for root avulsions were calculated. Sensitivity is the percentage of the roots, which intraoperatively were found to be avulsed and were correctly diagnosed with avulsion preoperatively (number of the “true positive” roots divided by number of roots intraoperatively found to be avulsed [“true positives” + “false negatives”] multiplied by 100). When the number of “false negatives” is small relative to the “true positives,” sensitivity approaches 100%.
Specificity is the percentage of the roots that intraoperatively were found not to be avulsed and that were correctly diagnosed as normal preoperatively (number of the “true negative” roots divided by number of roots intraoperatively found not to be avulsed [“true negatives” + “false positives”] multiplied by 100). When the number of “false positives” is small relative to the “true negatives,” specificity approaches 100%.
Positive predictive value is the percentage of the roots that were positive for avulsion preoperatively and that were found avulsed on the surgery from all number of roots positive for avulsion preoperatively (“true positives” + “false positive”). Negative predictive value is the percentage of the roots that were negative for avulsion preoperatively and that were found not avulsed on the surgery from all numbers of roots negative for avulsion preoperatively (“true negative” + “false negative”).
A total of 420 spinal nerves (from C4 through T1) were examined in 70 patients. Fifty-two patients (74.3%) had avulsion injury of brachial plexus. Intraoperatively, 135 roots (32.1%) were found avulsed, and 285 roots were not avulsed (67.9%). Accidence of each root avulsion is presented in Fig. Fig.1010.
One patient had avulsion of five roots from C5 through T1. Eleven patients (15.7%) had four roots avulsed. Fifteen patients (21.4%) had three roots avulsed. Sixteen patients (22.9%) had avulsion of two roots, and nine patients (12.9%) had one isolated root avulsion.
Plain myelography was performed in 67 patients, and 402 roots were evaluated. Accuracy of plain myelography was 85.3%, sensitivity was 71.0%, specificity was 92.3%, positive predictive value was 81.9%, and negative predictive value was 87.6%.
By CTM 348 roots were evaluated in 58 patients with accuracy of 89.4%, sensitivity of 83.2%, specificity of 92.1%, positive predictive value of 89.5%, and negative predictive value of 81.4%.
Sixty-five patients had preoperative electrodiagnostic studies. The accuracy, sensitivity, specificity, and positive and negative predictive values of electrodiagnostic studies were 76.2%, 39.5%, 93.2%, 69.1%, and 77.9%, respectively.
These data have shown that CTM was significantly more accurate and sensitive than electrodiagnostic studies as well as plain myelography versus electrodiagnostic studies (p<0.05) (Fig. 11). There was no statistically significant difference in accuracy and sensitivity between plain myelography and CTM.
The plain myelography and CTM had the least accuracy in the diagnosis of C8 and T1 lesions: 77.6% and 83.1%, respectively (Fig. 12). Both radiologic methods had an insignificant difference of accuracy rate in evaluation of the roots from C5 through C7.
Electrodiagnostic study was more accurate in the evaluation of C5 and T1 roots (83.1% and 72.3%) and less accurate at the C6, C7, and C8 (64.6%) levels.
Accuracy of each diagnostic method was also evaluated in regard to the age of the patients (Fig. 13). CTM showed best results in children older than 6 years (95.2%) in comparison to plain myelography and electrodiagnostic studies (81.5% and 70.7%, respectively).
The group of the babies younger than 6 months old has shown the lowest sensitivity rate for both radiologic methods, which was 63.2% for plain myelography and 76.1% for CTM. Both radiologic studies were superior to electrodiagnostic study in the group of the children 1–6 months old.
The electrodiagnostic studies in this group of patients were also low at 46.2%; nevertheless, the worst sensitivity of electrodiagnostic studies (33.3%) was found in patients who were 1–6 years old. In contrast, this group of the patients showed the highest sensitivity of plain myelography at 88.9%. The CTM was most sensitive in patients who were 6–14 years old at the time of the examination.
Differentiation of pre- and postganglionic injuries is the most important goal of examination of the patients with traction injuries of the brachial plexus. The radiological and electrodiagnostic studies play a crucial role. Since the first publication by Murphy et al. in 1947,17 plain or conventional myelography is considered to be useful in identification of cervical roots avulsions. Further studies confirmed the value of cervical myelography in the diagnosis of root avulsion by comparison of myelogram with the findings from direct observation of the cervical spine after laminectomy.18,28,29 On the basis of these studies, meningocele was considered as a pathognomonic sign of root avulsion.30 However, it was noted that myelography does not always correlate with clinical picture and intraoperative findings.22,31 The introduction of water-soluble contrast media in cervical myelography24 allowed more superior visualization of the normal roots and possible abnormalities that usually accompany nerve root avulsion. Nagano et al.27 (Fig. 8) classified the aberrations into six types and correlated them with intraoperative findings in 90 patients. Type N is a normal shadow, and A1 is a slightly abnormal root sleeve shadow, in which shadows of roots and rootlets can be recognized but are different from contralateral normal side. Type A2 is obliteration of the tip of the root sleeve with the shadow of root or rootlets showing. Type A3 is obliteration of the tip of the root sleeve with no root or rootlets shadow visible. Type D is defect instead of a root sleeve shadow, and type M is traumatic meningocele. Nagano et al. evaluated the condition of 369 roots according to their classification. Normal myelographic appearance was seen in 72 roots, of which 65 roots were found to be normal in surgery or had postganglionic injury (negative predictive value was 90.3%).
It is evident that the A1 type of root sleeve shadow is not a sign of preganglionic lesion. This is one of the reasons that the A1 type of root sleeve was not considered in our study as sign of nerve root avulsion. Types A2, A3, and D have high predictive value for a preganglionic root lesion. For differentiation of these shadows, a well-contrasted myelography is mandatory. Traumatic meningocele is the most characteristic myelographic finding of root avulsion. However, it was found in less than a half of avulsed nerve roots.
Our study showed accuracy of plain myelography as 86%, that of sensitivity as 79.0%, that of specificity as 91.6%, that of positive predictive value as 87%, and that of negative predictive value as 86% (Fig. 12). The plain myelography was less sensitive in infants younger than 6 months (63.2%) as compared with groups of older babies (70.6%), children (81.4%–88.9%), and adults (79.8%) (Fig. 13). This fact can be explained by difficulties in estimating small objects, such as shadows of nerve root sleeves and meningocele in small infants.32 Gilbert and Tassin4 evaluated 495 spinal nerves in 79 patients with obstetrical plexus palsy by plain myelography, which was compared with intraoperative findings and showed that conventional myelography was a reliable examination, with only 14 false positive results. Earlier, our institution also reported good correlation between plain myelography and intraoperative brachial plexus examination, with an accuracy of 80% in a group of 10 patients with obstetrical paralysis.33
It was reported that the shadows on myelograms may change according to the time elapsed from injury to examination.27 In our study, plain myelography was found to be less sensitive (55.6%) if the examination was performed earlier than 1 month after the injury. The reason for this could be leak of contrast medium through dural tears, which are usually obliterated by dural scarring late on. If the concentration of contrast medium is low, the shadows can be misjudged.
Regarding the level of lesion, plain myelography was less sensitive in the evaluation of C5, C6, and T1 spinal nerve (sensitivity of 61%, 75.4% and 67.2%) (Fig. 12), and there was no significant difference in sensitivity on the C7 (86.1%) and C8 (84.9%) levels. These data correlate with previous studies, which indicated that the relatively high incidence of incorrect predictions on the level of C5 and C6 is probably related to poor opacification of the cerebrospinal fluid space.24,25,27
CTM allows for the evaluation of the spinal nerve in the intervertebral foramina, ventral and dorsal nerve roots, and rootlets at the site of their attachment to the spinal cord. As mentioned earlier, traction of the brachial plexus will initially result in tearing of the arachnoidal and dural sheet, but rootlet avulsion may exist without meningocele. In addition, “complete” recovery is possible in children with multiple meningocele.34 The existence of meningocele on CTM can complicate visualization of roots and rootlets. Sometimes roots and rootlets are displayed very clearly in the middle of the meningocele, but it can also push the roots away—even intact rootlets at adjacent levels. Chow et al. compared preoperative myelograms with intraoperative findings and found that root avulsions were better predicted by identifying the absence of rootlets in a pseudomeningocele rather than in a pseudomeningocele where rootlets traversing the sac could not be identified.35 Other studies compared pseudomeningocele with intraoperative data and revealed its high specificity for root avulsions.32,36,37,38
In our study, CTM was found to be more accurate and sensitive than plain myelography in all levels of the cervical spine (Fig. 12). However, nerve root avulsions at the T1 level were correctly estimated by CTM only in 60 of 90 cases (sensitivity of 66.6%). This finding resembles data shown also by other authors.25,27,35,36,37 Marshal and De Silva25 first reported results of computerized axial tomography of the cervical spine in comparison with conventional myelography and surgical exploration of brachial plexus. They found that a CT scan with enhancement greatly improves diagnostic accuracy, particularly at C5 and C6 root levels.
As was mentioned earlier, the spinal nerves have progressively more oblique direction from C4 to T1. Thus, the lower cervical and first thoracic nerve and corresponding nerve roots are not usually seen on the same axial slice.27 Another reason for difficulty in evaluating T1 roots on CTM could be the significantly reduced quality of images at this level in patients with a relatively short neck and broad shoulders.35
In our study we came to a conclusion about root avulsion based on the following CTM picture: absence of dorsal or ventral or both rootlets without meningocele, absence of dorsal or ventral or both rootlets with meningocele, and meningocele that mask rootlets.
CTM, as the plain myelography, was less sensitive in the infants younger than 6 months (Fig. 13). In general, radiologic examination of the brachial in newborns is a challenge because nerve roots and rootlets are difficult to see even with 3-mm thin slices and enhancement. Thick 5-mm axial slices would not be adequate for recognition of nerve roots and rootlets. Such difficulties were noted by Chow et al.35 who could not determine the status of rootlets in 17% of the examined meningoceles. In our study, CTM was superior to plain myelography in all groups of age except children ages 1–6. In this group, there was no significant difference between sensitivity of both radiologic methods. However, in this group of patients (1–6 years) CTM had slightly superior specificity (84.6%) to plain myelography (80%).
Regarding the factor of time since the injury, the maximum CTM accuracy was shown in three patients who were examined earlier than 1 month following injury. In this group, all roots were correctly estimated with sensitivity and specificity at 100%. In contrast, plain myelography in these patients showed sensitivity as 55.6%. This data and previous studies indicate that CTM in fresh cases may be superior to plain myelography because contrast medium loss through the dural tear compromises the quality of images. Additionally, large traumatic meningocele, may collect big amount of contrast media and decrease its concentration on other levels.
Sometimes CTM overestimates the extent of the lesion, especially when there is contrast medium leak from adjacent meningocele. Although the timing of the study did not influence the sensitivity of CTM, specificity slightly decreased with the time elapsed since the injury.
Our comparison between the two radiologic methods leads us to the conclusion that despite their invasiveness, considerable exposure to radiation, and possible adverse reactions to the contrast medium, myelography alone or in conjunction with CT is a useful tool in the preoperative assessment of the level of brachial plexus injury. One of the advantages of myelography is its ability to delineate the entire cervical spine and spinal nerve sleeves. Plain myelography is not reliable in cases in which the examination is made earlier than 1 month after injury or in cases of large meningoceles, when a leak of contrast medium may compromise the quality of the images. In addition, conventional myelography is not so accurate at the C5 and C6 nerve root levels.27 CTM is superior to plain myelography in visualizing the nerve rootlets because of axial imaging. However, it is difficult to detect the entire extent of the injuries on axial slices, especially in the lower cervical and first thoracic segments, where nerve roots have the most oblique direction. Thus, combination of multiple slices and multiple planes is needed, which requires more time to complete examination, which becomes more difficult for the patients.
Electrophysiological examination was accurate in 81.8% of all evaluated levels of the cervical spine. Positive predictive value of this method for root avulsion was 87.6%, which was even higher than that shown by plain myelography. In spite of high level of accuracy, electrodiagnosis has shown significantly lower overall sensitivity rate per level compared with both radiologic methods (64%). Sensitivity of this method was found to increase from upper to lower segments of the spinal cord. At the level of C5, the sensitivity was as low as 51.2%. At the level of T1, sensitivity of electrodiagnosis was 74%, which is superior to both plain myelography and CTM (Fig. 12). These findings correlate with the value of electrodiagnosis encountered in previous publications.39,40,41
Our data confirmed common opinions that SNAP work best for assessment of C8 and T1 and less well for C5, C6, and C7 because of the overlap of their dermatomes.26,39,42,43 Knowledge of the segmental innervation of the skin of the involved limb is important for a topographical diagnosis. However, there is still controversy concerning the roots responsible for the sensory innervation of the upper limb.44,45,46,47,48,49 There is moderately good agreement that sensory fibers in the axillary and musculocutaneous nerves are derived from C5 and C6, and those of the index finger from C7 and of the little finger from C8. The main controversy concerns the roots responsible for innervation of the thumb. Thus, in a C5 and C6 injury, the index finger and even the thumb may have input from C7 as well as C6 and give a misleading positive trace. In addition, an absence of the SNAP with anesthesia in a corresponding segment is not pathognomonic sign of a postganglionic root injury because the lesion may involve two levels of the brachial plexus.
The detection of denervation potentials in the paraspinal muscles may indicate the involvement of the roots or the spinal nerves proximally to the origin of the posterior ramus supplying these muscles. The paraspinal muscles usually receive innervation from several dorsal rami, which have considerable overlap. This method does not allow an accurate delineation of the involved segment. The presence of denervation potentials does not indicate whether the root lesion is preganglionic or between the dorsal root ganglion and the origin of the dorsal ramus.39,50 In addition, the evaluation of paraspinal EMG requires complete relaxation, which can sometimes be difficult to obtain, especially in children, before the presence of fibrillation potentials can be ensured. EMG testing in children is very difficult because of the violent reaction the patients have to needles and stimulation. To make this test more reliable in our practice we have been wrapping the children in a sheet, immobilizing their contralateral normal extremities.39 However, lower sensitivity of electrodiagnosis (33.3%) was found in our study in patients between 1 and 6 years of age (Fig. 13). Even the tests, which were done in infants younger than 6 months old, were slightly more sensitive (46.2%). Regarding the time elapsed since injury, electrodiagnosis was less sensitive when performed earlier than 1 month postinjury (44.4%) and later than a year after the injury. The reasons for this are an absence of definitive degenerative changes in involved muscles and an unclear electrophysiologic picture a year after the injury, resulting from probable regeneration and overlap from less injured neighboring segments.
This patient was a 3-month-old child born to 31-year-old mother at 40 weeks of gestation after uncomplicated pregnancy by vaginal delivery and vertex presentation. Labor was complicated by shoulder dystocia. No forceps or suction were used. The birth weight was 4.5 kg. The child was diagnosed with right obstetrical brachial plexus palsy at birth and was referred to our center at the age of 2 months and presented with total paralysis of his right upper extremity. Electrophysiological studies revealed some motor unit activity in serratus anterior, supra-, and infraspinatus muscles. Absence of sensory nerve action potentials ruled out an avulsion at C8 and T1 roots. Preoperative CTM shows (Fig. 14A–C) normal C5 through C8 rootlets and (Fig. 14D) avulsion of the T1 root. During exploration, only the T1 root was found avulsed (Fig. 14E,F).
This child was born at 42 weeks' gestation. Mother developed maternal diabetes in her late pregnancy. The delivery was vaginal in vertex presentation and was complicated by shoulder dystocia that required excessive traction. Total paralysis of his left upper extremity was noted. The child was diagnosed with left obstetrical brachial plexus palsy. The patient had a course of physical therapy and gradually developed some improvement in the function of the shoulder. No surgical treatment had been done by the time the child was referred to our center at the age of 2 years and 5 months. Electrodiagnostic studies indicated C7 and C8 root avulsion. CTM showed C6–T1 root avulsion (Fig. 15A–D), and Intraoperative findings confirmed the diagnosis of C6–T1 roots avulsion (E,F).
This patient was born at 40 weeks gestation after uneventful pregnancy. Delivery was vaginal, and vertex presentation was complicated by shoulder dystocia and lasted 11 hours. No suction or forceps were used. The weight of the baby was 4.3 kg. He was diagnosed with right brachial plexus paralysis immediately after birth. There was no functional improvement during the first months of life. At age 4 weeks, the patient was referred to our center. Cervical myelography followed by CT scan were ordered. Plain myelogram showed C5–C7 avulsion (Fig. 16A, white arrows), and CTM found C5 (Fig. 16B), C6 (Fig. 16C), and C7 (Fig. 16D) roots avulsion. At 7 weeks old, the child underwent exploration of the right brachial plexus, which confirmed diagnosis of C5, C6, and C7 roots avulsion (Fig. 16E).
This child was born to 27-year-old mother after uneventful pregnancy at 41 weeks' gestation by vaginal delivery in vertex presentation. Labor was complicated by shoulder dystocia and lasted 13 hours. Forceps were used during delivery. Birth weight was 4.2 kg. Left brachial plexus paralysis was diagnosed immediately after the birth. He was referred to our service at age 7 years. Clinical and electrodiagnostic studies were in favor of C7, C8, and T1 roots avulsion. Patient underwent cervical myelography following a CT scan of the cervical spine that predicted C7–T1 avulsion, which was conformed during exploration of left brachial plexus.
Myelography and CTM are useful diagnostic tools in evaluating patients with obstetrical brachial plexus paralysis presenting with nerve roots avulsion injuries. For better diagnosis, they have to be combined.
The electrophysiological findings may be helpful, but they do not allow for reliable differentiation of a pre- and postganglionic nerve lesion in the children population.