Pathogenesis and Pathologies

In spinal concussion, there is a functional rather than pathologic disturbance, usually with return of function within hours [Tator, 1996]. Extravasation of potassium from neurons into the extracellular space, after direct injury of nerve cell membranes or secondary nerve cell damage from vascular disruption, is the most likely mechanism of spinal cord concussion. Ischemia of the spinal cord can occur from anterior spinal artery or other vascular compression during spinal cord trauma, but also can occur when the trauma does not involve the spinal cord directly. Hypotension, systemic shock, and vascular injury to the aorta or a vertebral artery are the most common causes of this type of injury. With compression of the anterior spinal artery, the anterior two-thirds of the spinal cord is affected, causing weakness and loss of pain and temperature perception, with preservation of proprioception and light touch. Contusion results from a blunt injury of the spinal cord without continued compression or disruption of nerve tissue, resulting in transient paresthesias and dysesthesias in the upper limbs, especially the hands (burning hands syndrome), accompanied by transient long tract signs. Recovery is incomplete, however, distinguishing contusion from concussion. Central cord necrosis (hemorrhagic or ischemic) is a common sequela that often can be seen on magnetic resonance imaging (MRI) [Bailes et al., 1991; Jellinger, 1976]. Compression of the spinal cord can be caused by vertebral dislocation or by pressure from bony fragments, a herniated disc, or extramedullary bleeding [Jellinger, 1976]. Laceration occurs when there is interruption of spinal cord tissue by bony fragments, knife or bullet wounds, fracture-dislocation, or severe stretching. In older children, the most frequent cause of laceration is a penetrating agent, whereas in newborns, hyperextension of the neck during breech delivery is the most likely cause [Allen et al., 1969; Bresnan and Abroms, 1974; Byers, 1975; Crothers, 1923]. Intramedullary hemorrhage, which can extend over several spinal cord segments (hematomyelia), is believed to result from direct injury to blood vessels or from altered vascular permeability, as from acidosis.

The pathology of intraspinal intramedullary injuries has been studied in detail in experimental animals and in human postmortem examination material [DeLaTorre, 1981; Dumont et al., 2001a; Jellinger, 1976; Kakulas, 1984; Norenberg et al., 2004; Park et al., 2004; Profyris et al., 2004]. To mimic most events that lead to various forms of human spinal cord injury, several experimental models have been developed, the most common being transection, compression, or contusion [Rosenzweig and McDonald, 2004]. Because there are striking pathophysiologic similarities between human spinal cord injury and experimental models of spinal cord injury, particularly in the rat, findings in experimental spinal cord injury are commonly extrapolated to human injury. Such extrapolation requires caution, however, because regulation of secondary events after spinal cord injury varies greatly among different animal species and strains [Profyris et al., 2004]. The pathologic changes observed in the spinal cord after trauma can be subdivided into early, late, and very late.

The pathophysiology of spinal cord injury is biphasic. First is the primary injury, which is mechanical (e.g., spinal fracture and/or dislocation), directly disrupting axons, blood vessels, and cell membranes. This is followed by the delayed onset of secondary injury, involving vascular dysfunction, edema, ischemia, excitotoxicity, electrolyte shifts, free radical production, inflammation, and delayed apoptotic cell death. While neurologic deficits are seen immediately following the initial injury, the secondary injury phase results in a protracted period of tissue destruction [Rowland et al., 2008].

Almost all of our detailed understanding of spinal cord injury pathophysiology has been obtained from animal studies. Important to understanding the pathophysiology of human spinal cord injury is that each injury is unique in cause and resultant damage. Further, the contrast between the homogeneity of experimental spinal cord injury models, as in widely used rodent models, and the highly heterogenous human spinal cord injury population adds to the challenge of successfully translating treatments to the clinic.

Phases of secondary injury

Secondary injury can be divided temporally into multiple contiguous phases: immediate, acute, subacute, intermediate, and chronic stages of spinal cord injury.

Immediate Phase (0–2 Hours)

During the immediate phase of spinal cord injury, which can last for up to 2 hours, traumatic severing of axons, the immediate death of neurons and glia, and accompanying spinal shock result in immediate loss of function at and below the level of complete injury. Pathological changes consist of swelling of the spinal cord, often with hemorrhage in the central gray matter and surrounding white matter, and necrosis (Figure 78-5) and spinal cord ischemia extending for many segments rostral and caudal to the injury site. Activation of microglial cells begins almost immediately, with the upregulation of the proinflammatory cytokines tumor necrosis factor (TNF)α and interleukin (IL)-β, detectable within minutes of injury. Also within minutes, levels of extracellular glutamate can reach excitotoxic levels.

Fig. 78-5 Severe traumatic injury to the spinal cord with secondary hemorrhage and coagulation necrosis.

Acute Phase (2–48 Hours)

It is in this phase that the secondary injury processes become dominant. For practical reasons, it is the phase likely to be most amenable to neuroprotective interventions, as it is typically the earliest point at which patients arrive at an appropriate treatment center.

The acute phase of spinal cord injury lasts from 2 to 48 hours. It is characterized by continuing hemorrhage, increasing edema and inflammation, and the onset of additional secondary injury processes, including free radical production, ionic dysregulation, glutamate-mediated excitotoxicity, and immune-associated neurotoxicity, which cause vascular disruption, hemorrhage, and resulting ischemia, with further axonal injury and cell death. The ischemia results in cytotoxic cell swelling, affecting neurons and glia, and axonal swelling. Loss of ion homeostasis immediately following spinal cord injury and excitotoxicity both contribute significantly to the propagation of cell injury after spinal cord injury. Ion dysregulation is also a central feature of necrotic and apoptotic cell death. Excitotoxicity results from excessive activation of glutamate receptors and is believed to play a role in the death of neurons and glia. Production of free radicals gives rise to lipid peroxidation, contributing to axonal disruption and death of neurons and glia. The modest protective effect of methylprednisolone treatment following acute spinal cord injury is believed to be due, at least in part, to the inhibition of lipid peroxidation [Kwon et al., 2004]. An additional acute-phase change is a marked increase in permeability of the blood–brain barrier. Death of neurons at all stages of injury occurs mainly from necrosis; by contrast, oligodendroglia readily undergo apoptosis. Loss of oligodendroglia results in axonal demyelination; persistent demyelination and other forms of axonal injury can then cause death of associated cell bodies. Animal studies have provided strong evidence that spared demyelinated axons are present following contusive spinal cord injury, and represent an important therapeutic target for spinal cord injury treatments that either remyelinate (cell transplants) or improve axonal conduction in demyelinated fibers (4-aminopyridine) [Hayes et al., 2004; Nashmi and Fehlings, 2001]. On the other hand, postmortem human studies and a limited number of animal studies [Lasiene et al., 2008] have not convincingly shown significant numbers of spared demyelinated axons to be present after human spinal cord injury [Kakulas, 2004; Norenberg et al., 2004].

Subacute Phase (2 Days to 2 Weeks)

The subacute phase of spinal cord injury lasts from 2 days to 2 weeks, and is the time period in which future cell-based therapeutic strategies are most likely to be applied. Transplantation of adult neural precursor cells 2 weeks following spinal cord injury in rats promotes remyelination and functional recovery, whereas cells transplanted at chronic time points either fail to survive, or fail to migrate and differentiate to promote functional recovery. A phagocytic response is maximal during the subacute phase; it helps to remove cell debris and may promote axon growth. An astrocytic response begins in the subacute phase, when astrocytes at the periphery of the lesion become hypertrophic and proliferative, and form an astrocytic (gliotic) scar. This scar presents both a physical and chemical barrier to axonal regeneration. Despite the production of a scar, astrocytes also serve beneficial functions following spinal cord injury, including re-establishment of ion homeostatis and reestablishing the integrity of the blood–brain barrier.

Intermediate Phase (2 Weeks to 6 Months)

The intermediate phase of spinal cord injury lasts from 2 weeks to 6 months. It is characterized by the continued maturation of the astrocytic scar and by regenerative axonal sprouting.

Chronic Phase (6 Months or More)

The chronic phase of spinal cord injury begins 6 months after injury and continues through the patients lifetime. Scar formation continues, and the central portion of the spinal cord is replaced by a cystic cavity extending for several segments above and below the point of impact, resulting in post-traumatic syringomyelia; this extension can continue over many years, significantly increasing the patient’s neurological deficit, including ascending paralysis, brainstem symptoms, and neuropathic pain [Barnett and Jousse, 1976; Profyris et al., 2004; Rowland et al., 2008; Tator, 1996]. Wallerian degeneration of ascending and descending tracts occurs within 6–12 months [Kakulas, 1999], and nerve roots at the level of the lesion become atrophic. The cord becomes thin in the area of injury, with fibrosis and thickening of the overlying meninges [Kakulas, 1984]. At this chronic phase, therapeutic strategies are directed to encouraging regeneration/sprouting of disrupted axons, promoting plasticity with rehabilitative strategies, and improving function of demyelinated axons. Despite much success with stem cell-based approaches when applied subacutely, use of stem cells during this chronic phase of injury has not been shown to be successful [Rowland et al., 2008].

Pathogenesis and Pathologies

The mode of spinal cord injury from intraspinal extramedullary conditions is primarily by spinal cord compression, although secondary vasospasm and ischemia also play a role. The causative pathologies include hemorrhage, infection, cystic fluid, and solid tissues. The degree of resultant injury to cord depends on the location and size of the mass and the duration of cord compression. Post-traumatic intraspinal extramedullary mass lesions occur relatively infrequently and, when present, are often small, causing little, if any, cord compression. Included among these disorders are spinal epidural, subdural, and subarachnoid hemorrhages; spinal epidural abscess; arachnoid cyst; epidermoid tumor; herniation of the nucleus pulposus; and cauda equina injuries.

Radiographic Evaluation

Radiographic evaluation of the child suspected of having spinal cord injury should begin in the emergency department with plain films of the entire spine. The purpose of the evaluation is to look for evidence of unstable fractures or vertebral dislocations that may require emergency surgical intervention. Radiography of the entire spine is important because approximately 15 percent of patients have injury at multiple sites [Hadley, 1992; Hadley et al., 1988]. Radiological evaluation of the spine is often neither possible nor desirable in an emergency situation, however, and may need to be deferred until the spine is stabilized. In such circumstance, only the spinal area clinically involved needs immediate imaging. In cases in which the patient cannot be assessed adequately by plain radiography, MRI may be required early in the evaluation. Stringer and Andersen suggested a diagnostic algorithm for spinal injury combining the findings on neurological examination and results of radiographic studies [Andersen and Stringer, 1996; Stringer and Andersen, 1996].

Children with spinal cord injury routinely have sandbag and tape fixation of the head and neck, making optimal radiographic examination more difficult. Nonetheless, it is crucial to obtain anteroposterior, lateral, and oblique views of the spine. All seven cervical vertebrae and the C7–T1 interspace must be visualized. Swimmer’s views are often necessary to visualize all of the lower cervical vertebrae, and an open-mouth view may be needed to visualize the odontoid process [Goldberg et al., 1990], although the open-mouth view can be safely omitted if the other views are normal [Dias, 2004]. Flexion/extension cervical x-rays and fluoroscopy are sometimes indicated to exclude ligamentous instability when there is still suspicion of cervical spine instability after static x-rays have been obtained [Hadley, 2002b]. Care must be taken not to misinterpret normal variations in cervical spine films of a child. These variations include pseudosubluxation of C2 on C3 (seen in 20 percent of normal children age 1–7 years) and the cartilaginous plate at the base of the odontoid (normally seen in children <3 years old and often mistaken for a recent fracture) [Cattell and Filtzer, 1965].

In spine-injured children, the upper cervical spine is especially susceptible to injury. Specific injuries to this region include:

Occipitoaxial dislocations can be identified on plain radiographs in some patients using the Powers ratio (distance of the basion [midpoint of the anterior border of the foramen magnum] to the posterior arch of C1, divided by the distance of the opisthion [midpoint of the lower border of the foramen magnum] to the anterior arch of C1). This ratio is usually 0.77. Ratios greater than 1 are abnormal. Most often, MRI is needed to identify occipitoaxial dislocations. Instability of the atlantoaxial region is suspected when the preodontoid space exceeds 5 mm in a child or 3 mm in an adult. Fractures of the atlas often can be stabilized by using a rigid (Philadelphia) collar [Alker et al., 1975]. Odontoid fractures and hangman’s fractures are rare in childhood and, if present, usually can be treated nonoperatively (halo brace) [Anderson and D’Alonzo, 1974].

Measurements made to assess alignment of the spinal column follow the “rule of 2”: that is, the allowable displacement between vertebrae in any direction should not exceed 2 mm in adults, with a greater allowance in children. The evaluation of bony integrity can be difficult at times, especially of the odontoid process and C1. If adequate quality cannot be obtained by plain radiographs, computed tomography (CT) is indicated [Andersen and Stringer, 1996; Stringer and Andersen, 1996].

Further limitations of plain radiographs of the spine include 40 percent false-positive and 20 percent false-negative interpretations. Also, significant spinal cord injury frequently occurs in children without radiographic abnormality (SCIWORA) (e.g., vertebral fracture or dislocation) [Burke, 1974; Pang and Wilberger, 1982], with this discrepancy observed in 15–70 percent of all pediatric spinal cord injuries [Hadley et al., 1988; Menezes et al., 1989; Pang and Wilberger, 1982]. Based on radiographic findings, spinal column injuries can be classified as follows:

SCIWORA, as first described, referred to cases of spinal cord injury with objective signs of myelopathy in the absence of abnormalities on static and dynamic flexion/extension films, CT imaging, and plain or CT myelography [Gore et al., 2009]. With modern MRI, many such cases show radiological evidence of injury to the spinal cord or spinal cord ligaments [Caviness, 2004; Davis et al., 1993; Matsumura et al., 1990; Proctor, 2002], with MRI abnormalities seen in about two-thirds of cases [Massagli, 2000]. Pang found MRI in 50 SCIWORA patients to have high prognostic value. The findings were complete transection or major hemorrhage (both with profoundly poor outcomes); minor hemorrhage (40 percent improved to mild disability); edema only (75 percent improved to mild disability); and normal (all recovered completely) [Pang, 2004].

Because the vertebral column in early life is quite elastic, allowing for significant flexion, hyperextension, and vertebral distraction without accompanying fracture or dislocations, SCIWORA is seen most often in infants and young children, in whom the frequency is three times greater than in children older than 10 years of age [Massagli, 2000]. Also predisposing to SCIWORA in young children is the greater elasticity of the spinal column than of the spinal cord; the bony spine can tolerate 5 cm of distraction, whereas the spinal cord can tolerate only 5 mm before disruption [Caviness, 2004]. Spinal cord ischemia from vertebral artery injury may also be contributory [Gore et al., 2009]. SCIWORA most often affects the cervical spine, typically producing signs of a central cord syndrome [Babcock, 1975]. In children less than 8 years, SCIWORA is more likely to be more rostral and more severe; from 8 to 16 years, it tends to be more caudal and less severe [Gore et al., 2009]. In some cases, transient neurologic symptoms may be the only indication that the cervical spinal cord has been injured; in other cases, permanent deficits may result, reflecting more severe spinal cord injuries that can be partial or complete [Caviness, 2004; Massagli, 2000]. In one study, 52 percent of children with SCIWORA had delayed onset of paralysis 4 days after their injury [Pang and Wilberger, 1982]. Children with spinal trauma must be examined frequently for signs of clinical deterioration in the days after their injury, even when plain radiographs of the spine are normal. Cervical immobilization of patients with SCIWORA is controversial. If dynamic films demonstrate stability of the cervical spine, the role of cervical immobilization is unclear. With only extraneural findings on MRI, neurological recovery, and no neck pain, 2 weeks of hard collar immobilization is recommended, followed by dynamic films; but with neural findings on MRI, the recommendation is for 12 weeks of hard collar immobilization before dynamic films are obtained [Gore et al., 2009]. Avoidance of high-risk activities for 6 months is recommended [Gunnarsson and Fehlings, 2003].

CT is the best means of detecting subtle spinal fractures and clarifying equivocal areas seen on plain films. CT also should be used when plain films are normal, but clinical suspicion of instability or fracture still exists [Andersen and Stringer, 1996; Stringer and Andersen, 1996], or when fracture/ displacement is seen on plain radiography [Caviness, 2004]. More recently, MRI of the spine has replaced CT and myelography as the preferred means of imaging the injured spinal cord (Figure 78-7) [Goldberg et al., 1988; Greenberg, 1991; Hyman and Gorey, 1988]. Many MRI sequences can be used, depending on the type of injury and the spinal cord area of concern, including spin-echo, fast-spin-echo, gradient-recall-echo, and inversion-recovery sequences. Excellent review articles on the use of these different sequences in specific situations are available [Andersen and Stringer, 1996; Stringer and Andersen, 1996]. MRI can distinguish intramedullary lesions from extramedullary ones, and early pathologic changes from later ones [Davis et al., 1993; Kalfas et al., 1988; Levitt and Flanders, 1991; Sett and Crockard, 1991]. In general, findings on MRI correlate well with the degree of deficit found on neurological examination [Bondurant et al., 1990; Davis et al., 1993; Flanders et al., 1990]. Additionally, the type and severity of spinal cord injury seen on MRI correlate well with the degree of recovery. In one study, MRI findings were classified into the following three categories:

type I, inhomogeneous on T1-weighted image, large central hypointensity with thin hyperintense rim on T2-weighted image

type II, normal T1-weighted image, hyperintense cord on T2-weighted image

type III, normal T1-weighted image, isointense center with thick hyperintense rim on T2-weighted image.

Fig. 78-7 Imaging studies of the thoracic spine in a 19-year-old male after a car accident in which he was an unrestrained passenger.

A, Computed tomography shows a fracture of the right lamina of T3, with a burst fracture of the body of T4, causing deformity and anteroposterior narrowing of the spinal canal. B, Magnetic resonance image of the thoracic spinal cord in the same patient, obtained the same day (T1-weighted image). There is a wedge fracture of T4 (arrow), producing compression of the underlying spinal cord.

Neurological recovery was insignificant in patients with intraspinal hemorrhage, while patients with cord edema or contusion recovered significant neurological function [Kulkarni et al., 1987] (Figure 78-8).

Fig. 78-8 Radiographs of the cervical spine in a 16-year-old boy who, 10 weeks previously, fell from a one-story roof, sustaining fractures of his 6th and 7th cervical vertebrae.

A, Computed tomography of the cervical spine, showing anterior subluxation of the body of C6 on C7 (arrow). B, CT scan of cervical spine showing fracture of the right articular mass of C6, with forward displacement of the C6 pedicle (circled). C, Magnetic resonance image of the cervical spine (T2-weighted image), showing narrowing and T2 hyperintensity of the cervical spinal cord at the C6 level (arrow), the site of the displaced C6 fracture.

In general, spinal cord injuries in which there is no identifiable MRI abnormality carry a better prognosis for recovery than injuries in which MRI abnormalities are found [Davis et al., 1993]. The usefulness of MRI in visualizing spinal cord injury in neonates, in whom neurologic abnormalities can be difficult to find and in whom significance can be difficult to interpret, has been documented [Lanska et al., 1990]. MRI also is useful in showing abnormalities that may not be seen with other imaging modalities, including ligamentous instability and disruptions, other soft-tissue injuries, disc herniations, intraspinal hemorrhage, and nerve root injuries [Kadish, 2001; Proctor, 2002]. In some cases, it may be necessary to use a combination of MRI and CT to identify coexisting injuries of the spinal cord and bony spine [Levitt and Flanders, 1991; Wittenberg et al., 1990]. MRI can be useful in predicting outcome in intramedullary lesions because it can differentiate hematoma (which often has a poor prognosis) from spinal cord edema, which usually has a better outcome [Schaefer et al., 1992]. In 49 patients who underwent MRI within 72 hours of sustaining a cervical spinal cord injury, the imaging features that correlated with a poor functional recovery included hemorrhage, long segments of edema, and high cervical lesions [Flanders et al., 1999].

Functional MRI has been used to assess the degree of spinal cord damage in 27 patients with complete and incomplete cervical and thoracic spinal cord injuries. With thermal stimuli applied to the inner calf (L4 sensory dermatome), activity in ipsilateral dorsal gray matter was diminished with complete and incomplete injuries; in the ventral regions, however, activity was increased with complete injuries, but with incomplete lesions, it was similar to or diminished compared with healthy subjects (i.e., it was notably less than in subjects with complete spinal cord injuries) [Stroman et al., 2004].

Serial MRI studies also can be helpful in following the evolution of traumatic spinal cord lesions [Sato et al., 1994; Sett and Crockard, 1991]. The need for metrizamide CT, used previously in evaluating post-traumatic spinal cord cysts and syringomyelia [Mace, 1985; Quencer et al., 1983; Rossier et al., 1983], has diminished substantially with the accuracy of MRI in showing these lesions. Also, because of the availability of MRI, the need for myelography to identify surgically remediable spinal cord injuries (e.g., extramedullary hematomas and major fracture-dislocations) has diminished greatly. When the patient’s clinical condition permits, most centers currently advocate early MRI for acute spinal trauma.

Electrophysiologic Evaluation

Motor-evoked potentials (MEPs) provide a means to assess descending spinal tract function. MEPs are induced by transcranial magnetic stimulation of the motor cortex and are recorded on muscles of interest using surface electrodes to determine the level and extent of the spinal cord injury. It has been shown that MEP amplitudes improve with spinal cord injury recovery but latencies do not [Xie and Boakye, 2008].

The usefulness of somatosensory-evoked potentials (SSEPs) and electromyography in evaluating the level of spinal cord injury and predicting clinical outcome is controversial. SSEPs are often useful in identifying the level of spinal cord injury [Louis et al., 1985] and in identifying areas of spinal cord damage that have been clinically unapparent [Toleikis and Sloan, 1987]. This test can be especially helpful in infants and young children in whom the degree of deficit often cannot be determined confidently by clinical examination [Kamimura et al., 1988]. It is unknown, however, whether SSEPs are superior to MRI in identifying the extent of a spinal cord injury and predicting clinical outcome. Some investigators have noted a high degree of correlation between abnormal SSEPs and potential for neurological recovery [Curt et al., 1997; Li et al., 1990; Perot, 1973; Rowed et al., 1976; Spiess et al., 2008; Ziganow, 1986], but others have not [Katz et al., 1991; York et al., 1983]. One study of experimental spinal cord injury revealed no correlation between the absence of SSEP waveforms and the degree of spinal cord injury [Singer et al., 1977]. In that study, animals demonstrated good return of function at 8–9 days, even though SSEP waveforms were not recordable 3–19 days after injury. Other studies have suggested, however, that patients with absent SSEPs after spinal cord injury are unlikely to have any meaningful recovery of neurologic function [Shurman et al., 1996].

The sympathetic skin response (SSR) is another noninvasive electrophysiological test, in which EMG electrodes placed on the skin surface record responses to electric or magnetic stimulation above the spinal lesion. These recordings can be used to assess damage to the spinal and peripheral efferent sympathetic pathways subserving corresponding skin areas. While SSR has been used less than MEPs and SSEPs to assess completeness of spinal cord injuries, the addition of sympathetic nervous system testing may provide a more complete picture of the spinal cord injury [Curt and Dietz, 1999; Xie and Boakye, 2008].

The presence of F-waves and H-reflexes depends on the integrity of the motor neuron pool. H-reflexes disappear in the first 24 hours of spinal shock, but return thereafter; F-waves disappear below the level of injury during spinal shock and reappear with its resolution [Horowitz and Patel, 2003]. F-wave and H-reflex latencies and electromyogram can show lower motor neuron unit abnormalities 3–4 myotome levels beyond the site of a spinal cord lesion [Shefner and Tun, 1991]. Based on such electrophysiologic findings, the extent of deficit, whether functional or anatomic, is often found to be more extensive than indicated by clinical examination. As with SSEPs, it is uncertain whether abnormalities on electromyogram can increase significantly the predictability of clinical outcome in spinal cord injury.

Lumbar Puncture

Lumbar puncture can provide useful diagnostic information in certain types of spinal injury. The presence of extramedullary hematomas should be suspected with bloody or xanthochromic cerebrospinal fluid. Conversely, spinal epidural abscess usually causes a mild cerebrospinal fluid pleocytosis with an elevated cerebrospinal fluid protein concentration and a normal glucose concentration. Very low cerebrospinal fluid pressure at the time of lumbar puncture suggests the presence of a spinal block.

Complete Spinal Cord Injuries

Complete loss of spinal cord function, as defined by the ASIA/IMSOP Spinal Cord Impairment scale, can be either physiologic or pathologic. In physiologic loss, there is no morphologic alteration of the spinal cord, which becomes dysfunctional after impact. This dysfunction may occur with transient compression of the cord by a dislocated vertebra. With more sustained compression of the spinal cord, anatomic disruption of spinal cord elements may result, with accompanying pathologic loss of function. The frequency of complete spinal cord injury in pediatric patients varies widely, depending on the series reviewed, ranging from 20 to 95 percent of all pediatric spinal cord injuries [Anderson and Shutt, 1980; Burke, 1974].

The initial phase of complete loss of spinal cord function is characterized by spinal shock [Green and Eismont, 1984; Kiss and Tator, 1993]. The truncal and extremity muscles below the level of the lesion are flaccid, deep tendon and superficial reflexes are depressed or lost, plantar responses are absent, anesthesia is present to all modalities, and autonomic dysfunction (hypotension and bradycardia) is present. The mechanism of spinal shock is unknown, but its physiologic effects must be differentiated from the more permanent pathologic effects of spinal cord injury. Motor and sensory deficits resulting from spinal shock alone resolve within 1 hour. Persistence of motor and sensory deficits beyond 1 hour implies that pathologic, more permanent injury has occurred. After 6–13 weeks in adults and within 1 week in children, tendon reflexes return, and muscle tone improves. The reflexes later become pathologically active, and spasticity ensues if the injury is more permanent, and there usually is no significant return of motor or sensory function. A four-phase model of the sequential clinical changes seen in spinal shock and the neuronal mechanisms that may underlie those changes has been described:

phase 1 (0–1 day), areflexia/hyporeflexia (from loss of descending facilitation)

phase 2 (1–3 days), initial reflex return (from denervation supersensitivity)

phase 3 (1–4 weeks), initial hyperreflexia (from axon-supported synapse growth)

phase 4 (1–2 months), final hyperreflexia (from soma-supported synapse growth) [Ditunno et al., 2004].

Autonomic dysfunction can persist for weeks to months. Also, ventilation may be compromised if the intercostal muscles are involved. Flaccid paralysis of the bladder can develop, with urinary retention and overflow incontinence. The gastrointestinal tract can be atonic. Vasomotor dysfunction can develop, with orthostatic hypotension, impaired shivering, sweating, and disturbances in temperature regulation persisting for weeks to months.

After complete injury of the cervical or upper thoracic spinal cord, an acute syndrome of bradycardia, hypotension, and hypothermia can occur, probably because of disturbed sympathetic outflow at the cervical and upper thoracic levels; most patients with quadriplegia after spinal cord injury have chronic hypothermia with poor temperature control related to loss of sympathetic peripheral vascular control [Green and Eismont, 1984]. Recovery from complete spinal cord injury is rare. The prognosis is poor. Tator estimated that less than 2 percent of patients with complete injury recover some distal cord function [Tator, 1996]. More encouraging was a comprehensive review of several large series of patients with complete spinal cord injury, which found that about 2 percent of the patients became ambulatory [Hansebout, 1982].

Incomplete Spinal Cord Injuries

Incomplete spinal cord injuries warrant special attention because children with these injuries often demonstrate significant recovery of function with a relatively good prognosis. Included among incomplete spinal cord injuries are cervical nerve root/brachial plexus neuropraxia, cervical spinal cord neuropraxia, the cervicomedullary syndrome, central spinal cord syndrome, syndromes of the anterior spinal cord and posterior spinal cord, Brown–Séquard syndrome, conus medullaris syndrome, and catastrophic spinal cord injuries (see p.1211).