ABSTRACT. WScJ 2: 90-97, 2010

Transcription

1 WScJ 2: 90-97, 2010 Using Magnetic Resonance Imaging to Accurately Assess Injury to the Posterior Ligamentous Complex of the Spine: A Prospective Comparison of the Surgeon and Radiologist Jeffrey A. Rihn, M.D., 1 Nuo Yang, Ph.D., 1 Charles Fisher, M.D., M.H.Sc., F.R.C.S.C., 4 Davor Saravanja, M.D., 4 Harvey Smith, M.D., 5 William B. Morrison, M.D., 2 Jam es Harrop, M.D., 3 Alexander R. Vaca ro, M.D., Ph.D. 1 1 Department of Orthopaedic Surgery, The Rothman Institute, and Departments of 2 Radiology and 3 Neurosurgery, Thomas Jefferson University Hospital, Philadelphia, Pennsylvania; 4 Combined Neurosurgical and Orthopaedic Spine Program, Vancouver General Hospital, Vancouver, British Columbia; and 5 Department of Orthopedic Surgery, Methodist Hospital, Houston, Texas ABSTRACT OBJECT: Magnetic resonance imaging has been proposed as a powerful technique for assessing the integrity of the posterior ligamentous complex (PLC) in spinal trauma. Because MR imaging is often used to determine appropriate treatment, it is important to determine the accuracy and reliability of MR imaging in diagnosing PLC disruption. The purpose of this study is to compare the ability of the radiologist and surgeon to assess disruption of the PLC in the setting of acute cervical and thoracolumbar trauma using MR imaging. METHODS: The components of the PLC in 89 consecutive patients with cervical or thoracolumbar fractures following acute spinal trauma were evaluated using MR imaging by both a musculoskeletal radiologist and an independent spine surgeon and assessed intraoperatively under direct visualization by the treating surgeon. The MR imaging interpretations of the musculoskeletal radiologist and surgeon were compared with the intraoperative report for accuracy, sensitivity, specificity, and positive and negative predictive values. A comparison between the radiologist s and spine surgeon s accuracy of MR imaging interpretation was performed. RESULTS: The agreement between both the spine surgeon s and radiologist s MR imaging interpretation and the actual intraoperative findings was moderate for most components of the PLC. Overall, the MR imaging interpretation of the surgeon was more accurate than that of the radiologist. The interpretation of MR imaging by the surgeon had negative predictive value and sensitivity of up to 100%. However, the specificity of MR imaging for both the surgeon and radiologist was lower, ranging from 51.5 to 80.5%. CONCLUSIONS: Comparison of the MR imaging interpretations between surgeon and radiologist indicates that the surgeon was more accurate for some PLC components. The relatively low positive predictive value and specificity for MR imaging in assessing PLC integrity suggests that both the surgeon and radiologist tend to overdiagnose PLC injury using MR imaging. This can lead to unnecessary surgeries if only MR imaging is used for treatment decision making. (DOI: / SPINE08742) KEY WORDS: Magnetic resonance imaging, Posterior ligamentous complex, Cervical, Thoracolumbar, Spinal trauma * Republished from: J Neurosurgery Spine 12: , 2010, with permission 90 World Spinal Column Journal, Volume 1 / No: 2 / May 2010

2 JA Rihn et al. Spinal stability was defined by White and Panjabi23 as the ability of the spinal elements to prevent the development of neurological injury and progressive deformity in the face of normal physiological forces. Stability is dependent on the integrity of both the bone and ligamentous components of the spine. Injury to either or both of these elements may result in spinal instability that requires surgical stabilization. The PLC is one of the main soft tissue structures that contributes to spinal stability.11,19 The PLC consists of the supraspinous ligament, the intraspinous ligament, the ligamentum flavum, the facet capsules, and the cervical/thoracodorsal fascia. Accurate and specific diagnosis of PLC disruption is important when assessing the stability of the spine and the need for surgical treatment. Although plain radiography and CT allow visualization of osseous anatomy, they provide limited information about soft tissue anatomy.1,2,12,17 Magnetic resonance imaging is a powerful tool that provides detailed information regarding soft tissue anatomy. Soft tissue injury is best demonstrated as high signal intensity on. T2-weighted MR imaging.3,8,13 Fat-suppressed T2-weighted MR imaging is routinely used to avoid false negative readings due to the presence of adjacent fat tissue, which also shows high signal intensity on standard T2-weighted MR imaging.7,10,16,18,24 The accuracy of MR imaging in diagnosing PLC injury in thoracolumbar trauma was reported by Lee et al. in These authors prospectively compared the diagnosis of PLC injury by palpation, plain radiography, and fat-suppressed T2-weighted MR imaging to intraoperative findings, and found that MR imaging was a highly sensitive, specific, and accurate method for evaluating the integrity of the PLC. Subsequent to their study, the use of MR imaging in the diagnosis of spinal trauma has been greatly expanded.6,20 Magnetic resonance imaging findings in the setting of spinal trauma are routinely used to guide treatment decisions. For this reason, it is important to understand the accuracy and reliability of MR imaging in diagnosing injury to the PLC. The purpose of this study is to compare the accuracy of the spine surgeon s and musculoskeletal radiologist s evaluation of cervical and thoracolumbar PLC injury based on MR imaging interpretations, with the decisive diagnosis provided by intraoperative observation. World Spinal Column Journal, Volume 1 / No: 2 / May 2010 METHODS Study Design This study is a prospective observational study of 89 consecutive patients who suffered acute cervical or thoracolumbar spinal trauma and presented to 1 of 2 participating Level I trauma centers. Institutional Review Board approval was obtained at both participating centers prior to patient enrollment. All patients provided informed consent to participate in the study. Patients requiring either posterior or anterior-posterior combined surgery for decompression and/or stabilization were included. Patients with pathological fractures or delayed imaging longer than 5 days from the injury were excluded. Patient Evaluation Standard of care treatment was provided to all patients, including preoperative clinical evaluation (such as medical history, physical examination, and American Spinal Injury Association impairment scale grade) and radiographic, CT, and MR imaging evaluation. The MR imaging included sagittal and axial T1-weighted and T2-weighted fat suppression sequences of the injured spinal level or levels. The following PLC components were individually assessed both on MR imaging and intraoperatively: the supraspinous ligament, the intraspinous ligament, the ligamentum flavum, the left facet capsular ligament, the right facet capsular ligament, and the cervical or thoracodorsal fascia. These components were characterized by MR imaging and intraoperative observation as intact, incompletely disrupted, or disrupted. Each preoperative MR image was interpreted by both a musculoskeletal radiologist (1 at each of the 2 institutions) and an independent, experienced spinal trauma surgeon (the same surgeon for both institutions) who was not involved with care of the study patients. Both the radiologist and the independent spine surgeon were blinded to patient information (such as mechanism of injury, American Spinal Injury Association impairment scale grade, and others), preoperative radiographs and CT scan data, and the intraoperative findings. Operative procedures were performed according to the current standard of care at the participating institutions, with additional emphasis placed upon direct intraoperative visualization of the PLC components. A standard midline posterior surgical approach was used with great caution 91

3 Using Magnetic Resonance Imaging to Accurately Assess Injury to the Posterior Ligamentous Complex of the Spine to avoid damage to the PLC upon exposure. The treating attending surgeon and secondary surgeon (coattending surgeon, spine surgery fellow, or chief resident), verified the status of each individual component of the PLC through a consensus process, with each component graded as intact, incompletely disrupted, or disrupted. Statistical Analysis Correlation between the intraoperative findings and the report of the surgeon or radiologist was defined by asymmetrical Somers D statistic, nonparametric Spearman rank correlation coefficient, and the Cohen kappa coefficient for interrater agreement. The Somers D statistic indicated how closely the intraoperative findings could be predicted from the MR imaging interpretation. It is interpreted as poor (0 0.2), moderate ( ), good ( ), and excellent ( ). The Spearman rank correlation coefficient is defined by a range of 1 to +1, with +1 representing a perfect positive correlation, 1 representing a perfect negative correlation, and 0 representing complete disagreement. A κ score from 0.00 to 0.20 represents slight agreement, fair agreement, moderate agreement, substantial agreement, and > 0.8 near perfect agreement.14 Data analysis also included the determination and comparison of overall percentage agreement, PPV, NPV, sensitivity, and specificity for each PLC component graded by the surgeon and radiologist. Statistical significance was set at a probability value of < RESULTS Of the 89 patients prospectively enrolled in this study, 68 (76.4%) were men and 21 (23.6%) were women, with a mean age of 40.3 years (range years). Causes of trauma were fall (29 patients), motor vehicle accident (38 patients), bicycle accident (8 patients), equestrian accident (4 patients), all-terrain vehicle accident (3 patients), and other (7 patients; injures such as sports, diving, and crush injuries, and train striking a pedestrian). Fracture patterns included compression (5 patients), flexion-distraction (36 patients), burst (26 patients), and translation-rotation (22 patients). The correlations of MR imaging interpretations by the independent surgeon and radiologist with the intraoperative report for each PLC component are demonstrated in Tables 1 and 2. Both the independent surgeon s and radiologist s interpretation of the MR imaging sequences demonstrated moderate Table 1: Assessment of agreement between MR imaging findings and intraoperative findings for each PLC component* PLC Component Somers D Spearman r Cohen κ SSL 0.72/ / /0.43 ISL 0.67/ / /0.44 LF 0.72/ / /0.36 L-CL 0.62/ / /0.35 R-CL 0.65/ / /0.35 C/T F 0.48/ / /0.28 *All values given as surgeon s correlation with intraoperative findings/ radiologist s correlation with intraoperative findings. Abbreviations: C/T F = cervical or thoracodorsal lumbar fascia; ISL = intraspinous ligament; L-CL = left facet capsular ligament; LF = ligamentum flavum; R-CL = right facet capsular ligament; SSL = supraspinous ligament. Table 2: Correlations between the MR imaging interpretations PLC Component Surgeon Radiologist p Value SSL ISL LF <0.001 L-CL R-CL C/T F agreement with the intraoperative findings (Table 1). The overall mean of the Somers D statistic was 0.64 (range ) for the surgeon s interpretation and 0.47 (range ) for the radiologist s interpretation. In addition, the overall mean of the Cohen κ was 0.58 (range ) for the surgeon and 0.37 (range ) for the radiologist. The MR imaging interpretation of the surgeon demonstrated a stronger correlation with the intraoperative findings than that of the radiologist, as analyzed by the Spearman correlation coefficient. The average correlation was 0.72 (range ) for surgeons and 0.52 (range ) for radiologists. The correlation of the surgeon s interpretation with the intraoperative findings was significantly higher than that of the radiologist for all components of the PLC with the exception of the cervical/ thoracodorsal fascia (Table 2). 92 World Spinal Column Journal, Volume 1 / No: 2 / May 2010

4 JA Rihn et al. A comparison of the percentage agreement, PPV, NPV, sensitivity, and specificity of the MR imaging interpretations of the surgeon and radiologist for each PLC component is demonstrated in Table 3. Sensitivity of MR imaging for the various PLC components ranged from 78.4 to 100%, specificity ranged from 51.5 to 80.5%, and PPV ranged from 59.0 to 90.7%. The MR imaging interpretations of the surgeon were statistically more accurate than those of the radiologist for the supraspinous ligament (NPV, p = 0.01; sensitivity, p = 0.03), interspinous ligament (percentage agreement, p = 0.02), ligamentum flavum (percentage agreement, p = 0.01; NPV, p < 0.001; sensitivity, p < 0.001), and left facet capsular ligament (percentage agreement, p = 0.01; NPV, p = 0.01; sensitivity, p = 0.03; Figs. 1 and 2; Table 3). DISCUSSION Spinal stability is based on the integrity of both the osseous and ligamentous components of the spine.4,5 Although plain radiography and CT can provide direct and detailed visualization of osseous anatomy,1,2,12,17 their assessment of the PLC is indirectly based on altered bone relationships (for example, splaying of the spinous processes and facet diastasis). Magnetic resonance imaging is a powerful diagnostic tool that provides direct, detailed visualization of soft tissue anatomy. Magnetic resonance imaging was reported as a highly sensitive, specific, and accurate method of evaluating PLC injury in thoracolumbar trauma by Lee et al.15 The T2-weighted fat-suppressed MR imaging sequence is believed to be the most useful sequence for diagnosing soft tissue injury.7,10,16,18,24 Since the report by Lee et al.,15 the use of MR imaging to evaluate the posterior ligamentous structures in cervical and thoracolumbar trauma has become the imaging technique of choice, but generalizability across specialists and specific anatomical validity have not been addressed.6,9,20 For years, controversy has surrounded the treatment of cervical and thoracolumbar spinal trauma. The recent classification systems for cervical and thoracolumbar trauma developed by the Spine Trauma Study group have attempted to standardize the classification of these injuries and provide guidelines for treatment. The Subaxial Cervical Spine Fracture Classification (SLIC) and Thoracolumbar Injury Classification and Severity Score (TLICS) systems rely on the assessment of 3 major components: morphology (fracture type), neurological World Spinal Column Journal, Volume 1 / No: 2 / May 2010 Table 3: Measures of accuracy of MR imaging for assessing the PLC PLC Component Surgeon (%) Radiologist (%) p Value SSL agreement PPV NPV * sensitivity * specificity ISL agreement * PPV NPV sensitivity specificity LF agreement * PPV NPV < 0.001* sensitivity < 0.001* specificity L-CL agreement * PPV NPV * sensitivity * specificity R-CL agreement PPV NPV sensitivity specificity C/T F agreement PPV NPV sensitivity specificity

5 Using Magnetic Resonance Imaging to Accurately Assess Injury to the Posterior Ligamentous Complex of the Spine Figure 1: Images obtained in a 40-year-old man who fell from a ladder from a height of 20 feet. A and B: Midline sagittal (left) and parasagittal (right) CT scan reconstruction (A), and consecutive axial CT images (B). These images demonstrate an L-1 burst fracture with involvement of the posterior column in the form of a right-sided lamina/facet fracture (arrows, B). C: Sagittal T2-weighted (left) and T1-weighted (right) MR images demonstrate the soft tissue structures. In this case, the radiologist read the interspinous ligament (dashed arrow) as intact and the supraspinous ligament (solid arrow) as disrupted. The surgeon, however, interpreted both the interspinous and supraspinous ligaments as intact. Intraoperatively, both the interspinous and supraspinous ligaments were noted to be intact D: Axial MR image demonstrating the disrupted right facet capsule (solid arrow) and the intact thoracodorsal fascia (open arrows), both of which the radiologist and surgeon agreed upon and the intraoperative report confirmed. status of the patient, and integrity of the soft tissue stabilizers of the spine.21,22 The integrity of the PLC is taken into consideration when making treatment decisions. For this reason, it is important to understand the accuracy of MR imaging in diagnosing PLC injury. In the present study, we compared both the surgeon s and radiologist s interpretation of the 6 components of the PLC on MR imaging with the intraoperative findings (Table 2). In addition, the accuracy, sensitivity, specificity, PPV, and NPV of their MR imaging interpretations of each of the PLC components was determined and compared (Table 3). It was found that the MR imaging interpretation of the surgeon better correlated with intraoperative findings than that of the radiologist. Furthermore, the surgeon s evaluation of PLC injury on MR imaging was more accurate, sensitive, and specific than the radiologists for many of the PLC components. Although it cannot be known for certain, this difference may be due to the exclusive exposure that an experienced spine surgeon has to decisive intraoperative observations, which may lead to a more accurate 94 World Spinal Column Journal, Volume 1 / No: 2 / May 2010

6 JA Rihn et al. Figure 2: Images obtained in a 30-year-old man who was an unrestrained passenger in a head-on motor vehicle accident. A: Midline sagittal (left), parasagittal left (center), and parasagittal right (right) CT scan reconstructions demonstrate bilateral perched facets at the level of C5 6, with significant angulation and subluxation of the C-5 vertebral body on the C-6 vertebral body. B: Midline sagittal (left), parasagittal left (center), and parasagittal right (right) T2-weighted MR images demonstrating significant interspinous widening (between the dotted arrows) with disruption of all components of the PLC. The facet capsules are disrupted bilaterally (solid arrows). In this case, there was complete agreement between the radiologist and surgeon regarding interpretation of the MR imaging findings and the intraoperative findings. interpretation of the PLC on MR imaging. The surgeon s opportunity to routinely compare MR imaging findings with intraoperative findings may lead to the improved accuracy demonstrated in this study. It is interesting to note that significant differences between the surgeon s and radiologist s correlations were observed for certain PLC components but not others. For example, the surgeon surpassed the radiologist in the accuracy of diagnosing disruption of the ligamentum flavum, left facet capsular ligament, supraspinous ligament, and interspinous ligament, but not the right facet World Spinal Column Journal, Volume 1 / No: 2 / May 2010 capsular ligament or cervical/thoracodorsal fascia (Tables 2 and 3). These data suggest that certain PLC components are more difficult than others to characterize based solely on the interpretation of MR imaging. With the advantage of direct intraoperative visualization of these structures and the ability to compare these direct observations to the MR imaging findings on a routine basis, the surgeon may have a diagnostic advantage when interpreting the MR imaging. It is important to note that the MR imaging interpretations of both the surgeon and radiologist showed 95

7 Using Magnetic Resonance Imaging to Accurately Assess Injury to the Posterior Ligamentous Complex of the Spine a relatively high sensitivity and NPV but lower specificity and PPV. This suggests that it is unlikely that an injury to the PLC will be missed by MR imaging, but that both the surgeon and radiologist may tend to overdiagnose PLC injury using this methodology. This relatively high rate of false positive MR imaging analyses may lead to more surgery than is necessary if the information provided by the MR imaging is weighted heavily when making decisions regarding treatment. This study represents the first comprehensive, prospective study directly comparing the MR imaging interpretation of all of the PLC components between a radiologist and an experienced spine surgeon to the intraoperative findings of both cervical and thoracolumbar spinal injuries. The study of Lee et al.15 was limited to the supraspinous and interspinous ligaments in patients with acute thoracolumbar trauma. We successfully demonstrated the role of MR imaging in the evaluation of the remaining PLC components, that is, the ligamentum flavum, facet capsular ligament, and cervical/thoracodorsal fascia. The surgeon may have improved accuracy compared with the radiologist when assessing the integrity of certain PLC components on MR imaging. The primary message of this study, however, is that both the surgeon and radiologist tend to overdiagnose PLC injury using MR imaging. CONCLUSIONS A comparison of the accuracy of MR imaging in the interpretation of PLC integrity between the surgeon and radiologist indicated that surgeons had a higher degree of accuracy when diagnosing injury to some components of the PLC in the setting of acute cervical and thoracolumbar trauma. This study confirmed a relatively high sensitivity of MR imaging in diagnosing PLC injury, indicating that PLC injury in the setting of acute trauma is unlikely to be overlooked. The most important finding of this study, however, is the tendency of both the surgeon and radiologist to overdiagnose injury to the PLC using MR imaging, as evidenced by the relatively low specificity and PPV of this modality. If MR imaging is used in isolation to determine a treatment plan, this high false-positive rate may lead to unnecessary surgery. Therefore, factors other than the MR imaging results, including the neurological status of the patient and the morphology of the injury, should be taken into consideration when deciding on treatment. DISCLOSURE This study was funded by an educational grant from Medtronic, Inc. Dr. Vaccaro receives royalties from Medtronic, Inc. REFERENCES 1. Brant-Zawadzki M, Miller EM, Federle MP: CT in the evaluation of spine trauma. AJR Am J Roentgenol 136: , Daffner RH, Deeb ZL, Goldberg AL, Kandabarow A, Rothfus WE: The radiologic assessment of post-traumatic vertebral stability. Skeletal Radiol 19: , Davis SJ, Teresi LM, Bradley WG Jr, Ziemba MA, Bloze AE: Cervical spine hyperextension injuries: MR findings. Radiology 180: , Denis F: Spinal instability as defined by the three-column spine concept in acute spinal trauma. Clin Orthop Relat Res 189:65 76, Denis F: The three column spine and its significance in the classification of acute thoracolumbar spinal injuries. Spine 8: , Emery SE, Pathria MN, Wilber RG, Masaryk T, Bohlman HH: Magnetic resonance imaging of posttraumatic spinal ligament injury. J Spinal Disord 2: , Georgy BA, Hesselink JR: MR imaging of the spine: recent advances in pulse sequences and special techniques. AJR Am J Roentgenol 162: , Goldberg AL, Rothfus WE, Deeb ZL, Daffner RH, Lupetin AR, Wilberger JE, et al: The impact of magnetic resonance on the diagnostic evaluation of acute cervicothoracic spinal trauma. Skeletal Radiol 17:89 95, Goradia D, Linnau KF, Cohen WA, Mirza S, Hallam DK, Blackmore CC: Correlation of MR imaging findings with intraoperative findings after cervical spine trauma. AJNR Am J Neuroradiol 28: , Henkelman RM, Hardy PA, Bishop JE, Poon CS, Plewes DB: Why fat is bright in RARE and fast spin-echo imaging. J Magn Reson Imaging 2: , James KS, Wenger KH, Schlegel JD, Dunn HK: Biomechanical evaluation of the stability of thoracolumbar burst fractures. Spine 19: , Kaye JJ, Nance EP Jr: Cervical spine trauma. Orthop Clin North Am 21: , Kliewer MA, Gray L, Paver J, Richardson WD, Vogler JB, McElhaney JH, et al: Acute spinal ligament disruption: MR imaging with anatomic correlation. J Magn Reson Imaging 3: , Landis JR, Koch GG: The measurement of observer agreement for categorical data. Biometrics 33: , Lee HM, Kim HS, Kim DJ, Suk KS, Park JO, Kim NH: Reliability of magnetic resonance imaging in detecting posterior ligament complex injury in thoracolumbar spinal fractures. Spine 25: , Magerl F, Aebi M, Gertzbein SD, Harms J, Nazarian S: A comprehensive classification of thoracic and lumbar injuries. Eur Spine J 3: , World Spinal Column Journal, Volume 1 / No: 2 / May 2010

8 JA Rihn et al. 17. McAfee PC, Yuan HA, Fredrickson BE, Lubicky JP: The value of computed tomography in thoracolumbar fractures. An analysis of one hundred consecutive cases and a new classification. J Bone Joint Surg Am 65: , McArdle CB, Crofford MJ, Mirfakhraee M, Amparo EG, Calhoun JS: Surface coil MR of spinal trauma: preliminary experience. AJNR Am J Neuroradiol 7: , Oxland TR, Panjabi MM, Southern EP, Duranceau JS: An anatomic basis for spinal instability: a porcine trauma model. J Orthop Res 9: , Terk MR, Hume-Neal M, Fraipont M, Ahmadi J, Colletti PM: Injury of the posterior ligament complex in patients with acute spinal trauma: evaluation by MR imaging. AJR Am J Roentgenol 168: , Vaccaro AR, Hulbert RJ, Patel AA, Fisher C, Dvorak M, Lehman RA Jr, et al : The subaxial cervical spine injury classification system: a novel approach to recognize the importance of morphology, neurology, and integrity of the disco-ligamentous complex. Spine 32: , Vaccaro AR, Lehman RA Jr, Hurlbert RJ, Anderson PA, Harris M, Hedlund R, et al: A new classification of thoracolumbar injuries: the importance of injury morphology, the integrity of the posterior ligamentous complex, and neurologic status. Spine 30: , White AA III, Panjabi MM: Clinical Biomechanics of the Spine, ed 2. Philadelphia: JB Lippincott, Zee CS, Segall HD, Terk MR, Destian S, Ahmadi J, Gober JR, et al: SPIR MRI in spinal diseases. J Comput Assist Tomogr 16: , 1992 World Spinal Column Journal, Volume 1 / No: 2 / May