Signal intensity patterns in intraspinal dermoids and epidermoids on MR imaging

Signal intensity patterns in intraspinal dermoids and epidermoids on MR imaging

ClinicalRadiology(1993)48, 405-413 Signal Intensity Patterns in Intraspinal Dermoids and Epidermoids on MR Imaging S. G U P T A , R. K. GUPTA, R. B. ...

5MB Sizes 0 Downloads 17 Views

ClinicalRadiology(1993)48, 405-413

Signal Intensity Patterns in Intraspinal Dermoids and Epidermoids on MR Imaging S. G U P T A , R. K. GUPTA, R. B. G U J R A L , P. MITTAL*, M. K U R I Y A L * and N. K R I S H N A N I t

Departments of Radiodiagnosis, *Neurosurgery and t Pathology, Sanjay Gandhi Postgraduate Institute of Medical Sciences, Lucknow, India Retrospective evaluation of MR images of nine intraspinai dermoid (n = 4) and epidermoid ( n = 5 ) tumours was undertaken. Tl-weighted images were available in all the patients while proton density (PD) and T2-weighted images were available in seven patients. Except for one lesion (only Tl-weighted images were available in this case), all the lesions were clearly delineated on M R images. Most (5 out of 7) were better appreciated on T1- and PD-weighted images than on T2-weighted images. On Tl-weighted images, the epidermoids were either isoto mildly hypointense to the cord (n = 3), isointense to CSF (n = 1) or markedly hyperintense (n = 1). On the PD images, the epidermoids were either isointense to cord (n = 3) or markedly hyperintense (n = 1). All the lesions were iso- to hyperintense to CSF on T2-weighted images. Of the four dermoids three showed a mixed intensity pattern on all sequences; the dominant component gave short TI and short T2 values, consistent with fat. The fourth lesion showed homogenous material with a signal intensity intermediate between that of CSF and cord on T1weighted images. Dermal sinus tracts were present in four of the nine patients; three patients had an intraspinal component. This study shows that although MR imaging is excellent for their detection, it cannot differentiate the two tumour types. Gupta, S., Gupta, R.K., Gujral, R.B., Mittal, P., Kuriyal, M. & Krishnani, N. (1993). Clinical Radiology 48, 405-413. Signal Intensity Patterns in Intraspinal Dermoids and Epidermoids on M R Imaging

Accepted for Publication 20 May 1993

Dermoids and epidermoids are congenital tumours that account for 1-2% of intraspinal tumours of all ages and 10% of intraspinal tumours below the age of 15 years [ 13]. CT myelography is excellent in the detection of intradural extramedullary tumours but is insensitive to intramedullary lesions which do not expand the core [4]. Sonography is useful in infants as the wide interlaminar spaces provide sonographic windows, allowing visualization of intra- and extramedullary tumours [5]. It is ineffective in older children. Although the efficacy of magnetic resonance (MR) imaging in the evaluation of various congenital spinal abnormalities is well documented [6,7], the literature describing the M R appearance of spinal dermoids and epidermoids is scanty [8-13]. We reviewed the M R findings of nine patients with intraspinal dermoid and epidermoid tumours in an attempt to evaluate the efficacy of M R in detection of these lesions and also to identify the signal intensity characteristics of each tumour type in an attempt to improve specificity.

superconducting system operating a 1.5 T (Magnetom SP; Siemens Medical System, Germany) using an oval spine coil. Sagittal Tl-weighted images (SE500/15) were obtained in all the patients. Sagittal proton density (PD) and T2-weighted images (Dual echo SE2500/20,80) were obtained in seven patients (cases 3-9). A slice thickness of 3 mm with inter-slice gap of 0.3 mm was used with a 256 x 256 image matrix. In one patient (case 8), 5 mm thick sagittal short TI inversion recovery (STIR 1500/100/ 28) images were also available. The MR studies were assessed for the size, location, signal characteristics and signal homogeneity of the lesions, presence and level of any associated cutaneous lesions, level ofconus medullaris and the location and signal characteristics of the sinus tract if present. The TI-, T2- and PD-weighted images were subjectively analysed to determine which sequence provided the highest lesion conspicuity, i.e. clearly differentiated the lesion from the cord and CSF. RESULTS

P A T I E N T S AND M E T H O D S M R studies of nine consecutive patients with intraspinal dermoid and epidermoid tumours were retrospectively evaluated. The study included four male and five female patients aged 6 months to 11 years. The diagnosis was confirmed by histological examination of the surgical specimens. The M R examinations were performed with a 2T Correspondence to: Dr R. K. Gupta, Associate Professor, MR Section, Department of Radiology, Sanjay Gandhi Postgraduate Institute of Medical Sciences,PB No. 375, Raibareli Road, Lucknow226001, India.

Table 1 summarizes the M R characteristics of the nine spinal lesions studied. Epidermoid tumours showed a variety of signal intensities on T 1-weighted images. Three lesions were iso- to mildly hypointense to the normal spinal cord. One of these lesions (case 4), in addition, showed a mixed intensity region in its cranial portion (Fig. 3a); chemical shift artifact was noted at the interface between the hyperintense and the isointense parts in the frequency encode direction. One lesion, being isointense to CSF (long TI) was invisible on Tl-weighted images; P D - a n d T2-weighted images were not available in this patient. Surgery, however, revealed small intradural

406

CLINICAL

RADIOLOGY

a

,,, ~ o

.

~+g~

+o~+

e~

S=~ o=,~ 9s o . ~

_

Z

Z

.~'.~

=

_

~

=

,,o

=

~,~ ~ o~

~ ~ o

=_~=

~s163

=

o= o = ~ o=&

= = ~e0 &

=

t~

gu = =~&

o~

~

g 6 ..

Z

o

~

~

Z

~

o

.~

~

~

~

~~

~

~

ol

&

.~

a.,

~s .~

~=~.o=~

~

o.~

~

o

<

.o7~= I= 9s r

<

~

co

.~

<

co

r-,

<

<

<

8 b~

o

a

6

=

-~

7==~

6~

"~ ~ . ~

~ a

~.~

g.=

o

6~

a3

7_=-~

&g

.~ -~

z

rr

c-4

<

t"q

I

t-,,i

t~

"-4 I

b~ r~ [-.

t,-q

tr

~

t13

~ID

t"--

o0

0",

MRI IN SPINAL DERMOIDS AND EPIDERMOIDS

407

defining the entire extent of the lesion. In the other two cases, T l-weighted images provided marginally superior lesion conspicuity than PD-weighted images which in turn were better than T2-weighted images. One dermoid was better appreciated on TI- and PD-weighted images than on T2-weighted images while two lesions were equally well shown on all the three sequences. Based on M R findings, the lesions were either intradural extramedullary (n = 3) or intramedullary (n = 2) or had both intra- as well as extramedullary components (n = 4). One of these lesions had an extraspinal extension into the soft tissues o f the back (Fig. 1). These locations were confirmed at surgery. One lesion (case 4), in addition, showed a small extradural component at surgery which was not appreciated on imaging. Dermal sinus tracts were seen on the MR images in four patients (three epidermoids and one dermoid). Subcutaneous portions of the tracts appeared as linear areas of hypointensity on all sequences. Intraspinal extent of the tract was detected in three patients. In one patient (case 2), in whom only Tl-weighted images were available, it was isointense to the cord. In another case, it was of cord intensity on T1- and PD-weighted images but showed a mixed intensity on T2-weighted images (Fig. 5). In case 4, it was not visible on Tl-weighted images, but was seen as a hypointense tract extending into the lesion on PD- and T2-weighted images (Fig. 3). DISCUSSION

Fig. 1 - Dermoid. T l-weighted sagittal image demonstrating a homogenous intermediate signal intensity mass involving and lying posterior to the cord. Note the extraspinal extension of the lesion.

extramedullary epidermoid at L3 vertebral level. Retrospective evaluation of the images also failed to reveal any signal intensity abnormality or mass effect on the nerve roots at the L3 level. One lesion exhibited a markedly hyperintense (short T1) signal (Fig. 4a). On the T2weighted images, the epidermoids were iso- to hyperintense in comparison with CSF. The mixed intensity lesion (Fig. 3c) showed a reversal of signal intensities; the area which was hyperintense on T 1-weighted images showed a decrease in signal intensity (suggestive of fat) while the hypointense area showed an increased signal intensity (suggesting a fluid-filled cyst). On the PD-weighted image, the lesions were either isointense to cord (n = 3) or markedly hyperintense ( n = l ) . Intratumoral heterogeneity of signal intensity seen as ill-defined hypo- or hyperintense areas was present in all the lesions and was demonstrated in all the three sequences. Of the four dermoids, three lesions presented a mixed signal intensity, the dominant component appearing hyperintense on Tl-weighted images with progressive decrease in signal intensity on PD- and T2-weighted images (Fig. 6). One case, in whom only Tl-weighted images were available, showed a homogenously hypointense lesion, the signal intensity being intermediate between that of CSF and cord (Fig. 1). Of the four epidermoids in which all three sequences were available, two were equally well seen on T 1- and PDweighted images, the T2-weighted images not clearly

Dermoids and epidermoids are congenital tumours arising from the embryonic rests sequestrated during the closure of the neural tube between the 4th week of intrauterine life; epidermoids arise from ectoderm while dermoids arise from ectoderm and mesoderm [1-4]. Histologically epidermoids have a collagenous wall lined with stratified squamous epithelium surrounding a mixture of desquamated debris, keratin, cholesterol crystals and water [14,15]. The wall and contents o f a dermoid are similar, but in addition, have dermal appendages such as hair follicles, hair, sebaceous glands and sweat glands [14-15]. The majority of intraspinal dermoid and epidermoid tumours are intradural and extramedullary [16]. In the present study three of the six epidermoids showed an intramedullary component, though no purely intramedullary lesion was seen. Two of the four dermoids were intramedullary in location. The reported incidence o f intramedullary dermoids varies from 30% to 40% [10]. Though several authors have documented the M R imaging characteristics of intracranial dermoids and epidermoids [I 7-20], little published literature is available concerning the M R appearances of these lesions when located in the spinal canal. Intracranial epidermoids are commonly characterized by long T 1 and T2 values with a signal intensity intermediate between that of CSF and brain on Tl-weighted images and greater than that of CSF on T2-weighted images [17,18]. In the report by Barkovich et al. [8], spinal epidermoids showed long T1 and T2 values appearing isointense to CSF on both T1and T2-weighted images. In the present series, though all the lesions exhibited long T2 values, only one of the five lesions had a long T1 (isointense to CSF). Cases with short T1 and short T2 lesions which appear hyperintense on Tl-weighted images with loss o f signal on

408

CLINICAL RADIOLOGY

(/,) (a)

(c) Fig. 2 - Epidermoid. An intradural mass lesion is seen in the lumbar region involving and expanding the conus medullaris. The lesion is slightly hypointense to the cord on (a) Tl-weighted image, isointense to the cord on (b) PD image and isointense to CSF on (c) T2-weighted image. Signal intensity inhomogeneity is evident in all three images.

409

MRI IN SPINAL DERMOIDS AND EPIDERMOIDS

(b) (a)

(c) Fig. 3 Epidermoid. (a) T l-, (b) PD- and (c) T2-weighted sagittal images demonstrating a mixed signal intensity intramedullary mass in the lower dorsal cord. The mixed intensity component shows a reversal of signal intensities: the area which is hyperintensc on Tl-weighted image (small arrowhead), appears hypointense on T2-weighted image (large arrowhead), and the posterior hypointensr area (open arrowhead) becomes hyperintense (arrow). Note the chemical shift artifact (small arrow).

410

CLINICAL RADIOLOGY

(a)

(b)

(c) Fig. 4 - Epidermoid. (a) T 1-, (b) PD- and (c) T2-weighted sagittal images revealing a well-defined homogenous intradural mass in the lumbar region that is hyperintense on all three images. On T2-weighted image, the lesion is partly obscured by the surrounding increased signal intensity of the spinal fluid.

i

\I (a)

(c) Fig. 5 Epidermoid. (a) Tl-weighted sagittal image demonstrating a well-defined intradural mass, slightly hypointense to the cord in the lumbo-sacral canal with involvement of the conus. The subcutaneous (arrow) and intraspinal (arrowhead) portions of the dermal sinus tract are well shown. (b) On PD image, the mass is isointense to the cord. (c) Most of the lesion is of CSF intensity on T2-weightedimage but with heterogeneity of signal intensity in its caudal portion (open arrowhead).

(b)

T2-weighted images have also been reported [17,20]. The basis o f the diversity o f signal characteristics exhibited by epidermoids is incompletely understood. Generally, the long T1 values have been attributed to the solid nature o f cholesterol crystals [19]. ReCently, H o r o w i t z et al. [21] correlated M R images with in vitro 13-C spectroscopy o f epidermoid cyst specimens, and found T1 shortening associated with cyst with high lipid contents o f mixed triglycerides and unsaturated fatty acids, but no cholesterol. T I prolongation was present in cysts with reduced lipid content consisting o f cholesterol but no triglycerides or fatty acids. Presence oflipids can account for the short T1 and short T2 c o m p o n e n t o f the mixed intensity lesion seen in the present study; this is further substantiated by the demonstration o f the chemical shift artifact. However, the persistently hyperintense signal on all sequences exhibited by the lesion in case 6 cannot be attributed to lipid content. This M R signal, to the best o f our knowledge, has not been reported previously in intracranial or intraspinal epidermoids. The cord-like signal intensity on T 1 - W t seen in three o f the five epidermoids in the present series is also very unusual. A more extensive study is required to elucidate the cause o f these variable signal intensities seen inepidermoids.

412

CLINICAL RADIOLOGY

,f

(b) Fig. 6 - Dermoid. (a) T 1-weighted SE(500/15) and (b) STIR(1500/100/ 25) sagittal images demonstrating a heterogenous mixed intensity mass in the lumbar region. The major portion of the lesion is hypcrintense on Tl-weighted image and shows a decrease in signal intensity on STIR image consistent with lipid.

Inhomogeneity of internal texture observed in all of our patients with epidermoid tumours has also been observed in intracranial epidermoids [17,18]; it is believed to reflect the complexity of the internal matrix. It may also b e related to the presence o f desquamated keratinized epithelium within the cyst [22]. The dominant component in three of the four spinal

dermoids exhibited short T 1 and T2 values, in agreement with the usual appearances described in intracranial dermoids. It represents the fat (triglyceride) content of the lesion [17,23]. The mixed signal intensity with parts of the tumours showing T1 and T2 lengthening as seen in these three cases has also been reported [23]. These signal intensity changes may be due to the presence of other components like hair, glandular elements and other debris. It could also be due to increased water content as a result of secretions from sweat glands [24]. Absence of fat with increased water content could explain the homogenously hypointense appearance on Tl-weighted images encountered in one dermoid (ease 1). Considerable overlap was found in the signal intensity characteristics of intraspinal dermoids and epidermoids. While the presence of characteristic fat signal in part of the lesion suggests the diagnosis of dermoid tumours (cases 7, 8 and 9), this may also occasionally be observed in epidermoids (case 4). Similarly absence of fat signal does not necessarily rule out a dermoid tumour (case 1). Recent reports have shown the superiority of MR imaging in defining intramedullary epidermoid and dermoid cysts [11-13]. Barkovich et al. [8] in their study of five intraspinal tumours (3 epidermoids, 2 dermoids) observed that standard SE sequences are excellent in demonstrating the intramedullary tumours, but insensitive in the detection of extramedullary lesions. In the present series, except for one lesion (case 2), all the intraspinal tumours, whether intramedullary or extramedullary were clearly delineated on M R imaging. One reason for this discrepancy could be that unlike Barkovich's cases, none of the lesions in our series were isointense to CSF on all the three spin-echo sequences. On Tl-weighted images only one lesion showed a signal intensity similar to CSF and was missed; however, PD and T2-weighted images were not available in this case. In all other cases, the entire extent of the lesion was clearly delineated on the Tl-weighted images. P D images were either as good as (n = 5) or only slightly inferior (n = 2) to Tl-weighted images in visualizing the lesion. On T2weighted images, though all the epidermoids were isointense to CSF resulting in suboptimal visualization of the lesions, presence of intratumoral heterogeneity of signal intensity suggested the presence of an intraspinal lesion in all the patients. The dermoids, on the other hand, were well appreciated on T2-weighted images because of the T2 shortening effect of their lipid content. Dorsal dermal sinuses are epithelium-lined tracts which extend inwards from the skin surface for a variable distance [4]. Barkovich et aL [8] commented on the difficulty in adequate delineation o f the intraspinal extension o f the dermal sinus tracts with MR. In the present series, however, MR clearly depicted the entire tract (extra- as well as intraspinal portion) in the three cases with intraspinal extension of the dermal sinus. In conclusion, the study shows the wide spectrum of signal intensity changes shown by spinal dermoid and epidermoid tumours. Though all three sequences are complementary to one another in characterizing the lesion, T 1- and PD-weighted images are ideal for optimal visualization of the entire extent of the lesion. MR imaging also permits adequate visualization of any associated dermal sinus tract. However, M R cannot always differentiate between dermoid and epidermoid tumours as there is considerable overlap in their signal characteristics.

MRI IN SPINAL DERMOIDS AND EPIDERMOIDS REFERENCES l Guidetti B, Gagliardi FM. Epidermoid and dermoid cysts: clinical evaluation and late surgical results. Journal of Neurosurgery 1977;47:12-18. 2 Lombardi G, Passerini A. Congenital tumours of the spinal cord. Acta Radiologica: Diagnosis 1966;5:1047 1050. 3 Lunardi P, Missori P, Gagliardi FM, Fortuna A. Long term results of the surgical treatment of spinal dermoid and epidermoid tumours. Neurosurgery 1989;25:860 864. 4 Naidich TP, McLone DG, Harwood-Nash DC. Spinal dysraphism. In: Newton TH & Potts DG, eds. Modern neuroradiology, Vol. I. Computed tomography of the spine and spinal cord. San Anselmo, CA: Clavadel Press, 1982:299 353. 5 Kangarloo H, Gold RH, Diament M J, Boechat MI, Barret C. High resolution spinal sonography in infants. American Journal of Neuroradiology 1984;5:191-195. 6 Davis PC, Hoffman JC Jr, Ball TI, Wyly JB, Braun IF, Fry SM et al. Spinal abnormalities in pediatric patients: MR imaging findings compared with clinical, myelographic and surgical findings. Radiology 1988;166:679-685. 7 Raghavan N, Bardovich A J, Edwards MSB, Norman D. MR imaging in the tethered cord syndrome. American Journal of Neuroradiology 1989;10:27-36. 8 Barkovich JA, Edwards MSB, Cogen PH. MR evaluation of spinal dermal sinus tracts in children. American Journal of Roentgenology 1991;156:791-797. 9 Benzil DL, Epstein MH, Knuckey NW. Intramedullary epidermoid associated with an intramedullary spinal abscess secondary to a dermal sinus. Neurosurgery 1992;30:118 121. 10 Graham DV, Tampieri D, Villemure J-G. Intramedullary dermoid tumour diagnosed with the assistance of magnetic resonance imaging. Neurosurgery 1988;23:765-767. 11 Hatfield MK, Udesky RH, Strimling AM, Kim BH, Silbergleit R. MR imaging of a spinal epidermoid tumour. American Journal of Neuroradiology, 1989;10:95-96. 12 Penisson-Besnier I, Guy G, Gandon Y. Intramedullary epidermoid

413

cyst evaluated by computed tomographic scan and magnetic resonance imaging: case report. Neurosurgery 1989;25:955-959. 13 Roux A, Mercier C, Larbrisseau A, Dube L-J, Dupuis C, Carpio RD. Intramedullary epidermoid cysts of the spinal cord: case report. Journal of Neurosurgery 1992;76:528-533. 14 Burger PC, Fogel FS. Surgical pathology of the central nervous system andits coverings. New York: John Wiley and Sons, 1982:117 123. 15 Russell DS, Rubenstein LJ. Pathology of tumours o f the nervous system, 5th ed. Baltimore: Williams and Wilkins, 1989:652 659. 16 Barkovich AJ, Naidich TP. Congenital anomalies of the spine. In: Barkovich A J, ed. Contemporary neuroimaging. Vol. 1, Paediatrics neuroimaging. New York: Raven Press, 1990:236-238. 17 Newton DR, Larson TC III, Dillon WP, Newton TH. Magnetic resonance characteristics of cranial epidermoid and teratomatous tumours. American Journal of Neuroradiology 1987;8:945. 18 Steffey DJ, Fillipp GJ, De Spera T, Gabrielsen TO. MR imaging of primary epidermoid tumours. Journal of Computer Assisted Tomography 1988;12:438-440. 19 Tampieri D, Melanson D, Ethier R. MR imaging of epidermoid cysts. American Journal o f Neuroradiology 1989;10:351-356. 20 Vion-Dury J, Vincentelli F, Jiddane M, Van Bunnen Y, Rumeau C, Grisoli F et al. MR imaging of epidermoid cysts. Neuroradiology 1987;29:333-338. 21 Horowitz BL, Chari MV, James R, Bryan RN. MR of intracranial epidermoid tumours. Correlation of in vivo imaging with in vitro 13C spectroscopy. American Journal of Neuroradiology 1990; 11:299-302. 22 Gupta RK, Jena A, Kumar S. Iophendylate or spillage from epidermoid a diagnostic dilemma on cranial MR imaging. Magnetic Resonance Imaging 1989;7:293 295. 23 Hahn FJ, Ong E, McComb RD, Mawk JR, Leibrock LG. MR imaging of ruptured intracranial dermoid. Journal of Computer Assisted Tomography 1986; 10:888-892. 24 Davidson HD, Ouchi T, Steiner RE. NMR imaging of congenital intracranial germinal layer neoplasms. Neuroradiology 1985;27:301303.