Experimental Studies

An acute traumatic injury to the spinal cord may leave varying amounts of physiologically viable tissue in an incomplete lesion that may progressively deteriorate due to secondary injury.^{44,53,54,80–83} The amount of viable tissue that remains and the ability of the tissue to function partly depend on the primary mechanism of the injury such as the characteristics of the initial force, as well as on the secondary mechanisms of injury caused by the combination of elevated levels of intracellular calcium, endogenous opiates, and inflammatory mediators, lipid peroxidation, edema, glutamatergic excitotoxicity, and mechanical factors producing continued compression and cord ischemia.^†
Several experimental models have been developed to determine how much injury the cord can tolerate and how long it can be compressed anteriorly and still recover neurologic function.^‡ Allen,²² in 1911, developed a standardized weight-drop model to produce graded spinal cord lesions in vivo. Pathologic evaluation of the injured cords revealed central hemorrhage and progressive necrosis that was thought to be the underlying mechanism behind neurologic deterioration in these dogs. Some dogs regained the ability to ambulate after myelotomy and evacuation of the hemorrhagic tissue. Other animal experiments have subsequently been done using the weight-drop method, and others have used dynamic loading with a piston device or a graded constriction method using clips, constriction bands, or balloons.* Collectively, these experiments suggest that surgical decompression, preferably early but also late, may result in varying degrees of neurologic recovery.

^†References 20, 21, 25, 27, 28, 46, 58, 82–85.

^‡References 22, 39, 44, 53, 54, 58, 80, 81, 86–97.

The duration of compression and timing of decompression are not the only factors affecting the potential for improved neurologic function in these animal studies. Other factors such as force of the initial compression, amount of cord displacement, and the kinetic energy of the applied load affect the severity of the injury. Dolan and colleagues⁹⁰ and Guha and colleagues⁹² used modified aneurysmal clips to constrict the T1 cord of rats with graded forces over different time periods. Clinical neurologic recovery varied exponentially according to the force of injury and linearly according to the duration of compression. They concluded that the major determinant of recovery was the initial force of injury. Although earlier decompression produced significantly more neurologic recovery, later decompression has the potential to salvage physiologically viable tissue that survived the initial injury and subsequent compression. Sommerson and Stokes⁹⁸ showed that compression injuries with less than 1-mm deformation of the spinal cord in rodents result in linear displacement of the cord relative to the applied force. This linear relationship becomes nonlinear if the cord displacement exceeds 1 mm. The result of this study suggests the possibility of a similar relationship between a traumatic compressive lesion and neurologic deterioration. Late decompression may, therefore, prevent further neurologic deficit owing to increasing cord deformation from progressive deformity or instability. Kobrine and colleagues^88,89 compressed the T6 spinal cord level in monkeys using an inflatable extradural balloon at different compression rates. Monkeys who had slow balloon inflation had return of function based on spinal cord evoked potentials even after several minutes of compression.⁸⁸ In contrast, monkeys that had rapid balloon inflation and acute cord compression showed return of response in 1 hour if compression lasted for 1 minute but no recovery up to 1 hour if compression lasted for 3, 5, 7, or 15 minutes.⁸⁹ Although this study can be criticized for not checking responses beyond 1 hour, the authors concluded that recovery of spinal cord function depended on the rapidity and length of time of compression. Therefore lower energy injuries or progressive deficits due to cord compression after an acute injury may have potential for neurologic recovery even with late decompression.

Bohlman and colleagues^53,54 developed an animal model to demonstrate that late decompression of a traumatic incomplete spinal cord injury that occurred after either cord contusion or compression may result in neurologic recovery. Incomplete paralysis was produced in eight beagles by remote control of a pressure transducer (Figs. 79-1 to 79-4) and maintained for 3 to 8 weeks until recovery plateaued. After cord compression was relieved, all animals recovered completely by 10 weeks after injury. Recovery was faster in the five with complete decompression versus the three with lesser degree of ongoing compression from the transducer. Microscopic examination was normal in two dogs, central gray necrosis in two, peripheral demyelinization in two, and lacerations in three. The weight-drop method was used to contuse the cord in seven dogs using 257 to 722 gm/cm of energy. Variable levels of quadriparesis occurred, and all dogs recovered by 6 weeks after injury. Thirteen dogs developed central cord syndromes (forepaw paralysis greater than hindpaw) with either method. Central gray matter damage was the most common finding. The gray matter seemed more sensitive to deforming forces of external degeneration of the ascending sensory posterior columns, whereas others showed damage of the anterior and lateral corticospinal tracts, indicating descending wallerian degeneration. However, histopathologic damage of the spinal cord did not always correspond to the degree of clinical paralysis in these dogs. For example, there was normal histology in three dogs, two with quadriplegia for 1 to 6 weeks, and one with grade 3 paralysis. Because these dogs recovered neurologic function after contusion and compression, it appeared that deformation of the cord produced a physiologic block but the cord remained viable. In the contusion model, greater contusion forces caused proportionately increased neurologic damage and recovery time, whereas in the compression model anterior decompression allowed for faster and more complete recovery.

FIGURE 79–1 Newer design of pressure transducer with solid polyethylene piston and incorporated electrical pressure sensor to measure pressure directly applied to the spinal cord. The transducer piston is remotely controlled and monitored after insertion into the anterior cervical spine.

(From Bohlman HH, Bahniuk E, Raskulinecz G, Field G: Mechanical factors affecting recovery of incomplete cervical spinal cord injury: A preliminary report. Johns Hopkins Med J 145:115-125, 1979.)

FIGURE 79–2 A, Intraoperative photograph of the site of the laminectomy of the sixth and seventh lumbar vertebrae showing the constriction band with 50% constriction of the cauda equina. B, Illustration of the photograph shown in A. The constriction band was always applied at the seventh lumbar level, and the roots of the seventh lumbar nerve were included in the constriction. C, Illustration of the canine cauda equina (dura removed). Note that the conus medullaris extends to the level of the fifth and sixth lumbar vertebrae and that the cauda equina begins at the sixth lumbar vertebra. D, Nylon constricting band. Each catch-ratchet is 0.35 mm apart, allowing precise measurements of constriction.

(From Delamarter RB, Bohlman HH, Dodge LD, Biro C: Experimental spinal stenosis. J Bone Joint Surg Am 72:111, 1990.)

FIGURE 79–3 Cortical evoked potentials after bilateral stimulation of the posterior tibial nerve in a dog that had 50% constriction. Increased latency and decreased amplitude occurred immediately after operation and 1 month postoperatively, with substantial recovery by the second postoperative month (amplitude axis, 2.5 µV per division).

(From Delamarter RB, Bohlman HH, Dodge LD, Biro C: Experimental spinal stenosis. J Bone Joint Surg Am 72:113, 1990.)

FIGURE 79–4 A dog that had 50% constriction. A, Note the hourglass appearance, with excellent filling of the vessels, except for the peripheral radicular vessels. B, The root of the seventh lumbar nerve, just caudal to the constriction band, shows venous distention and a curlicue pattern. The large central radicular artery is normal. C, The dorsal root ganglion of the seventh lumbar nerve shows vascular congestion and leakage of dye.

(From Delamarter RB, Bohlman HH, Dodge LD, Biro C: Experimental spinal stenosis. J Bone Joint Surg Am 72:114, 1990.)

Clinical Studies

Although there are likely significant differences between the mechanical and biologic properties of the spinal cord in human versus animals, the animal studies suggest that the characteristics of the initial force and the amount and duration of cord compression alter neurologic recovery. Persistent compression may cause progressive mechanical injury and diminished blood flow to the cord in the region of the anterior spinal artery, thus preventing improvement with nonoperative treatment. Early decompression of incomplete cord injury restores blood flow and provides faster and more complete recovery, but late decompression can achieve varying degrees of the same result in cases of residual compression. Decompression in the region of the lesion in a patient with an incomplete spinal cord injury may provide functional recovery of motor roots and the spinal cord at that level and below. Patients with complete cord injuries are not expected to improve distal neurologic function from late decompression, but they may recover nerve roots near or above the level of injury and prevent ascending nerve root loss. The majority of clinical studies that have tried to determine the optimum timing of surgery for traumatic spinal cord injury are retrospective^{16,31–40,43,48,49} with only a few prospective studies.^{16,17,40,43,99} These studies have inconsistent definition of early versus late surgery that may have contributed to conflicting findings in which some studies found a benefit from early decompression and stabilization, whereas others found no advantage. However, if 4 weeks is considered late, there are studies, albeit retrospective, that have shown reversal of neurologic deficits in incomplete lesions.*

^‡References 6, 14, 15, 38, 39, 55–57, 59–67, 39.

Cervical Spine Lesions

Bohlman and colleagues^15,55 demonstrated that later anterior decompression can result in reversible neurologic function in patients with incomplete cord injuries (Fig. 79–5). In one study, 58 patients with incomplete cord injuries underwent late anterior cervical decompression and fusion with iliac crest autograft an average of 13 months (range, 1 month to 9 years) from injury. Twenty-nine patients who could not walk preoperatively became functional ambulators, and six patients who could walk before surgery had improvement in the ability to walk. Thirty-nine patients had improvement in motor root function in the upper extremities. Improvement was less in the patients who had decompression more than 12 months after injury.

FIGURE 79–5 A, Lateral intraoperative radiograph of a man who sustained an axial loading compression fracture of C5 with bone protrusion into the spinal canal. Initially he had an anterior cord syndrome with sparing only of perianal-sacral sensation and a one-sided toe wiggle. In addition, he had spared position and vibratory sense. He was treated conservatively for 18 months and recovered only minimal function in one leg. B, Lateral myelogram demonstrating anterior compression of the dural sac at C5-6. C, Anteroposterior myelographic view demonstrating marked central compression of the spinal cord in the dye column. D, Lateral radiograph of the patient 6 years and 5 months after surgery. Note the incorporation of the anterior iliac bone graft at the level of decompression. E, The patient recovered almost completely with slight residual left arm weakness. He was able to walk with a cane and had normal bladder, bowel, and sexual function. This case demonstrates the feasibility of late anterior decompression for an incomplete spinal cord injury and potential neurologic recovery.

In a similar study by Benzel and Larson,⁶¹ 41 of 51 (81%) patients with incomplete myelopathies and whose neurologic function had plateaued had improved spinal cord function after decompression and fusion as late as 115 and 173 days in 2 patients. Among eight patients with motor complete lesions preoperatively with no useful function, six regained useful function, with four able to walk with assistance. The return of neurologic function in these patients began promptly after decompression, suggesting a cause-and-effect relationship. These studies suggest that late anterior decompression and stabilization can result in reversible neurologic deficits of both upper and lower extremity roots and distal cord function in the lower extremities.

Reduction of pain and restoration of even a single functional nerve root level may greatly improve the quality of life in complete quadriplegic patients. Return of sensory perception improves the chances of preventing pressure sores, and arm functional recovery of one or more levels may allow independent ambulation. Several studies^{38,39,55–57,61,62,69} have shown that these are realistic goals even when decompression is done as late as after 21 years.⁶² Patients with complete cord injuries have far less predictable and dramatic recovery. The mechanisms of recovery are poorly understood, but it appears that the injury falls along a continuum of an incomplete lesion where spinal cord fibers are anatomically but not functionally intact. Dimitrijevic¹⁰¹ proposed the term discomplete quadriplegia to describe this concept. Multiple factors including persistent compression, instability, and progressive deformity may diminish the healing environment of the cord and cause secondary injury. Nevertheless, decompression appears to have a limited but positive effect on recovering function of remaining viable motor roots and cord tissue.

Anderson and Bohlman⁶⁹ performed anterior cervical decompression and fusion with iliac crest autograft in 51 patients with complete motor quadriplegia in an attempt to gain improvements of the motor roots in the upper extremities and thereby to improve function. Five patients died in less than 1 year. The average time from injury to decompression was 15 months (range, 1 month to 8 years), and 46 patients were followed for an average of 5 years (range, 2 to 13 years). Neurologic improvement of at least two new functional motor root levels occurred in 7 patients, and 18 patients had one-level recovery. Six patients had increased motor strength. Therefore anterior decompression resulted in neurologic improvement in 67% of the patients. Results were better in patients younger than age 53 years, those who had decompression less than 18 months after injury, and those who had neurologic levels cephalad to the site of skeletal injury rather than those with functioning motor roots at or caudad to the level of injury.

Benzel and Larson⁶¹ reported on 35 patients with complete quadriplegia who also underwent late surgery to improve neurologic function. Twenty-five patients underwent anterior decompression with or without posterior fusion and foraminotomy, whereas 10 had posterior fusion alone. Recovery of nerve root function only occurred in the patients who had anterior decompression (15 of 35 patients). No patient with both motor and sensory complete lesions recovered long tract function. However, of the 11 patients with motor injuries and sensory perception, 5 improved significantly and 2 were able to walk. It is highly unusual to have return of long tract motor function in patients with motor complete injuries, but it highlights the potential for late decompression to allow the cord maximal opportunity to recover neurologic function.

Thoracolumbar and Cauda Equina Lesions

The factors affecting neurologic recovery in the thoracic spine are the same as in the cervical spine. Early anterior decompression in a stable patient is the preferred treatment to maximize the opportunity for improved motor root and cord function and pain relief, but late decompression is also beneficial.

Benzel and Larson⁶¹ described the results of anterior decompression and fusion via the lateral extracavity approach in 105 patients with thoracic and lumbar spinal cord injuries. There was significant improvement in the neurologic status of patients with incomplete lesions. Four of 10 patients with motor complete injuries with some sensory perception improved neurologically, with one able to ambulate. The anterior approach was deemed superior to a nonoperative or posterior stabilization approach for anterior compressive lesions.

Similarly, Maiman and colleagues⁵⁶ reported the cases of 20 patients with incomplete thoracolumbar spinal cord injuries who underwent late anterior decompression via a lateral extracavity approach. Operations were done on average at 23 months and as late as 5 years after the initial injury or prior surgery. Thirteen patients had a mass in the canal, and seven patients had kyphotic deformities. Seventeen patients had substantial neurologic improvement. All but three of the nonambulatory patients with canal masses and the seven patients with kyphosis became ambulatory. The authors concluded that anterior decompression with appropriate stabilization affords the best opportunity for neurologic improvement in thoracolumbar spinal cord injuries.

Injuries to the cauda equina and conus medullaris with ongoing compression may also benefit from late decompression. The peripheral nerve roots of the cauda equina are fairly resilient, and conus medullaris injuries have significant potential for neurologic recovery, especially bladder control after decompression.^57,59 Bradford and McBride⁶⁴ studied 59 patients with incomplete deficits after thoracolumbar and cauda equina injuries. Eighteen patients who had anterior decompression were treated an average of 15.8 months (range, 1 to 79 months) after their injury. Anterior decompression was shown to be superior to posterior decompression, with 88% versus 64% neurologic improvement, respectively.

Although laminectomy is rarely indicated in spinal cord injuries, Ramamurthi and colleagues⁶⁶ observed improvement in bladder function in 14 of 27 patients and motor function in 8 of 27 patients who had decompressive laminectomy between 6 and 12 weeks after conus and cauda equina injuries. Compression was due to displaced lamina, subarachnoid adhesions after traumatic fractures involving T12 to L4.

Late Surgical Procedures

Neural Compression, Deformity, and Pain

Patients may present late with neural compression, deformity, and pain because of failure to adequately stabilize a fracture, decompress the nerve roots or cord, or reduce dislocations. Despite controversy between the benefits of early versus late surgery, anterior decompression and stabilization leads to improved neurologic recovery over that expected with either a conservative, nonoperative approach or a posterior stabilization-only approach in patients with anterior compressive lesions.* Laminectomy is contraindicated in most instances because it does not adequately decompress the spinal cord when the compression is anterior and predisposes the patient to spinal instability and further neurologic deterioration.^55,59,105

In the cervical spine, anterior cervical discectomy and interbody fusion using the Robinson technique safely and reliably treats central and lateral disc herniations.^8,110 Vertebral body burst fractures with cord compression due to malunion or kyphosis respond well to corpectomies via an anterior approach. Supplemental posterior fixation may be required depending on the associated posterior osseous and ligamentous injuries.

An anterior approach is also indicated in patients with a healed malunion of the odontoid or axis body fracture and anterior cord compression with myelopathy. We prefer the extrapharyngeal¹¹¹ approach to the transoral¹¹² approach with supplemental posterior C1 to C3 arthrodesis (Fig. 79–6). Alternatively, if the posterior atlantodental interval is less than 13 mm, it may be treacherous to fuse the axis to the atlas by passing instruments beneath the atlas posteriorly. This maneuver risks further neurologic injury that could be fatal. Instead, these patients can be treated with laminectomy of the arch of the atlas followed by arthrodesis between the occiput and C3. Use of C1 lateral mass screws and C2 translamina screws offer alternative fixation techniques as well (Fig. 79–7).

FIGURE 79–6 A, Lateral radiograph of an old C1-2 dislocation secondary to an odontoid fracture. The patient had developed syncopal episodes and drop attacks. B, Posterior fusion and wiring were carried out. However, in the recovery room it was noted that the patient was quadriparetic with a central cord syndrome. The subluxation had recurred, and the fusion was allowed to heal in the displaced position. C, Two years after the original procedure the patient had recovered to ambulatory status but remained with a Brown-Séquard syndrome and left hemiparesis. A cervical myelogram revealed severe compromise of the spinal cord at the C1-2 level. This necessitated a transoral anterior decompression, as well as a posterior decompression by removal of the arch of the atlas and a portion of the posterior fusion as separate procedures. The patient eventually regained some motor function. D, The transoral microscopic decompression was inadequate, and the patient continued to have pain without major recovery of neurologic function. This necessitated a retropharyngeal anterior resection of the vertebral body of C2 and iliac fusion from C1 to C3. This lateral tomogram after surgery demonstrates the adequacy of anterior decompression and the incorporated iliac strut graft. E, Lateral radiograph 2 years after retropharyngeal decompression demonstrating the totally incorporated iliac graft. The patient had recovered further neurologic function, and most of his pain was relieved. F, Key features of the extramucosal anterior approach to the atlas and axis. Through a submandibular incision on the right side the submandibular gland is resected and the digastric tendon divided. It should be noted that the lower limb of the incision is not used unless concomitant exposure to the midcervical levels is required. G, After the superficial layer of the deep cervical fascia is incised along the anterior border of the sternocleidomastoid muscle, the superior thyroid artery and vein are divided. The hypoglossal and superior laryngeal nerves are mobilized. Additional branches of the carotid artery and internal jugular vein are ligated to allow mobilization of the contents of the carotid sheath laterally as the hypopharynx is mobilized medially. H, The first step of the anterior spinal decompression is a meticulous removal of the disc between the second and third cervical vertebrae. The longus colli muscle is dissected in a lateral direction, exposing the second cervical vertebral body and the anterior arch of the atlas. Removal of the body of the second cervical vertebra can then be performed with a high-speed burr. I, The most frequent non-neoplastic cause of compression at the level of the second cervical vertebra is rheumatoid pannus, associated with atlantoaxial subluxation and basilar invagination of the odontoid process. The goal of the anterior decompression and stabilization is to provide an arthrodesis from the clivus to the subaxial spine.

(F to I from McAfee PC, Bohlman HH, Riley LH Jr, et al: The anterior retropharyngeal approach to the upper part of the cervical spine. J Bone Joint Surg Am 69:1374-1379, 1987.)

FIGURE 79–7 A and B, Flexion and extension lateral radiographs of a 65-year-old male who presented to the emergency department with loss of leg strength and ability to ambulate. He had spasticity and clonus in his lower extremities. Thirty years ago he had an odontoid fracture treated in a halo. C, Sagittal T2-weighted magnetic resonance imaging view shows myelomalacia of the spinal cord and the odontoid nonunion. D and E, Anterior and lateral plain radiographs show C1-C2 stabilization with fixation in the C1 lateral masses and C2 lamina.

Patients presenting with late root compression from an unreduced cervical facet subluxation, dislocation, or fracture are best approached posteriorly. A burr is used to remove the superior articular facet to assist foraminotomy and reduction followed by posterior arthrodesis.

Thoracic and midlumbar neural compression are treated with the same principled approach as in the cervical spine. The site of neural compression dictates the approach, with anterior decompression preferred when there is proven anterior pathology.^{59,104,105,110} The deformity pattern and the osseous and ligamentous pathology dictate stabilization procedures. Treatment can include a single anterior¹¹³ or posterior^114–116 approach, a combined anterior and posterior approach, and a simultaneous anterior and posterior approach with the patient in the lateral decubitus position.¹¹⁷

Late pain and paralysis may occur after thoracolumbar fractures with canal compromise in patients with or without neurologic deficits at the time of their injuries.^57,62,107 Bohlman and colleagues⁶² performed anterior decompression of 45 patients with late pain and paralysis at an average of 4.5 years (range, 3 months to 21 years) after the initial injury. Twenty-one of the 25 patients with neurologic deficits noted neurologic improvement. Patients with less than severe paralysis had greater recovery. Forty-one of the 45 (91%) patients noted pain relief, with complete relief in 30 and partial in 11. In a similar study, Transfeldt and colleagues⁵⁷ noted pain relief in 83% of their population after delayed anterior decompression of thoracolumbar and cauda equina injuries. In rare instances, isolated intercostal pain without thoracic cord compression may respond to posterior hemilaminectomy and foraminotomy to decompress the nerve root. At the L4 and L5 levels, the anterior retroperitoneal approach may be technically challenging, so a laminectomy and foraminotomies may be performed and the dural sac gently mobilized to remove bone or disc material.

Spasticity

Uninhibited muscle contractions below the level of injury in spinal cord–injured patients lead to spasticity of the lower abdominal wall, extremities, or bladder. Surgery aims to reduce the amount of local sensory input to the lower motor cells that have lost their regulatory input from the central nervous system (Fig. 79–8). The most common operation is a dorsal “T”-shaped myelotomy through the two dorsal columns to disrupt the reflex arc between the entering sensory axon as it traverses to the motor neuron. This can be selective to counteract the abdomen, extremity, or bladder. Some surgeons prefer a percutaneous dorsal rhizotomy. Patients choosing surgery for a spastic bladder should undergo preoperative urologic evaluation and be counseled on the possibility of losing erectile function or reflex bladder emptying. The spastic bladder can be controlled by sectioning the second, third, and fourth sacral segments or only the second and third sacral nerve roots in selected patients. Sphincterotomy before the neurosurgical procedures may relieve the bladder symptoms.⁷⁰

FIGURE 79–8 Lateral radiograph of a 16-year-old boy who sustained an axial loading compression fracture of C5 in a diving injury. Initially he had a severe anterior cord syndrome with poor motor function. B, The patient underwent an anterior fusion using Caspar plates without decompression. He recovered some motor function but by 6 months was unable to ambulate. This lateral cervical myelogram demonstrates continued compression of the cervical spinal cord. In addition, he was extremely spastic. C, Enhanced computed tomography scan demonstrates continued bone compression of the spinal cord and some dye in the central portion of the spinal cord. D, The patient underwent removal of the Caspar plates and corpectomy of C5 with iliac fusion from C4 to C6. Immediately postoperatively, cervical magnetic resonance imaging (MRI) reveals a central contusion or myelomalacia of the spinal cord. E, Axial MRI reveals central myelomalacia with a surrounding shell of the spinal cord. F, Postoperative lateral radiograph shows incorporation of the iliac graft. The patient has become fully ambulatory with the aid of crutches.