European Journal of Radiology, 12 (1991)

219-225

219

Elsevier

EURRAD

00169

Spinal cord cavities: differential-diagnostic imaging P. Schubeus, W. Schemer,

criteria in magnetic resonance

N. Hosten

and R. Felix

Department of Radiology, University Clinic Rudolf VirchowlCharlottenburg,

(Received

Key words: Spinal cord, neoplasm;

19 September

1990; accepted

Magnetic resonance

after revision

imaging, syringomyelia;

Free University of Berlin. Berlin

19 January

1991)

Magnetic resonance

imaging, contrast

medium

Abstract

MRI examinations of 30 patients with idiopathic syringomyelia and 10 patients with cavities associated with an intramedullary neoplasm were evaluated with respect to typical MRI features in both groups. All tumor-associated cases resembled the idiopathic syringomyelias in some portions of the cavity. At the tumor site, however, tumor-associated cases demonstrated typical findings: the cavities showed abrupt changes of diameter (IO/lo) and position (S/10) and the surrounding spinal cord demonstrated an uneven thickness (lo/lo), an increased signal intensity on T2-weighted images (lo/lo) and pathological contrast enhancement (7/7). Displacement of cerebellar tonsils below the level of the foramen magnum (21/30) and enlargement of the spinal canal (7/29) were characteristic features of idiopathic cases. In conclusion, MRI provides valuable criteria to differentiate between idiopathic and tumor-associated cavities.

Introduction Intramedullary cavities can be of idiopathic origin or can be induced by trauma, inflammation or tumor. Differential diagnosis of syringomyelia is clinically important, since treatment of idiopathic and tumor-associated cases is different: idiopathic cases are often referred to shunt procedures, whereas neoplasms require radical removal or radiotherapy. Differentiation of both groups using clinical findings may be difficult and often requires a long-time follow-up of the symptoms [ 11. In order to refer the patient for the appropriate therapy as early as possible, imaging techniques have to be applied. In recent years MRI proved to be a highly sensitive imaging technique in the diagnosis of spinal cord cavities [ 2-41. Few artifacts, high contrast between the spinal cord and the cerebrospinal fluid (CSF) and multiplanar imaging modalities are advantages of MRI over CT in visualizing the cavity and the surrounding spinal Address for reprints: Dr. Peter Schubeus, Department University Clinic Rudolf Virchow/Charlottenburg, Damm 130. 1000 Berlin 19, F.R.G. 0720-048X/91/$03.50

ofRadiology, Spandauer

0 1991 Elsevier Science Publishers

B.V.

cord. However, up to now only few investigators have focussed on the differential-diagnostic values of MRI findings of intramedullary cavities (5). The aim of the present study was to establish criteria that may differentiate between idiopathic and tumor-associated cavities by evaluating MRI of patients showing strong clinical evidence that their signs were either idiopathic or induced by an intramedullary neoplasm. Patients and Method Cavities of the spinal cord were found in about 60 cases from a total of more than 10000 MRI examinations. Four idiopathic and 7 tumor-associated cavities were proven by surgery. If there was no surgical proof, the inclusion to the groups was based on clinical and radiological findings. Rapid progressive pareses, elevated CSF protein and a suspected tumor at CT are known to be indicative for an intramedullary neoplasm [ 11. Cases were included in the idiopathic group, if none of these criteria was found in a long-time clinical follow-up. If all these criteria were met, on the other hand, cases were included in the tumor-associated group. In this way 40 patients could be included in one

220

of the groups. The remaining cases showed inconsistent clinical/radiological findings and were excluded from the study. The idiopathic group consisted of 30 patients, 8 men and 22 women, ranging in age from 19 to 82 years. 19 patients had dissociated anaesthesia, 16 painful sensory symptoms in the arms, 15 a paresis of the arms with lower motor-neuron signs, and 4 a paraparesis of the legs. The patients have had a clinical follow-up for 10 years in average. All but one idiopathic case had been examined by CT and 4 patients underwent shunt surgery. Ten patients, 5 men and 5 women, had cavities associated with an intramedullary neoplasm. The age of the patients was in the range of 7 to 80 years. The clinical symptoms of tumor-associated cases were different from those of the patients with idiopathic syringomyelia. Eight patients with tumor-associated cavities had a progressive paresis of the arms and/or legs with upper and lower motor-neuron signs. Dissociated anaesthesia was not found in this group. Seven patients with tumorassociated cavities underwent surgery. Histological diagnoses included 2 haemangioblastomas, 1 astrocytoma, 2 ependymomas, and 2 gliomas. All examinations were performed with an 0.5 T whole body scanner (Magnetom, Siemens). Usually a head coil was used for examinations of the cervical area and a surface coil for examinations of the rest of the spine. In all patients sagittal Tl-weighted spin echo images (SE) with a repetition time (TR) of 400 ms and an echo time (TE) of 30 ms (SE 400/30) or sagittal Tl-weighted gradient echo images (FLASH) with a TR of 315-500 ms, a TE of 14 ms and a flip angle of 90” (FLASH 3 15-500/14) were performed. T2-weighted spin echo images with a TR of 1600-2000 ms and a TE of 70-90 ms (SE 1600-2000/70-90) were generated in 29 cases. Axial views were obtained in 35 cases. A 256 x 256 matrix was used for data acquisition and the field-of-view was 300 x 300 mm. The slice thickness usually was 5 mm. In 12 cases Tl-weighted images were repeated following administration of 0.1 mmol Gadolinium (Cd)-DTPA/kg body weight (Magnevist, Schering). Examinations were evaluated with respect to the shape of the intramedullary cavity, the spinal cord, the spinal canal and the craniovertebral junction. Cavities were examined for localization, length (in vertebral segments), maximal a.-p. diameter, abrupt changes of diameter or position (on axial views) and signal intensity, as compared to normal spinal cord tissue. The spinal cord surrounding the cavity was inspected for enlargement, uneven contours, uneven thickness and its signal intensity on plain and contrast enhanced images,

as compared to normal spinal cord tissue. The spinal canal was examined for enlargement and disappearance of the CSF cleft (the morphology of the spinal canal was not evaluated in 3 cases, which had been examined after surgery). The craniovertebral junction was inspected for tonsillar ectopia, basilar impression and other malformations. Results MRI findings of idiopathic and tumor-associated cases are summarized in Table 1. Cavity Idiopathic cavities showed a length of 4 to 19 vertebral segments (mean = 10) and a maximal diameter of 2 to 13 mm (mean = 6.0 mm). All cases involved the cervical region. The length of the tumor-associated cavities was in the range of 5 to 21 vertebral segments (mean = 12) and the maximal diameter ranged between 2 to 16 mm (mean = 7.8 mm). Tumor-associated

TABLE

1

MRI findings in 30 patients with idiopathic patients with tumor-associated cavities

Cavity Abrupt change of diameter Abrupt change of position’ Spinal cord Enlargement - diffuse - circumscribed Uneven outside contours Uneven thickness Increased signal intensity on TZ-weighted images - surrounding the cavity - diffuse Contrast enhancement Spinal canal b Enlargement Disappearance

of CSF cleft

Craniovertebral junction Tonsillar ectopia - 3 mm or more - 5 mm or more Basilar impression

syringomyelia

and 10

Idiopathic cases

Tumor-associated cases

o/30 2125

lo/loa s/10=

19/30 o/30 o/30 l/30

7110 3/10” 2/1oa lo/lo=

6119 o/19 O/5

2/10 10/1oa 717”

7129 l/29

l/8 318”

21130 15/30 2/30

O/l0 O/l0 O/l0

a At the tumor site. b Not evaluated in cases that had been operated ’ Only evaluated on axial images.

on.

Fig. 1. Typical appearance of idiopathic syringomyelia on sagittal MR images (5 1 years, female, clinical symptoms for 19 years, MR follow-up for the last 4 years). (a) SE 1600/90, precontrast: the TZ-weighted image demonstrates an enlarged spinal cord with an intramedullary cavity extending from Cl to TlO (shown here from Cl to T2). Note slightly increased signal intensity of the spinal cord surrounding the cavity (arrows). (b) FLASH 315/14, precontrast: an even thickness of the surrounding spinal cord and caudal displacement of cerebellar tonsils (arrow) is demonstrated on the plain Tl-weighted image. (c) FLASH 315/14, postcontrast: the enhanced Tl-weighted image does not display pathological contrast enhancement.

cavities involved the cervical cord in 9/10 cases. One tumor-associated cavity was localized in the lower thoracic area. Idiopathic cavities did not show abrupt changes of diameter (Fig. 1). An abrupt eccentric cavity position was seen in 2/25 idiopathic cases. All tumor-associated cavities resembled the idiopathic cases in some portions. However, adjacent to the suspected or proven tumor localization all cases demonstrated abrupt changes of cavity diameter (Figs. 2 and 4) and 8/10 cavities showed an abrupt eccentric position (Fig. 4). Idiopathic as well as tumor-associated cavities were hypointense on Tl-weighted images, as compared to normal spinal cord tissue. On T2-weighted images cavities were hypointense in 6/19 and O/10 cases, isointense in 6119 and 4/10 cases and hyperintense in 7119 idiopathic and 6/10 tumor-associated cases, respectively. Spinal cord Idiopathic syringomyelias showed diffuse spinal cord enlargement in 19/30 cases (Fig. 1). All cases demonstrated even outside contours and an even thickness of the spinal cord surrounding the cavity was found except for one case. 7/10 tumor-associated cases

showed diffuse spinal cord enlargement. The appearance of the surrounding spinal cord was similar to idiopathic cases in large portions of all tumor-associated cavities. However, at the suspected or proven tumor site all cases displayed characteristic findings : Uneven contours, circumscribed enlargement and uneven thickness of the spinal cord (Figs. 3 and 4) were found in 2/10, 3/10, and lO/lO cases, respectively. 6/19 idiopathic cases displayed an increased signal intensity on TZweighted images surrounding the cavity as a small hyperintense rim (Fig. 1). A diffusely increased signal intensity on T2-weighted images was not seen in this group. In all tumor-associated cases an increased signal intensity of spinal cord tissue on T2weighted images was demonstrated at the tumor site (Fig. 2). This increased signal intensity was of diffuse appearance and could be distinguished from the small hyperintense rim surrounding some idiopathic cavities. Five idiopathic cases were also examined after administration of Gd-DTPA. On postcontrast images there was no pathological enhancement in spinal cord tissue in these cases (Fig. 1). In contrast, all tumors showed a pathological enhancement after administration of Gd-DTPA (7/7). Tumor delineation was improved in all cases (Fig. 2-4).

Fig. 2. Characteristic features of a tumor-associated cavity on sagittal MR images (28 years, female, histologically proven ependymoma). (a) SE 1600/90, precontrast: the TZ-weighted image shows a hyperintense intramedullary cavity extending from the medulla oblongata to C6/7 and increased signal intensity ofthe spinal cord (arrows) adjacent to the cavity. (b) FLASH 315/14, precontrast: changes of cavity diameter between C3 and C5 (arrows) are demonstrated on the plain Tl-weighted image. There is no displacement of cerebellar tonsils. (c) FLASH 315/14, postcontrast: the enhanced Tl-weighted image shows a pathological contrast enhancement ofthe spinal cord between C3 and C5, corresponding with a histologically proven ependymoma (arrows). There is no contrast enhancement in the spinal cord tissue surrounding the associated cavity.

Spinal canal

Spinal canal was enlarged in 7129 idiopathic syringomyelias but only in l/8 tumor-associated case. Disappearance of the CSF cleft adjacent to the suspected or proven tumor was observed in 3/8 tumor-associated cases. Craniovertebral junction

In idiopathic syringomyelias the cerebellar tonsils were displaced 3 mm or more below the foramen magnum in 21/30 cases and 5 mm or more below the foramen magnum in 15/30 cases (Fig. 1). Tonsillar ectopia was not found in tumor-associated cases 2130 idiopathic cases de(Fig. 2). Furthermore, monstrated a basilar impression. Other hind-brain malformations like abnormalities of the 4th ventricle or platybasia were not found in this study. Discussion In the diagnosis of syringomyelia ruling out an underlying neoplasm is one of the main goals of every imaging technique [4]. The exclusion of a tumor can be achieved either by demonstrating specific findings of

idiopathic syringomyelia or excluding features which are frequent in tumor-associated cavities. In this study a circumscribed uneven thickness of spinal cord surrounding the cavity, as well as an abrupt change of cavity diameter and cavity position were frequent in tumor-associated cases, whereas these features were rare in idiopathic cavities. Thus, these criteria seemed to be useful to verify or exclude a underlying tumor. Williams et al. [ 51 reported, that uneven margins of an intramedullary cyst indicate an underlying neoplasm. Our data confirmed this finding. Small tumors, like haemangioblastomas, showed discrete changes of the medullary contours, that could have been overlooked on routine sagittal imaging using a slice thickness of 5 mm. In these cases axial images were helpful to demonstrate changes of cavity position or spinal cord thickness. An increased signal intensity of the spinal cord on T2-weighted images was another helpful criterion in the differentiation of idiopathic and tumor-associated cavities. All tumor-associated cases showed a diffuse increase in signal intensity on T2-weighted images at the tumor site, according to other investigator’s results [6,7]. In contrast to that, idiopathic cases only displayed a small hyperintense rim surrounding the cavity

223

Fig. 3. Characteristic features of a tumor-associated cavity on axial MR images (same patient as in Fig. 2). (a) FLASH 500/14, precontrast, C2-level: the plain Tl-weighted image at the level of C2 demonstrates a cavity with an even thickness of the surrounding spinal cord, resembling an idiopathic syringomyelia. (b)FLASH 500/14, postcontrast, CZ-level: the enhanced Tl-weighted image at the CZ-level does not show pathological contrast enhancement. At surgery no tumor tissue was found at this level. (c) FLASH 500/14, precontrast, C4-level: at the level ofC4 an uneven thickness ofthe spinal cord surroundingthe cavity (arrow) is demonstrated on the plain Tl-weighted image. (d) FLASH 500/14, postcontrast, C4-level: the enhanced Tl-weighted image at the C4-level reveals a diffuse contrast enhancement in the spinal cord tissue (arrows), corresponding with a histologically proven ependymoma.

in some cases. This feature might represent surrounding edema or gliosis [ 81. In this way, T2-weighted images were important to demonstrate or exclude a tumor in patients with intramedullary cavities. However, exact delineation of the tumor extent proved to be difficult in

several cases, because cavities as well as tumors appeared hyperintense on T2-weighted images. In the present study all tumors examined with Gd-DTPA showed contrast enhancement, whereas there was no contrast enhancement in idiopathic cases

Fig. 4. Changing aspect of a tumor-associated cavity on contiguous axial MR images (35 years, male, intramedullary tumor). (a-d) FLASH 315/14, postcontrast: the enhanced Tl-weighted image at the level of C3 (a) shows a centric cavity position (white arrows) and no pathological enhancement. More caudal localized contiguous sections (b-d) demonstrate an abrupt change of cavity diameter and cavity position (white arrow) adjacent to the contrast enhancing tumor mass (black arrows).

examined after administration of Gd-DTPA. Contrast enhancement generally improved tumor delineation and made it easier to recognize small tumors. Most intramedullary tumors, even low-graded astrocytomas, show contrast enhancement [9-l 11, whereas the surrounding spinal cord in idiopathic syringomyelias does not enhance [lo]. These results suggest, that adminis-

tration of Gd-DTPA is an important tool to verify or exclude an underlying tumor when syringomyelia is found. Less frequent, but also quite specific findings in tuI-nor-associated cavities were circumcribed enlargement and irregular outside contours of spinal cord and disappearance of the CSF cleft adjacent to the tumor. How-

225

ever, since these findings were infrequent, the absence of these findings cannot exclude an underlying tumor in syringomyelia. In this study enlargement of the spinal canal was frequent in idiopathic cases and can indicate a primary syringomyelia [ 121. On the other hand, an enlarged spinal canal was also seen in one 7-year-old patient with an intramedullary tumor. In the present study, tonsillar ectopia was a frequent and specific finding in idiopathic syringomyelia. The role of ectopic cerebellar tonsils in the pathogenesis of syringomyelia is still unclear [ 13-181. Furthermore, cerebellar tonsils can be localized slightly below the for-amen magnum in normal controls as well. Aboulezz et al. claimed, that only a displacement of cerebellar tonsils of 5 mm or more below the level of the foramen magnum should be called tons&u ectopia [ 191. We also found idiopathic cases without ectopic cerebellar tonsils. On the other hand, a coincidence of tonsillar ectopia and an intramedullary tumor can occur in rare cases [ 81. For this reason, exclusion of an underlying tumor in syringomyelia should not rely on the presence of tonsillar ectopia alone. In conclusion, MRI provides valuable criteria to differentiate between idiopathic and tumor-associated cavities. MR examinations should cover the whole cavity and include axial and sagittal Tl-weighted images, TZweighted images and contrast enhanced Tl-weighted images to demonstrate all features of the cavity, the surrounding spinal cord, the spinal canal, and the craniovertebral junction. References 1 Schliep G. Syringomyelia and syringobulbia. In: Vinken PJ, Bruyn GW, eds. Handbook of Clinical Neurology, Vol. 32. Amsterdam: Elsevier/North-Holland Biomedical Press 1978; 255-328. 2 Yeates A, Brant-Zawadski M, Norman D, Kaufman L, Crooks LE, Newton TH. Nuclear magnetic resonance imaging of syringomyelia. AJNR 1983; 4: 234-237.

3 Pojunas K, Williams AL, Daniels DL, Haughton VM. Syringomyelia and hydromyelia: magnetic resonance evaluation. Radiology 1984; 153: 679-683. RD, Manning JJ, Deck MDF. MR 4 Lee BCP, Zimmermann imaging of syringomyelia and hydromyelia. AJNR 1985; 6: 221-228. 5 Williams AL, Haughton VM, Pojunas KW, Daniels DW, Kilgore DP. Differentiation of intramedullary neoplasms and cysts by MR. AJNR 1987; 8: 527-532. 6 Goy AMC, Pinto RS, Raghavendra BN, Epstein EJ, Kricheff II. Intramedullary spinal cord tumors: MR imaging with emphasis on associated cysts. Radiology 1986; 161: 381-386. 7 Scotti G, Scialfa G, Colombo N, Landoni L. Magnetic resonance diagnosis of intramedullary tumors of the spinal cord. Neuroradiology 1987; 29: 130-135. 8 Sherman JL, Barkovich AJ, Citrin CM. The MR appearance of syringomyelia: new observations. AJNR 1986; 7: 985-995. 9 Bydder GM, Brown J, Niendorf HP, Young IR. Enhancement of cervical intraspinal tumors in MR imaging with intravenous Gadolinium-DTPA. J Comput Assist Tomogr 1985; 9: 847-851. 10 Dillon WP, Norman D, Newton TH, Bolla K, Mark A. Intradural spinal cord lesions: Gd-DTPA-enhanced MR imaging. Radiology 1989; 170: 229-237. 11 Sze G, Krol G, Zimmerman RD, Deck MDF. Intramedullary disease of the spine: diagnosis using Gadolinium-DTPAenhanced MR imaging. AJNR 1988; 9: 847-858. 12 Hertel G. The width of cervical spinal canal and the size of the vertebral bodies in syringomyelia. Eur Neurol 1973; 9: 168-182. 13 Williams B. The distending force in the production of”‘communieating syringomyelia”. Lancet 1969; ii: 189-193. 14 Ball MJ, Dayan AD. Pathogenesis of syringomyelia. Lancet 1972; 2: 799-801. 15 DuBoulay G, Shah SH, Currie JC, Logue V. The mechanisms of hydromyelia in Chiari type I malformation. Br J Radio1 1974; 47: 579-587. 16 Gardner WJ, Angel J. The mechanisms of syringomyelia and its surgical correction. Clin Neurosurg 1975; 6: 131-140. 17 Aboulker J. La syringomyelie et les liquides intra-rachidiens. Neurochirurgie 1979; 25, Suppl 1: 1-144. in the diagnosis and 18 Cahan LD, Bentson JR. Considerations treatment of syringomyelia and the Chiari malformation. J Neurosurg 1982; 57: 24-31. 19 Aboulezz AO, Sator K, Greyer CA, Gado MH. Positions of cerebellar tonsils in the normal population and in patients with Chiari malformation: a quantitative approach with MR imaging. J Comput Assist Tomogr 1985; 9: 1033-1036.

Spinal cord cavities: differential-diagnostic criteria in magnetic resonance imaging.

MRI examinations of 30 patients with idiopathic syringomyelia and 10 patients with cavities associated with an intramedullary neoplasm were evaluated ...
5MB Sizes 0 Downloads 0 Views