Presented at ACVR scientific meeting, San Antonio, TX, October 22, 2008
Address correspondence and reprint requests to Gabriela Seiler, at the above address. E-mail: firstname.lastname@example.org
Using cadaveric dogs, we established the ultrasonographic landmarks for performing paravertebral injections around the brachial plexus nerve roots in the dog, and assessed the accuracy and regional spread of the aliquots. A mixture of methylene blue dye and an iodinated contrast medium was used as the injectate. A 0.3 ml volume was used to assess accuracy and a 3.0 ml volume was used to assess regional spread. Accuracy and regional spread were assessed from computed tomography (CT) images acquired after injection by measuring the distance from each foramen to the nearest edge of contrast medium, and the dimensions of spread of contrast medium in each anatomic plane, respectively. The mean distance of small volume injections from each foramen was 0.9 cm (standard deviation [SD] 0.56 cm). The mean spread of contrast medium for the small volume injections measured 1.7 cm (SD 0.6 cm) cranial-to-caudal, 1.2 cm (SD 0.4 cm) dorsal-to-ventral and 7.4 cm (SD 0.4 cm) medial-to-lateral. The mean spread of contrast medium for the combined three large volume injections measured 7.4 cm (SD 1.7 cm) cranial-to-caudal, 3.1 cm (SD 0.8 cm) medial-to-lateral, and 2.8 cm (SD 0.5 cm) dorsal-to-ventral. After the CT studies, staining of each nerve root and any other regional structure was assessed grossly. Based on our results, ultrasound can be used to guide injections around the nerve roots of the brachial plexus in dogs.
Regional sensory nerve desensitization with local anesthetics has many uses in dogs and cats.1–3 For the brachial plexus, desensitization is typically performed at the level of the shoulder joint using one of two methods. The anesthetic agent can be injected without image guidance using the cranial aspect of the shoulder joint as a landmark. Or, anesthetic agent can be injected after locating the motor branches of the brachial plexus using an insulated needle and a nerve stimulator.4–6 With these methods, local anesthesia at the level of the brachial plexus allows analgesia for procedures involving, or below, the elbow. However, desensitization of the nerves innervating the humerus and other aspects of the brachium requires introducing anesthesia at the level of the spinal nerve roots.2,7,8 Although electrolocation of spinal nerve roots is possible, administration of the anesthetic agent outside the perineural sheath cannot be recognized during injection. Consequently, there is a need for improving the accuracy of local anesthesia infiltration at the level of the cervical spinal nerve roots.
Ultrasound-guided injection of local anesthesia is well-established in humans.9–15 This allows direct visualization of target tissues, precise introduction of the anesthetic agent either into a nerve sheath or around it, and avoidance of other major structures. It also allows control of the volume of anesthetic agent injected into or around the nerve.9 In animals, ultrasound has been used to image the brachial plexus and other peripheral nerves in the dog and horse, as well as the structures of the neck and cervical spine in the dog.16–19 We were not able to find information on the use of ultrasound to guide injections at the level of the cervical and thoracic spinal nerve roots. Therefore, we identified the ultrasonographic landmarks for locating the C6, C7, and C8 cervical spinal nerve roots and assessed the feasibility and precision of paravertebral brachial plexus injections in dogs.
Materials and Methods
Eleven cadaveric dogs weighing between 20 and 30 kg were obtained through the Educational Memorial Program of the University of Pennsylvania. Most were studied within 24 h of death, though in some cases use was delayed up to 3 days post-mortem, in which instance the body was kept refrigerated at 1.1°C. The hair along the lateral aspect of the neck was clipped from the transverse processes of C4, across the brachial plexus area to the lateral aspect of the shoulder joint. Ultrasound imaging was performed with an 8 MHz microcurved probe.*
Bony ultrasound landmarks for locating and imaging the intervertebral foramina of the caudal cervical spine were established using a canine cervical spine skeleton immersed in clear gel. This was done with the transducer aligned in a dorsal plane, parallel to the axis of the spine at the level of the intervertebral foramina. Soft tissue landmarks were added subsequently by imaging a canine cadaver in lateral recumbency with the transducer in the same orientation parallel to the axis of the spine and injecting 1% methylene blue† around the bony landmarks. Verification of these landmarks was done through gross dissection and visualization of the methylene blue.
Ten canine cadavers were then imaged using the established landmarks. The cadavers were in lateral recumbency and the nondependent thoracic limb was pulled as far caudally as possible. This was done to improve visualization of the C7 transverse process, which was often partially obscured by the scapula. To visualize the C7 transverse process in some dogs, it was necessary to angle the transducer caudally, while maintaining the parallel orientation to the axis of the spine. Ultrasound-guided injections were then performed using a mixture of 1% methylene blue and Iohexol 350 mgI/ml,‡ in a 2:1 ratio, as a contrast agent. Three injections were performed along each side of the neck, targeting the left and right nerve roots of the C6, C7, and C8 spinal nerves. A different volume was used for each spinal nerve pair: a volume of 0.2 ml 1% methylene blue and 0.1 ml of Iohexol was used on the left side and a volume consisting of 2.0 ml of 1% methylene blue and 1.0 ml of Iohexol was used on the right. The smaller volume aliquots (0.3 ml) on the left side were used to assess the precision of the injections. The larger volume aliquots (3.0 ml) on the right side were used to assess the extent of regional spread.
All injections were performed using a 3.8 cm 22-G needle introduced from the cranial end of the transducer in a craniolateral to caudomedial direction with the probe positioned parallel to the spine, such that at least two, if not all three, transverse processes could be seen in the same longitudinal imaging plane (Fig. 1). For the C6 spinal nerve root, the needle tip was directed adjacent to the cranial edge of the base of the C6 transverse process. For the C7 spinal nerve root, the needle tip was directed adjacent to the caudal edge of the base of the C6 transverse process. For the C8 spinal nerve root, the needle tip was directed along the caudal edge of the C7 transverse process. When performing the injections, the needle tip was placed deep to the hyperechoic line representing the fascial plane of the scalenus muscles (Fig. 2).
Computed Tomography (CT)§ was performed immediately after the injections on one side of the neck. Each cadaver was in dorsal recumbency and lateral and dorsal scout images were obtained to assess alignment. In some dogs, the gantry was angled to compensate for a steeply angled caudal cervical spine. One millimeter transverse slices were acquired using 130 mA and 120 kVp at 1 mm intervals from C4 to T3 with an 8 cm2 field-of-view for eight of the dogs, and a 12 cm2 FOV for two larger dogs. After the initial CT study, the injections on the opposite side of the neck were completed, followed by a second CT scan using the same protocol.
CT images were viewed in a bone window using the same window width and level (2000/400). Injection precision was assessed by measuring the proximity of the small volume injections to the nerve roots. This was quantified on transverse images by measuring the shortest distance from the medial aspect of each intervertebral foramen to the nearest edge of the contrast medium for each segmental spinal nerve root (Fig. 3). Regional spread was measured for both small and large volume injections. Reformatted images were used to visualize and measure the longest distances of contrast medium spread in the cranial-to-caudal, dorsal-to-ventral, and medial-to-lateral directions. The dimensions of the small volume injections were measured individually, but measuring the dimensions of the large volume injections individually was not possible due to extensive overlap and confluence of adjacent aliquots; therefore the extent of the combined large volume injections (3.0 ml each, total of 9.0 ml) was measured.
After the second CT scan, each cadaver was dissected and staining of the targeted nerve roots was confirmed visually. Staining of a nerve root was recorded as present or absent, and was not quantified with respect to degree of staining. The presence of intraneural staining was not evaluated. Any staining of the phrenic nerve with 1% methylene blue and any spread to the epidural space or the mediastinum was recorded.
The most consistent ultrasonographic landmarks for guiding paravertebral injections were the transverse processes of C5, C6, and C7 and the bodies of the superficial and deep scalenus muscles. The transverse processes were identified by their consistent locations with respect to the scapula and by their characteristic appearances. The C5 transverse process has a flattened profile on ultrasound that is angled in a caudodorsal to cranioventral direction. The cranial portion of the C6 transverse process has a focal, pointed profile that merges ventrally into the sled-like caudal portion, which has a broad, uniform profile orientated in a craniodorsal to caudoventral direction. The C7 transverse process has a narrow profile and is typically visualized at the same level as the pointed, craniodorsal prominence of the C6 transverse process. At this level, a hyperechoic line extends between the transverse processes of C5, C6, and C7 that is formed primarily by the fascial margins of the scalenus muscles, although it also includes contributions from several other muscles that insert on or arise from the transverse processes, including the longissimus cervicis and thoracis, serratus ventralis, and intertransversarius muscles (Figs. 1 and 2).
With increasing experience, the nerve roots of the 6th–8th cervical segmental spinal nerves were identified sonographically in several cadavers as thin, rounded structures with heterogeneously echoic centers surrounded by hyperechoic sheaths. They coursed between the transverse processes deep to the scalenus muscles in a consistent manner. The 6th cervical spinal nerve runs obliquely across the cranial aspect of the sled-like base of the C6 transverse process, in a craniodorsal to caudoventral direction. The 7th cervical spinal nerve root runs approximately midway between the transverse processes of C6 and C7, orientated more parallel to the first rib. The 8th cervical spinal nerve root runs immediately caudal to the base of the C7 transverse process, orientated perpendicular to the axis of the spine (Fig. 2). Distension of the scalenus muscle fascial plane during injection was observed in several cadavers. On dissection, both large and small volume injections resulted in 100% staining of the targeted nerve roots. The averaged measurements obtained from each of the small volume aliquots, and from the dimensions of the individual small and combined large volume aliquots, are summarized in Tables 1–3. The average distance each small volume aliquot extended from the foramen increased from cranial to caudal. The extent of spread of the contrast medium was greatest in the cranial-to-caudal direction for both the small and the large volume injections. The incidence of contrast medium spreading to other regional structures of the neck for both large and small volume aliquots is summarized in Table 4. The probability of such staining was greater for the large volume aliquots than for the small volume aliquots, and staining of the phrenic nerve occurred only with large volume aliquots.
Table 1. Distance between Each Intervertebral Foramen and the Small Volume Contrast Medium Injections
Contrast Medium to Foramen Distance Mean (Standard Deviation)
Mean and standard deviation of the distance between the medial aspect of each intervertebral foramen and the nearest edge of contrast medium for each small volume aliquot (0.3 ml) injected at C5–6, C6–7, and C7–T1.
6.2 mm (3.7 mm)
9.0 mm (6.2 mm)
12.0 mm (5.6 mm)
Table 2. Spread of Small Volume Contrast Medium Injections
Direction of Spread
Spread of Contrast Medium at Each Location Mean (Standard Deviation)
Mean and standard deviation of the extent of the individual small volume aliquots (0.3 ml each) injected at C5–6, C6–7, and C7–T1, measured in dorsal, sagittal, and transverse image planes.
0.84 cm (0.31 cm)
0.87 cm (0.35 cm)
0.94 cm (0.29 cm)
0.53 cm (0.28 cm)
0.61 cm (0.24 cm)
0.63 cm (0.29 cm)
0.57 cm (0.26 cm)
0.49 cm (0.2 cm)
0.53 cm (0.15 cm)
Table 3. Spread of the Large Volume Contrast Medium Injections
Direction of Spread
Spread of Contrast Medium Mean (Standard Deviation)
Mean and standard deviation of the extent of contrast distribution after injection of large volume aliquots at C5–6, C6–7, and C7–T1, measured measured in dorsal, sagittal, and transverse image planes.
7.4 cm (1.7 cm)
2.8 cm (0.5 cm)
3.1 cm (0.8 cm)
Table 4. Nerve Root Staining and Regional Spread of Contrast
Small Volume Aliquots (%)
Large Volume Aliquots (%)
The incidence of staining of the targeted brachial plexus nerve roots and other regional structures of the neck for the two different volumes of contrast medium.
Targeted nerve roots
In humans, ultrasound guidance is the standard of care for administration of local anesthesia.20–23 For application of brachial plexus local anesthesia in animals, only anatomic landmarks or electrostimulation have been used to guide injections.6,8,24,25 However, the use of anatomic landmarks is inherently imprecise, especially if the nerve root is targeted as it exits the intervertebral foramen. With electrostimulation, there is a trade-off between using a higher current through the shielded needle vs. using a lower current. Higher currents lack relative specificity and can result in injections on the other side of fascial planes from the targeted nerves, whereas lower currents lack relative sensitivity and lead to increased risk of intraneural injections.15,25 With such close proximity to the spinal cord, any inaccuracy can result in serious complications. Furthermore, while in humans electrostimulation might be performed with an awake patient, animals require general anesthesia or profound sedation for electrical stimulation to be used.2,6,24 Ultrasound-guided techniques could potentially be performed in awake or slightly sedated dogs.
We demonstrated the feasibility of ultrasound-guided injections in the vicinity of the nerve roots of the brachial plexus and defined ultrasonographic landmarks: the transverse processes of C5–7 and the fascial plane of the scalenus muscles. The merged bellies of the superficial scalenus muscle cover the main spinal nerves comprising the brachial plexus as they course in a caudoventral direction between the transverse processes of C5–T1. Making the injection medial to this muscle resulted in containment of the aliquot since the nerves of the brachial plexus run in the superficial and deep portions of the scalenus muscle for approximately 2 cm before emerging and entering the thoracic limb. This is homologous to the interscalene groove in humans, which is used in a similar fashion to localize and constrain anesthetic agents for supraclavicular brachial plexus blocks.9–13,26,27
Using the described landmarks, all targeted nerve roots were stained with methylene blue following either small or large volume injections. For small volumes, aimed to assess the accuracy of the ultrasound landmarks, the distance of extension of the injected contrast medium from each foramen increased from cranial to caudal. This was due to the caudal nerve roots being less accessible due to the increased depth and angle required to target the nerve, as well as the presence of the scapula. The ideal distance from the foramen for a paravertebral injection of local anesthetic in the dog is unknown, but it is likely dependent on the volume of anesthetic agent used. The proximity of the contrast medium to each foramen in this study resulted in staining of all nerve roots and therefore the precision of the injections was considered adequate.
The large aliquot injections were aimed to simulate a standard volume of anesthetic agent for brachial plexus analgesia.6 However, there are no established volumes for paravertebral local anesthesia in the dog, neither for single nor multiple injections. Previously, when 3–5 ml of methylene blue dye were injected blindly at each of the C5–6, C6–7, C7–T1, and T1–2 nerve roots, the C8 spinal nerve was stained in only 66% of the cadavers.8 Using ultrasound guidance in our study, nerve root staining was observed in 100% of cadavers using a significantly smaller volume of contrast medium; this illustrates the increased precision made possible using ultrasound guidance. Ultrasound guidance may also allow reduction of volume of anesthetic agent used for brachial plexus blocks, although this requires further testing.
There can be clinical complications from brachial plexus anesthesia.28–32 Diaphragmatic hemiparesis secondary to phrenic nerve anesthesia in humans, while rarely clinically significant, can be the result of C4 nerve root involvement or aberrant spread of anesthetic agent outside the prevertebral fascia of the anterior scalenus muscle, and is therefore dependent on the anatomic location of the needle and the volume of anesthetic used.27,32–38 That phrenic nerve staining was observed in two out of 10 dogs when the 3 ml volume was used suggests that a similar volume of anesthetic agent could lead to complications. Accidental epidural or intrathecal injection of anesthetic agent during brachial plexus blockade in humans is typically accompanied by signs of cervical neuraxial anesthesia characterized by bilateral cervical and thoracic blockade with difficulty in breathing.29,35 Though relatively infrequent, staining of the epidural space was observed with the small volume aliquots in our study. The fact that the larger volumes resulted in a higher frequency of epidural and phrenic nerve staining compared with the smaller volume aliquots suggests that these problems could potentially be mitigated through dose reduction and ultrasound-guidance. The use of the scalenus muscle as a landmark was especially useful, as placement of contrast medium medial to the fascial plane helped contain the aliquot around the nerve roots. Other complications of brachial plexus blockade are due to long-term nerve damage resulting from direct trauma or pressure necrosis.28,31,37,38 Because cadavers were used in this study, it was not possible to assess that complication.
The use of cadavers was also associated with other limitations. Both the appearance of the nerve roots and the spread of the contrast media could be different than in live animals, especially in cadavers that were not studied immediately after death. The ultrasonographic appearance of peripheral nerves in both dogs and humans has been described as hypoechoic round structures with a hyperechoic surrounding sheath.9,12,17,19 In our study, the hyperechoic sheath was a consistent identifying feature; however, the nerves themselves had heterogeneously echoic centers. A similar appearance has also been noted in cadavers and live dogs where a high-frequency transducer (11–14 MHz) was used.19 While our high-frequency transducer was sufficient to visualize the nerves in most dogs, there was a learning curve involved in localizing and visualizing the deeper nerve roots, especially in larger cadavers. Using the described bone and soft tissue landmarks, however, the injections could still be guided into close proximity of the nerve roots. It is anticipated that in live patients, accurate placement of injections will be facilitated by the more readily identifiable nerves and vessels.
We conclude that ultrasound can be used to precisely guide injections around the nerve roots of the brachial plexus in dogs. Establishing landmarks for ultrasound guidance of paravertebral injections is the first step toward creating a practical and efficient clinical application of ultrasound for local brachial plexus analgesia in the dog. This could be useful for performing local anesthesia of structures proximal to the elbow, such as for adjunct anesthesia during surgery or perioperative pain control. Further investigation is needed to ascertain whether the technique can be applied in the clinical setting and to establish the optimal volume of anesthetic agent.
*GE Medical LOGIQ 9, General Electric Medical Systems, Milwaukee, WI.
†American Regent Inc., Shirley, NY.
‡GE healthcare Inc., Princeton, NJ.
§GE Prospeed, General Electric Medical Systems, Milwaukee, WI.