♦ Clinical criteria and the nature of the injury determine who needs imaging
♦ Plain films are still commonly employed though CT finds more fractures
♦ Alignment, bony contour, cartilage (Disc and facets) and soft tissue are assessed in turn (ABC’S)
♦ CT is superior to MRI in assessing the bony configuration of fracture
♦ MRI is superior to CT in assessing the ligament tears and associated disc herniations
♦ Plain films have little role in the assessment of more chronic back pain and radiculopathy
♦ Sacral insufficiency fractures may be misdiagnosed as metastases on MRI.
The aims of this chapter are to:
1) Describe how to clear the spine
2) Describe the common injury patterns and radiological findings.
The concept of risk of spinal injury is very important in helping to decide the appropriate imaging strategy. This is often difficult as a good history of mechanism and potential injury pattern is not always available.
The NEXUS study retrospectively applied the following clinical criteria in relation to potential cervical trauma: absence of midline tenderness, no focal neurological deficit, normal level of consciousness, no evidence of intoxication, and no evidence of distracting injury. If all of these criteria were negative then imaging was not considered necessary. The study included 34 000 patients and found that eight of 818 (1%) cervical spine fractures would have been missed; two of the eight were deemed serious. Their conclusion was that if the clinical criteria listed were all negative there was no need to perform any imaging.
The Canadian Cervical spine Rules (CCR) studies, through the application of clinical criteria, identified the following: persons of high risk who all required imaging, patients of low risk on whom it is safe to assess the active range of movement of the cervical spine, and finally patients with the ability to actively rotate their heads 45 degrees in both directions who required no imaging.
Prospective studies have been applied to both the NEXUS and CCR and found CCR to have a sensitivity of 99.4% for clinically important injuries, versus 90.7% for NEXUS.
In 2002 The British Trauma Society published a series of guidelines for the management and assessment of spinal injury that was, in essence, a hybrid of the conclusions of NEXUS and CCR, concluding that if a patient cannot be clinically cleared then imaging should be performed to exclude cervical spine injury.
Once the decision to image has been made, several imaging options need to be considered. Plain radiography is universally available. Traditionally the three view series (anteroposterior (AP), lateral, and peg views) has been employed. An essential requirement of the lateral view is that the cervicothoracic junction is demonstrated, as up to 20% of cervical injuries involve this segment. If traction on the arms fails to depress the shoulders sufficiently, a swimmer’s view may be necessary. If clinical concern persists, computed tomography (CT) is indicated. The five-view series (the three-view series augmented by oblique views) is now largely redundant, as the obliques rarely add useful diagnostic information. CT scanning has largely replaced other x-ray techniques.
The plain film
The plain radiograph remains the essential first step in the assessment of the spinally injured patient. The majority of vertebral injuries can be accurately diagnosed and classified using AP and lateral views. Some fractures can be subtle and will only be detected by an ordered evaluation of the radiographs and by recognizing subtle changes in the normal anatomical relationships. Five per cent of cervical trauma radiographs are misinterpreted, resulting in delays in diagnosis with death or further neurological deterioration in up to 30% of the missed injuries. In half of these patients a complete three-series set of films had not been obtained.
Analysis in the cervical spine begins with the lateral view proceeding to the AP and peg views.
In turn, alignment, bony contour, cartilage, and soft tissue spaces (ABCs) are examined. Normal alignment on the lateral is confirmed by the integrity of three lines; the anterior vertebral line, the posterior vertebral line, and the spinolaminar line (Figure 12.37.1). Apparent subluxation of C2 on C3 is often identified in children and is physiological. This can persist up to the mid-teens. The most important portion of the spinolaminar line is between C1 and C3. A straight line joining the spinolaminar arcs of C1 and C3 should pass within 2mm of C2. This relationship is disturbed in patients with displaced fractures of C2 (Figure 12.37.2).
Assessment of ‘cartilage’ includes the disc spaces and facet joints. Loss of disc height is occasionally the only clue to a fracture through the disc, although this is most commonly a result of degenerative disc disease. The presence of traction spurs arising from the adjacent vertebral body margins is an associated sign in degenerative disease. The posterior facets should be superimposed one side upon the other if a true lateral has been obtained. Loss of parallelism is a feature of unifacet dislocation and should prompt a search for this injury. Even if alignment is normal, a careful assessment of bony contour is essential.
Soft tissue swelling can be a clue to underlying injury, but it is neither sensitive nor specific. Appreciable soft tissue shadowing does not develop until 6h after injury. At the level of the peg, the normal precervical space measures 5mm in adults and up to 7mm in children (Figure 12.37.1). Between C2 and C4, the normal precervical space depends on both age and weight according to the formula:
Less than 5mm is a useful guide. Below C4, an AP diameter of 22mm, or approximately the equivalent of vertebral AP diameter, is normal in adults. The normal atlantoaxial distance is less than 3mm in adults and less than 5mm in children. In some patients the precervical fat stripe can be seen as a black line within the precervical soft tissues. Anterior displacement of this line is a subtle sign of precervical haemorrhage or oedema. Posteriorly, the spaces between the spinous processes should be symmetric. Separation, which is often exaggerated on flexion, implies interosseus ligament rupture, which can be readily confirmed by magnetic resonance imaging (MRI). Widening of the interspinous distance also occurs with subluxed or perched facets.
The upper cervical spine
The distinctive anatomy of the upper cervical spine renders it prone to a different pattern of injury compared with the lower cervical spine. The upper cervical segments are also more commonly injured in children. A variety of measurements have been used to describe the normal relationships between the occiput and the atlas. Most of these are now superseded by sagittal CT reconstructions.
The atlantoaxial relationship can be examined using the ratio of the height of the atlantal spinolaminar line to the atlantoaxial interspinous distance. This ratio is normally 2. The distance between the spinolaminar line and the articular pillars can also be used to assess unifacetal dislocation. Abrupt alteration of this distance suggests spinal rotation and has a strong association with unifacetal dislocation following hyper flexion injury. Another common injury to C1–C2, rotary subluxation, will be discussed in relation to the peg view later in this chapter.
The posterior arch of C1 can be found anywhere between the occiput and the spinous process of C2. Angulation between the anterior arch and the dens is also commonly observed in normal individuals. The posterior arch of C1 is often not ossified and absence must not be misconstrued as a fracture. Abnormal widening may be seen (>10mm) between the spinous processes of C1 and C2 in patients with significant cord injury without other plain film findings.
The C2 vertebral body may show several fracture patterns, and careful review may be necessary to detect them. Small avulsions from the anteroinferior body of C2 may be the only clue. It is useful to look carefully at the apparent dense ring (Figure 12.37.3) projected over the C2 vertebral body, as disruption may be a clue to an underlying fracture, although in many cases this line is incomplete without injury. In a review of 165 injuries, 41% were dens fractures, 38% were traumatic spondylolistheses (hangman’s fractures), 13% were extension teardrop fractures, and 6% were hyperextension dislocations. The remainder were fractures of the lamina and spinous processes. Consistent with other series, no type 1 dens fractures were reported, suggesting that a type 1 fracture, if it exists, is a rarity.
Dens fractures (Figure 12.37.4)
There are three types
♦ Type I: a fracture of the tip of the odontoid (rare)
♦ Type II: through the base of the odontoid
♦ Type III: through the body of the odontoid (commonest, 69%).
The lower cervical spine
The pattern of injury of the lower cervical spine differs from that of the upper two levels. Before diagnosing injury, a number of variations have to be recognized. The fourth and the fifth vertebral bodies can normally be slightly smaller than the adjacent third and sixth vertebral bodies without pathology being present. This can be particularly marked in children, possibly as a result of exaggerated hypermobility during ossification. In general, the difference between the anterior and posterior heights of a vertebral body should be no greater than 3mm. More than this implies an anterior wedge fracture.
The clay shoveller’ fracture involves the spinous process of C6 or C7 (Figure 12.37.5). In hyperflexion sprain, it is a stable injury. The plain radiograph is often normal, and widening of the interspinous distance may be seen. The flexion teardrop is seen as a small anteroinferior fragment. It indicates an unstable injury and is often associated with abnormal neurology, particularly the anterior cord syndrome. The hyperextension teardrop is smaller and is more common in the upper cervical spine. Displacement is unusual and neurological abnormalities are less frequent. The plain film in hyperextension sprain is also normal unless views are obtained in extension.
The most obvious sign of facet dislocation on the lateral plain film is vertebral body subluxation. Half of these injuries are unilateral facet injuries and half are bilateral. In bifacetal dislocation, subluxation can reach 50% (Figure 12.37.6). In unifacet dislocation displacement is usually less than 25% and can often be difficult to detect, particularly if the film is rotated. Once suspected, a careful examination of the facets will reveal the abnormality. Below a dislocation the facets are seen in the true lateral projection with the right facet overlapping the left. Above a unifacet dislocation, the facets on one side will lie more anterior to the other. The lateral mass is seen obliquely and resembles a bow tie. These injuries are frequently associated with fracture of the facet, best seen on CT. At the level of the injury, the superior articular process of the inferior vertebral body does not articulate with the corresponding inferior process of the vertebral body above. This is called the naked facet sign. Oblique views show loss of the normal ‘tiled roof’ appearance of the overhanging facets. Facet fracture–dislocation represents approximately 7% of cervical spine fractures. The incidence of neurological abnormality is high, up to 90% in some series, and is particularly common in bilateral injuries. MRI is necessary to exclude an associated disc prolapse which can cause neurological deterioration following facet reduction.
The diagnosis of a cervical injury may depend on the detection of abnormal alignment between vertebra. Immediately following injury, the presence of muscle spasm may cause the vertebra to maintain their normal relationships and obscure injury. Persistent pain following an injury is an indication of missed injury, and some centres employ flexion and extension views to exclude abnormal segmental motion. If flexion is adequate, the mandible should form an angle of at least 66 degrees with the horizontal. The film should be examined for angulation between the posterior border of the vertebral bodies and for fanning of the spinous processes. The latter is present when the interspinous distance is more than 2mm greater than the spaces above and below. Angulation of more than 3 degrees between vertebral bodies is considered abnormal. If the angle reaches 11 degrees the injury is considered unstable.
The anteroposterior view
The standard cervical AP view should be scrutinized for alignment along both lateral borders and the spinous processes. Abrupt loss of alignment of the spinous processes is seen in unifacet dislocation and may be the most obvious sign of this injury.
Subtle fractures of the lateral pillars may only be appreciated on the AP view. Sagittal oriented fractures appear normal on the lateral view in up to a third of cases and the majority are unstable. Fractures of the upper ribs imply a high-velocity injury which may be associated with lower cervical vertebral fractures and rupture of the great vessels.
The pedicles are also best seen in the AP projection. Absence of a pedicle may be due to destruction by tumour or congenital absence, although this is more common in the thoracic or lumbar spine. The AP view can be obtained with caudal tilt to show the posterior arch, or cranial tilt to give better delineation of the vertebral bodies. In cases where better definition is required, CT is usually employed to better effect.
The peg view
The AP projection of the peg shows its relationship to the lateral masses of C1 and C2. The combined distance between the peg and the lateral masses should be no more than 7mm (Figure 12.37.7). Displaced fractures of the arch of C1 (Jefferson fractures) are recognized by an increase in this measurement and by lateral displacement of the lateral mass of C1 with respect to C2 (Figure 12.37.8).
The classical Jefferson fracture involves four breaks in the arch of C1, two anterior and two posterior, although fewer are frequently seen (Figure 12.37.9).
The injury can also be unstable if there are fewer breaks but the posterior longitudinal ligament is torn. Many injuries of this type are not obviously displaced on initial plain films.
Asymmetry between the right and left spaces without an absolute increase in the total measurement occurs when there is rotation of C1 on C2. Abnormal rotation can occur following trauma, viral infection, or torticollis, particularly in children. Asymmetry may be seen in normal individuals even with correct positioning of the head. Failure to correct with 15 degrees of rotation to either side has been termed fixed atlantoaxial rotatory subluxation (ARS), and is said to distinguish traumatic rotation from other causes, but only if combined with positive clinical signs. Rotation also causes an apparent increase in the size of one lateral mass (that furthest from the film) and a decrease in the size of the other. In true fixed ARS, axial CT sections may show an associated fracture of the facet. In the absence of a fracture, haemorrhage, or synovial entrapment within the facet may account for the fixed rotation. CT is also used to classify fixed ARS based on the peg–anterior arch distance. A distance greater than 5mm implies an unstable lesion which requires surgery.
Undisplaced peg fractures can be difficult to diagnose in the early stages on standard plain films. Fractures are usually oriented in the axial plane and therefore can be overlooked on standard axial CT sections. Coronal CT reconstructions will show the fracture line; however, thin sections have to be acquired to provide sufficient resolution to avoid reformatting artefact. These problems can be overcome with multislice CT and orthogonal voxel acquisition (Figure 12.37.4).
Plain film vs computed tomography
Plain film radiography has been the mainstay of spinal imaging for decades, but there is no doubt that even with excellent radiographic technique and accurate interpretation, a significant number of bony injuries are not detected. The percentage of missed injuries is age dependent and may be as much as 10–20%. With this high miss rate there is a strong argument for bypassing plain film radiography in favour of multislice CT in all apart from the low-risk groups, particularly as the multiply traumatized patient often requires CT imaging of the brain, chest, and abdomen and other injuries. The main counterargument to screening with CT is radiation burden, particularly to the thyroid gland, which is very radiosensitive.
When single slice CT was first introduced it was used as a problem solving tool for areas like the craniocervical or cervicothoracic junctions that were not well visualized on plain film. Helical slice CT and subsequently multislice CT followed with the latter permitting 1-mm slice thickness and isotropic voxels enabling multiplanar image reformatting, have tended to be used to image the entire cervical spine between the foramen magnum and T4. This aids in the perception of difficult subtle fractures and dislocations particularly fractures through the pedicles, pillars and posterior arches. Sensitivities of 95% and specificities of 93% are quoted
Improvements in CT technology and computing power have facilitated the development of isotropic voxel imaging using multislice CT. This means that slice thickness is now so thin that multiplanar reformatting can be performed on the original voxel data set (Figure 12.37.10). Not only are sagittal and coronal reformats used routinely, but oblique and three-dimensional modelling now become possible. The three main imaging plains axial, sagittal, and coronal are all complementary and useful in the detection and analysis of different fractures (Figure 12.37.11). Plain film radiography has no equivalent to the axial plane in which the CT data set is acquired and here lies the greatest advantage of CT over plain film and this is that the bony canal can be clearly demonstrated and the relationship of bony fragments to the canal can be directly delineated(Figure 12.37.12).
Magnetic resonance imaging vs computed tomography
MRI is used in the assessment of the cord and disc herniation in patients with neurological deficit and the assessment of ligamentous structures.
The four main indications for MRI are suspected spinal cord injury are, suspected ligamentous injury, radiculopathy, and progressive neurological deficit, which is an indication for emergency MRI. Sequence optimisation is particularly important in maximizing diagnostic yield. However, the basic examination of the cervical spine should include a T1-weighted sagittal sequence, a T2 or more preferably a fat saturated T2 sagittal sequence and axial images with either T1, T2, or GRE T2* images, or a combination of these. Further sequences can be obtained depending on the clinical scenario and bespoke investigation can be tailored as appropriate. For example if nerve root avulsion is suspected, specific axial oblique T2 or T2*-weighted images are obtained through the exiting nerve roots.
The T1 images are particularly useful for displaying anatomy, for example the sagittal T1 sequence beautifully demonstrates the anterior longitudinal ligament. Likewise the STIR (short-tau inversion recovery) or T2 fat sat images are crucial for detecting bone marrow oedema and microfracture. CT and plain films miss 0.5% of significant injuries in the obtended patient.
Other indications for MRI in trauma include:
♦ Diffuse idiopathic skeletal hyperostosis (DISH) and ankylosing spondylitis
♦ SCIWORA (spinal cord injury without ‘radiological’ abnormality)
♦ Vertebral artery trauma, traumatic meningocoeles, CSF leaks after nerve root avulsions, post-traumatic syrinx, syringohydromyelia, and myelomalacia.
Posteriorly the interspinous ligaments lie between the spinous processes and lamina, they are not as conspicuous as the ALL and PLL on sagittal images but abnormal signal or haematoma in their location infers disruption and potential instability when taken into context with the other injuries seen. As well as assessing alignment of the longitudinal and interspinous ligaments, the paravertebral soft tissues and cord (Figure 12.37.13) should be examined, as vertebral fractures may be associated with paravertebral haematomas. The MRI signal characteristics of blood are complicated and evolve with time, as the oxygenated haemoglobin released from the red blood cells undergoes degradation via deoxyhaemoglobin and methaemoglobin to haemosiderin, all of which have different and definable signal characteristics.
Injuries to the thoracolumbar spine account for approximately 60% of all spinal injuries. In adults, 90% are between T11 and L4, and the majority of these occur between T12 and L2. In children T4 and T5 also figure prominently. The standard projections are AP and lateral views, which can be supplemented by cone lateral views of the injured area.
On MRI, the axial images are best suited for assessing the paravertebral areas, canal and spinal cord. Haematoma is not uncommonly seen within the spinal canal, usually in the epidural space. Epidural haematomas may be compressive and reactive cord oedema is identified as high signal on the fluid sensitive sequences within the cord itself. Differentiation from cord haemorrhage can be achieved by considering the different signal characteristics of oedema and haemorrhage. Finally, the osseous structures should be considered, as previously stated it is well known that MRI is relatively insensitive to detecting fractures, even if STIR sequences are employed. The classic signal characteristics of a fracture are low signal lines on T1 imaging (Figure 12.37.14) and associated high signal on T2 (Figure 12.37.15) and other fluid sensitive sequences. If fat saturation techniques are employed, the fatty signal of the marrow can be negated to exaggerate the fluid signal associated with fracture and micro fracture, improving the pickup.
On the AP projection, the margins of the vertebral bodies can be difficult to distinguish owing to overlap with the posterior elements. A careful scrutiny of the superior and inferior margins will demonstrate wedging; however, this is much easier to assess on the lateral view. The upper thoracic vertebra can be difficult to assess on the lateral view; therefore particular attention must be paid to these on the AP view. Signs of vertebral injury on the AP view also include paravertebral haematoma, apical capping, and mediastinal widening. These are also features of vascular injury. Aortic rupture causes a variety of plain radiograph changes, but the most reliable are tracheal or oesophageal deviation to the right, apical capping, and depression of the left mainstem bronchus. The right paravertebral line lies in close proximity to the vertebral column throughout its length; the left is slightly more laterally placed as a result of displacement by the aorta and can be up to 1cm from the vertebral column. A focal increase in these distances can occur with a paravertebral collection of blood or pus. Displacement due to an ectatic aorta or large osteophytes is the more common cause and tends to be diffuse rather than focal. Below the diaphragm, the psoas shadows may assist in the diagnosis of a paravertebral mass or haematoma. The left psoas shadow can be identified in more than 70% of individuals. The right is more variable in its appearance and is not seen in up to 60% of normal subject. Blurring or enlargement of the psoas shadow may be a sign of retroperitoneal haemorrhage.
Evaluation of the AP view should also include an assessment of the costovertebral joints, the pedicles, and the transverse and spinous processes. The pedicles are normally oval in shape and thespinous processes are teardrop shaped. The cortical margins of both structures should be carefully scrutinized for fracture. There is a gradual increase in the interpediculate distance between T1 and L5. An increase in this distance, following trauma, to more than 4mm greater than adjacent levels indicates involvement of the posterior column and a burst fracture. The presence of a fracture of the transverse process should prompt a search for associated renal injury.
The lateral view
The anterior and posterior margins of the vertebral body should be compared on the lateral view. Minor anterior wedging is acceptable, but the anterior vertebral margin should not differ from the posterior vertebral height by more than 3mm (Figure 12.37.16). More significant wedging causes retropulsion of the posterior vertebral margin into the spinal canal. The most common site of retropulsion is from the posterosuperior margin of the vertebral body. Signs of fragment retropulsion include a convex configuration of the posterior vertebral margin which may be associated with loss of the normal cortex and an increase in the posterior vertebral bony angle to more than 100 degrees. Fragments from both superior and inferior margins are next in frequency. CT will assess the precise nature of the retropulsed fragment and the degree of spinal canal compression; however, the latter does not necessarily correlate with the degree of neurological deficit. A limbus vertebra may mimic a compression fracture. It can be differentiated by its more rounded contour and the corticated fragment of bone isolated by herniated nuclear material (Figure 12.37.17).
When the fulcrum of force is anterior to the vertebral body, a ‘lap-belt’ or Chance fracture ensues (Figure 12.37.18). Distraction forces may result in tearing of the interspinous ligament, facet subluxation, or horizontal fractures through the vertebral body, pedicles, or laminas. Vertebral body height is characteristically increased.
The most common cause of pathological thoracic vertebral fracture is osteoporosis. The most common cause is postmenopausal osteoporosis, but myeloma as a cause of osteoporosis should be considered and excluded. Multiple osteoporotic fractures are more likely to be contiguous. Plain film changes suggestive of malignancy include involvement of the pedicles (usually spared in multiple myeloma), cortical destruction and a soft tissue mass. MRI is the investigation of choice in distinguishing benign from malignant causes and in determining whether osteoporotic fractures are acute or chronic.
Displaced fractures of the pars interarticularis are readily identified on plain films. Undisplaced fractures may be identified on the lateral view, or, more easily, on the 45-degree oblique projections. This technique has been replaced by CT (Figure 12.37.19) and MRI. The authors consider MRI to be the more justifiable imaging technique, as it is more sensitive and does not involve ionizing radiation. These patients are often young, and a negative plain film in suspected pars defects often does little more to progress the differential diagnosis. Sagittal T1-weighted MRI shows the defect as a focus of the intermediate signal within the pars and is the method of choice for the initial evaluation of the pars. In the majority of cases an interruption to the normal bright marrow signal within the pars is seen on the sagittal T1-weighted images. MRI also has the advantage of demonstrating the adjacent exit foramen in patients with associated root compression secondary to spondylolisthesis (Figure 12.37.20). Degenerative disc changes at the level above the pars defect can also be diagnosed. Traumatic spondylolisthesis can be distinguished from the degenerative type by a relative increase in the AP canal diameter. Sagittal STIR or fat-suppressed images may demonstrate oedema in the adjacent marrow secondary to attempts at repair; however, these changes are uncommon. In some cases, high signal is demonstrated within the pars in the absence of an established fracture and this is termed stress response. Differentiation of injury at this occult stage may be important in the training athlete, as modification of the training programme may prevent progression to established fracture. In patients with known pars fracture, CT is superior to MRI in determining healing. Skeletal scintigraphy, preferably with single-photon emission CT (SPECT), shows increased uptake at these sites of repair and is also sensitive to the prefracture stage.
The sacrum, sacroiliac joints, and coccyx
The sacrum can be difficult to assess on plain radiographs, as anatomical detail can be obscured by overlying soft tissue. Particular attention should be paid to the lateral margin of the bone and the integrity of the arcuate lines (Figure 12.37.21). If either of these landmarks are blurred or interrupted, cross-sectional imaging with CT (Figure 12.37.22) or MRI may be necessary.
Fractures of the sacrum are frequent in the osteoporotic patient. One or both ala may be involved and a linking fracture may cross at the S2 level giving the classical H or Honda sign configuration. The fractures are easier to see on MRI than plain films. Fractures of the pubic rami are frequently present in these patients.
Coccydynia is a common problem in the general population, usually following trauma, particularly parturition. In these cases imaging is of little help. If sinister pathology is suspected by the clinical history and rectal examination, MRI or CT is the investigation of choice.
Bohrer, S.P., Chen, Y.M., and Sayers, D.G. (1990). Cervical spine flexion patterns. Skeletal Radiology, 19, 521–5.Find this resource:
De Smet, A.A., Robinson, R.G., Johnson, B.E., and Lukert, B.P. (1988). Spinal compression fractures in osteoporotic women: patterns and relationship to hyperkyphosis. Radiology, 166, 497–500.Find this resource:
Gisbert, V., Hollerman, J., Ney, A. (1989). Incidence and diagnosis of C7-T1 fractures and subluxations in multiple trauma patients: evaluation of the Advanced Trauma Life Support guidelines. Surgery, 106, 702–9.Find this resource:
Griffen, M.M., Frykberg, E.R., Kerwin, et al. (2003). Radiographic clearance of blunt cervical spine injury: plain radiograph or computed tomography scan? Journal of Trauma-Injury Infection and Critical care, 55(2), 222–7.Find this resource:
Sliker, C.W., Mirvis, S.E., Shanamuganathan, K. (2005). Assessing cervical spine stability in obtunded blunt trauma patients: review of the medical literature. Radiology, 234(3), 733–9.Find this resource: