Australas Radio1 1992; 36: 198-203
A Comparison Between M.R.I. and C.T. in the Investigation of Neurological Deterioration in Longstanding Spinal Trauma MORRY SILBERSTEIN, M.B., B.S. University of Melbourne Fellow in Radiology and Magnetic Resonance Department of Radiology OLIVER HENNESSY, F.R.C.R. Department of Radiology Austin Hospital, Heidelberg, Victoria, Australia. BRIAN TRESS, F.R.A.C.R. Department of Radiology Royal Melbourne Hospital, Parkville, Victoria, Australia.
ABSTRACT MRI at 0.3 T and CT with myelographic contrast (CTM) were compared in the retrospective evaluation of 35 patients investigated for the development of new neurological symptoms following longstanding spinal cord injury. Compared with MRI, CTM was relatively accurate for the demonstration of spinal cord compression, but failed to identify 23% of patients with spinal cord atrophy, and 43% of patients with post-traumatic syrinx formation. However, 5 patients had unsatisfactory MR imaging, either due to motion o r metallic artifact, and in 3 of these, CTM demonstrated a syrinx. Although MRI is the method of choice in the investigation of this
Key words: MRI, CT, Spine Trauma, Syrinx Address for correspondence: M. Silberstein University of Melbourne Fellow in Radiology and Magnetic Resonance Department of Radiology Austin Hospital Heidelberg Victoria Australia 3084
Droblem. CTM mav still be reauired kor patients with i n unsatisfictory MR examination. Magnetic Resonance (MR) imaging is now an established technique for imaging the spine, with accurate depiction of the spinal cord, as well as the adjacent soft tissues (1, 2). However, the cost of this technique, and its as yet limited availability in Australasia, h a s resulted in th e necessity to demonstrate its superiority over other imaging modalities fo r any specific clinical problem (3). O n e of th e m a j o r a r e a s of impact of MR has been in the investigation of the problem of acute neurological deterioration in patients with past spinal tr a u ma (4, 5, 6). Some of these patients will have treatable causes of deterioration, either a post-traumatic syrinx, o r spinal cord compression (6),and MR can be used to image these conditions (7), which, until recently, were investigated with computed tomography with myelographic contrast medium (CTM),(8,9). In the largest reported study comparing MR and CTM, Betz et al (6) reported the results of 43 children who were investigated u p to 48 months fro m original injury, of whom 24 had CTM. They concluded that although MR was indicated for the evaluation of s u b a c u te a n d chronic spinal cord injured patients, CT-myelography remained essential before considering surgery for spinal cord decompression (6). We retrospectively evaluated the results of MR and CT-myelography in 35 patients who developed new symptoms following past spinal cord injury to determine whether there was still any place for CTM in the evaluation of this problem.
MATERIALS AND METHODS The 35 patients (29 males and 6 females) were aged between 21 and 70 years (mean 39.9 years). The original mode of injury was: motor vehicle accident (27), fall (3,and diving (3). Twenty four patients had injuries to the cervical spine and 11 to the thoracic spine. Patients included in the study had a single neurologic injury level, above the conus medullaris, initial admission to the Hospital Spinal Injuries Unit, and subsequent development of new symptoms following neurological stabilization. Patients presented between 4 months and 408 months from injury. The indications for imaging were: increasing myelopathy (16), ascending neurologic level (8), pain (6), increasing muscular spasms (4), and recent onset of hyperhidrosis (1). MR imaging was performed with a 0.3 T resistive magnet (Fonar; B3000) on all patients within 3 weeks of presentation, using a spin echo (SE) technique consisting of 2 sagittal sequences (repetition time msec/echo time msec=500/16 for T1 weighted images, and 2500/80 for T2 weighted images) and 2 axial sequences (500/16 and 2500/80), in view of the recognized occasional false negative diagnosis of a small syrinx on sagittal scans alone (10). Slice thickness was 5mm with 2mm interslice gap.
Submitted for publication on: 17th September, 1991 Accepted for publication on: 21st January, 1992
Australasian Radiology, Vol,36, No. 3, August, 1992
MRI AND CT IN NEUROLOGICAL DETERIORATION SPINAL TRAUMA
FIGURE 1A - False positive CTh4 diagnosis of normal spinal cord. Axial CT image following myelography through C7. No contrast accumulation in the cord. FIGURE 1B - Sagittal T1 weighted MR. Longitudinal syrinx extends from C5 to T1.
CT was performed on all patients (Siemens;Somatom DR3) following installation of intrathecal contrast via lumbar puncture under fluoroscopic guidance. Contiguous 4mm scans were then obtained within 1 hour, between 4 and 6 hours, and between 12 and 24 hours from lumbar puncture, through the suspected clinical level. The region studies was extended to include the whole extent of any abnormality demonstrated, and if the study was considered to be normal, further non-contiguous scans were performed to include the whole of the cervicothoracic spinal cord. The MR images were assessed for: the presence of spinal cord compression - non-visualisation of any portion of the CSF space around the cord; cord atrophy - an anteroposterior diameter of less than 6mm in the cervical cord and 5mm in the thoracic cord on a midsagittal slice (11); myelomalacia - cord hyperintensity on T2 weighted images with either normal or low intensity on T1 weighted images, and ill-defined margins; and syrinx - a well defined area of CSF intensity on T1 weighted images, regardless of intensity on T2 weighted images, in view of signal alterations induced by flow effects (7). Australasian Radiology. Vol. 36,No. 3 , August, 1992
Normal Cord 3
Cord atro = cord atrophy Myel mala = myelomalacia Cord comp = cord compression
PredicVal Sens Spec Pos Neg ACC 77 40 100 80 NormalCord 100 83 88 63 95 87 CordComp 91 84 87 94 CordAtrophy 77 0 100 0 90 90 Myelomalacia syrinx 57 100 100 100 70
The CT images were assessed for the same characteristics: spinal cord compression and spinal cord atrophy defined as above, and assessed on axial images: myelomalacia defined as poorly defined contrast accumulation within the spinal cord: and syrinx - a well defined area of contrast accumulation within the spinal cord. MR was used as the standard of reference for all calculations. Five patients subsequently underwent spinal surgery, and the findings were available for correlation with the imaging data.
In the remaining 30 patients the findings, based on MR observations, were: normal spinal cord - 4 patients (13%); cord compression - 6 patients (20%); myelomalacia - 1 patient (3%); and syrinx - 21 patients (70%). Sensitivity,specificity, positive and negative predictive values, and overall accuracy of CTM for each of the 5 imaging characteristics are presented in Table 2.
RESULTS Five patients had unsatisfactory MR imaging, either due to motion (3) or artifact from adjacent metal (2). The CTM findings for these patients are presented in Table 1.
The high false positive rate of CTM for the diagnosis of normal cord was due to the failure to identify the patient with myelomalacia and some patients with a syrinx (Figure 1). CTM was relatively accurate in the diagnosis of 199
M. SILBERSTEIN et al
FIGURE 2A - Spinal cord compression, secondary to cord expansion by syrinx. Four axial CTM images extending cranially from C7 (top left) to C6 (bottom right). Contrast outlines atrophic cord inferiorly, with complete obstruction to passage of contrast at C6.
cord compression (Figure 2 ) and cord atrophy, although in the latter category CTM had a 23% false negative rate. There were no false positive diagnoses of syrinx by CTM, every patient with contrast accumulation in the spinal cord being confirmed to have a syrinx on MR (Figures 3, 4). The 5 patients who underwent surgery had complete agreement with the CTM and MR diagnoses of syrinx. However, CTM failed to demonstrate 43% of patients with a syrinx (Figure 5).
FIGURE 2B - Sagittal T1 weighted MR. Cord expansion by syrinx results in anterior cord compression opposite C6-C7 vertebral bodies.
DISCUSSION Although uncommon, late neurological deterioration following past spinal cord injury requires investigation to exclude post-traumatic syringomyelia, which frequently improves following surgical treatment
(12). The incidence of this condition varies between 0.5% and 2% in patients with past spinal trauma (12). Until the advent of MR, this diagnosis was difficult to make, with a number of techniques in use, including standard myelography, CT-myelography, a combination of these techniques with imaging in prone and supine positions to demonstrate variations in spinal cord size, and endomyelography (8,9). Apart from being invasive, and usually requiring hospital admission (3, the major problem arising from these techniques was that the most sensitive technique - CTM - was associated with uptake of contrast by normal spinal cords in some circumstances (13). In addition, CTM had a significant false negative rate for the diagnosis of syrinx. In the series of Betz et a1 CTM failed to identify 2 out of 5 patients with a post-traumatic syrinx Australasian Radiology, Vol. 36, No. 3, August, 1992
MRI AND CT IN NEUROLOGICAL DETERIORATION SPINAL TRAUMA
FIGURE 3A - Mid-thoracic syrinx on CTM, confirmed on MR. Axial CTM image opposite T7.Dense contrast accumulation in cord, consistent with syrinx.
FIGURE 3B - Axial T1 weighted MR image at similar level confirms diagnosis of syrinx.
FIGURE 4A - Small syrinx opposite T5. Axial CTM image. Focal contrast accumulation in spinal cord, consistent with syrinx.
FIGURE 4B - Axial T1 weighted MR image confirms small syrinx cavity.
Australasian Radiology. Vol. 36. No. 3 . August. I992
FIGURE 5A - False negative CTh4 diagnosis of syrinx. Axial post-myelogram CT image at T5-T6. No contrast is present within the cord. Note, however, that the cord is expanded, and of low density. Although suggesting intramedullary cyst formation, the appearances do not fit into the inclusion criteria for syrinx.
on MR (6). In a series of 10 patients who had both CTM and MR, Quencer e t a1 failed to identify 2 out of 6 patients with a syrinx on MR using CTM (4). Significantly, another 2 patients who were thought to have a syrinx on CTM had features consistent with myelomalacia on MR (4). In a series of 9 patients who had CTM and MR, all of whom had contrast uptake by the spinal cord on CTM, Gebarski et al found 1 patient in whom MR showed myelomalacia, and although the other 8 had features consistent with a syrinx, the information from MR was superior to that from CTM (5).
Clearly, MR has significant advantages over other imaging modalities in post-traumatic syringomyelia. In our series, CTM failed to identify 9 out of 21 patients in whom MR showed a syrinx. MR is non-invasive, does not require hospitalisation, and does not employ ionising radiation. However, certain problems require attention. Five of our patients had significant compromise to their MR images, rendering diagnosis impossible. Three of these patients had a syrinx on CTM, including 1 patient with cord compression. Another 2 patients with 202
FIGURE 5B - Sagittal T1 weighted MR image. Syrinx extends both inferiorly and superiorly from level of fracture at T7.
cord compression were identified on CTM in this group. In the series of Betz et al, artifact from metallic fixation devices was present in 6 of 23 MR scans, and in 2 of these, the metal totally obscured diagnostic detail (6). Of course, CT may also be non-diagnostic in these circumstances, although we did not find this to be a problem. Motion is a significant source of degradation of MR images, especially in the thoracic spine, but newer imaging sequences are likely to alleviate this problem in the future (14).
There may also be a group of patients in whom MR is falsely negative. In the series of Quencer et al, 1 patient in whom MR was interpreted as showing a syrinx was found to have myelomalacia using intraoperative sonography (4). In the series of Stevens et al, 15% of patients thought to have a syrinx on myelography of CTM had no evidence of this at operation (8). In view of the high protein content of these cysts, it is possible that there is T1 shortening within the cyst fluid making it hyperintense to CSF on T1 weighted images, and since T2 weighted images are not
always reliable for the diagnosis of syrinx, because of flow effects, the lesion could be interpreted as myelomalacia. In the diagnosis of cord compression and cord atrophy, CTM had a relatively high accuracy in our series 87% for each of these findings. The 17% false negative rate of CTM for the diagnosis of cord atrophy may be due to the method used to measure the spinal cord. Only slight angulation of scan plane, be it sagittal for the MR images, or axial for the CT images, can result in oblique plane of section distorting the spinal cord in these images and result in an inaccurate measurement. In addition, the normal spinal cord diameter varies along its length, and it is easier to determine the narrowest anteroposterior diameter of the spinal cord in a mid-sagittal slice than in a series of axial images where each level must be measured and related to the associated vertebral body level (15). There is a high possibility of observer error with the latter method, and, as we made the determinations independently, this might explain our false negative cases of cord atrophy on CTM. Australasian Radiology, Vol. 36, No. 3?August, 1992
MRI AND CT IN NEUROLOGICAL DETERIORATION SPINAL TRAUMA
Although our results indicate that MR is the method of choice in the investigation of patients presenting with delayed neurological deterioration following spinal cord injury, this imaging modality may not yield all of the answers in every patient. Intraoperative sonography is a highly accurate technique for assessing the spinal cord in patients with cord injury, and may be required for confirmation of MR findings prior to myelotomy for shunt insertion in patients with post-traumatic syrinx (7). Finally, computed tomography with myelographic contrast cannot be considered obsolete in the investigation of this problem at this time, since some patients will have a failed MR examination (14% in our series) and will require an alternate means of investigation to make the diagnosis of posttraumatic syrinx.
Australasian Radiology, Vol. 36, N O . 3 . August, I992
Modic MT, Masaryk TJ, Paushter DM. Magnetic resonance imaging of the spine. Rad Clin North Am 1986; 2 4 229-245. Hyman RA, Gorey MT. Imaging strategies for MR of the spine. Rad Clin North Am 1988; 26: 521-534. Benness GT. (editorial) MRI assessment programme: a clearer picture? Med J Aust 1991; 154: 229-230. Quencer RM, Sheldon JJ. Donovan Post MJ er al. Magnetic resonance imaging of the chronically injured cervical spinal cord. AJRN 1986; 7: 457-464. Gebarski SS. Mavnard FW.Gabrielsen TO. Knake JE, LataLk JT, Hoff JT. Post-trau: matic progressive myelopathy. Radiology 1985; 157: 379-385. Betz RR, Gelman AJ, DeFilipp GJ, Mesaarzadeh M. Clancv M. Steel HH. MagGetic resonance imaging in the evaluation of spinal cord injured children and adolescents. Paraplegia 1987; 25: 92-99. Quencer RM. The injured spinal cord: evaluation with magnetic resonance and intraoperative sonography. Rad Clin North Am 1988; 26: 1025-1045. Stevens JM, Olney JS, Kendall BE. Posttraumatic cystic and non-cystic myelopathy. Neuroradiology 1985; 27: 48-56.
9. Quencer RM, Green BA, Eismont FJ. Posttraumatic spinal cord cysts: clinical features and characterization with metrizamide computed tomography. Radiology 1983; 146: 415-423. 10. Dowling RJ,Tress BM. MRI - The investigation of choice in syringomyelia? Aust Rad 1989; 33: 337-343. 11. Yamashita Y,Takahashi M, Matsuno Y et a l . Chronic injuries of the spinal cord: assessment with MR imaging. Radiology 1990; 175: 849-854. 12. Vernon JD, Silver JR, Ohry A. Post-traumatic syringomyelia. Paraplegia 1982; 20: 339-364. 13. Dubois PJ, Drayer BP, Sage M, Osbome D, Heinz ER. Intramedullary penetrance of metrizamide in the dog spinal cord. AJNR 1981; 2: 313-317. 14. Kricun R, Kricun ME, Dalinka MK. Advances in spinal imaging. Rad Clin North Am 1990 28: 321-339. 15. Sherman JL, Nassaux PY, Citrin CM. Measurements of the normal cervical spinal cord on MR imaging. AJNR 1990; 11: 369-372.