➢ Magnetic resonance imaging (MRI) is used to assess soft-tissue structures following spine trauma; yet, in order to achieve optimum diagnostic accuracy, interpretation must also consider the patient history, physical examination findings, radiographic findings, and computed tomography (CT) findings.
➢ The indications for MRI following spinal trauma include progressive neurological deficit with normal radiographic and CT findings, the presence of a neurological deficit (especially a neurological deficit inconsistent with the fracture level), and prior to surgical stabilization in the setting of a facet dislocation. MRI also may be considered for the assessment of obtunded patients and neurologically intact patients with the possibility of instability or facet subluxation-dislocation (relative indications).
➢ Moderate concordance exists between MRI and intraoperative findings, and care should be taken to avoid overdiagnosis of injury to the posterior ligamentous complex structures, particularly in patients with minimal or no neurological deficit.
➢ Several issues concerning the utility of MRI findings following spine trauma remain debatable. Clinicians should plan the most prudent strategy on the basis of an individualized approach.
Magnetic resonance imaging (MRI) provides high-resolution images without ionizing radiation. It has substantially improved our diagnostic capability following spine trauma, especially for assessing soft-tissue structures. Strong, changing magnetic fields are used to generate cross-sectional images on the basis of the anatomy and chemical composition of the tissues. Nevertheless, compared with radiographs and computed tomography (CT) scans, MRI has several disadvantages, including inadequate visualization of bones, the considerable amount of time required for the procedure (rendering this method impractical in emergency situations), logistical considerations (including the availability of infrastructure and skilled technicians and the necessity for patient transfer), and high cost. Moreover, MRI is routinely performed with the patient in the supine position and therefore may not detect instability associated with loading (upright) positions. Finally, MRI is contraindicated for patients with metal devices such as cochlear implants, pacemakers, and vascular clips as well as those with retained metallic fragments (such as retained gunshot fragments near vital structures).
MRI is undoubtedly gaining an increasing role in the decision-making algorithms for the treatment of spine trauma. The accuracy and reliability of MRI findings depend on the strength of the magnetic field, the severity of the lesion in question, the interpreter, and the reference standard upon which the test is based1-4. Therefore, given these limitations, it should be emphasized that MRI findings should be interpreted within the context of the overall clinical scenario, including patient characteristics, injury mechanism, associated lesions, neurological status, and the findings of other imaging and non-imaging diagnostic modalities.
MRI Protocol Following Spine Trauma
The optimal MRI protocol following spine trauma is debatable and depends on the region, anatomical structure, and lesion type5. Earlier investigators recommended the use of the turbo-spin-echo (TSE) technique following spine trauma because of its shorter imaging time and lower possibility for motion-induced artifacts compared with conventional spin-echo techniques6, which are rarely used nowadays. Nevertheless, recent sequences such as gradient-echo techniques are faster than TSE and allow for better visualization of spinal cord hemorrhage7. In addition to routine T1 and T2-weighted images, lipid suppression (or fluid-sensitive) techniques such as short-tau inversion recovery (STIR) or fat saturation are applied to enhance the visibility of edema on T2-weighted images6,8. These techniques better identify injuries to the posterior ligamentous complex by differentiating signals of fat and soft-tissue edema, which appear bright on non-fat-suppressed T2-weighted images9-11 (Fig. 1). Compared with fat saturation sequencing, STIR imaging provides more uniform fat signal suppression and is less affected by inhomogeneity of the magnetic field at its sides where the magnets are located, near the metallic implants, or within air-filled cavities such as the chest. Moreover, because of the additive effects of T1 and T2 prolongation, the conspicuity of blood products is improved on STIR imaging. However, STIR has longer acquisition time and has inherently lower signal-to-noise ratio, which is typically compensated for by decreasing the resolution via a wider field of view, a smaller imaging matrix, and/or thicker slices. Although non-selective fat suppression associated with STIR imaging decreases the resolution of the spinal cord8, STIR is still the most sensitive sequence for the detection of abnormal cord signals and, therefore, is highly recommended for patients with spine trauma.
Cervical Spine Clearance
Physical examination of victims of major trauma who have neck pain or tenderness, altered mental status, neurological deficit, and distracting injuries is often unreliable, and therefore imaging is essential to evaluate the cervical spine12. CT scanning is the screening imaging modality of choice, but it lacks sensitivity for spinal cord and soft-tissue injuries. Newer-generation multi-detector CT scanners provide higher resolution and have improved the screening role of CT scanning; however, more evidence is required to assess their strength and potential limitations13,14. MRI is the most sensitive tool for identifying soft-tissue and spinal cord injuries and therefore is indicated for all patients with neurological deficits12. For alert and awake patients with neck pain in whom CT does not reveal abnormal findings, the guidelines of the Eastern Association for the Surgery of Trauma (EAST) suggest that MRI can optionally be performed to determine whether to cease or to continue collar immobilization12. Nonetheless, anecdotal reports have described devastating consequences of not performing MRI before the removal of a cervical collar from alert patients with normal CT findings15. However, a recent prospective study did not demonstrate additional value of MRI for the evaluation of alert patients with neck pain, tenderness, or focal sensory neurological deficits who had already undergone screening multi-detector helical CT scanning16. Perhaps the most controversial clinical scenario is in obtunded patients with blunt trauma and negative CT findings. Opponents of routine MRI question the likelihood of clinically substantial injuries in patients with normal physical examination findings17-20. They argue that MRI is time-consuming, costly, and technically challenging. Furthermore, they argue that MRI is inappropriate for hemodynamically unstable patients and that supine positioning during MRI can increase the intracranial pressure, imposing unnecessary risks in vulnerable patients. Advocates reason that MRI can exclude the presence of major soft-tissue injuries and can help to avoid the prolonged use of a collar and its related inconvenience (e.g., pressure ulcers and interference with nursing care, physical therapy, pulmonary hygiene, and access to the airway or central venous line)21. A meta-analysis regarding cervical spine clearance in clinically suspicious or non-evaluable patients demonstrated a negative predictive value of 100% for MRI on the basis of clinical follow-up22. Inconsistencies in the definition of clinically substantial injury and differences in management protocols between institutions contribute to this debate. Nonetheless, precise screening criteria may be helpful for identifying patients with a higher risk of clinically substantial injury that CT scanning would miss. High-risk patients and patients with substantial cervical spondylosis may benefit from an additional MRI evaluation even when a CT scan of the cervical spine reveals negative findings21,23,24. Because no precise criteria have been accepted, the use of MRI in obtunded patients should follow a case-by-case approach.
Cervical spine clearance in the pediatric population deserves specific considerations. The reliability of physical examination in the pediatric population can be a concern; the injury patterns in the pediatric population are different from those in adults, and, in most cases, soft-tissue injuries are more common than fractures. Therefore, CT scanning would lead to unnecessary radiation exposure25, and the accuracy of radiographs is substantially influenced by age and skeletal maturity. Additionally, children may not cooperate for MRI scans. Currently, the only available guidelines for cervical spine clearance in pediatric trauma are from the Trauma Association of Canada Pediatric Subcommittee National Pediatric Cervical Spine Evaluation Pathway26. These guidelines recommend MRI for all patients with an abnormal neurological examination, for patients requiring special investigation of the soft tissues and spinal cord, and for patients with a persistent unreliable clinical examination (i.e., patients who do not become alert and cooperative after twenty-four to seventy-two hours). Of note, a recent study suggested that multi-detector CT technology and MRI have comparable ability for detecting unstable cervical spine injuries in pediatric patients and could be interchangeably used depending on clinical feasibility27.
Cervical Facet Dislocation
The routine use of pre-reduction MRI in patients with unilateral or bilateral cervical facet dislocation has long been a matter of substantial debate28. Supporters of pre-reduction MRI cite the potential risk of neurological damage due to displacement of an undiagnosed disrupted disc into the spinal canal during closed reduction29. Opponents question the relationship between increased prevalence of post-reduction disc herniation and neurological safety30. They consider the importance of immediate realignment of the spine to decompress the spinal cord as well as the fact that, in alert and cooperative patients, successful awake closed reduction can safely be achieved31. Both sides agree that pre-reduction MRI is unnecessary for patients with a complete spinal cord injury. Nevertheless, established guidelines are currently unavailable, and considerable inconsistencies in practice patterns exist in terms of the indications for pre-reduction MRI32. According to one report, even when pre-reduction MRI was available for the assessment of discal integrity, the surgeons disagreed as to whether closed or open reduction was indicated32. Nonetheless, it remains unclear and subjective which findings on pre-reduction MRI are associated with an increased risk of neurological deterioration during closed reduction (Fig. 2).
Vertebral artery injury occurs in 0.5% of patients with cervical spine trauma33. Symptomatic vertebral artery injury is less common and occurs when bilateral or dominant arteries are damaged34. MRI and magnetic resonance angiography (MRA) are helpful for detecting vertebral artery injury (narrowing, intramural hematoma, and pseudo-aneurysms) with accuracy comparable with that of CT angiography. Certain injuries, such as subluxation or translation injuries, fractures extending into transverse process foramina, and C1-C3 fractures, are associated with considerable risk of vertebral artery injury33,35. However, the sensitivity of MRA is low for blunt vascular injuries of the neck, and there is controversy regarding the implication of MRI findings for planning treatment36.
The ligamentous structures appear hypointense (low signal or dark) on both T1 and T2-weighted sequences, and structural changes are associated with hyperintense signals.
The utility of MRI in assessing ligamentous stabilizers of the craniocervical junction (alar and transverse ligaments, tectorial and atlantooccipital membranes) in patients with whiplash syndrome is controversial. One study demonstrated correlation between hyperintense signals in these structures and poor outcomes in cases of chronic whiplash syndrome37. However, other investigators have questioned the specificity and clinical relevance of MRI because similar findings have been reported in subjects without trauma or neck pain38-40. Anatomic variation (orientation, shape, thickness, and fiber density), lack of objective criteria with high interobserver agreement, overlap of degenerative and subtle traumatic changes, and lack of a reference standard definition contribute to this controversy38-40. Therefore, it remains unclear how additional MRI sequences specific for craniocervical ligaments are helpful for treatment. Interestingly, evidence has shown the predictive role of posttraumatic cervical muscle fatty infiltration for poor functional outcomes of whiplash syndrome41-43, although interventional studies to prove its therapeutic implications are lacking.
MRI is helpful for determining lesions of the anterior longitudinal ligament and the posterior ligamentous complex of the neck, which result in clinically important segmental instability following hyperextension and flexion-type injuries, respectively. Maeda et al., in a retrospective study of eighty-eight patients with acute spinal cord injury and no major osseous lesion who had cervical hyperextension instability confirmed with dynamic fluoroscopy at the time of admission, reported that both anterior longitudinal ligament disruption (50% of cases) and the surface area of prevertebral hyperintensity (due to fluid collection or hemorrhage) were significantly associated with segmental instability and lower American Spinal Injury Association (ASIA) motor scores (p < 0.01)44. Nonetheless, despite its high specificity, MRI has low sensitivity (48%) for anterior longitudinal ligament lesions based on intraoperative findings45.
Earlier studies demonstrated that MRI has substantial limitations in terms of excluding fractures of the posterior elements in cervical vertebrae, with a sensitivity of 46% to 55% in comparison with CT46-49. However, it is possible that with modern technologic advances, its sensitivity for detecting fractures has improved.
The assessment of the posterior ligamentous complex (the supraspinous, interspinous, and capsular ligaments; ligamentum flavum; and thoracolumbar fascia) is critical for determining the stability of the spine in patients with flexion-type injuries50. It is therefore important to know how accurately MRI demonstrates the components of the posterior ligamentous complex.
While the biomechanical importance of the posterior ligamentous complex has been long recognized51, the ability of surgeons to evaluate posterior ligamentous complex integrity has proven challenging. Vaccaro et al. identified the posterior ligamentous complex as one of the major determinants in the Thoracolumbar Injury Classification System (TLICS)52. More recently, Schroeder et al. reported that the interobserver reliability of identifying posterior ligamentous complex injuries in patients with AO Thoracolumbar Spine Injury Classification System53 Type-A (compression-type) fractures was only slight (κ = 0.11)54.
Despite low interobserver reliability, multiple systems have been developed to aid in the assessment of the posterior ligamentous complex on MRI. Haba et al. reported that the simultaneous presence of (1) discontinuity of the black stripe on sagittal images (representing supraspinous ligament injury) and (2) high signal intensity at the interspinous space on sagittal T2-weighted images (showing hemorrhage or scar formation) is strongly indicative of posterior ligamentous complex injuries10 (Fig. 1). The sensitivity and specificity of MRI when discontinuity of the black stripe and high signal intensity were considered together were 95% and 93%, respectively10. In addition, Pizones et al. proposed a so-called “damage sequence” model for posterior ligamentous complex injury with use of a progressive scale of AO classification (with the damage sequence being described, in order of increasing severity, as capsular distraction, rupture of the interspinous ligament, rupture of the supraspinous ligament, and rupture of the ligamentum flavum) in which supraspinous ligament injury was interpreted as the threshold for posterior ligamentous complex instability55,56. They defined intraoperative findings and radiographic follow-up at two years as the standard for surgically and nonsurgically managed patients and reported that, for the integrity of the posterior ligamentous complex as a whole, the sensitivity and specificity of MRI, according to their definition model, were 91% and 100%, respectively. For isolated posterior ligamentous complex component lesions, the sensitivity and specificity were between 92% and 100%, except for lesions of facet capsules, for which MRI had specificity of 52%57.
In most studies, the accuracy of MRI for the detection of posterior ligamentous complex injuries was compared with intraoperative findings. On the basis of intraoperative findings, two studies initially demonstrated that MRI had an accuracy of 88% to 97% for detecting posterior ligamentous complex injuries in the thoracolumbar region10,11. However, other investigators found lower sensitivity (79% to 90%) and specificity (53% to 67%) and called for more accurate MRI criteria to define clinically important posterior ligamentous complex injuries1,3,58. As one would expect, the accuracy of MRI for identifying posterior ligamentous complex injuries appears to depend on injury severity, and, in cases of subtle injuries with minimal or no neurological deficit, MRI may not be as reliable1. A 2012 meta-analysis indicated that MRI was better for ruling out rather than confirming a clinically substantial posterior ligamentous complex lesion2.
MRI is more sensitive for the detection of thoracolumbar fractures, especially lower lumbar fractures, than it is for the detection of cervical fractures. In the study by Ganiyusufoglu et al., MRI had an 87% sensitivity for the detection of lumbar stress fractures, with CT being used as the standard59. There are certain types of fractures for which MRI can be helpful. Band-like bone marrow edema parallel to the end plates suggests an insufficiency fracture, particularly if accompanied by a linear low-signal-intensity shadow within the edema zone, representing the fracture line. MRI can be helpful for detecting age-indeterminate, occult, or pathological vertebral fractures, particularly in elderly patients with osteoporosis60,61. Malignant lesions should be considered when the signal change involves the entire vertebral body, when lytic lesions are observed, when the surrounding soft tissues are affected, when the posterior elements are involved, or when additional lesions are present. The so-called “fluid sign” is defined as the appearance of linear fluid signal intensity at the fracture cleft of vertebral body as a result of negative pressure causing transudation62. The Kümmel phenomenon (the presence of vacuum or gas) may be observed in the vertebral body. The signal can be hypointense on T1 and T2-weighted images or hyperintense on T2-weighted images if the cavity is filled with fluid as a result of positional shifts (e.g., prolonged supine position). Osteonecrosis and pseudarthrosis have been postulated to contribute to this incompletely understood phenomenon63. Both the fluid sign and the Kümmel phenomenon indicate that infiltrative processes, such as metastasis, are less probable62.
For athletic adolescents with back pain, MRI can serve as a reliable first-line imaging technique for the diagnosis of juvenile spondylolysis64. Hollenberg et al. described an MRI-based grading system for spondylolysis that is based on consecutive pathophysiologic stages of an evolving stress fracture of the pars interarticularis65 (Table I). This classification system correlates well with findings on CT and single-photon emission CT (SPECT) and has high interobserver and intraobserver reliability59,64. With the combination of CT and SPECT being used as the standard, the sensitivity of MRI for distinguishing pars defects was 73%, but it improved to 98% when secondary signs of abnormality such as widened sagittal spinal canal diameter, posterior vertebral body wedging, and reactive marrow edema in the adjacent pedicles were considered64. The major limitation of MRI is in evaluating cortical integrity, which may lead to incorrect grading of incomplete fractures that require physical activity restrictions66. Moreover, axial MRI sections angled to target the disc may miss some pars defects67. Stress fractures of the pedicles or sacrum are less common but can also be visualized on MRI sequences59,60. Interestingly, MRI had substantially better sensitivity than CT scanning (98% compared with 69%, respectively) for detecting sacral insufficiency fractures68.
Ankylosing Spinal Disorders
Patients with ankylosing disorders (such as ankylosing spondylitis and diffuse interstitial skeletal hyperostosis) are at high risk of spine fracture. These commonly subtle fractures occur even with minor or no clearly identifiable trauma and can be highly unstable because of the levering action of the fused segments and poor bone quality. Delayed diagnosis can lead to catastrophic neurological injuries3,69,70. MRI is not as reliable as high-definition multi-detector CT for detecting fractures in these patients71. However, MRI may reveal critical soft-tissue lesions such as epidural hematoma or occasionally fractures that were undetected with CT71,72 (Fig. 3). Nevertheless, in symptomatic patients with ankylosing disorders, a high index of suspicion for fracture should always be maintained even though initial advanced imaging studies (CT and MRI) do not reveal a lesion73.
MRI and Management
MRI findings can have considerable influence on decisions related to the treatment of thoracolumbar spine fractures74,75. Previous investigators have reported that the use of MRI would have changed the treatment plan from nonoperative to operative in 16% to 24% of patients with acute thoracolumbar fractures74,75. Considering the higher sensitivity of MRI for ligamentous and spinal cord lesions, it is not surprising that adding MRI findings to the initial CT findings substantially affects the grading according to the TLICS or the AO classification system. Furthermore, surrogate indicators of osseous injury on CT scans, such as vertebral body kyphosis, loss of height, and canal compromise, are not adequate substitutes for the precise evaluation of the posterior ligamentous complex and spinal cord with MRI. The only exception is adjacent-segment translation of >3.5 mm, which has been associated with posterior ligamentous complex injury76.
In patients with osteoporotic compression fractures, MRI can help to localize acute fractures that could have been missed on radiographs or CT scans. Some studies have suggested that MRI should be performed preoperatively to determine the levels for cement augmentation techniques more accurately, particularly in patients with multiple osteoporotic compression fractures77,78 (Fig. 4). According to Park et al., MRI information changed the diagnostic profile in 22% (twenty-eight) of 125 patients with single osteoporotic fractures and 65% (twenty-eight) of forty-three patients with multiple osteoporotic fractures77.
MRI can help to more precisely depict the spine injury and to make thorough therapeutic decisions. Nonetheless, it can be excessively sensitive and inadequately specific for ligamentous injuries and, in order to avoid overdiagnosis (and, consequently, overtreatment), its findings should be interpreted in the context of other clinical and radiographic details.
MRI and Outcomes
Another potential utility of MRI in cases of trauma is the prediction of clinical outcomes. The correlation of MRI findings with histological patterns supports the use of MRI for predicting long-term outcomes79,80. However, there is no direct evidence showing that the use of MRI leads to better outcomes in patients with spine trauma. Several studies have demonstrated that, in combination with other clinical parameters, the intramedullary MRI signals (no signal change, signal consistent with edema, or signal indicating hemorrhage) correlate with initial, short-term and long-term outcomes of spinal cord injury6,81,82. Wilson et al., with use of two large databases that included information on patients with spinal cord injury, proposed a prediction model for functional independence measures one year after a traumatic event; the model was based on age, ASIA motor and impairment scales, and signal changes on MRI scans made within the first three days82. Other studies have evaluated the utility of quantifiable MRI parameters for the prognostication of functional outcomes (such as the severity and size of signal intensity changes, measures of maximum canal compromise, and cord compression)81,83-86. The sequential changes of MRI lesions (in terms of signal type or size) also correlate with the severity of cord injury and with clinical outcomes87. The use of multivariate models combining MRI findings with clinical, biomarker, and electrophysiological parameters may further improve outcome predictions.
The predictive potential of MRI is not restricted to spinal cord injury. Signals consistent with bone and soft-tissue injuries also can have prognostic implications. Posttraumatic kyphosis, which may affect the long-term outcome following thoracolumbar injuries, depends on vertebral comminution, posterior ligamentous complex damage, and disc injury, for which MRI is the best available diagnostic method. Oner et al., in a study of fifty-three patients who were evaluated with MRI, found that end-plate comminution in the anterior half and involvement of the vertebral body of more than one-third were highly predictive of increasing kyphosis in patients who were managed nonoperatively and that posterior ligamentous complex injury, the severity of end-plate comminution, and vertebral body involvement were correlated with postoperative kyphosis in patients who were managed operatively88. MRI should be interpreted in the context of the major determinants of deformity progression, such as bone quality and initial surgical correction57. Finally, T2-weighted hyperintense signals in the pedicles adjacent to pars defects suggest an early-stage evolving spondylolysis with greater potential for healing with nonoperative treatment89. Resolution of such signals in subsequent MRI scans can be associated with symptom relief90.
Cost-Effectiveness of MRI
Considerable controversy exists regarding the cost-effectiveness of MRI in cases of spine trauma. The economic implications of MRI are considerable, and the justification for routine use of MRI in cases of spine trauma therefore requires well-designed cost-analysis studies. Unfortunately, scarce direct evidence exists in this regard, and published studies have mainly utilized indirect data and hypothetical models91-93. In a meta-analytical decision model based on patients with an unreliable clinical evaluation, Halpern et al. concluded that the empirical application of a semi-rigid collar for ten days after trauma was more cost-effective than any single screening imaging modality (flexion-extension radiography, CT, or MRI) because the modalities were not adequately sensitive91. Even if these modalities were combined, their increased sensitivity would not justify the cost. On the contrary, a study of pediatric patients showed that implementing an MRI-based cervical clearance protocol decreased the time to cervical clearance, the length of stay in the intensive care unit, and the overall length of stay in the hospital, leading to a savings of $7700 per patient92. In another study of neurologically intact trauma patients with neck pain after trauma, the costs associated with MRI were compared with those associated with cervical collar placement and outpatient follow-up93. The authors found that, considering the cost with and without Medicare reimbursement, the expense of MRI was offset after two and seven days of median lost wages, respectively.
MRI should be performed for all patients who have a neurological deficit after trauma, including those without radiographic evidence of fracture, patients with discrepancy between the level of neurologic injury and the fracture level, and patients with injuries resulting in discontinuity of the longitudinal axis of the canal (fracture-dislocations)6,94,95. Moreover, MRI should be considered for patients with persistent pain or suspected instability. Instability can potentially exist in certain injury patterns (such as burst fractures, flexion-distraction, and hyperextension injuries), when dynamic flexion-extension radiographs are suggestive of disruption of the anterior longitudinal ligament or posterior ligamentous complex as well as in patients with cervical facet dislocations that are reduced or nonreduced prior to stabilization96.
MRI provides unparalleled information on the status of soft tissues and bone, and its application has substantially altered the modern approach to the diagnosis, classification, and treatment of spinal trauma. MRI does have limited accuracy and reliability when used for certain purposes, such as the exclusion of fractures of posterior vertebral elements and evaluation of less-severe injuries involving the paraspinal soft tissues. Therefore, MRI findings should be interpreted with caution in terms of the implications for treatment strategy. Several questions regarding the utility of MRI following spine trauma remain unresolved and highly controversial: What are the MRI criteria for clinically important posterior ligamentous complex lesions? Should MRI be used to clear obtunded patients with potential cervical trauma? Is pre-reduction MRI helpful in cases of cervical facet dislocation? How should MRI findings be utilized to predict the outcome following spine trauma? And, finally, what is the most cost-effective algorithm to indicate the use of MRI in cases of spinal trauma? Pending the resolution of such questions, clinicians are advised to make the most prudent decisions on the basis of the individual clinical scenarios and the availability of resources.
Source of Funding: No external funds were received for the present study.
Investigation performed at the Rothman Institute at Thomas Jefferson University, Philadelphia, Pennsylvania
Disclosure: None of the authors received payments or services, either directly or indirectly (i.e., via his or her institution), from a third party in support of any aspect of this work. One or more of the authors, or his or her institution, has had a financial relationship, in the thirty-six months prior to submission of this work, with an entity in the biomedical arena that could be perceived to influence or have the potential to influence what is written in this work. No author has had any other relationships, or has engaged in any other activities, that could be perceived to influence or have the potential to influence what is written in this work. The complete Disclosures of Potential Conflicts of Interest submitted by authors are always provided with the online version of the article.
- Copyright © 2015 by The Journal of Bone and Joint Surgery, Incorporated