Pediatr Radiol (2001) 31: 677±700 Ó Springer-Verlag 2001

R E V I E W A RT I C L E

Spinal trauma in children

Claire Roche Helen Carty

Received: 13 January 2000 Revised: 21 September 2000 Accepted: 6 March 2001

)

C. Roche ´ H. Carty ( ) Radiology Department, Royal Liverpool Children's NHS Trust-Alder Hey, Eaton Road, Liverpool, L12 2AP, UK e-mail: [email protected] Tel.: + 44-1 51-2 52 54 32 Fax: + 44-1 51-2 52 55 33

Abstract Evaluation of the child with suspected spinal injury can be a difficult task for the radiologist. Added to the problems posed by lack of familiarity with the normal appearances of the paediatric spine is anxiety about missing a potentially significant injury resulting in neurological damage. Due to differences in anatomy and function, the pattern of injury in the paediatric spine is different from that in the adolescent or adult. Lack of appreciation of these differences may lead to over investigation and inappro-

Introduction Spinal injury in children, both minor and serious, is fortunately rarer than in adults. Children account for 1±10 % of all spinal injuries [1±5]. The mortality among spine-injured children is higher than in adults and is estimated at 25±32 % [5±8]. Death is most often due to associated injuries to other organs, including the brain. Reports of associated neurological damage vary, depending on the reporting centres, being higher in tertiary referral centres. The quoted incidence is between 25 and 50 % [5, 9±14], which is similar to adult studies [3, 15, 16]. Practical experience in a hospital such as ours, which serves a population of about 400,000 for primary trauma, suggests that serious neurological sequelae occur in less than 1 % of all children admitted with a history of spinal injury. In neurologically impaired survivors the trauma is at the C1±2 level or in the lower cervical or thoracic spine. Fractures below L1 may be associated with nerve root damage but as the cord normally ends at L1 level, cord damage does not occur. Fifty percent of children with neurological damage have complete sen-

priate treatment. This review attempts to clarify some of the problems frequently encountered. It is based on a review of the literature as well as personal experience. The normal appearances and variants of the spine in children, the mechanisms and patterns of injury are reviewed highlighting the differences between children and adults. Specific fractures, a practical scheme for the assessment of spinal radiographs in children, and the role of cross sectional imaging are discussed.

sory and motor loss below the level of the cord injury with a poor prognosis for recovery [5, 8, 10, 11, 13, 17] Children with `incomplete' neurological lesions fare better than adults [5, 8, 10, 11, 18] with up to 90 % showing significant and 60 % showing full recovery [5, 14]. Spinal injury in children less than 8 years old is mainly in the upper cervical spine [5, 10, 12, 14, 18, 19] and is associated with a high risk of neurological damage [5, 6, 14, 19, 20], as much of the serious trauma is at atlantooccipital level. Neurological injury resulting from trauma at C1±2 is dependent on the degree of subluxation sustained at the time of injury. The cervical canal is at its widest at this level, providing a measure of protection, and subluxation may take place without impinging on the cord. Injury at a lower level, where the canal is smaller, carries a greater risk of cord damage. After the age of 8 years, the fulcrum of movement changes from C2±3 to C5±6, the adult fulcrum. The types and patterns of spinal injury in those over 8 years of age reflect this. Children with spinal injuries have associated head injury in 25±50 % of cases [7, 13, 21, 22], extremity fractures in 30 % [13] and chest and abdominal injuries in

678

21 % [13]. Abdominal injuries, particularly of the small bowel, are associated with `seatbelt' fractures of the thoracolumbar spine [13]. Neurological status is impossible to assess in the unconscious patient and victims of severe trauma are considered to have spinal injuries until proven otherwise. The initial trauma series includes a lateral view of the cervical spine. The cervical collar remains in situ until the clinical examination and spinal X rays have been deemed normal. The incidence of cervical spine fracture in blunt trauma victims requiring admission is estimated at 1±2 % for both paediatric and adult populations [7, 23]. Spinal injury often occurs at more than one level [5, 13, 14, 19, 24]. While most injuries occur at contiguous levels, 16 % occur at different levels in the spine [13, 14]. It is therefore recommended that if spinal injury is identified at one level, the entire spine should be imaged, especially in high-risk situations, and is best achieved by sagittal T1-weighted (T1W) and T2-weighted (T2W) MRI of the whole spine. MRI is mandatory when there is neurological deficit. In reported series with multiple levels of injury, the other fractures are mainly compression fractures that do not alter acute management. Long-term sequelae of compression fractures include kyphosis, especially if the injury involves the growth plates.

Aetiology Road traffic accidents (RTAs) account for 36±54 % of spinal injury in children, [5, 7, 10, 13, 21, 25], followed by falls and sports injuries. In children under 12 years the majority of incidents are motor vehicle/pedestrian/ cycle accidents. With increasing age, injury sustained as car passengers increases. Above the age of 8 years, boys are affected more than twice as often as girls [5, 7, 10, 12, 13, 17, 26, 27]; below this the sex incidence is equal. In the USA, gunshot injuries are a frequent cause, accounting for 22 % of injuries in one study of 277 patients with spinal trauma [4]. Birth trauma is a rare, but well recognised cause of cervical spine injury [3, 18±20, 28±30] with up to 75 % occurring with breech deliveries [28]. Cadaver studies show that the spinal column can be longitudinally stretched to about 2 inches without disruption in a neonate, whereas the cervical cord will rupture if stretched beyond 1/4 inch [31, 32]. Many cases are `without radiographic abnormality' but have profound neurological damage. NAI is a rare cause of spinal trauma [6, 13, 29, 33±36] and occurs mainly at the thoracolumbar junction and lumbar spine, though Hangman's fractures are recorded. The mechanism is hyperflexion. Compression fractures are commonest, but vertebral dislocation also occurs.

Congenital anomalies of the cervical spine, such as `os odontoideum', block vertebrae, Klippel-Feil syndrome and Down's syndrome increase the risk of cervical spinal trauma [12]. Atlanto-axial instability is reported to occur in 10±20 % of individuals with Down's syndrome [36, 37], but is symptomatic in about 3 %. There are 31 reported cases of atlanto-axial dislocation in Down's syndrome up to 1987. Most of these had a minimum of a 1-month history of neurological signs before major difficulties ensued [36].

Differences between adult and paediatric spinal injury The patterns of spinal injury in children, especially in the cervical region, relate to changing anatomy [1]. Other factors include the child's resilience to trauma and potential for growth and recovery, which allows for restoration of vertebral body height after anterior wedging, but may rarely result in progressive spinal deformity if end plate injury or paralysis has occurred [1]. In the child less than 8 years old, cervical spine fractures tend to occur from the occiput to C2 [5, 10, 13, 14, 17, 19, 37]. Possible reasons for this include: · Fulcrum of movement located at C2±3 in the child, C5±6 in the adult. · Relatively large head and weak neck muscles · Ligamentous and joint capsule laxity · Horizontal orientation of the facet joints in younger children · Underdeveloped uncinate processes · Mild physiological anterior `wedging' of vertebral bodies · Incomplete ossification of the odontoid process [1, 31] Late spinal deformity after spinal injury may occur in children and often appears at times of growth spurts [1]. In general, children who injure their spines do not sustain occult injury [1]. Neurological trauma is usually apparent at the time of presentation.

SCIWORA The term `SCIWORA' was coined in 1982 by Pang and Wilberger [29] and is defined as ªspinal cord injury without evidence of vertebral fracture or malalignment on plain radiographs and computed tomographyº [29, 38±40]. Though radiographs are normal, MRI has shown significant pathology in many of these patients [39, 41] (Fig. 1). The overall reported incidence of SCIWORA amongst spinal cord-injured children varies from 5 to 65 % [6, 8, 17, 20, 29, 38]. SCIWORA is commoner in

679

younger children and in this group is often severe with a poor prognosis for recovery [5, 14, 17, 29, 30, 38±40]. SCIWORA is most frequent with cervical injury but has also been reported with thoracic injury [29, 38, 39]. It is postulated that excessively elastic ligaments and other developmental features of the immature spine allow transient excessive movement during trauma, resulting in cord distraction or compression [10, 25, 29, 38±42]. Cord ischaemia is another postulated cause [29, 38, 41] due to direct vessel injury or hypoperfusion. One quarter of the 24 children with SCIWORA described by Pang and Wiberger [29] had significant hypotension on admission. Delayed onset of neurological damage is reported to occur in 6±54 % of cases [5, 29, 38, 39], usually within 48 h. Such patients, if they are able to give a history, often describe transient neurological symptoms such as numbness, paraesthesia or `shock-like' sensations in the extremities occurring shortly after the trauma [29, 38]. Such symptoms should be taken seriously and, once suspected, the child should have MRI (to include T1W, T2W and gradient-echo T2W images) to distinguish cord oedema from haemorrhage and to assess ligament injury. Increased signal in the ligaments indicates disruption. Further assessment with flexion and extension films, either by supervised radiographs or fluoroscopy, and CT is required to exclude fracture and instability. MRI will identify treatable injuries, such as extrinsic compression of the cord by epidural haematoma or retropulsed bony fragments [38]. The outcome after SCIWORA is related to the degree of cord damage, as shown by MRI. Cord transection and major haemorrhage is associated with poor outcome; minor haemorrhage or oedema with moderate-to-good recovery; and no cord abnormality with complete recovery [43]. SCIWORA with more than one focus of injury to the spinal cord has been reported [39].

Atlanto-occipital dislocation Atlanto-occipital dislocation is an injury seen rarely in RTAs, falls, direct trauma and forceps delivery and is usually fatal [1, 3, 6, 7, 19, 44±50]. Most survivors have severe neurological damage [1, 46, 47]. Though rare, it occurs 2.5 times more frequently in children than in adults [1, 9], particularly amongst younger children [1, 7, 14, 19]. It is found at autopsy in 15 % of cases of fatal spine injury [6] and is often associated with severe brain injury. The young child is at particular risk because of relatively small occipital condyles and relative horizontal orientation of the atlanto-occipital joint. These features render the atlanto-occipital articulation less stable [1]. Disruption of the apical and alar ligaments and of the tectorial membrane allows the cranium to move with re-

spect to the spine. There is usually severe damage to the medulla oblongata, often complete transection. Other susceptible structures include the caudal cranial nerves and the upper three cervical nerves [1, 45]. Children with this injury often have cardiorespiratory arrest due to brain-stem injury, which carries an appalling prognosis for any child who survives resuscitation, with, almost invariably, gross neurological sequelae [47]. There is a high incidence of severe injuries elsewhere in the body and submental laceration and mandibular fractures [44, 47]. On the lateral cervical spine radiograph, there is malalignment between the cranium and the spine (Fig. 2). Anterior dislocation of the cranium on the cervical spine and distraction occurs in 65 % [48]. There is usually massive prevertebral soft-tissue swelling [46, 49] with displacement of the nasogastric tube if one is in situ. Because of rupture of ligaments, the normal relationship of the basion (anterior lip of foramen magnum, base of clivus) to the odontoid is disrupted. There are four described methods of assessment of this injury on the lateral radiograph: 1. Widening of the gap between the occipital condyles and the condylar surface of the atlas to more than 5 mm [45, 47]. 2. The `Wackenheim' clivus line: a line drawn along the posterior clivus should intersect or be tangential to the odontoid [1, 2, 50] (Fig. 3 a). 3. The `Powers Ratio' requires identification of four bony landmarks. A is the anterior tubercle of the atlas; B is the basion; C is the spinolaminar line of the atlas; O is the opisthion (posterior lip of foramen magnum). The ratio BC/AO should be less than 1 [47] (Fig. 3 b). 4. Harris' methods: [50] (Fig. 3 c) a. The basion should lie within 12 mm of the superior continuation of a line drawn along the posterior cortex of the body of the axis (the posterior axial line). b. The distance between the basion and the tip of the odontoid should be less than 12 mm. All these techniques have limitations, especially in children [47, 49]. The bony landmarks required for these assessments are often not clearly seen, particularly the opisthion (posterior margin of foramen magnum), rendering the Powers ratio unhelpful. The odontoid is not fully ossified until 12 years of age [1]; therefore, the Wackenheim clivus line and Harris' basion-odontoid interval are unsuitable for use in children. Widening of the atlanto-occipital articulation to more than 5 mm and Harris' `posterior axial line' method appear to be useful [48, 50]. In all 50 normal children included in Harris' report, the basion lay 0±12 mm anterior to this line [50].

680

1

2

3a

3b

Fig. 1 SCIWORA. Follow-up T1W sagittal MRI of the cervicothoracic junction in a child with SCIWORA showing damage to the spinal cord in two places. The plain radiographs at the time of injury were normal. Note the slight separation of the spinous processes of C7 and T1, indicating that the injury was due to hyperflexion Fig. 2 Atlanto-occipital dislocation. Note the prevertebral soft tissue swelling. The distance between the occipital condyles and the atlas is pathologically increased Fig. 3 a±c Diagrams illustrating the lines used to assess atlanto-occipital dislocation

3c If the child survives, CT shows the bony disruption and the frequently associated subarachnoid haemorrhage. MRI shows cord and ligament injuries, as in the series of Grabb et al. [49], and distinguishes partial from complete atlanto-occipital dissociation.

681

Fractures of the atlas In children, atlas fractures are less common than subluxations at the atlanto-occipital and atlanto-axial levels [1, 19]. Difficulties in interpretation of radiographs arise because of the ossification pattern of the atlas and its numerous normal variants [1]. There are separate ossification centres for the lateral masses. A third centre for the anterior arch is ossified in only 20 % of neonates and may not appear until 1 year of age. Ossification of the posterior arch is sometimes incomplete, resulting in partial spina bifida. These congenital defects can be distinguished from acute fractures by their well-corticated margins [1]. Overriding of up to two-thirds of the C1 anterior arch above the odontoid tip occurs in up to 20 % of normal children when the neck is extended [25, 51]. The atlas reaches adult dimensions by age 4 years. The spinal canal diameter is approximately 22 mm at this level [1]. Atlas fractures could therefore be expected to cause little spinal cord impingement, unless the transverse ligament is also damaged, resulting in atlanto-axial subluxation [15]. Axial compression forces transmitted via the skull result in a `Jefferson' burst fracture [1, 25]. Fractures typically occur in four places, two in the anterior arch and two in the posterior arch. If the transverse ligament remains intact, the injury is stable [25]. This fracture is rare in children, typically occurring in the teenage years after diving accidents or motor vehicle accidents, where the head impacts on the car roof. Associated hyperextension force leads to fracture at the weakest points of the posterior arch where it is grooved by the vertebral artery [1, 25]. With posterior arch fractures there is usually little anterior soft-tissue swelling [1, 25]. The open-mouth odontoid radiograph shows bilateral offset of the lateral masses of the atlas with respect to the odontoid [l]. Total offset of more than 6 mm in children is highly suggestive of rupture of the transverse ligament or avulsion from its bony attachments [15, 45, 48, 52]. This results in an unstable injury [25]. In 40 % of cases of atlas fracture there is an associated fracture of the axis [50]. CT demonstrates these fractures well, as the ring of the atlas is shown in the axial plane (Fig. 4).

Atlanto-axial subluxation Traumatic rupture of the transverse ligament with resultant atlanto-axial subluxation is rare in children since the more vulnerable odontoid usually fails before the ligament does [1, 25, 45]. The atlas moves forward on the axis, increasing the distance between the anterior arch of the atlas and the odontoid and decreasing the spinal canal space [1] (Fig. 5). The child usually presents with a painful torticollis [25]. Lateral cervical spine ra-

a

b Fig. 4 a, b C1 fracture. a Axial CT at the level of the atlanto-occipital articulation shows the fracture through the left lateral mass of C1. b Note the acute subarachnoid haemorrhage in the cervical spinal canal

diographs show increased distance between the anterior arch of the atlas and the odontoid to greater than 5 mm. The measurement is made from the posteroinferior margin of the base of the anterior arch of the atlas to the odontoid [53, 54]. There is usually anterior soft tissue swelling, but this may be subtle. Oedema may be more evident on MRI. Although atlanto-axial subluxation is rare in children, children with Down's syndrome, skeletal dysplasias (especially Morquio's syndrome and diastrophic dwarfism), juvenile chronic arthritis, pharyngeal infec-

682

simple analgesia. In a few cases the deformity becomes fixed and irreducible owing to an underlying disorder of rotation at the atlanto-axial joint. This rare condition has been variously termed `atlanto-axial rotatory subluxation' or `atlanto-axial rotatory fixation' [57]. Fixation is probably the better term, as in most cases the joint is fixed within the normal range of motion of the joint rather than truly subluxed [57, 58]. To put things in perspective, although we see ten cases of torticollis per month in our A&E department (70,000 A&E attendances per year), there have been only four cases of atlanto-axial rotatory fixation in the past 4 years, all referred from other centres. The length of symptoms ranged from 1 to 7 months; none had a history of trauma. Although rotatory fixation is rare, it is important to recognise it because early treatment results in improved outcome [59, 60]. Torticollis may be due to a wide variety of conditions. The causes may broadly be divided into two distinct groups: 1. Disorders of rotation of the atlanto-axial joint resulting in fixed or limited rotation of the neck 2. Other disorders causing limited rotation of the neck without primarily involving the atlanto-axial joint Fig. 5 Anterior atlanto-axial subluxation

tion and hypoplasia of the odontoid are at increased risk with neck injury [1, 18, 23, 25, 26]. Controversy exists on the management of children with Down's syndrome who wish to participate in sports and whose radiographs show an increased atlanto-axial distance [37, 55, 56]. It has been recommended by the Down's association that children should not participate in sports such as trampolining, diving and gymnastics until they have flexion and extension views of the cervical spine demonstrating no instability. These views may be difficult to obtain. Fluoroscopy may be a more efficient method of examining the spine in these children than flexion/extension views if the child's activities are to be restricted until these views are done. One study showed no differences in outcome after 1 year in children with restricted versus non-restricted activity [55].

Torticollis and rotatory deformities of the atlanto axial joint Acquired torticollis or `wry neck' is a common clinical problem in children. Most cases settle spontaneously when treated conservatively using a cervical collar and

The first group is rare and will be discussed in the next section. The second group includes conditions where the primary abnormality lies in the sternomastoid muscle, such as congenital fibrosis of the sternomastoid and acquired benign paroxysmal torticollis, and locally painful neck conditions, such as lymphadenitis, inflammation or tumour in the cervical spine, cord or posterior fossa. In this group, contraction of the sternomastoid muscle is the deforming force resulting in torticollis. It is the `short' sternomastoid, on the side opposite the direction of head rotation that is contracted. Conversely, when torticollis is due to an underlying disorder of atlanto-axial rotation, sternomastoid spasm occurs on the side to which the head is turned. The spasm in this case represents an attempt to correct the deformity [59, 60±62].

Atlanto-axial rotatory fixation (AARF) This rare condition may occur spontaneously, following minor trauma, or in association with congenital anomalies of the cervical spine, or arthritis [2, 57, 59, 61±64]. Grisel's syndrome is atlanto-axial subluxation due to ligamentous laxity from hyperaemia following infection or surgery in the head and neck area [60, 62]. Seventyfive to eighty percent of reported cases of AARF occur in children less than 13 years of age [61, 64]. The principal motion at the atlanto-axial joint is rotation and 50 % of total neck rotation occurs here [59].

683

Fig. 6 Diagram of atlanto-axial joint (reproduced from [51], with permission)

Fig. 7 Fielding's classification of atlanto-axial rotatory fixation (reproduced from [59] with permission)

The facet joints are almost horizontal, allowing excellent rotation at the expense of bony stability. The transverse ligament prevents excessive anterior motion of Cl on C2 and the paired alar ligaments, running from the posterolateral tip of the odontoid to the occipital condyles, prevent excessive rotation [57, 58] (Fig. 6). CT studies of normal adult volunteers show the physiological range of rotation of C1 on C2 to be 29±50o to either side [65]. A recent study of 32 normal children using CT reported a similar range of motion [66]. In atlanto-axial rotatory fixation the normal rotation of C1 on C2 becomes limited or fixed, but the cause is not known [54, 67]. Fielding and Hawkins [57] postulated that swollen capsular and synovial tissues and muscle spasm prevent reduction in the early stages and that capsular and ligament contractures develop later, caus-

ing fixation. Others postulate a capsular tear with invagination of inflamed synovium [67, 68]. The angle between the odontoid and the facets is steeper in children less than 10 years old than in adults. Cadaver studies on infants have shown meniscus-like folds attached to the capsule of the atlanto-axial joints. No such folds were found in adults [69]. These anatomical differences, which allow greater movement, may help to explain the increased incidence of atlanto-axial rotatory fixation in children. Fielding and Hawkins [57] have classified rotatory fixation into four categories (Fig. 7): · Type I: Rotatory fixation without anterior displacement of C1. The odontoid acts as the pivot and the transverse and alar ligaments are intact. This is the

684

Fig. 8 The `cock robin' position. Typical `cock robin' position of the head in atlanto axial rotatory fixation, `like a bird listening for a worm' Fig. 9 a, b Atlanto-axial rotatory fixation. a Lateral cervical spine view. Because of rotation and tilt of the head, the C1±2 articulation is obscured. There is lack of overlap of the posterior arch of C1. b Open-mouth peg view. There is asymmetry of the lateral masses and apparent reduction in C1±2 joint space on the right, due to rotation

8

9a

9b most common type and occurs within the normal range of motion of the joint. · Type II: Rotatory fixation with anterior displacement of 3±5 mm. One of the articular masses acts as the pivot. This is the second-most-common type and is associated with deficiency of the transverse ligament. · Type III: Rotatory fixation with anterior displacement of more than 5 mm. This degree of displacement implies deficiency of both the transverse and the alar ligaments. · Type IV: Rotatory fixation with posterior displacement of C1. This is the rarest type and occurs with a deficient odontoid. Types II±IV are more serious, as neurological compromise due to narrowing of the spinal canal is more likely [57]. Presentation is typically with sudden onset of torticollis without any antecedent symptoms, or there is a history of a sudden click or minor trauma. The child

shows the typical `cock robin' appearance with the head rotated to one side and tilted to the other, like a bird listening for a worm [2, 57] (Fig. 8). Sternomastoid spasm, if present, occurs on the side to which the head is turned. The radiographic features of rotatory fixation may be confusing because of difficulty in positioning the patient and interpreting the radiographs [57, 59]. C1 and C2 are often obscured on the lateral cervical radiograph because of the tilt and rotation of the head (Fig. 9). Peg views are often impossible to obtain. The peg view shows asymmetry of the lateral masses of C1 with respect to the odontoid. The lateral mass that has rotated forwards appears wider and closer to the odontoid [57, 58, 67, 69]. The facet joint between C1 and C2 may be obscured because of apparent overlapping [57, 61, 70, 71] (Fig. 9 b). On the lateral cervical radiograph the right and left parts of the posterior arch of C1 no longer superimpose because of the head tilt. Depending on the degree of rotation, one lateral mass of C1 may be visi-

685

a b

c Fig. 10 a±c Atlanto-axial rotatory fixation. a Axial CT shows rotation as well as anterior subluxation of C1 on C2 (Fielding type 3). b 3D reconstruction. c 2D coronal reconstruction

ble, anterior to the odontoid, in the expected position of the anterior arch [52, 69]. These radiographic features are not pathognomonic for atlanto-axial rotatory fixation, however, as similar appearances are found in normal children whose heads are voluntarily rotated and in those with torticollis due to other causes [60, 68, 69]. The diagnosis is established by demonstrating fixation of C1 on C2. A dynamic test is required to prove that C1 and C2 no longer rotate independently of each other, but as a unit. CT with coronal and 3D reconstruction is the best and simplest method of demonstrating the lesion. Axial slices show the rotation of C1 on C2. The odontoid appears to lie eccentrically between the lateral masses of C1 [61, 62, 68, 72]. When there is associated anterior or posterior shift of C1 on C2 (Fielding Types II±IV) static CT is diagnostic (Fig. 10). Because the CT appearances in Fielding Type I atlanto-axial rotatory fixation are identical to the CT appearances in those with torticollis and in normal subjects with voluntary head rotation, a dynamic study is needed [59, 62, 65, 68]. Patients are initially scanned at rest and the scan is then repeated with maximal voluntary contralateral rotation of the head. If rotatory fixation is present there will be no significant motion of C1 on C2 whereas normal subjects and those with torticollis due to other causes show a reduction or

a reversal of the rotation [62, 68]. Some authors have recently advocated dynamic CT with general anaesthesia and full muscle relaxation in order to make a confident diagnosis of atlanto-axial rotatory fixation [73]. Early diagnosis is important as it leads to improved outcome with non-operative management. If the diagnosis is established within a month of onset, then traction combined with muscle relaxants often corrects the deformity [59, 60, 67]. Surgical fusion may be required when the deformity persists and if there is anterior displacement of C1 or neurological involvement [57, 60]. Children whose torticollis fails to settle within 1 week require dynamic CT investigation using CT. This approach will avoid over-investigation and over-treatment, yet will still detect atlanto-axial rotatory fixation early enough to achieve a good outcome.

Odontoid fracture Reports of the frequency of odontoid fractures in children vary [1, 2, 9, 10, 19]. Generally rare, one report quotes fractures of the odontoid as being ªprobably the most common cervical spine injuryº in children [10]. There is a difference in fracture pattern and in outcome between children under 7 years of age and adults [1, 74]. Developmentally, the odontoid is separated from the body of the axis by a cartilaginous synchondrosis which fuses between the ages of 5 and 7 years, but a remnant may persist for several years [19, 25, 51]. This cartilaginous synchondrosis lies within the body of the axis, below the level of the vascular supply to the lower part of the odontoid. Fractures in children younger than 7 years occur through this growth plate and unite readily. Fractures in adults and older children usually occur at the base of the odontoid, at or above the level of the superior facets. This type of fracture interferes with the blood supply to the odontoid and is often com-

686

Fig. 11 a±c Low odontoid fracture. a The normal sclerotic `axis ring' is disrupted, indicating a fracture (arrow). The `axis ring' is a composite density formed anteriorly by the cortex of the junction between pedicles and body, posteriorly by the cortex of the axis body, and superiorly by the junction of odontoid and body [76]. b Axial CT shows the fracture line extending into the body of C2. c T2W MRI shows prevertebral soft tissue oedema (arrow)

b

a

c plicated by non-union and pseudarthrosis [l, 15, 18, 22, 74]. Odontoid fractures cause variable symptoms including torticollis, nuchal rigidity and occipital pain due to trauma to the greater occipital nerve [1, 25]. The fracture may be displaced, resulting in spinal cord damage [1, 25, 74]. The lateral cervical spine radiograph usually shows prevertebral soft tissue swelling [1]. The fracture line lies below the level of the superior articular facets [1, 25, 74]. The course of this low odontoid fracture through the body of the axis disrupts the normal sclerotic `axis ring'. Such disruption is a subtle sign of axis fracture (Fig. 11). Plain film tomography may be useful. CT in the axial plane may miss undisplaced odontoid fractures [15, 45]; therefore, scans should be reconstructed in coronal and sagittal planes [18]. Following suspected or definite odontoid fractures, follow-up radiographs should be obtained to demonstrate whether union has occurred or whether unforeseen problems have arisen [1].

Os odontoideum Anomalies of the odontoid include aplasia, hypoplasia and `os odontoideum'. Os odontoideum is an oval or round ossicle of variable size with a smooth cortical border located in the expected position of the odontoid process [1, 75]. Although there is debate about whether the os odontoideum is a congenital or acquired lesion, the balance of opinion is that it is due to non-union of an odontoid fracture, often unrecognised [1, 12, 18, 74, 75]. There was a history of a significant episode of neck trauma in 17 of 35 cases and there are reports of an `os odontoideum' developing in a patient with a previously radiographically normal odontoid process [1, 75]. The `os odontoideum' is associated with instability of the atlanto-axial joint, present in up to 83 % of cases [74, 75] (Fig. 12), and confirmed or excluded by radiographs in flexion and extension. MRI confirms the anomaly, if unclear on the radiographs and shows any cord compression. There is an increased incidence of `os odontoideum' in Down's syndrome, Klippel-Feil syndrome and multiple epiphyseal dysplasia [75].

687

a

b

Fig. 12 a±c Os odontoideum. This child sustained cord injury while playing on a bouncy castle. a The os odontoideum is difficult to appreciate on the lateral radiograph. b The separate os odontoideum is well shown on the T1W MRI scan (arrow). c The T2W image shows high signal in the upper cervical cord due to oedema. Note that C1, together with the mobile os odontoideum, are displaced anteriorly on C2. causing narrowing of the spinal canal

The ossiculum terminale is a secondary centre of ossification for the tip of the odontoid which appears at 3 years and fuses with the body of the odontoid at age 12 years [1, 75]. This ossiculum terminale is present in 26 % of normal children aged 5±11 years and should not be mistaken for a fracture [51]. In some cases it never fuses with the odontoid and remains as a separate ossicle and is distinguished from the os odontoideum by its higher position.

c (Fig. 14). This is an unstable injury [1, 25]. Neurological deficit is rarer than might be expected, as the fracture leads to `autodecompression' rather than impinging on the cord [15]. If the fracture extends to the vertebral foramina, damage to the vertebral artery may cause neurological impairment [15]. The injury is visible on the lateral cervical spine radiograph in approximately 90 % of cases [47]. There are three `normal variants' that may lead to confusion. The cartilaginous synchondrosis between the odontoid process and the body of the axis has already been discussed. Its typical location and smooth edge should help distinguish it from a fracture of the body of the axis. The very rare `variant' is congenital spondylosis of the axis, seen as smooth corticated defects in the pars interarticularis. Only 11 cases had been reported up to 1989, 5 associated either with congenital cervical fusion or pyknodysostosis [76]. The commonest `variant' to cause confusion is pseudosubluxation of C2 on C3 [1, 31, 53, 77], discussed below.

Fractures of the body and neural arch of the axis (C2) Fractures of the body and neural arch of the axis in children are rare compared with fractures of the odontoid process and atlas [1, 17, 19, 37, 45]. Like other injuries from the occiput to C3, they are commoner in younger children. Associated atlas and/or upper thoracic spinal injury is reported in 10 % [15] (Fig. 13). The mechanism of injury is hyperextension which results in a `hangman's fracture' with bilateral fracture of the pars interarticularis of the axis, horizontal tearing of the C2±3 disk and anterior subluxation of C2 on C3

Injuries of the mid and lower cervical spine The classification and appearances of these injuries follows the adult pattern [18]. They increase in frequency as the spine matures to adult-type [1±3, 10±12, 18, 19, 25] and in a study of 152 vertebral fractures in children accounted for 14 % of all fractures [11]. Because of ligamentous laxity, facet disruption without associated fracture may occur in children [l]. Compression fractures are rare [1] and posterior spinal ligament damage

688

13 a

13 b

14 a

14 b

Fig. 13 a±c Fracture dislocation C2. a Radiograph shows fracturedislocation of C2. b Axial CT shows rotational deformity of C1 on C2. c Follow-up MRI performed 4 months after the injury shows persistent malalignment of C2 and CSF leak around the fractured C2 (arrow). Note the complete cord transection at the cervicothoracic level Fig. 14 a, b Hangman's fracture. a There is a subtle lucency in the posterior `pars interarticularis' part of C2 (arrow) and anterior subluxation of C2 on C3. b Congenital spondylolysis of C2 in a patient with pyknodysostosis. Note that the defect has a smooth edge

13 c

should be suspected, if found. If there is greater than 50 % of anterior wedging, then instability due to posterior disruption is likely [1, 25] (Fig. 15). Wedging is due to under-ossification of the growth plates and must not be confused with fractures. Vertebral end plate injury is usually secondary to hyperextension and is mostly reported in adolescents. The injury usually involves the inferior end plate, perhaps because the uncinate processes protect the superior region. The `ring apophyses' or secondary centres of ossification usually appear between 10 and 12 years of age [1] and must not be mistaken for fractures. Type 1 injury or complete transection of the growth plate is usually as-

689

of a vertebral body, are clues to the diagnosis [1, 18]. Flexion and extension radiographs or fluoroscopy of the neck should be taken to supplement the standard views and will usually show the instability [22, 23, 78, 79].

Cervicothoracic junction injury

a

b Fig. 15 a, b Fracture of C5. a Lateral radiograph shows involvement of both the superior and inferior end plates. b Axial CT shows the fracture through the body of C5

sociated with severe cord and ligament damage and tends to occur in younger children. Type III injuries are reported in adolescents and characteristically show displacement of a corner of the ring apophysis. These heal well. The anterior longitudinal ligament, which is displaced, is osteogenic in children and leads to rapid bony fusion anteriorly resulting in bridging osteophytes [1]. Purely ligamentous injuries may occur in children and result in delayed instability even when the initial radiographs are normal [1, 18]. Affected children present with persistent pain, stiffness or muscle spasm, or deformity. Prevertebral swelling, loss of lordosis, widening of interspinous distance and the occasional dimple fracture

Injury to the cervicothoracic junction is rare [10, 17, 19, 20, 39] and almost never reported in those under 10 years of age [10, 19], but is the commonest affected site of cord injury in breech deliveries, many of whom present as SCIWORA (see preceding section) [80]. In contrast, cervicothoracic injury accounts for 9 % of all cervical spinal injury amongst adults [23] and approximately 20 % of these injuries are missed on initial evaluation [3, 81, 82]. There were no reports of an `asymptomatic' cervical spine injury in the paediatric literature to 1992 [77]. The report by Orenstein et al. [79] of nine children in whom there were delays in diagnosing cervical spine injury included only one case of cervicothoracic junction injury. None of the children in his study were asymptomatic. A recent report by Baker et al. [42] of 72 children with cervical spine injury noted that 18 % were `asymptomatic' with respect to the cervical spine, but had both a high-risk injury mechanism and other painful distracting injuries. For practical purposes, therefore, if a child is alert, able to give a history and asymptomatic without neurological signs, the likelihood of missing a significant injury at C7/T1, if not clearly seen on the lateral radiograph, is small. Multiple attempts to repeat views of this area with arm traction or swimmer's views can cause distress in the conscious child. Oblique views are preferable to multiple attempts at lateral views but if the AP view shows no loss of alignment and there is no neurological deficit, one must question the need for multiple attempts to achieve adequate radiographic demonstration of this difficult area.

Thoracic spine injury Approximately 30 % of all spinal injury in children occurs in the thoracic spine [5, 11, 15, 30, 83]. They are flexion/extension injuries. Associated injuries to the chest and abdomen are common in severe trauma [15, 18]. Fractures of the transverse processes of T1 and T2 are associated with brachial plexus injury [15] and `seatbelt' fractures of the lower thoracic and upper lumber spine have associated visceral or small bowel injury in up to 50 % of cases [13, 15, 84]. Road traffic accidents and falls account for most of the injuries, but thoracic spinal injury has also been reported in non-accidental injury (NAI) [33±35]. A thoracic level of injury accounts for 20±30 % of all cases of SCIWORA [29, 40].

690

Fig. 16 a, b Fracture of T5. a CT shows a burst fracture of T5 involving the body and posterior elements. Note the paraspinal haematoma. b MRI shows the acute angular kyphosis due to the fracture of T5 Fig. 17 a, b Fracture of T6. a MRI at the time of injury shows anterior wedging of T6. b Follow-up radiographs of the thoracic spine over subsequent years show almost complete restoration of vertebral body height

16 a

17 a The most frequent fracture in the thoracic spine is a vertebral body compression fracture almost invariably due to a fall [1, 15, 24, 85]. Anterior wedging is seen on the radiograph [1, 24]. A difference in height of more than 3 mm between the anterior and posterior cortices of the vertebral body is considered significant and distinguishes this from the normal variant appearances of slight anterior vertebral wedging [7]. Most fractures are

16 b

17 b visible on the lateral projection, though a small number are seen only on the AP view. Wedging of more than 25 % is suggestive of significant posterior ligamentous damage and resultant instability [15]. If this is at multiple levels, there may be a kyphosis. Acute angular kyphosis is seen in association with fracture dislocation (Fig. 16). Most thoracic fractures in children occur from T4 to T12 and are multiple [1, 20, 24] and often underes-

691

Fig. 18 a, b `Chance' fracture of lumbar vertebra. a AP view shows wide separation between the spinous processes of L2 and L3 with a fracture line visible through the pedicles (arrow), resulting in an apparently `empty' vertebral body. (Vertical linear opacity overlying L2±3 disc level is an artefact.) b The horizontal fracture line through the upper vertebral body and posterior elements is well shown on the lateral view Fig. 19 a, b `Burst' lumbar vertebral fracture. a AP radiograph shows fracture dislocation of L2. Note widened interpedicular distance and separation of the spinous processes. b CT defines the full extent of the `burst' fracture and shows the retropulsed bony fragments in the spinal canal. Despite the severity of the injury, this child remained neurologically intact as the injury was below the level of the cord

18 a

18 b

19 b

19 a

692

Fig. 20 a, b Posterior limbus fracture. a The postero-inferior endplate of L4 is fractured and displaced posteriorly into the spinal canal (arrow). b T2W MRI shows high-signal oedema in the marrow at the site of injury. Note how the L4/5 disc also protrudes into the spinal canal

a timated as, on radiographs, there may be wedging of only one or two vertebral bodies. Scintigraphy or MRI will show more extensive injury [24]. Thoracic fractures secondary to NAI are usually of the wedge compression type and are thought to represent hyperflexion injuries [33, 35]. Avulsion fractures of the spinous processes, due to hyperflexion injury, also occur [36]. Restoration of vertebral body height and shape often occurs after compression fractures in children [1, 24]. The amount of restoration is proportional to the severity of the fracture and the age of the child [1] (Fig. 17). Injury to the vertebral end plate may lead to a lack of vertebral growth and subsequent deformity [1].

Lumbar spinal injury Seventeen to twenty-seven percent of spinal injury in children occurs in the lumbar spine [5, 11, 30]. Fractures at the thoracolumbar junction are associated with neurological deficit in up to 40 % [1, 45]. The spinal cord ends at the level of L1, so fractures distal to this, though they may be associated with injury to the cauda equina, cause less neurological injury. The relatively high incidence of injury at the thoracolumbar junction is due to the large range of motion at this level combined with a lack of supporting structures and the changing orientation of the facet joints [45]. Fractures are commonly wedge compression type and usually multiple [1, 24]. `Seatbelt' fractures tend to

b occur at the thoracolumbar junction and associated visceral or mesenteric injury is frequent, occurring in up to 50 % [1, 13, 15, 24]. The anterior fulcrum of flexion results in disruption of the posterior elements and extension of the fracture horizontally into the vertebral body. In the `chance' fracture there is horizontal fracture through the spinous process, pedicles and posterior vertebral body. A `Smith' fracture is similar, but the bony posterior elements are spared, injury occurring to the posterior spinal ligaments instead [15, 24]. On the AP radiograph, there may be an apparent `empty vertebral body' because of distraction and displacement of the posterior elements which are normally visible through the vertebral body (Fig. 18). The lateral view shows the fracture and any associated displacement. CT is essential to fully demonstrate the extent. As the fracture tends to be in the horizontal plane, sagittal reformatting is necessary in addition to the axial slices. `Burst' fractures occur with axial compression forces, causing endplate rupture and herniation of the disc into the body of the vertebra. Retropulsed fragments, usually from the postero-superior aspect of the vertebral body, can cause spinal canal impingement. A burst fracture can be distinguished from a simple compression fracture by noting disruption of the posterior cortex of the vertebral body and by widening of the interpedicular distance on the AP view [1, 15] (Fig. 19). Isolated fractures of the neural arch and transverse processes are usually due to direct trauma or lateral hyperflexion and extension [1]. Upper lumbar transverse process fractures are associated with renal injury and

693

b

a Fig. 21 a±c Lumbar spondylolysis. a Lateral radiograph shows spondylolysis at L5/S1 with grade I spondylolisthesis. b Isotope bone scan in another patient shows increased uptake in the pars interarticularis (confirmed on oblique bone scan images) of L4. c Reverse-angle CT in the plane of the pars shows the well-corticated bilateral pars defects

lower lumbar transverse process fractures with injury to the lumbosacral plexus [15]. Lumbar fractures in NAI are typically of the anterior compression type and occur from transmitted forces, as the child is forcibly thumped on its bottom with hyperflexion or as it impacts and hyperflexes against a solid wall as it is thrown away [33, 35, 36]. Lumbar apophyseal injury is increasingly recognised as a cause of low back pain in adolescents, particularly in gymnasts and weight lifters. The injury involves avulsion of the posterior portion of a lumbar apophysis, usually a type 3 or 4 growth mechanism injury. The disc and ring apophysis and sometimes a metaphyseal fragment are displaced posteriorly into the spinal canal and may be seen on a lateral radiograph, but usually requires CT (Fig. 20). If left to heal conservatively, the displaced fragment ossifies in its abnormal location, resulting in narrowing of the spinal canal at that level [1]. Intervertebral disc prolapse is rare in childhood. Trauma is the usual mechanism and there may be an associated fracture of the posterior end plate.

Spondylolysis Spondylolysis is a stress fracture through the pars interarticularis [1] due to repetitive flexion and extension in

c the adolescent spine. It is more frequent in athletes than in sedentary children and has been reported in 11 % of female gymnasts [1]. Spondylolysis may also occur in a dysplastic form and is due to congenital abnormalities of the lower lumbar spine, for example an elongated isthmus, or partial sacralisation of the vertebral body. When unilateral, as it is in 40 % of cases, there is often sclerosis of the contralateral pedicle, leading to an erroneous diagnosis of osteoid osteoma. If spondylolysis is obvious on AP and lateral views, oblique radiographs add little further information. If it is not evident on the AP and lateral views, then oblique views should be added. Bone scintigraphy, both planar and SPECT, can identify occult cases. Reverse-angle CT slices orientated in the plane of the pars show the defects most accurately (Fig. 21). The defect is shown to best advantage on CT, which is indicated for treatment planning, as it shows whether the defect is bilateral, the extent of the defect and the bony anatomy for planning screw fixation. Spondylolisthesis is graded according to the severity of forward slip. Grade I indicates forward slip of less than one quarter of the vertebral body diameter and grade V indicates complete forward slip. Complications include nerve root compression and spinal canal stenosis.

Imaging in spinal trauma Injuries in childhood range from the minor to the severe and appropriate management and radiological investigation vary accordingly. Most Accident and Emergency departments have a standard protocol to deal with patients who are victims of polytrauma or high-velocity injury. Implicit in the trauma resuscitation is the need to protect the spinal cord from iatrogenic injury [22]. The child is put in a rigid cervical collar until full clinical and radiological assessment is made.

694

A supine, lateral cervical spine radiograph, which should include the entire cervical spine from the occiput to the cervicothoracic junction, a supine chest radiograph and full abdomen and pelvis radiograph complete the initial survey. If difficulty is experienced in demonstrating the cervicothoracic junction, the shoulders may be pulled down or a `swimmer's view' performed. The chest radiograph often affords a good AP view of the lower cervical spine. Subsequent radiological investigation is tailored to the relevant clinical question. Though traditionally an `open-mouth' peg view is recommended, the reality is that axial CT with coronal reconstruction has now replaced `peg views' when the mechanism of injury is such that it is needed. A recent study by Swischuk et al. [87] questions whether the open mouth odontoid view is necessary in children under 5 years if the lateral view is normal, the conclusion of which was a fracture miss rate of 0.007 per practice year, if not done. Children with obvious neurological damage need full radiological assessment of the spine, including CT, to delineate bony injury and MRI to assess damage to the cord [22, 31]. Imaging requirements in the less severely traumatised child are more difficult to define. The adult literature reports that only 50 % of cervical spinal injuries are identified prospectively on the basis of history and physical examination [22, 23]. Children are even more difficult to assess [77]. No single clinical predictor has a 100 % sensitivity for predicting spinal injury, but a complaint of significant neck pain, neurological findings, or a history of high risk mechanism of injury, appear to be the best predictors [12, 26, 41]. The lateral decubitus cervical radiograph is reported as having a sensitivity of between 75 and 85 % in the adult literature [15, 16, 22, 23, 47, 86] and 79±85 % in the paediatric literature [31, 43]. Overall, up to 13.9 % of all spinal fractures are missed on initial assessment [82]. Most missed injuries in the cervical spine occur in the upper cervical spine and at the cervicothoracic junction [3, 31]. The addition of an AP and an open-mouth view increases the sensitivity to 95 % under optimal conditions in adult patients [15, 22, 47], and similar sensitivity of 94 % was reported in the study of children by Baker et al. [42]. If spinal injury is identified at one level, the entire spine should be imaged, as multilevel injury occurs in 16 % of cases [13, 14], but this must be related to the mechanism of injury. MRI is the best method of assessment. The use of CT increases the sensitivity for fracture detection in cervical spine injury in children to 93 % [43]. Injuries difficult to visualise on CT are fractures in the axial plane such as undisplaced odontoid fractures, spinous process, superior facet, and laminar fractures [45, 48]. Sagittal reformatting helps. Ligamentous injury is best assessed using flexion and extension views

and MRI [31]. The flexion/extension manoeuvre should be performed by the patient without assistance, under supervision by a radiologist or orthopaedic surgeon. The role of CT in spinal injury is two-fold. Firstly, CT will further define the bony injury in cases where the initial radiographs have already shown a fracture. CT provides more detailed anatomical information and allows assessment of impingement on the thecal sac and spinal cord from extradural sources such as retropulsed bony fragments or haematoma [15, 50, 88]. Secondly, CT may detect injuries that are not readily apparent on plain radiographs, especially in cases where there is a high index of suspicion of bony injury. Sensitivity figures of 93 % in children are quoted for CT [41]. CT is especially useful for those parts of the spine where fractures are commonly missed such as the occiput/C1±2 level and the cervicothoracic junction [3, 22, 81, 88]. Even in cases where injury has been detected on the initial radiographs CT often reveals other occult injuries [12, 89]. CT is particularly useful in assessing Jefferson fractures, atlanto-axial rotatory disorders, burst fractures with retropulsed fragments and injuries to the cervicothoracic junction [48]. The CT examination should be tailored to the plain radiography findings, and the clinical setting. Thin slices of 2±3 mm are recommended with sagittal and coronal and 3D reformatting. It is useful to angle the gantry in the plane of the posterior arch if injury is suspected there. The radiation dose from CT may be reduced by using reduced mAs if bony injury is the main problem. Spinal cord, disc and ligament injury is best imaged by MRI [15, 45, 48, 90, 91]. MRI may be impracticable or impossible in the acute trauma setting [15, 48]. Where MRI is possible in the acute setting its greatest potential role is in the identification of remediable intraspinal problems in those patients with an incomplete neurological deficit [90, 92]. The soft-tissue components of spinal trauma, unsuspected acute trauma-related disc herniation, epidural haematoma and vascular injury to the vertebral arteries are often injured in cervical spine fractures [91, 93] (Fig. 22). The sequences used should include sagittal images of the whole spine with spin-echo T1W and T2W sequences. A gradient-echo T2W sequence will differentiate cord oedema from haemorrhage, cord oedema having a better prognosis than haemorrhage [92]. Axial images using T1 W, T2 W and gradient-echo sequences are required at areas of bony injury or where an abnormality is seen on sagittal views. The appearance of the spinal cord on MRI allows prediction of neurological outcome [39, 94±98]. Cord transection and major haemorrhage are associated with a poor outcome, minor haemorrhage or oedema with moderate-to-good recovery, and absence of abnormal

695

Fig. 22 a, b Sagittal MRI show high signal in the prevertebral soft tissues on both T1W (a) and T2W (b) images indicating haemorrhage (arrows)

a cord signal with complete recovery. Direct sagittal imaging depicts overall spinal alignment and facet alignment and perched or locked facets are more easily appreciated than on CT [91]. In the chronic setting, MRI allows detection of the late sequelae of spinal injury such as the development of syringomyelia, cord atrophy and delayed cord compression [90].

Review of radiographs An ordered review is essential. Though discussed in part in the individual sections, the problem areas are reviewed again here for ease of reference. Cervical spine Lateral projection All seven cervical vertebrae and the cervicothoracic junction should be visualised. Parallelism of vertebrae is essential. The most important line in children is the spinolaminar line, a line joining the bases of the spinous processes which should be a smooth arc (Fig. 23). The problem is that this line does not always intersect the base of the spinous process of C2, which even in normal children is apparently `out of alignment', lying posterior to the spinous processes of C1 and C3. See `pseudosubluxation' in the section on normal variants. Other lines are:

b · The anterior spinal line joining the anterior borders of the vertebral bodies · The posterior spinal line joining the posterior borders of the vertebral bodies (Fig. 23) These last two are useful in children above the age of about 7 years. Below this, pseudosubluxations of C2 on C3 or C3 on C4 are common normal variants and cause disruption of these lines [10, 31, 51, 77]. The facets should all lie parallel. The distances between individual spinous processes should show no significant widening or `fanning'. The distances are variable, being least at C4±5 level and greatest at C6±7 [31]. There are no specific figures as to the maximum width in children, but in adults a difference of 2 mm between three contiguous levels, measured at the spinous process base, is considered significant [15]. On the AP view, interspinous distance which measures more than 1.5 times the one above and more than 1.5 times the one below indicates anterior cervical dislocation [96] (Fig. 24). A lack of uniform angulation at the interspaces in flexion may occur in normal children [49]. The atlanto-axial distance, measured from the postero-inferior margin of the anterior arch of C1 to the odontoid should be no more than 5 mm [15, 54]. The basion (base of clivus/anterior margin of foramen magnum) should lie within 12 mm of the superior continuation of a line drawn along the posterior cortex of the body of C2 [52] (Fig. 4). The anterior vertebral body height should be no more than 3 mm less than the posterior height [15, 31, 51].

696

a

Fig. 23 Normal cervical spine (a spinolaminar line, b posterior spinal line)

Soft tissue measurements are of limited value in predicting injury. Significant spinal injury may be present without any increase in the prevertebral soft tissues, especially posterior spinal injury [23]. The cervicocranial prevertebral soft tissue contour, unless modified by adenoidal tissue, should follow the contour of the anterior cortex of the atlas, the axis and the caudal portion of the clivis [97]. A width of 7 mm [15, 19, 45] or 3/4 vertebral body width at the C2 level is quoted. The soft tissues anterior to C6 should be less than 14 mm in children [15]. At the cervicothoracic junction, the normal contour of the prevertebral soft tissues should also follow the contour of the vertebral bodies and appear to `tuck' or `dip' into the thoracic inlet [97]. However, the width of the prevertebral soft tissues in children, especially at and above the level of the glottis, is heavily influenced by neck posture and by crying.

b Fig. 24 a, b Widened interspinous distance in a patient who sustained a flexion injury with fracture of C7. a Note the increased distance between the spinous processes of C6 and C7. b Note the `fanning' or widening between the spinous processes of C6 and C7 (asterisk)

AP projection

AP open-mouth odontoid

On the AP view, the spinous processes and facets should be in straight alignment and evenly spaced. The interspinous distances must vary by no more than 50 % from level to level [96]. The spinous processes should form a straight line; any offsetting may indicate unilateral facet dislocation.

The lateral masses of C1 may extend beyond the lateral margins of the articular surface of C2 in most children up to 4 years of age, including over 90 % of those aged 1±2 years, and is termed `pseudospread'. The total lateral offset in cases of `pseudospread' is usually less than 6 mm, although a total of 8 mm has been recorded in two children who did not have any injury. `Pseudo-

697

Normal variants Normal variants may cause difficulties for the inexperienced radiologist in the assessment of the spine. There are variations due to displacement of vertebrae that may resemble subluxation, variations of curvature that may resemble spasm and ligamentous injury and variations related to growth centres that may be confused with fractures.

Fig. 25 Fracture dislocation of T3±4. Note the widened distance between ribs 3 and 4. The spinous process of T3 is `missing', leading to an apparently `empty' vertebral body. The left pedicle of T4 is also missing

spread' occurs less commonly as the child gets older, but is still found in 18 % of 7 year olds [51]. Thoracic and lumbar spine Lateral projection The anterior, posterior and spinolaminar lines should show parallel alignment. A difference of more than 3 mm between anterior and posterior vertebral body height is considered abnormal [15, 31, 51]. Alternatively, the ratio of anterior vertebral body height to posterior should be greater than 0.95 [24]. AP projection The thoracic paraspinal lines should be scrutinised for evidence of paraspinal haematoma. The posterior elements should be clearly visible through the vertebral body. An apparently `empty' vertebral body indicates fracture-dislocation [19, 25] (Fig. 25).

1. Absence of normal cervical lordosis occurs in 14 % of normal children [31, 51]. 2. `Pseudospread' of the atlas on the axis of up to 6 mm is common up to age 4 and occurs up to age 7 years [52]. 3. `Pseudosubluxation' or anterior displacement of C2 on C3 of up to 4 mm occurs in 40 % of normal children under 7 years of age and in 24 %, overall, of those aged under 16 years [10, 31, 53, 77, 98]. It is thought to result from ligamentous laxity and is termed `pseudosubluxation'. In contrast to pathological subluxation, this malalignment is not exaggerated and often improves with flexion and extension [31]. When anterior displacement of C2 on C3 is noted on the lateral radiograph, a helpful line, devised by Swischuk, aids differentiation of pseudosubluxation from pathological subluxation. A line drawn connecting the anterior cortices of the spinous processes of C1 and C3 should intersect or come within 1 mm of the anterior cortex of the spinous process of C2 (Fig. 26). In pathological subluxations this line misses the spinous process of C2 by more than 2 mm [31, 77]. Pseudosubluxation of C3 on C4 also occurs in 20 % of those less than 7 years of age and 9 % of those less than 16 years of age [51]. 4. Anterior `wedging' of up to 3 mm may occur in normal children [31]. 5. Over-riding of up to two-thirds of the anterior arch of C1 above the odontoid occurs in up to 20 % of children under 7 years of age [51]. 6. There may be a lack of uniform angulation at the interspaces in flexion in 16 % of those under 16 years of age [31, 51]. 7. The synchondrosis at the base of the odontoid persists in 76 % of those under 7 years of age and 30 % of those under 16 years of age [1, 51]. 8. The apical odontoid epiphysis is visible in 26 % of those under 7 years and 9 % of those under 16 years [31, 51]. 9. Secondary centres of ossification of the spinous processes may resemble avulsion fractures [51]. 10. There may be incomplete ossification of the posterior arch of C1 (Fig. 27). 11. A bifid spinous process can cause loss of alignment on the AP view of the cervical spine.

698

27 b

26

27 a

Fig. 26 `Pseudosubluxation' of C2 on C3. A line joining the anterior cortices of the spinous processes should intersect or come within 1 mm of the anterior cortex of the spinous process of C2 Fig. 27 a, b Incomplete ossification of the posterior arch of C1 demonstrated on a plain radiography and b CT

13. Unfused ring apophyses, which usually appear between 10 and 12 years of age, may be confused with injuries [1, 2]. End-plate fractures are usually displaced and associated with soft tissue swelling.

12. Primary spondylolysis of C2 is a rare normal variant, usually associated with congenital anomalies of the cervical spine or pyknodysostosis [76, 99] (Fig. 14 b).

References 1. Ogden JA (1990) Skeletal injury in the child. In: Spine, 2nd edn. Saunders, Philadelphia, pp 57l-562 2. Renton P (1994) Trauma. In: Carty HM, Brunelle F, Shaw D, et al (eds) Imaging Children. Churchill Livingstone, Edinburgh, pp 1167±1175 3. Bohlman H (1979) Acute fractures and dislocations of the cervical spine. J Bone Joint Surg Am 61: 1119±1142 4. Haffner DL, Hoffer MM, Wiedbusch R (1993) Etiology of children's spinal injuries at Rancho Los Amigos. Spine 18: 679±684 5. Hamilton MG, Myles ST (1992) Pediatric spinal injury: review of 174 hospital admissions. J Neurosurg 77: 700±704 6. Hamilton MG, Myles ST (1992) Pediatric spinal injury: review of 61 deaths. J Neurosurg 77: 705±708

7. Dietrich AM, Ginn-Pease ME, Bartkowski HM (1991) Pediatric cervical spine fractures: predominantly subtle presentation. J Pediatr Surg 26: 995±1000 8. Farley FA, Hensinger RN, Herzenberg JE (1992) Cervical spinal cord injury in children. J Spinal Disord 5: 410±416 9. Babcock JL (1975) Spinal injuries in children. Pediatr Clin North Am 22: 487±500 10. Hill SA, Miller CA, Kosnik EJ, et al (1984) Pediatric neck injuries. J Neurosurg 60: 700±706 11. Anderson JM, Shutt AH (1980) Spinal injury in children: a review of 156 cases seen from 1950 through 1978. Mayo Clin Proc 55: 499±504 12. Jaffe DM, Binns H, Radkowski MA (1987) Developing a clinical algorithm for early management of cervical spine injury in child trauma victims. Ann Emerg Med 16: 270±276

13. Mann DC, Dodds JA (1993) Spinal injuries in 57 patients 17 years or younger. Orthopaedics 16: 159±164 14. Hadley MN, Zabranski JM, Browner CM (1988) Pediatric spinal trauma. J Neurosurg 68: 18±24 15. Murphey MD, Batnitzky S, Bramble JM (1989) Diagnostic imaging of spinal trauma. Radiol Clin North Am 27: 855±872 16. Gerrelts BD, Petersen EU, Mabry J (1991) Delayed diagnosis of cervical spine injuries. J Trauma 31: 1622±1625 17. Kewalramani LS, Tori JA (1980) Spinal cord trauma in children: neurologic patterns, radiologic features and pathomechanics of injury. Spine 5: 11±18 18. McSweeney T (1988) Spinal injuries. In: Alpar EK, Owen R (eds) Pediatric Trauma, pp 131±145

699

19. Apple JS, Kirks DR, Merten DF, et al (1987) Cervical spine fractures and dislocations in children. Pediatr Radiol 17: 45±49 20. Ruge JR, Sinson GP, McLone DG (1988) Pediatric spinal injury: the very young. J Neurosurg 68: 25±30 21. Kewalramani LS, Krays JF, Sterling HM (1980) Acute spinal-cord lesions in a pediatric population: epidemiological and clinical features. Paraplegia 18: 206±219 22. Vandemark RM (1990) Radiology of the cervical spine in trauma patients: practice pitfalls and recommendations for improving efficiency and communication. AJR 155: 465±472 23. Harris MB, Waguespark AM, Kranlage S (1997) ªClearingº cervical spine injuries in polytrauma patients: is it really safe to remove the collar? Orthopaedics 20: 903±907 24. Hegenbarth R, Ebel RD (1976) Roentgen findings in fractures of the vertebral column in childhood. Pediatr Radiol 5: 34±39 25. Bonadio WA (1993) Cervical spine trauma in children: Part II. Mechanisms and manifestations of injury, therapeutic considerations. Am J Emerg Med 11: 256±278 26. Rachesky I, Boyce WT, Duncan B (1987) Clinical prediction of cervical spine injuries in children. Am J Dis Child 141: 199±201 27. Burke DC (1971) Spinal cord trauma in children. Paraplegia 9: 1±14 28. Abrams IF, Bresnan MJ, Zuckerman JE (1973) Cervical cord injuries secondary to hypertension of the head in breech presentation. Obstet Gynecol 41: 369±378 29. Pang D, Wilberger JG Jr (1982) Spinal cord injury without radiographic abnormalities in children. J Neurosurg 57: 114±129 30. Hacher HJ (1977) Spinal cord injury in children and adolescents: diagnostic pitfalls and therapeutic considerations in the acute stage. Paraplegia 15: 55±64 31. Bonadio WA (1993) Cervical spine trauma in children. I. General concepts, normal anatomy, radiographic evaluation. Am J Emerg Med 11: 158±165 32. Leventhal H (1969) Birth injuries of the spinal cord. J Pediatr 56: 447 33. Swischuk LE (1969) Spine and spinal cord trauma in the battered child syndrome. Radiology 92: 733±738 34. Rooks VJ, Sisler C, Burton B (1998) Cervical spine injury in child abuse: report of two cases. Pediatr Radiol 28: 193±195

35. Cullen JC (1975) Spinal lesions in battered babies. J Bone Joint Surg Br 57: 364±366 36. Davidson RG (1988) Atlantoaxial instability in individuals with Down syndrome: a fresh look at the evidence. Pediatrics 81: 857±865 37. Stevens JM (1994) Congenital lesions of the vertebral column. In: Carty HM, Brunelle F, Shaw D, et al (eds) Imaging children. Churchill Livingstone, Edinburgh, p 1422 38. Dickman CA, Zabramski JM, Hadley MN et al (1991) Pediatric spinal cord injury without radiographic abnormalities: report of 26 cases and review of the literature. J Spinal Disord 4: 296±305 39. Kriss VM, Kriss TC (1996) SCIWORA (spinal cord injury without radiographic abnormality) in infants and children. Clin Pediatr (Phila) 35: 119±124 40. Yngve DA, Harris WP, Herudon WA (1988) Spinal cord injury without osseous spine fracture. J Pediatr Orthop 8: 153±159 41. Grabb PA, Pang D (1994) Magnetic resonance imaging in the evaluation of spinal cord injury without radiographic abnormality in children. Neurosurgery 35: 403±414 42. Baker C, Kadish H, Schwik J (1999) Evaluation of pediatric spine injuries. Am J Emerg Med 17: 230±234 43. Duprez T, De Merlier Y, Clapuyt P (1998) Early cord degeneration in bifocal SCIWORA. Pediatr Radiol 28: 186±188 44. Bucholz RW, Burkhead WZ (1979) The pathological anatomy of fatal atlantooccipital dislocations. J Bone Joint Surg Am 16: 248±250 45. Pathria MN, Petersilge CA (1991) Spinal trauma. Radiol Clin North Am 29: 847±865 46. Harris JH, Carson GC, Wagner LK (1994) Radiologic diagnosis of traumatic occipitovertebral dissociation. 2. Comparison of three methods. AJR 162: 887±892 47. Maves CK, Souza A, Prenger EC (1991) Traumatic atlanto-occipital disruption in children. Pediatr Radiol 21: 504±507 48. El-Khoury GY, Kathol MH, Daniel WW (1995) Imaging of acute injuries of the cervical spine: value of plain radiography, CT and MR imaging. AJR 164: 43±50 49. Grabb BC, Frye TA, Hedlund GL, et al (1999) MRI diagnosis of suspected atlanto-occipital dissociation in childhood. Pediatr Radiol 29: 275±281

50. Harris JH, Carson GC, Wagner LK (1994) Radiologic diagnosis of traumatic occipitovertebral dissociation. 1. Normal occipitovertebral relationships on lateral radiographs of supine subjects. AJR 162: 881±886 51. Cattell HS, Filtzer DL (1965) Pseudosubluxation and other normal variations in the cervical spine in children. J Bone Joint Surg Am 47: 1295±1309 52. Suss RA, Zimmerman RD, Leeds NE (1983) Pseudospread of the atlas: false sign of Jefferson fracture in young children. AJR 140: 1079±1082 53. Locke GR, Gardner JI, Van Epps EF (1966) Atlas-dens interval (ADI) in children AJR 97: 135±140 54. Lusted LB, Keats TE (1972) Atlas of roentgenographic measurement, 3rd edn. Year Book, Chicago 55. Cremers MJ, Bol E, de Roos F, et al (1993) Risk of sports activities in children with Down's syndrome and atlantoaxial instability. Lancet 342: 511±514 56. Editorial (1989) Atlantoaxial instability in Down's syndrome. Lancet 1: 24 57. Fielding JW, Hawkins RJ (1977) Atlanto-axial rotatory fixation. J Bone Joint Surg Am 59: 37±44 58. Wortzman G, Dewar F (1968) Rotary fixation of the atlantoaxial joint: rotational atlantoaxial subluxation. Radiology 90: 479±487 59. Phillips WA, Hensinger RN (1989) The management of rotatory atlanto-axial subluxation in children. J Bone Joint Surg Am 71: 664±668 60. Weinstein SL (1994) The pediatric spine, vol 1. Raven Press, New York, pp 408±420 61. Lukhelc M (1996) Atlanto-axial rotatory fixation. S Afr Med J 86: 1549±1552 62. Maheshwaran S, Sgouros S, Jeyapalan K, et al (1995) Imaging of childhood torticollis due to atlantoaxial rotatory fixation. Childs Nerv Syst 11: 667±671 63. El-Khoury GY, Clark CR, Gravett AW (1984) Acute traumatic rotatory atlantoaxial dislocation in children. J Bone Joint Surg Am 66: 774±777 64. Mathern G, Batzdorf U (1989) Grisel's syndrome. Clin Orthop Rel Res 244: 131±146 65. Dvorak J, Hayek J, Zehnder R (1987) CT ± functional diagnostics of the rotatory instability of the upper cervical spine. Spine 12: 726±731 66. Viras I, Zubieta JL, Arriagada C, et al (1999) Physiological rotatory C1-C2 subluxation in children (abstract). Scientific poster No. 12±038 presented at ECR 1999 Vienna 67. Johnson D, Fergusson G (1986) Early diagnosis of atlantoaxial rotatory fixation. J Bone Joint Surg Br 68: 698±701

700

68. Kowalski M, Cohen W, Cooper P, et al (1987) Pitfalls in the CT diagnosis of atlantoaxial rotatory subluxation. AJNR 8: 697±702 69. Kawabe N, Hirotani M, Tanaka O (1989) Pathomechanism of atlantoaxial rotatory fixation in children. J Pediatr Orthop 9: 569±574 70. Hohl M, Baker M (1964) The atlantoaxial joint. J Bone Joint Surg Am 64: 1739±1752 71. Penning L (1978) Normal movements of the cervical spine. AJR 130: 317±326 72. Fielding J, Stillwell W, Chynn K, et al (1978) Use of computed tomography for the diagnosis of atlanto-axial rotatory fixation. J Bone Joint Surg Am 68: 1102±1104 73. Flodmark OC, Grane P, Gustavsson B, et al (1999) Atlanto-axial rotatory fixation: a new understanding of an old disease (abstract). Abstract book, Society for Pediatric Radiology, pp 60±61 74. Harris JH Jr, Mirvis SE (1996) The radiology of acute cervical spine trauma, 3rd edn. Williams & Wilkins, Baltimore, pp 42l ±474 75. Fielding JW, Hensinger RN, Hawkins RJ (1980) Os odontoideum. J Bone Joint Surg Am 62: 376±383 76. Currarino G (1989) Primary spondylolysis of the axis vertebra (C2) in three children, including one with pyknodysostosis. Pediatr Radiol 19: 535±538 77. Swischuk LE (1977) Anterior displacement of C2 in children: physiologic or pathologic? Radiology 122: 759±763 78. Making the best use of a department of clinical radiology (1998) Royal College of Radiologists, London

79. Orenstein JB, Klein BL, Oschsenschlager DW (1992) Delayed diagnosis of pediatric cervical spine injury. Pediatrics 89: 1185±1188 80. Franken EA (1975) Spinal cord injury in the newborn infant. Pediatr Radiol 3: 101±104 81. Evans DK (1983) Dislocations at the cervico thoracic junction. J Bone Joint Surg Br 65: 124±127 82. Reid DC, Henderson R, Saboe L (1987) Etiology and clinical course of missed spine fractures. J Trauma 27: 980±986 83. Turgut M, Akpinar G, Akala N, et al (1996) Spinal injuries in the pediatric age group: a review of 82 cases of spinal cord and vertebral column injuries. Eur Spine J 5: 148±152 84. Dehner JR (1971) Seatbelt injuries of the spine and abdomen. AJR 3: 833±843 85. Holdsworth F (1970) Fractures, dislocations and fracture-dislocations of the spine. J Bone Joint Surg Am 52: 1534±1551 86. Shaffer MA, Doris PE (1981) Limitation of the cross table lateral view in detecting cervical spine injuries: A retrospective analysis. Ann Emerg Med 10: 508±513 87. Swischuk LE, John SD, Hendricke P (1999) Is the open mouth odontoid view necessary in children younger than 5 years? (abstract). Abstract book, Society for Pediatric Radiology, p 61 88. Blacksin MF, Lee HJ (1995) Frequency and significance of fractures of the upper cervical spine detected by CT in patients with severe neck trauma. AJR 165: 1201±1204 89. Lynch D, McManus F, Ennis JJ (1986) Computed tomography in spinal trauma. Clin Radiol 37: 71±76

90. Kerslake RW, Jaspan T, Worthington BS (1991) Magnetic resonance imaging of spinal trauma. Br J Radiol 64: 386±402 91. Mirvis SE (1998) MR imaging for assessing acute vertebral trauma. Semin Musculoskelet Radiol 2: 27±43 92. Terk MR, Hume-Neal M, Fraipont M (1997) Injury of the posterior ligament complex in patients with acute spinal trauma: evaluation by MR imaging. AJR 168: 1481±1486 93. Lee RR (1996) MR imaging and cervical spine injury. Radiology 201: 617±618 94. Mascalchi M, Dal Prozzo G, Dini C (1993) Acute spinal trauma: prognostic value of MRI appearances at 0.5 T. Clin Radiol 48: 100±108 95. Schaefer DM, Flanders A, Northrup BE (1989) Magnetic resonance imaging of acute cervical spine trauma. Spine 14: 1090±1095 96. Naidich JB, Naidich TP, Garfein C (1977) The widened interspinous distance: a useful sign of anterior cervical dislocation in the supine frontal projection. Radiology 123: 113±116 97. Harris JH Jr, Mirvis SE (1996) The radiology of acute cervical spine trauma, 3rd edn. Williams & Wilkins, Baltimore, pp 180±212 98. Shaw M, Burnett H, Wilson A, et al (1999) Pseudosubluxation of C2 on C3 in polytraumatized children ± prevalence and significance. Clin Radiol 54: 377±380 99. Williams J, Baker D, Miller W (1999) CT appearance of congenital defect resembling the Hangman's fracture. Pediatr Radiol 29: 549±550