Acute spinal trauma: Prognostic value of MRI appearances at 0.5 T

Clinical Radiology(1993) 48, 100-108

Acute Spinal Trauma: Prognostic Value of MRI Appearances at 0.5 T M. MASCALCHI*w G. DAL POZZO*, C. DINI*, V. ZAMPAw M. D ' A N D R E A t , M. M I Z Z A U t , F. LOLLI~:, D. CARAMELLAw and C. BARTOLOZZIw

* Unita' RMN, Sezione di Radiodiagnostica, Dipartimento di Fisiopatologia Clinica, Universita' di Firenze, Firenze, t Unita' Spinale and ~Servizio di Neurofisiopatologia, Ospedale di Careggi, Firenze, and w di Radiologia, Universita" di Pisa, Pisa, Italy Thirty-two patients with post-traumatic myelopathy were examined with a 0.5 T MRI system within 4 days of injury and the MRI findings analysed with respect to the immediate and residual functional deficit and (in 20 patients) the MRI appearances of the spinal cord in the chronic phase. In the acute phase a normal spinal cord was associated with only slight clinical deficit in four patients. Signal abnormalities in the spinal cord at the site of trauma were identified on T2weighted spin-echo or T2*-weighted gradient-recalled echo images in 28 patients. The 12 most functionally impaired patients showed focal low signal suggestive of intramedullary haemorrhage: the other 16 had homogeneous high signal consistent with diffuse oedema. Swelling of the spinal cord and mild persistent cord compression following reduction were noted in 17 and 26 patients respectively. All patients were treated conservatively other than undergoing surgical decompression. Four died of complications. No patient with low signal in the spinal cord on initial MRI showed w clinical improvement. Five whose spinal cord was hyperintense remained unchanged, whereas nine made a significant recovery, as did all patients with normal-appearing spinal cords. Cord compression on the initial examination was not relevant to clinical outcome. lntramedullary scars were identified at follow-up in 18 patients and were more extensive in those with haemorrhagic acute lesions. Haemorrhagic contusion of the spinal cord can be demonstrated in the acute phase with midfield MRI and is a valuable predictor of the functional outcome in patients with traumatic myelopathy. Mascalchi, M., Dal Pozzo, G., Dini, C., Zampa, V., D'Andrea, M., Mizzau, M., Lolli, F., Caramella, D. & Bartolozzi, C. (1993). Clinical Radiology 48, 100-108. Acute Spinal Trauma: Prognostic Value of MRI Appearances at 0.5 T

Accepted for Publication 23 February 1993

Trauma is one of the main causes of chronic spinal cord dysfunction because of its frequency and the frequent residual functional deficit. Severa~ groups have reported lack of association between bony changes on X-ray or computed tomography (CT) and clinical deficit in patients with spinal trauma [1,2]. With the advent of magnetic resonance imaging (MRI) which provides unique information about intervertebral disc changes, paravertebral soft tissues and spinal cord, more comprehensive analyses of the morphological features in patients with spinal trauma have been possible [3-7]. They indicate that haemorrhagic contusion of the spinal cord, as revealed by MRI, is associated with the most severe functional compromise, whereas a purely oedematous intramedullary lesion is associated with less severe clinical deficit. In the acute phase of trauma, intramedullary haemorrhage gives a characteristic low signal on T2-weighted images, but this association has been systematically investigated using only high field MRIo systems [4-6]. Magnetic susceptibility effects of deoxyhaemoglobin responsible for the T2 shortening depend on the strength of the magnetic field [8], and these effects would be expected to be less conspicuous and possibly undetectable at lower fields. Correspondence to: Dr Mario Mascalchi, Cattedra di Radiologia, Universita' di Pisa, Via Roma 67, 56100 Pisa, Italy.

We wished to assess whether acute haemorrhagic contusion of the spinal cord can be detected using a midfield MRI unit and whether its identification has a prognostic value. PATIENTS AND M E T H O D S The MRI examinations of 32 consecutive patients (six females, 26 males; age range 15-76 years) with spinal cord dysfunction following a spinal trauma were examined with a 0.5 T MRI unit within 4 days of injury (12 on the first day, six on the second day, nine on the third day and five on the fourth day). Most had been referred from a spinal rehabilitation unit to exclude persistent soft tissue spinal cord compression (disc herniation, extradural haematoma) after reduction. Patients without symptoms and signs of spinal cord dysfunction, i.e. those with local pain or radiculopathy only, were excluded. Trauma involved the cervical spine in 25 patients, the thoracic spine in six, and the upper lumbar spine in one. Nine had suffered from road traffic accidents, 12 from falls, and 11 from accidents in recreational activities (diving, skiing, riding). Plain radiographs or CT showed vertebral fractures in 22 patients; 10 patients with no evidence of fracture had clinical deficits, at the cervical level in nine and thoracic in one.

MRI OF ACUTE SPINAL TRAUMA

101

(a)

(c)

(b)

(a)

Fig. 1 - (a,b) MRI 4 days after trauma in a 15-year-old boy with complete tetraplegia (Frankel grade A). Sagittal Tl-weighted (350/30/6) SE image (a) shows deformity and retropulsion of the C5 and C6 vertebral bodies whose signal intensity is lower than that of those above and below. At the same level the spinal cord appears swollen and slightly compressed but gives normal signal. High signal in the anterior and posterior extradural space at C5 and C6 could be due to venous stasis or to a circumscribed haematoma. On a sagittal T2-weighted (1500/200/2) SE image (b), the C5 and C6 vertebral bodies give high signal, presumably due to oedema or haemorrhage in the marrow (compression fractures). The spinal cord shows low signal centrally, with peripheral high signal. Note high signal in the prevertebral soft tissue from C3 to C5, and the soft tissue posterior to the spinal canal at C2-C3 and C3C4, indirect evidence of ligamentous injury. (c,d) M R122 months later; the clinical status was unchanged. Sagittal T 1-weighted (360/30/6) SE image (c) shows persistent retropulsion of the C4, C5 and C6 vertebral bodies. At the same level a heterogeneous area within the spinal cord, in part of signal intensity similar to that of the CSF, indicates development of a cystic lesion. On a gated, PD-weighted (1600/50/2) SE image (d) the injured portion of the cord gives increased signal, indicating myelomalacia.

Twenty patients underwent follow-up M R I over an interval o f from 10 days to 45 months after the accident (mean 7 months). All were clinically stable at the time of the follow-up examinations and were studied to investigate the evolution of the spinal cord lesions detected in the acute phase.

Magnetic Resonance Imaging All patients were examined after reduction and nonsur-

gical stabilization. Seven were examined in nonferrous halo traction by means of a pulley device attached to a sand bag. N o patient was imaged while having assisted ventilation. Examinations of the cervical spine were carried out using a standard transmitting and receiving volume coil which shows down to the T I - T 2 level. For the examination of the thoracic and lumbar spine, circular or rectangular receiver surface coils were used with the body volume coil as transmitter. After coronal scout 600/30/I (TR/TE/excitations)

102

CLINICAL RADIOLOGY

(a)

(d)

(b)

(c)

(e)

Fig. 2 - (a) CT at C4 2 days after trauma in a 15-year-old boy with tetraplegia and sparing of the vibration sense (Frankel grade B). A dense lesion consistent with intramedullary haemorrhage is seen in the anterior portion of the spinal canal. (b-d) MRI l day later. On a sagittal Tl-weighted (460/30/4) SE image (b), the spinal cord appears swollen from C3 to C5 and an area of low signal is seen behind C4. On a T2-weighted (I 820/200/2) SE image (c) the C4, C5 and C6 vertebral bodies give a high signal due to compression fractures. An area of low signal anteriorly in the spinal cord at C4, surrounded by diffuse high signal is evident. An axial T I-weighted (253/15/4) gradient echo image with 70 ~ flip angle (d) confirms low signal in ,the cord, whose shape closely resembles that of the dense area on CT. (e) MRI 8 months after trauma; the patient's clinical status was unchanged. A sagittal Tl-weighted (520/30/2) SE image shows persistent malalignment of the cervical spine and an inhomogeneous low signal in the anterior portion of the spinal cord consistent with extensive myelomalacia and atrophy.

103

MRI OF ACUTE SPINAL TRAUMA

Data Analysis and Statistics

images with 10 mm slice thickness over a 128 x 128 pixel matrix, 350-600/30/4 images with 5 7 mm slice thickness for TI weighting and 1600-1960/50-100/2 images with 7 mm slice thickness for proton density (PD) and T2 weighting were obtained in the sagittal plane with multislice spin-echo (SE) sequences and a matrix size of 192 • 256 pixels. Additional Tl-weighted images in axial and occasionally also coronal planes were usually acquired. In most examinations gradient modifications partially compensating for flow-related CSF signal loss (first order flow compensation) were used for the long TR-long TE images. In 15 initial and 13 follow-up examinations long T R images were acquired using cardiac gating. Supplementary gradient-recalled echo (GRE) 330 450/30/4 6 images with 5 7 mm slice thickness and a flip angle of 70 ~ for T1 weighting and 300/3060/4-6 with 5-7 mm slice thickness and a flip angle of 15~ for T2* weighting were also acquired in 16 initial (and seven follow-up) examinations.

The initial MRI examinations were retrospectively reviewed by two observers (M.M., V.Z.) blind to the clinical features for the following aspects: presence and type of intramedullary signal abnormalities; spinal cord swelling; persistent cord compression; signal changes in the vertebral marrow, bony deformity and fractures; residual bone dislocation; disc herniation or bulging; spondylosis; ligamentous injuries; extradural haematoma. The severity of the clinical deficit at the time of the initial MRI and at follow-up was scored using the Frankel classification [9] which identifies five broad functional levels: A, complete deficit; B, preserved sensation only; C, preserved nonfunctional motor; D, preserved functional motor; E, no deficit or complete recovery. Associations between MRI and clinical findings were analysed in contingency tables using the X2 test with

(a)

(b)

Fig. 3

104

CLINICAL RADIOLOGY

(c)

(d)

(e) Fig. 3 - (a) MRI 1 day after trauma in a 46-year-old man with tetraparesis (Frankel grade C). Sagittal-gated, T2-weighted (1820/150/2) SE image shows that the cervical spinal canal is narrowed by multiple vertebral spurs and bulging discs. A thick band of high signal (*) indicating extensive soft tissue oedema is present in the upper prevertebral space. The spinal cord is compressed and gives a high signal from C4 to C5. No intramedullary low signal is seen on a T2*-weighted (498/30/4) G R E image with 15~ flip angle (b) on which the intramedullary lesion appears less extensive. (c e) MR! 4 months later; strength had improved in the limbs (Frankel grade D). On a sagittal-gated T2-weighted (1500/150/2) SE image (c), a small area of high signal is seen behind the C4 C5 disc. The intrameduUary lesion (arrowhead) gives slightly higher signal than the normal cord and the CSF on a sagittal-gated PD-weighted (1500/50/2) SE image (d). On an axial T l-weighted (330/29/4) SE image (e), the intramedullary lesion appears as an eccentric area o f lower intensity than the normal cord but higher than the CSF. These changes are consistent with focal myelomalacia.

105

MRI O F A C U T E S P I N A L T R A U M A

Intramedullary signal changes were identified on T2weighted SE or T2*-weighted G R E images in 28 patients. Twelve showed central low signal, surrounded by extensive peripheral increased signal in 11 (Figs 1, 2), and 16

showed h o m o g e n e o u s high signal (Fig. 3). On T I weighted SE images one patient o f the former g r o u p showed low signal corresponding to the low intensity on T2-weighted images (Fig. 2), whereas all the others showed no alteration in intensity (Fig. 1). PD-weighted images usually confirmed the hyperintensity on the T2weighted images, but did n o t demonstrate low signal intensity in any spinal cord. W h e n patients were examined with both SE and G R E sequences the former were invariably superior for demonstrating intramedullary high signal (Fig. 3). In three patients intramedullary low signal was seen equally well on SE and G R E sequences, but the latter allowed detection o f unquestionably abnormal low signal within the spinal cord in three further patients with central areas o f intermediate signal and peripheral high signal on SE images (Fig. 4). N o n e o f the patients with intramedullary low signal on T2- or T2*weighted images had intraspinal bone fragments on CT. In four patients the spinal cord showed a n o r m a l signal at the site o f t r a u m a on all sequences. Spinal cord swelling was better appreciated on T1weighted images (Figs 1, 2) and was noted in 17 patients, all but one o f w h o m had intrameduUary signal changes,

(a)

(b)

Yates' continuity correction. The Fisher exact test was employed whenever the expected frequencies fell below n = 5 in a cell. All tests were two-sided. Values o f P below 0.05 were considered significant.

RESULTS Acute Phase

Clinical Findings At the time o f the initial M R I the patients were distributed according to the Frankel classification as follows: A = 14; B = 7; C = 6; D = 5; E = 0,

Magnetic Resonance Imaging

Fig. 4 - MRI 2 days after trauma in a 16-year-oldboy with complete tetraplegia (Frankel grade A). Sagittal-gated T2-weighted (1600/100/2) SE image (a) shows a crushed, slightly retropulsed C4 vertebral body with hyperintense marrow. The spinal cord is swollen from C3 to C5 and a central area of intermediate signal surrounded by diffuse high signal is seen behind the C4 vertebral body. On a sagittal T2*-weighted (300/50/4) GRE image with 15~ flip angle (b), the high signal in bone marrow and the spinal cord is less conspicuous, whereas a focal area of low signal is demonstrated at the site of the intermediate signal in (a), indicating the presence of deoxyhaemoglobin. Further low signal foci are seen above and below,

106

CLINICAL RADIOLOGY

Persistent cord compression caused by dislocation, spondylosis or herniated disc was identified in 26 patients, 23 of whom also had intramedullary signal changes (Figs 1, 3). The degree of compression was, however, mild in all instances (Figs 1, 3). The frequency of the other MRI findings is reported in Table 1. Despite the MRI examination having been prompted in several cases by a CT suggestive of extradural haematoma we did not find the expected signal changes in any case [6]. However, in many patients circumscribed areas of high or intermediate signal on T lweighted images and high signal on PD and T2-weighted images were detected behind the vertebral bodies, often in conjunction with protruded or herniated discs and were interpreted as congested veins (Fig. 1). The statistical analyses of possible associations between the initial M R I findings and the functional level are shown in Table 2. A significant association between intramedullary signal change and the severity of the deficit was observed. Patients with intramedullary low intensity on T2- or T2*-weighted images had the most severe deficits, patients with only high intensity had variably severe deficits and patients with no signal change were least impaired. Cord swelling was also associated with greater degrees of impairment. No significant association was observed for the other MRI features.

complications, three within the first week and one 2 months later. Mean clinical follow-up for the surviving patients was 13 months (range 2-41 months). Fifteen patients improved by at least one grade of the Frankel system, while 13 remained unchanged. All 11 surviving patients with the most severe functional deficit (Frankel grade A) were unchanged. One grade B patient was also unchanged, but six recovered partially, three rising to grade C and three to grade D. All grade C patients improved to grade D. Of the five patients with the slightest deficit (grade D), four recovered completely (grade E) and one was unchanged. Eventually, 15 patients were left with major deficits (grades A-C), i.e. bedridden or confined to a wheelchair, and 13 with minor deficits (grade D, E), i.e. could walk with or without support. Table 3 shows the correlation between acute MRI findings and residual functional level. A significant association was observed between major clinical deficit at follow-up (Frankel grades A-C) and intramedullary low intensity on T2- or T2*-weighted images in the acute phase. Cord swelling and marrow signal changes, bony deformity and fractures were also statistically associated with more severe degrees of residual functional deficit.

Magnetic Resonance Imaging

Follow-up Clinical Findings All patients were treated conservatively and none underwent surgical decompression procedures after MRI. Four patients (three grade A, one grade C) died of Table 1 - MRI in 32 patients with acute traumatic myeiopathy

Intr0medullary signal changes on T2- or T2*-weighted images Low signal + high signal High signal alone Cord swelling Cord compression Marrow signal changes, bone deformity ahd fractures Dislocation Disc herniation or bulging Spondylosis Ligamentous injuries

28 12 16 17 26 17 17 14 8 16

Eighteen of the 20 patients undergoing follow-up MRI showed abnormal signal intensity in the spinal cord. Although it was possible in several cases to distinguish solid from cystic intramedullary lesions, since the former were more intense than CSF (Fig. 3) and the latter isointense with CSF (Fig. 1) on PD-weighted images, this distinction was difficult in some cases and solid and cystic components coexisted in many cases (Figs 1, 2). Intramedullary low signal consistent with haemosiderin deposition was never observed on T2- or T2*-weighted images. Intramedullary signal changes extended longitudinally for more than one vertebral body in 10 patients (Figs 1, 2) and for one vertebral body or less in eight (Fig. 3). The former appearance was associated with more severe residual functional deficit (Table 4). Persistent cord compression was identified at follow-up

Table 3 - MRI findings in the acute stage and residual functional deficit in 28 surviving patients

Table 2 - MRI findings and functional level in the acute stage

Frankel grades

Frankel classification

A (14) lntramedullary signal on T2/T2*-WI Low signal (12) __+high signal High signal alone (16) Normal (4) Cord swelling (17) Cord compression (26) Marrow signal changes, bony deformity and fractures (17) Ligamentous injuries (16) NS, Non-significant.

B (7)

C (6)

D (5)

P

I1 3 0 9 13

1 6 0 6 5

0 4 2 0 4

0 3 2 2 4

l vertebral body (8) Persistent cord compression (12)

A, B and C (12)

D and E (8)

9 3 8

1 5 4

}

0.03 NS

NS, Non-significant. Table 5 - M R I in acute and chronic phases

Longitudinal extent of the spinal cord lesion

Intramedullary signal on T2/T2*-WI Low signal (6) ___high signal High signal alone (10) Normal (2) Cord swelling ( I 1) Cord compression (17) Marrow signal changes, bony deformity and fractures (10) Ligamentous injuries (12)

1 vertebral body (8)

1 8 1 5 10

5 2 1 6 7

4

6

NS

6

6

NS

0.04 NS NS

NS, Non-significant.

in 12 patients but was not apparently related to residual function (Table 4). Table 5 shows the correlations between acute and follow-up MRI findings. Acute intramedultary low signal on T2- or T2*-weighted images was associated with a longitudinal extent of the signal changes over more than one vertebral body at follow-up (Figs 1, 2). Conversely, in patients with only intramedullary high signal on the initial MRI the extent of the lesion at follow-up was often less than one vertebral body, and considerably reduced in several patients (Fig. 3). DISCUSSION Intramedullary low signal was seen at the site of trauma on T2- or T2*-weighted images in many of our patients, notably those with the most severe clinical deficit. Although direct surgical or pathological verification was not available, data obtained at 0.5 T in human intracranial haemorrhage [10], an animal model of spinal trauma [11], and in a patient dying a few days after spinal injury [12], support the hypothesis that the low signal is attributable to susceptibility effects of deoxyhaemoglobin. Hence, post-traumatic spinal cord lesions exhibiting this signal change can be assumed to have relatively abundant haemorrhage. We found that demonstration of acute intramedullary haemorrhage at 0.5 T is enhanced by G R E sequences, which is not surprising as they are known to be more sensitive than SE sequences to magnetic susceptibility effects of deoxyhaemoglobin [8, 13, 14]. Our experience therefore significantly differs

107

from that of others groups, who did not find signal changes indicative of haemorrhagic contusion in acutelyinjured patients imaged at 0.5 T [7,15]. In all but one of our patients with focal low signal on T2- or T2*-weighted images there was diffuse high signal in the adjacent spinal cord, presumably related to oedema, which in some cases may hinder identification of the low signal. As in previous studies [4-6], our patients with only high signal on T2- or T2*-weighted images had less severe, although variable, clinical deficits. The considerable reduction of the extent of the abnormal signal in some of these patients on follow-up examinations supports the interpretation of this high signal as representing oedema. An association between cord swelling and severity o f initial clinical deficit was reported by Flanders et al. [6]. Our data confirm this observation. Unlike Flanders et aL [16], however, we did not find persistent cord compression after reduction to be significantly associated with more severe clinical deficits. This may be related to our inclusion criteria: we excluded patients with vertebral fractures with a normal neurologic examination (whereas these were included by Flanders et al.).

Few data on the relationship between acute MRI findings and residual clinical deficit are available and no detailed report o f the therapeutic options is given. This is surprising since treatment of acute traumatic myelopathy is very controversial [2,6], there being no proven benefit for emergency surgical decompression, at least in patients with clinically-complete cord transection [16,17]. All our patients were managed conservatively before and after MRI. This uniform treatment, based on absence of significant compression on MRI, was advantageous in evaluation of the prognostic value of acute MRI since it eliminates possible bias inherent in different treatments. When two gross categories of residual functional deficit were considered, a strong association was seen between intramedullary signal patterns and clinical outcome. Patients with haemorrhagic lesions invariably failed to improve, whereas some patients with oedematous lesions, and those without abnormal signal intensity showed significant recovery. We considered marrow signal changes, bony deformity and fractures as a single comprehensive variable which correlated weakly with clinical outcome. A similar association has been reported by Kulkarni et al. [5] and the apparent conflict with data from plain film and CT studies [1,2] is probably explained by the marrow signal changes, frequently observed on MRI and not detected by the other techniques. In the chronic phase most of our patients, despite variable clinical outcome, showed MRI findings consistent with myelomalacia or a circumscribed intramedullary cyst, without a clear relationship to the appearance of the spinal cord in the acute phase. Like Takahashi et al. [15] we found a relationship between the extent of the sequelae of trauma on follow-up MRI and clinical status. Syringomyelia was not identified in any of our patients, which may be due to the relatively short follow-up [7]. Longer follow-up on larger series may be needed to identify the acute changes associated with this late complication. The capacity of M R I to characterize acute traumatic damage to the spinal cord has little effect on treatment. It is however conceivable [2] that to discriminate haemorrhagic and non-haemorrhagic injuries could favourably

i 08

CLINICAL RADIOLOGY

influence selection of candidates for surgical decompression, who are likely to be those with persistent severe cord compression without intramedullary signal change or with oedema alone. This distinction might also help explain the different responses to bolus intravenous administration of steroids observed in large clinical trials

9

10

[171. I1

REFERENCES

12 1 Atlas SW, Regenl~ogen V, Roger LF, Kim KS. The radiographic characterization of burst fractures of the spine. American Journal o f Roentgenology 1986;147:575 582. 2 Hackney DB. Denominators of spinal cord injury. Radiology 1990; 177:18-20. 3.Mirvis SE, Geisler FH, Jelinek JJ, Joslyn JN, Gellad F. Acute cervical spine trauma: evaluation with 1.5-T MR imaging. Radiology 1988; 166:807-816. 4 Kulkarni MV, McArdle CB, Kopanicky D, Miner M, Cotler HB, Lee KF et al. Acute spinal cord injury: MR imaging at 1.5 T. Radiology 1987; 164:837-843. 5 Kulkarni MV, Bondurant FJ, Rose SL, Narayana PA. 1.5 tesla magnetic resonance imaging of acute spinal trauma. Radiographics 1988;8:1059-1082. 6 Flanders AE. Schaefer DM, Mishkin MM, Gonzalez CE, Northrup BE. Acute cervical spine trauma: correlation of MR imaging findings with degree ofneurologic deficit. Radiology 1990; 177:25-33. 7 Yamashita Y, Takahashi M, Matsuno Y, Sakamoto Y, Oquai T, Sakac T et al. Chronic injuries of the spinal cord: assessment with MR imaging. Radiology 1990;175:849-854, 8 Brooks RA, Di Chiro G, Patronas N. MR imaging of cerebral

13 14 15 16 17

18

hematomas at different field strengths. Journal r Assisted Tomography 1989;13:194-206. Frankel HL, Hancock DO, Hyslop G, Walsh L, Michaelis J, Melzack D. The value of postural reduction in initial management of closed injuries of the spine with paraplegia and tetraplegia. Paraplegia 1969;7:179 192. Zimmerman RD, Heier LA, Snow RB, Liu DPC, Kelly AB, Deck MDF. Acute intracranial hemorrhage and intensity changes on sequential MR scans at 0.5 T. American Journal o f Roentgenology 1988; 150:651-661. Schouman-Claeys E, Frija G, Cuenod CA, Begon D, Pararire F, Martin V. MR imaging of acute spinal cord injury: results of an experimental study in dogs. American Journal o f Neuroradiology 1990; 11:959-965. Beers GJ, Raque GH, Wagner GG, Shields CB, Nichols GR, Johnson JR et al. MR imaging in acute cervical spine trauma. Journal of Computer Assisted Tomography 1988; 12:755-76 I. Edelmann RR, Johnson K, Buxton R, Shoukimas G, Rosen BR, Davis KR et al. MR of hemorrhage: a new approach. American Journal of Neuroradiology 1986;7:751-756. Mascalchi M, Dal Pozzo G. Mottled appearance of a hyperacute intracranial hemorrhage at 0.5 T. Journal of Computer Assisted Tomography 1990;14:387 390. Takahashi M, Izunaga H, Satoh R, Shizato J, Korogi Y, Yamashita Y. Correlation of sequential MR changes of the injured spinal cord with prognosis. Radiology 1991; 181(P): 172. Janssen L, Hansebout RR. Pathogenesis of spinal cord injury and newer treatments. Spine 1989; 14:23-32. Hansebout RR. A comprehensive review of methods in improving spinal cord recovery after injury. In: Tator CH, ed. Early managemerit o f acute spinal cord injury. New York: Raven Press, 1982:181 196. Bracken MB, Shepard M J, Collins WF, Holdorf TR, Young W, Baskin DS et al. A randomized, controlled trial of methylprednisolone or naloxone in the treatment of acute spinal cord injury. New England Journal of Medicine 1990;322:1405 141 I.