Thoracolumbar Spine

Chapter 32

Thoracolumbar Spine

Surgery of the thoracolumbar vertebral column is indicated for spinal cord decompression, vertebral column stabilization/realignment, various neuro-oncologic procedures, and spinal canal exploration. Accurate neuroanatomic localization and knowledge of disease pathophysiology are required for proper selection of appropriate surgical candidates.47 Before surgery, physical examination–based neurologic scoring with a validated system allows the clinician, in many instances, to provide important prognostic valuation.9,126 Specific anatomic detail obtained from advanced cross-sectional imaging studies is required for development of an adequate surgical plan.

Once surgery is deemed appropriate, timing should be considered. In certain situations, immediate intervention is paramount to mitigate ongoing spinal cord damage. Those animals, for example, with acute onset of severe signs, or that have rapid deterioration, may need surgery immediately.9,166 Other animals may benefit from a delay in surgery if concomitant illness, metabolic derangements, hydration status, or cardiorespiratory abnormalities require treatment.9

Structural detail obtained with the use of magnetic resonance imaging (MRI) or computed tomography (CT) is important in selection of a surgical approach for most thoracolumbar lesions. The approach should allow the surgeon direct access to the lesion with minimal spinal cord manipulation and soft tissue disruption. The specific location of a lesion (e.g., lateral, dorsal, ventral) within the spinal canal dictates which surgical technique is used and which side is approached. Animal positioning on the surgical table is important in allowing the surgeon optimum exposure and in facilitating technique. Animals with unstable vertebral column injuries should be moved extremely carefully to minimize spinal movement and should be positioned in a manner that aids in realignment. White tape or a deflated bean bag positioning device helps improve and secure positioning throughout the procedure.196

Thoracolumbar Vertebral Column Anatomy

Vertebrae are composed of a vertebral body, paired pedicles, laminae, and the dorsal spinous process. Articular processes arising from the pedicle at the cranial and caudal aspects form synovial joints with adjacent vertebrae, and the vertebrae themselves vary tremendously in shape and size (Figure 32-1). The vertebral articular process joint is more correctly referred to as the zygapophyseal joint. Accessory processes extend caudolaterally from the pedicles and are the attachment site for the tendon of the longissimus lumborum musculature. Transverse processes arise from the junction of the pedicle and body and extend laterally to craniolaterally throughout the lumbar vertebral column. The thirteen thoracic vertebrae articulate with the ribs. The spinous processes of the cranial thoracic vertebrae (to T10) slant caudally; those of the last two thoracic vertebrae slant cranially. The site of this change in direction (which may vary) is termed the anticlinal vertebrae.200 Thoracic vertebrae are similar between T11 and T13, except that no transverse processes are present. The zygapophyseal joints begin to diminish in size from T11 to T12 and are much less prominent cranial to T10-T11 in the thoracic spine. The articulation between vertebrae in the mid to cranial thoracic spine is less conspicuous and lies flush with the dorsal spinous process. Seven lumbar vertebrae are present; the transverse process of L1 is the smallest and may be obscured by the last rib. The more caudal lumbar transverse processes are narrower and longer.200

The intervertebral foramen is an opening between each pair of vertebrae ventral and slightly cranial to the zygapophyseal joints to allow spinal nerves and vasculature to pass through the vertebral column. A venous sinus lies along the ventrolateral aspect of the spinal canal. This vascular structure is often encountered during spinal surgery (Figure 32-2).59

The vertebral body lies ventral to the pedicles and is connected to adjacent vertebrae by intervertebral discs, which have three components: annulus fibrosus, nucleus pulposus, and cartilaginous end plate. The annulus fibrosus is made of concentric lamellae of collagen and encircles the central nucleus pulposus. It provides biomechanical support and allows multidirectional movement. The nucleus pulposus originates from remnant notochord and chondrocyte-like cells. It is normally well hydrated, contains glycosaminoglycans such as chondroitin sulfate and keratan sulfate, and has a dispersed matrix of type IV collagen. The cartilaginous end plate attaches the outer annulus to the bony end plate; it functions to allow nutrient entry to the avascular disc (Figure 32-3).

Biomechanical stability of the vertebral column relies on passive and active measures. The zygapophyseal joints, intervertebral disc space, and multiple muscle tendon units and ligaments provide endogenous biomechanical stability. Of these, the intervertebral disc has been shown to account for a majority of the stability.203 Three long ligaments of the spine (the supraspinous ligament and the dorsal and ventral longitudinal ligaments) and three short ligaments (intraspinous, intratransverse, and yellow ligaments), along with the intercapital ligament (T2-T11), contribute to stability of the spine.60 The dorsal longitudinal ligament lies directly on the ventral floor of the vertebral canal. The intercapital ligament runs from rib head to rib head directly under the dorsal longitudinal ligament, and is thought to be one of the reasons that intervertebral disc rupture is much less common cranial to T11. Disruption of disc components, especially in combination with zygapophyseal joint removal, alters spinal biomechanics.203 The clinical significance of this depends on the case scenario. For example, disc fenestration may be appropriate unilaterally in combination with a hemilaminectomy, but may cause instability when combined with a bilateral hemilaminectomy,6 or may be more risky in a large-breed versus a small-breed dog. 201

Approaches to the Thoracolumbar Vertebral Column

Surgery of the vertebral column requires a meticulous approach through overlying soft tissues. Various techniques have been described to provide maximum exposure, while minimizing unnecessary soft tissue dissection. For instance, relatively more exposure may be required in the setting of vertebral column stabilization than for a single-site hemilaminectomy.173 Of utmost importance is determining the precise vertebra number. This can be accomplished by utilizing endogenous landmarks (last rib, transverse process of the first lumbar vertebra, anticlinal vertebra, sacrum), by injecting a dye (such as new methylene blue) to mark the tissues, or by placing a marker needle under radiographic guidance or with the use of intraoperative radiography or fluoroscopy. It is important to note that a great deal of interanimal anatomic variation exists, including transitional vertebrae and hypoplastic, aplastic, or asymmetric ribs. Before surgery, the surgeon should carefully study high-quality images of the patient; during the procedure, articulated and disarticulated anatomic specimens of the spine should be available to the surgeon for reference purposes.

Dorsal Approach to the Cranial Thoracic Vertebral Column

The dorsal approach the cranial thoracic vertebral column is used when dorsal laminectomy, hemilaminectomy, or vertebral column realignment/stabilization is indicated in the region of T1-T5.32,169,173,196 The patient is secured on the surgical table in sternal recumbency with elbows adducted. For unilateral procedures, slight axial rotation of the patient so that the affected side is more dorsally positioned (with dorsal spinous processes pointing slightly away from the surgeon) facilitates access to laminae and articular processes on the affected side.

Access to the cranial thoracic (T1-T5) region is typically gained through a dorsal parasagittal (approximately 2 to 3 cm from midline in a medium-size dog) skin incision made from approximately the midcervical to the midthoracic vertebral region. After hemostasis of the skin edges, sub-draping with sterile huck towels or a water-impermeable barrier is done to minimize the risk of drape puncture by a surgical instrument. The subcutaneous tissues are incised sharply until the dorsal tendinous raphe is identified. This tendon is incised on midline, sharply or with the use of monopolar cautery.169 Dissection is done bilaterally or unilaterally, depending on the exposure required. The supraspinous ligament should remain intact when possible.

The trapezius and rhomboideus muscles are retracted using self-retaining retractors. Underlying tissues include the splenius muscle cranially and the serratus dorsalis muscle caudally. These muscles should be incised very close to the dorsal spinous processes at the level of the median tendinous raphe. Lateral retraction of these muscles exposes the nuchal ligament cranially (which originates from T1 and T2 dorsal spinous processes) and the longissimus cervicis and longissimus thoracis et lumborum muscles (Figure 32-4).

Elevation of the epaxial musculature from the attachment to the dorsal spinous processes and the vertebral laminae is done with periosteal elevators (Figure 32-5). Careful elevation should be done at the level of the vertebral pedicles between C7 and T1 to prevent injury to the nearby vertebral artery.173 The interarcuate space between C7 and T1 is relatively large compared with other interspaces; care should be taken at this location to avoid inadvertent spinal canal penetration. If removal of the dorsal spinous processes of T1-T3 is required, the origin of the nuchal ligament will be disrupted; however, clinically, this is not significant.32 Alternatively, the nuchal ligament can be longitudinally divided on midline, leaving intact its attachment to the supraspinous ligament, or the dorsal tip of the spinous process can be cut from the remainder of the process, preserving the attachment of the nuchal ligament to the supraspinous ligament.32,169,173

Careful multilayer closure is required to minimize seroma formation. Each muscle layer should be closed separately, and dead space should be obliterated. A light compressive bandage applied postoperatively may lessen the risk of seroma formation.173

Dorsal Approach to the Thoracolumbar Vertebral Column

The dorsal approach to the thoracolumbar vertebral column is used for dorsal laminectomy, hemilaminectomy, pediculectomy, mini-hemilaminectomy, intervertebral disc fenestration, lateral corpectomy, and vertebral realignment/stabilization for vertebrae from T6 through L6.32,173,181,196 Conceptually, this technique varies little from the approach to the cranial thoracic spine. However, fewer muscles are encountered because this region is caudal to the attachments of the thoracic limb muscles.

The dog is positioned in sternal recumbency with limbs in a neutral flexed position and firmly secured to the surgery table (Figure 32-6). Slight rotation of the dorsal aspect of the dog toward the unaffected side can be done to facilitate exposure for unilateral procedures (affected side is dorsolaterally positioned). A dorsal skin incision is made from about three vertebrae cranial to three vertebrae caudal to the affected site. Making the skin incision 1 to 2 cm parasagittally prevents the incision from being directly over the dorsal spinous process postoperatively and may reduce risk of wound breakdown caused by skin tension directly over the spinous processes,201 although many surgeons will incise directly over the midline. After hemostasis, sub-draping is done to minimize the risk of drape puncture. Sub-draping involves using Backhaus towel clamps, Michel skin clips, or suture on a taper point needle to secure the drape and completely cover the skin surfaces. If intraoperative fluoroscopy will be used, suture is preferred. If an adhesive plastic drape (e.g., an Ioban drape) has been used before skin incision, this step may not be necessary. Incising the subcutaneous tissue and underlying adipose tissue exposes the dorsal thoracolumbar fascia. The next few steps can be performed bilaterally or unilaterally depending on the exposure necessary. The dorsal thoracolumbar fascia is incised close to midline, exposing the multifidus musculature (Figure 32-7). Periosteal elevators are used to elevate the multifidus muscle from adjacent dorsal spinous processes. Next, the tendinous attachment of the multifidus muscle along the caudolateral aspect of each dorsal spinous process is released with scissors. This step is repeated for each of the vertebra within the surgical field. Elevation of this muscle should continue ventrally to the level of the vertebral lamina. A thick tendinous attachment of the multifidus muscle to the articular process exists at each zygapophyseal joint. This tendon should be incised close to the articular process. Care should be taken, especially in relatively small dogs, to sever only the tendon of the multifidus muscle while avoiding damage to surrounding soft tissues. After this tendon is released, the multifidus muscle is elevated to the level of the accessory process.

The longissimus lumborum muscle attaches via a prominent caudolaterally directed tendon to the accessory processes from T11-L7. This can be incised to enhance exposure of the lateral vertebral column. The spinal nerve and associated vasculature lie just ventral and cranial to this tendon (Figure 32-8). The accessory process is an important landmark for identifying the ventral aspect of the spinal canal. The longissimus lumborum muscle between the articular process and the accessory process can be removed to aid in exposure. Self-retaining retractors (e.g., Gelpi retractors) can be used to retract soft tissues. This exposure allows ready access to the dorsolateral aspect of the vertebral column.

Closure requires reapposition of the thoracolumbar fascia, the subcutaneous tissue, and the skin. Subfascial muscles do not need to be sutured.

Lateral Approach to the Thoracolumbar Vertebral Column

The lateral approach to the thoracolumbar vertebral column is indicated for lateral fenestration of intervertebral discs and lateral corpectomy from T10 to L5.32,65,151,173,194 It provides lateral exposure to the vertebral body and disc.

The animal is positioned in lateral recumbency with the affected side dorsal and secured to the surgery table. The spine is positioned so it is slightly laterally bent during the procedure to open disc spaces by placing a sandbag under the dog at the level of the thoracolumbar junction (Figure 32-9). An oblique skin incision is made from the lateral aspect of the dorsal spinous process of T9 toward the ventral border of the wing of the ilium, stopping at about L5.65 The superficial and deep thoracolumbar fasciae are incised to reveal the epaxial musculature (serratus dorsalis caudalis, longissimus lumborum, and iliocostalis lumborum). Accurate counting is necessary to approach the correct vertebra. Often, the thirteenth rib (when present) is a good anatomic landmark from which to begin counting.

The ribs and transverse processes are palpated at this point during surgery. Blunt dissection through the muscles just dorsal to the transverse process allows a window to be created. The annulus fibrosus of the disc lies just cranial to the base of the transverse process of the lumbar vertebrae and just cranial to the rib head in the thoracic spine. The corresponding nerve root and blood vessels of each disc should be retracted cranially to avoid damage (Figure 32-10). Care should be taken to avoid injury to the spinal nerve and to prevent entry into a body cavity (thorax or abdomen).65 Closure of the deep and superficial thoracolumbar fascia, subcutaneous tissue, and skin is required.

Dorsolateral Approach to the Thoracolumbar Vertebral Column

The dorsolateral approach to the thoracolumbar vertebral column is an alternative surgical technique for approaching the lateral aspect of the vertebral column from T9 through L7.32,173,233 It is adequate to perform hemilaminectomy, lateral intervertebral disc fenestration, lateral corpectomy, pediculectomy, and mini-hemilaminectomy.65,151,233 Complications associated with this procedure have ranged from mild to severe and include pneumothorax, hemorrhage, transient neuromuscular dysfunction, scoliosis, worsening of neurologic grade, and myelomalacia.14

The animal is positioned in sternal recumbency with the hindlimbs flexed to maintain normal curvature of the spine. Some surgeons prefer slight dorsal rotation of the affected side as described earlier for the dorsal approach. A parasagittal skin incision 1 to 2 cm lateral from midline toward the affected side is made from about two vertebrae cranial and two vertebrae caudal to the target site(s).32 Subcutaneous and adipose tissues are incised and retracted laterally to expose the thoracolumbar fascia covering the spinalis et semispinalis and multifidus musculature. The fascia is incised 5 to 10 mm lateral and parallel to midline to reveal the underlying epaxial musculature. The intermuscular septum of the multifidus musculature (medially) and the longissimus musculature (laterally) is identified. The division between these muscles is bluntly dissected and the multifidus musculature retracted dorsomedially to reveal the tendinous attachments of the longissimus muscle on each accessory process. Elevating the multifidus musculature dorsally from the transverse process may allow greater exposure of the disc space.233

Approaches to the Thoracolumbar Spinal Cord

Once the vertebral column has been exposed properly, bone removal is necessary to expose the spinal cord. Techniques for bone removal should be based on lesion characteristics and goals of surgery. Generally, removing as little bone as possible to achieve surgical goals is recommended. With a pneumatic drill, the outer cortical, medullary, and inner cortical bone is carefully removed down to the level of the periosteum. A dental spatula or rongeurs (pediatric Lempert rongeurs, Lempert rongeurs, or Love-Kerrison rongeurs, depending on the size of the laminectomy and the level of access to the laminectomy) are used to enter the spinal canal (Figure 32-11). An understanding of vertebral anatomy and lesion distribution within the spinal canal is important for selecting the optimal technique. Anatomic differences among vertebrae, as described earlier, within the same animal and between species must be recognized.

At times, variation or modification of standard technique is required to achieve adequate exposure for a given surgical problem. This may include combining various techniques to enhance access to the spinal canal.203 Knowledge of vertebral column biomechanics is important in determining when surgical stabilization procedures are required to reinforce the vertebral column. Briefly, the normal spinal column must be able to withstand forces acting on it, including dorsoventral and lateral bending, rotation, shear, and axial loading. When determining whether a surgical approach (sometimes combined with a previous traumatic episode) has destabilized the spine enough to warrant surgical stabilization, the surgeon should evaluate the three major portions of the spine that provide stability. The vertebral body acts as a buttress to resist bending and axial loading, the articular processes resist all forces, and the intervertebral disc is an important stabilizing factor against rotation and lateral bending. Compromise of two or more of these components, particularly bilaterally, traumatically (as opposed to surgically), or in a large dog, may suggest the need for additional stabilization.202


The hemilaminectomy93,181,201 is the most common procedure used to expose the vertebral canal in dogs and cats, and intervertebral disc disease is the most common indication for this approach. It exposes the ventral, dorsal, and lateral (unilateral) aspects of the spinal cord and vertebral canal. Typically, a dorsal approach to the vertebral column is used (see previous description); however, use of the dorsolateral approach has also been reported.233

Once the vertebral column is exposed, the correct intervertebral disc space should be confirmed by counting from a known anatomic reference point. Often, the last rib is a good anatomic landmark. The presence (or absence) of a rib on each side should be confirmed via radiographic imaging before surgery. For caudal lumbar procedures, counting from the sacrum is useful. Radiographic screening for transitional vertebrae should be done, as this could alter counting. Occasionally, surgeons mark the dorsal spinous process of a specific vertebra with a small volume of dye (methylene blue) injected under the guidance of fluoroscopy before surgery.

Once the correct location has been identified, the articular processes are removed with rongeurs. For multilevel hemilaminectomies, all of the articular processes can be removed before drilling. Anatomic landmarks outlining the hemilaminectomy defect include the ventral aspect of the accessory process, the base of the dorsal spinous process, and the base of the articular processes at the cranial and caudal extent of the defect (see Figure 32-8). Modifications in the defect can be made depending on the location of the lesion. The authors recommend extending the bony defect ventrally to include the dorsolateral aspect of the vertebral body cranial and caudal to the affected disc space. This allows more ventral exposure and aids in removing herniated disc material from the ventral spinal canal (Figure 32-12).

High-speed pneumatic drill technique (see Figure 32-5, C) as used to gain access to the vertebral canal varies among surgeons. Generally, holding the drill with two hands, while resting at least one hand on a solid surface, is recommended to ensure drill stability. The outer cortex of the vertebral lamina should be removed along the entire length of the defect. Distinction between cortical and medullary bone can be made on the basis of color—cortical bone is white and medullary bone is red. This distinction is not prominent in smaller patients, especially cats, and may not exist at the level of the articular processes. Drilling continues until a thin layer of inner cortical bone remains. The final thin layer of bone, the joint capsule of the articular processes, and the yellow ligament can be removed with a spinal pick, a small curette, or rongeurs, as described previously (see Figure 32-12). Some surgeons prefer to remove all of the bone with rongeurs instead of using a high-speed drill, particularly in patients weighing less than 10 kg.

Variations of the hemilaminectomy include the mini-hemilaminectomy and the pediculectomy, which are meant to preserve the zygapophyseal joint, thus resulting in less mechanical instability. A pediculectomy (also called a partial pediculectomy) alone is performed when extruded disc material is identified over the body of one affected vertebra; advantages of the procedure are that the intervertebral foramen and associated blood supply are avoided, and the procedure is faster because less hemorrhage occurs. Technically, “mini-hemilaminectomy” is actually a combined pediculectomy, as the lamina of the vertebral bodies is preserved because the articular process is preserved. Disadvantages of mini-hemilaminectomy and pediculectomy include decreased exposure of the spinal cord, which may make orientation difficult for the less experienced surgeon, and the possibility of leaving disc material behind because of decreased exposure.147 Both of these procedures can be carried out via the dorsal or dorsolateral approach to the vertebral column.201 If accurate imaging is available (MRI), these procedures are likely to be successful because bone can be removed with precision over the site of disc herniation.

Dorsal Laminectomy

Dorsal laminectomy is indicated for access to the dorsal and/or lateral compartments of the spinal canal. Depending on the features of the lesion and the extent of bone removal required, various modifications of this technique can be used, including Funkquist A, Funkquist B, and modified dorsal laminectomy.68 These techniques have been named according to the extent of bone removal. Funkquist A involves removing the dorsal spinous process, laminae, articular processes, and approximately half of the pedicles of a vertebra to gain access to the spinal canal. It provides maximum exposure of the spinal canal.68 Funkquist B leaves intact the articular processes and pedicles, but removes the dorsal spinous process and laminae. A modified dorsal laminectomy is midway between Funkquist A and Funkquist B. For the modified dorsal laminectomy, the lamina, dorsal spinous process, and caudal articular process are removed. The pedicles are undercut to enhance exposure, but the articular processes remain intact (Figures 32-13 and 32-14).196

The dorsal approach to the vertebral column is necessary for any dorsal laminectomy technique. Careful counting to locate the precise surgical site should be done before the procedure is begun. Rongeurs are used to remove the dorsal spinous process to the level of the laminae of the target vertebra(e). Generally, a high-speed drill is used to remove the outer cortex of the lamina. Medullary bone can be identified by its red color; however, medullary bone is not present at the intervertebral space near the yellow ligament (interarcuate ligament). Drilling is carried out until a thin intercortical layer of bone remains. The interarcuate ligament will be present dorsally in the intervertebral space. Once the inner cortex is relatively thin, this ligament can be removed sharply to gain entry to the spinal canal. Rongeurs are used to remove remaining cortical bone and to expand the laminectomy defect. The extent of bone removal can be modified on the basis of location of the lesion and surgical goals. For instance, focal dorsal spinal cord decompression may be accomplished through a Funkquist B procedure, but access to more extensive neoplastic lesions may require more extensive bone removal.

As an alternative to a Funkquist procedure, thoracolumbar dorsolateral laminectomy with osteotomy of the spinous process has been reported.67 Indications for this approach included large lateral or dorsolateral lesions; the advantage was less muscle dissection. The approach as described earlier for hemilaminectomy was used. The fascia on the lesion side was opened in a paramedian fashion to avoid damage to the supraspinous ligament. The base of the spinous processes of the vertebrae and the cranial and caudal processes overlying the lesion in the sagittal plane were cut with an osteotomy or pneumatic drill. The osteotomized spinous processes were retracted laterally to expose the lamina, and a drill and rongeurs were used to remove the lamina. The articular processes at the center of the lesion on the operated side were completely removed and the laminectomy site extended ventrally to the floor of the spinal canal by unilateral excision of the vertebral arch, including the articular processes (facetectomy), the accessory process (foramenotomy), and the pedicle (pediculectomy). The ventrolateral aspect of the spinal cord was inspected by placement of stay sutures in the lateral and dorsal dura. With gentle traction, the spinal cord could be tilted to the left or right. A fat graft was not placed over the defect, and the dorsolumbar fascia and fat were closed routinely, with no attempt made to stabilize the osteotomized processes. Technique complications were not observed in 14 dogs at up to 4 years’ follow-up.

Intervertebral Disc Fenestration

Fenestration of the intervertebral disc196 has been suggested as a way to remove the nucleus pulposus to prevent its herniation into the spinal canal. This technique alone is not generally recommended as a treatment for disc extrusion because it does not allow spinal cord or nerve root decompression and does not enable the surgeon to remove herniated disc material. Currently, evidence does not support the use of this procedure prophylactically in dogs without a history of clinical disc herniation. Controversy remains regarding (1) whether fenestration reduces the likelihood of recurrence (at the same site) or a second episode (at another site) in dogs requiring surgical treatment for intervertebral disc herniation, and (2) which sites should be selected for fenestration. See further discussion later in this chapter.

A dorsal, dorsolateral, or lateral approach to the thoracolumbar spine can be used for this procedure. The lateral aspect of the annulus fibrosus is identified ventral and slightly cranial to the articular process. A scalpel blade or high-speed drill is used to create a window in the annulus (see Figure 32-10). Disc material can be removed with a small curette, dental scraper, or curved spatula. Every effort should be made to remove as much nucleus pulposus as possible while taking care not to injure surrounding structures. If fenestration is being performed concurrently with decompression, the spinal canal should be explored after fenestration but before closure to ensure that no disc material has been displaced dorsally into the spinal canal. Closure is routine. Complications associated with fenestration include discospondylitis, pneumothorax, iatrogenic damage to the spinal cord and nerve roots, and spinal instability.

Thoracolumbar Vertebral Column Imaging

Next to neuroanatomic localization, diagnostic imaging represents the most important step in determining the presence of clinical disc herniation in dogs. In human and veterinary medicine, an evidence-based approach has been advocated to assist clinicians in selecting appropriate imaging studies. Essential to this approach is an understanding of the concepts of sensitivity, specificity, prevalence, and accuracy, as well as an appreciation for the incidence of adverse effects associated with each modality. Additionally, interpretation of literature-based evidence relies on an understanding of appropriate study design and the limitations associated with gold standards.

Sensitivity and specificity determine how well an imaging study rules out and rules in disease, respectively.121 Sensitivity is defined as the number of positive animals identified by a test divided by the total number of animals with the disease (true-positives/true-positives + false-negatives). Essentially, higher sensitivity leads to a lesser chance of false-negative results. Specificity is defined as the number of animals without disease that test negative divided by the total number of animals without disease (true-negatives/true-negatives + false-positives). In other words, it is the ability of a test to correctly identify those individuals that do not have disease. Or, thought of another way, high specificity implies few false-positives. Prevalence is the proportion of animals in a population affected by a disease and will influence the accuracy and predictive value of a test. Accuracy is related to sensitivity, specificity, and prevalence and is best thought of as the proportion of correct (negative or positive) diagnoses a test provides (sensitivity*prevalence + specificity*[1-prevalence], or true-positive + true-negative/all animals).

Defining an appropriate gold standard for many vertebral column diseases is challenging. In studies of disc herniation, surgical confirmation is typically the gold standard but is biased by results of presurgical imaging.102 Reliance on necropsy as a gold standard may not be realistic, may affect associations with outcome data, and will bias the population toward those animals that were difficult to diagnose or were severely affected; necropsy is not necessarily more accurate than imaging. Finally, data have to be considered in light of study design. In some instances, group assignment is nonrandom, or the power of a study is insufficient to show clinically significant differences.

Traditional Imaging Modalities


Vertebral column radiography has been used in veterinary medicine since the early 20th century to diagnose thoracolumbar disc herniation. Radiographs are perhaps best used as a screening diagnostic to rule down vertebral fracture/luxation, discospondylitis, and other large osseous lesions. The diagnostic utility of radiographs in confirming the presence of thoracolumbar disc herniation is questionable. In a retrospective population of dogs with various thoracolumbar vertebral column disorders, the accuracy of radiographs for determining the presence of single-site disc herniation was 51% to 61% when myelography and surgery were used as the gold standard.120 A prospective study of dogs with surgical thoracolumbar disc herniation found radiography to be 57.1% accurate in determining the primary lesion site.162 Radiographs do not allow identification of the spinal cord or determination of the distribution of disc material within the vertebral canal.

Radiographic features of disc herniation include narrowing of the intervertebral disc space, wedging of the intervertebral disc space, increased articular process overlap, mineralized material within the intervertebral foramen or vertebral canal, and reduced intervertebral foramen diameter.79 Narrowing of the intervertebral disc space has the highest reported sensitivity for detecting disc herniation among these imaging features.120 Vacuum phenomenon, which is defined as gas opacity within the intervertebral disc due to degeneration, is uncommonly seen in dogs with disc herniation, but was a highly specific finding in one report.120 Radiographically evident spondylosis deformans does not appear to be associated with thoracolumbar disc extrusion and likely has only weak associations with disc protrusion.127


Myelography in human beings was first described by Sicard and Forestier in 1923.205 Widespread use did not occur in veterinary medicine until the technique was refined and popularized by Hoerlein in the 1950s.94 Myelography is performed by injecting iodinated contrast within the lumbar cistern or cisterna magna to facilitate radiographic identification of the subarachnoid space. Although myelography does not allow direct identification of the spinal cord, extradural compression can be recognized so long as the subarachnoid contrast column is obstructed. Animals with spinal cord edema or necrosis secondary to disc herniation may develop an intramedullary pattern. Intradural, extramedullary patterning is uncommon in disc herniation, but has been reported in one dog with disc extrusion that penetrated the dura mater.136

Myelography is believed to be an accurate means of diagnosing disc herniation in veterinary species. When surgical confirmation is used as a gold standard, the reported accuracy of myelography for identifying the correct site of thoracolumbar disc herniation in dogs is 85.7% to 98%.117,140,162 A more recent study suggests that the relative sensitivity of myelography for detecting thoracolumbar disc herniation is 83.6%.99 Myelography has been reported to correctly determine the lateralization of disc-associated compression, when compared with surgical findings, in 89% to 100% of dogs.22,74,117 Oblique views and paradoxical contrast column obstruction (predominant loss of the ventrodorsal contrast column contralateral to lateralized extradural compression) may assist in determining lateralization.22,74,117 Finally, myelography may provide prognostic information in dogs with thoracolumbar disc herniation lacking deep nociception; in this population, animals with an intramedullary pattern/L2 ratio ≥5 had a 26% recovery rate (defined as voluntary motor function at discharge) compared with a 61% recovery rate for dogs with ratios <5. Spinal cord swelling was measured by calculating the ratio of the length of loss of the myelographic contrast column to the length of the second lumbar vertebra.58 Intraparenchymal contrast uptake may indicate a guarded prognosis, as it was present in six of seven of dogs with ascending-descending myelomalacia secondary to thoracolumbar disc herniation.138

Data concerning the accuracy of myelography are typically generated in populations with surgically confirmed or necropsy-confirmed disc herniation. It must be remembered that the diagnosis of many vertebral column diseases with similar clinical signs as disc herniation may be challenging or impossible with myelography. For example, the presence of an intramedullary pattern is a nonspecific finding and may occur for a variety of reasons, including fibrocartilaginous embolic myelopathy, myelitis, disc herniation, and neoplasia. Also, myelographic abnormalities may be uncommon with certain disease processes. In one report, 74.1% of dogs with presumptive fibrocartilaginous embolic myelopathy had normal myelographic studies, and 25.9% had an intramedullary pattern.72

Myelographic studies are associated with patient adverse effects. These include seizures,13,135 myelopathy,135 apnea, cardiac arrhythmia,29 meningitis, subarachnoid hemorrhage,167 and death. Seizures appear to be a particularly common adverse event associated with myelography and are estimated to occur in 10% to 21.4% of dogs.13,135 Heavier dogs and dogs receiving cisternal contrast delivery seem at higher risk.13,135

Myelographic artifacts can complicate the recognition of clinically significant disc herniation.193 Dogs with compressive obstruction of the subarachnoid space may be challenging to inject and can have epidural contrast leakage. Subdural filling causes a drapery-like, thick appearance to the contrast columns and can preclude identification of compressive lesions. Subdural patterns are most commonly seen dorsally on lateral projections.171,193 Central canal filling (“canalogram”) can occur as the result of penetration of the spinal cord, leakage of contrast into the central canal, or existing communication between the conus medullaris and subarachnoid space.193 Central canal filling typically does not interfere with assessment of disc herniation but may be confused with intraparenchymal contrast uptake in the setting of a severe myelopathy. Central canal filling is not necessarily associated with a poor outcome, but may be associated with temporary worsening of neurologic deficits.118

Multiplanar Imaging Modalities

Computed Tomography (CT)

Computed tomography is a noninvasive, rapid, x-ray–based technique for generating multiplanar images. Bone detail is excellent, rendering this modality particularly useful in the setting of vertebral fracture or osseous neoplasia. Computed tomography is receiving increasing attention as a means to identify thoracolumbar disc herniation in dogs.87,99,164 Potential advantages of CT compared with myelography include speed of examination (median of 4 minutes with helical CT), avoidance of adverse effects associated with myelography, ability to visualize vertebral column structures in multiple planes, and enhanced detection of lateralized disc herniation.87,99,164 Imaging findings associated with disc herniation include loss of epidural fat opacity surrounding the compressed spinal cord, visible spinal cord compression, mineral-dense material within the vertebral canal, and soft tissue dense material within the epidural space consistent with hemorrhage.87,99,164 In all reported cases, extruded mineralized disc material causing compression has been hyperdense to spinal cord parenchyma.87,99,164

The sensitivity and accuracy of CT for determining the site of thoracolumbar disc herniation and the lateralization of disc material have been investigated in the setting of surgically confirmed disc herniation. As with myelography, certain diseases such as myelitis, neoplasia, and fibrocartilaginous embolism may be challenging to detect on CT and may present similarly to disc herniation. Thus, CT may not be the ideal means of imaging all dogs with vertebral column disease. In two prospective studies, including 20 dogs with acute, surgical thoracolumbar disc herniation, CT was 90%165 accurate and 94.7% to 100%87 accurate at determining the site of compression; myelography had similar reported accuracy. Lateralization of disc material was correctly appreciated in 96% of dogs using CT and 92% of dogs using myelography.165 A large retrospective study found that CT and myelography had similar relative sensitivity for detecting the site of disc herniation when performed as the primary imaging modality (81.8% and 83.6%, respectively).99 In that report, CT had significantly higher relative sensitivity compared with myelography in dogs with chronic lesions, but significantly lower relative sensitivity in small dogs.99 All dogs with disc herniation detected by CT had mineral-dense material within the epidural space.99

Magnetic Resonance Imaging (MRI)

Magnetic resonance imaging has largely replaced myelography and CT as the first-line vertebral column imaging technique in human beings.112 Similar to CT, MRI is noninvasive, generates multiplanar images, and can be performed reasonably quickly with appropriate field strength, receiver coils, and software algorithms. It offers the advantage, however, of superior soft tissue contrast, which facilitates identification of structural changes within the spinal cord, epidural space, and intervertebral disc (Figure 32-15).112 Unlike myelography and CT, MRI allows the classification of both disc degeneration and disc herniation in dogs.18 Disc herniation has been categorized as disc bulge, disc protrusion (Figure 32-16), dispersed disc extrusion, nondispersed disc extrusion, sequestered disc extrusion, and noncompressive nucleus pulposus extrusion.18,44 Clinical disc herniation usually results in visible spinal cord compression with loss of surrounding cerebrospinal fluid and epidural fat signal (Figure 32-17). Although disc material is hypointense on T1-weighted (T1W) and T2-weighted (T2W) images, associated hemorrhage has variable signal intensity that varies with the stage of hematoma evolution. Canine disc material may enhance contrast on postgadolinium T1W images.3,112,198 Short tau inversion recovery (STIR)- and fluid-attenuated inversion recovery (FLAIR)–weighted images may be helpful in revealing low-volume disc extrusion by subtracting signal from epidural fat and cerebrospinal fluid, respectively (Figure 32-18). In animals with high signal lesions on T2W images, T2* (gradient echo) images may show a signal void if the lesion contains hemorrhage (Figure 32-19). Single-shot turbo spin echo pulse sequences can be used for rapid image acquisition and appear to be an effective way to assess pathologic involvement of the subarachnoid space secondary to compression or other disease.170 MRI myelography may be another means to assess the subarachnoid space for disc-associated compression (Figure 32-20).

Data are not currently available that compare the sensitivity or specificity of MRI with other modalities for detecting disc herniation in veterinary species. Most studies of human beings concerning detection of disc herniation with MRI are older and do not reflect recent improvements in MRI field strength, coils, protocols, and software. In one report, MRI was roughly equivalent to CT-myelography for detecting lumbar disc herniation (82.6% versus 83%, respectively) when surgical exploration was used as the gold standard.150 In another study on human lumbar disc herniation, MRI was more specific but less sensitive than CT-myelography (86.5% versus 78.9% and 64.3% versus 72.8%, respectively) when surgical exploration was used as the gold standard.102 Magnetic resonance imaging is likely superior to other modalities for detecting many of the neoplastic, inflammatory, and vascular diseases that can mimic disc herniation, although supporting evidence is based on data from human medicine, limited veterinary data, and the experiences of those with access to all modalities.130 Data are available for fibrocartilaginous embolic myelopathy, in which 78.8% of dogs have recognizable MRI lesions that conform to a particular intramedullary signal pattern.45

Signal patterns within the spinal cord have been associated with initial neurologic grade and functional outcome in dogs with disc herniation and other vertebral column diseases.43,100,130 The presence of T2W hyperintensity within the spinal cord has been assessed in two reports of dogs with surgical thoracolumbar disc herniation.101,130 In a single-center prospective study,100 all dogs lacking T2W signal changes within the spinal cord recovered voluntary ambulation. In contrast, only 20% of dogs with T2W hyperintensity >3 times the length of the second lumbar (L2) vertebral body (T2W length ratio) returned to ambulatory status. A more recent multicenter, retrospective report130 supported these findings and demonstrated that the T2W length ratio was an independent predictor of functional outcome, with the odds of recovery reduced by 1.9-fold per unit of hyperintensity. In dogs with acute noncompressive nuclear extrusion addressed nonsurgically, a percent cross-sectional area of intramedullary T2W hyperintensity ≥90% had an 86% sensitivity and 96% specificity for determining poor long-term functional outcome.44 Dogs with percent cross-sectional area <90% had a 93% chance of returning to voluntary ambulation.44 A single-center retrospective study of dogs with thoracolumbar disc herniation did not find an association between the degree of spinal cord compression visible on T2W transverse images and severity of initial neurologic dysfunction or functional outcome.172

Cerebrospinal Fluid Analysis

With advances in imaging technology, many institutions no longer collect cerebrospinal fluid routinely in dogs with thoracolumbar spinal cord injury. This approach is understandable, as results of cerebrospinal fluid analysis are often not etiology specific, in-house analysis may not be possible, obtaining cerebrospinal fluid is invasive, and temporal delays are associated with acquisition.20,220,221 Nonetheless, the authors of this manuscript believe that cerebrospinal fluid analysis remains an important diagnostic tool in the evaluation of dogs suspected of disc herniation.

Several potential advantages are associated with routine cerebrospinal fluid collection in dogs with myelopathies. First, abnormalities identified on imaging studies may not always be indicative of clinically significant disease. For example, Axlund et al.7 demonstrated that in neurologically normal large-breed dogs, L7-S1 disc protrusion as assessed by CT is relatively common.7 And cadaveric studies in dogs have suggested that most necropsy-identified cervical disc herniations were not associated with neurologic impairment.86 Therefore, cerebrospinal fluid has the potential benefit of acting as an additional correlate to clinical and imaging findings. This is especially important in the setting of central nervous system inflammatory disease, where MRI does not always demonstrate signal changes within central nervous system parenchyma, and myelography is classically normal or less commonly shows nonspecific intramedullary patterning.81 Second, abnormalities on imaging might be consistent with several disease processes. For example, ventral extradural compression overlying several vertebral articulations can occur as the result of disc herniation, epidural empyema, hemorrhage, neoplasia, or other causes. In these cases, presurgical cerebrospinal fluid data may provide essential information that will alter the clinical approach. Third, intrathecal iodinated contrast delivery is contraindicated in dogs with inflammatory central nervous system disease, as it can result in clinical deterioration; therefore, cerebrospinal fluid analysis before myelography may be beneficial.228 It must be remembered, however, that increases in cerebrospinal fluid protein and nucleated cell count can occur in diseases that do not have a primary inflammatory basis, such as neoplasia, disc herniation, vascular myelopathies, and trauma.20,220 Finally, biomarkers within the cerebrospinal fluid can provide information that is prognostic or diagnostic across several causes, including disc herniation.

Cerebrospinal fluid can be acquired from the cerebellomedullary cistern or the lumbar cistern in dogs. In general, cerebellomedullary cistern cerebrospinal fluid is easy to obtain, free from blood contamination, and is reasonably easy to interpret, considering the extensive number of diseases for which data are available. Disadvantages of cerebellomedullary cistern cerebrospinal fluid include potential higher hazard associated with collection compared with lumbar cistern cerebrospinal fluid and location of the cerebellomedullary cistern cranial to the thoracolumbar spinal cord. The flow of cerebrospinal fluid within the subarachnoid space is rostral to caudal, and sampling cranial to pathology might be less likely to reflect abnormalities.220,221,228 Collection of cerebrospinal fluid from the lumbar cistern may be advantageous in that this site is caudal to thoracic and all but the most caudal lumbar lesions. Although iatrogenic injury is possible with lumbar cistern collection, in most instances nerve roots or the conus medullaris will be affected rather than the caudal brainstem, as is the case with cerebellomedullary cistern acquisition. Limitations associated with lumbar cistern collection include increased likelihood of blood contamination compared with cerebellomedullary cistern sampling, a small amount of sample volume available for analysis, and more limited published data associating findings with particular origins.

Several recent reports have described cerebrospinal fluid characteristics in dogs with thoracolumbar disc herniation (Table 32-1). When cerebellomedullary cistern cerebrospinal fluid is acquired, abnormalities in protein or nucleated cell count occur in 26% to 49% of cases.133,231 Pleocytosis (elevated nucleated cell count [>5 WBC × 106/L {WBCs/µL}]) is infrequently identified and, if present, is usually neutrophilic and of low cell count (<50 nucleated cells/µL). In dogs with thoracolumbar disc herniation, lumbar cistern cerebrospinal fluid has elevated protein concentration in 66% of samples and pleocytosis in 61% of samples, as compared with normal values for protein in lumbar cistern cerebrospinal fluid, which for this study were below 35 mg/dL.230 Similar to cerebellomedullary cistern cerebrospinal fluid, when pleocytosis is identified in lumbar cistern cerebrospinal fluid, it is typically of low cell count. However, lumbar cistern cerebrospinal fluid frequently contains a high proportion of lymphocytes, in contrast to what has been recognized in cerebellomedullary cistern cerebrospinal fluid.230 Abnormalities in cerebrospinal fluid protein and cell count appear to be more common in dogs with acute and severe neurologic impairment due to thoracolumbar disc herniation than was previously reported, making differentiation from inflammatory disease difficult without imaging.221

< div class='tao-gold-member'>

Jul 18, 2016 | Posted by in PHARMACOLOGY, TOXICOLOGY & THERAPEUTICS | Comments Off on Thoracolumbar Spine

Full access? Get Clinical Tree

Get Clinical Tree app for offline access