Dynamic MRI of meningiomas and schwannomas: is differential diagnosis possible?

Neuroradiology (1997) 39: 633 638 Springer-Verlag 1997 DIAGNOSTIC NEURORADIOLOGY I. Ikushima Y. Korogi J. Kuratsu T. Hirai S. Hamatake M. Takahashi Y. Ushio Dynamic MRI of meningiomas and schwannomas: is differential diagnosis possible? Received: 17 June 1996 Accepted: 11 October 1996 I.Ikushima ()) Y. Korogi T.Hirai S.Hamatake M. Takahashi Department of Radiology, Kumamoto University School of Medicine, 1-1-1 Honjo, Kumamoto 860, Japan J. Kuratsu Y. Ushio Department of Neurosurgery Kumamoto University School of Medicine 1-1-1 Honjo, Kumamoto 860, Japan Abstract We studied 23 patients with meningiomas and 14 with schwannomas using dynamic spin-echo (TR/TE 200/15 ms) MRI. Histologically the meningiomas were classified according to the 1993 WHO classification. Serial images were obtained every 30 s for 210 s after rapid injection of gadopentetate dimeglumine (0.1 mmol/kg). The contrastenhancement ratio (CER) was divided into three patterns; a sharp rise with a peak within 60 s (A), a relatively rapid increase with a peak between 60 and 210 s (B), a slow increase without a peak (C). The patterns were correlated with the histology of the tumors. The signal intensities of the tumours on T2-weighted images were also analyzed and correlated with the dynamic patterns. Meningiomas had more varied dynamic patterns than schwannomas. Almost half of the meningiomas showed pattern A, and one third pattern C. Of six meningothelial meningiomas showed pattern A; all schwannomas and fibrous meningiomas showed pattern C. Various patterns were observed in transitional meningiomas. Of the 8 meningiomas showing pattern C, only one gave high signal on T2-weighted images, and could not be differentiated from the schwannomas. Thus, one third of meningiomas could not be differentiated from schwannomas by the dynamic contrast enhancement alone. However, when this was combined with the signal intensity on T2-weighted images, most meningiomas could be differentiated from schwannomas. Key words Meningioma Schwannoma Magnetic resonance imaging Introduction With the development of rapid MRI techniques, dynamic studies have been used clinically. There are reports of dynamic MRI of sellar or parasellar lesions [1 3] and intracranial tumours [4, 5]. In a previous report on a relatively small series [6], the contrast-enhancement ratio (CER) of meningiomas showed rapid rise with gradual decline, whereas that of schwannomas showed a slow but steady increase, without a peak within 210 s. However, we have seen several meningiomas in which the pattern of dynamic contrast enhancement was similar to that of schwannomas. The purpose of this study was to reevaluate the usefulness of dynamic MRI in differentiating meningiomas and schwannomas in a larger group of patients and to correlate the dynamic pattern with signal intensities on T2-weighted images. We also analysed differences between subtypes of meningiomas. Materials and methods We studied 23 patients with meningiomas and 14 with schwannomas, using dynamic spin-echo MRI. Histological proof was obtained in all patients. The histological subtypes of meningioma were classified according to the 1993 WHO classification: the tumours were meningothelial in six cases, transitional in six, fibrous

634 a b c d Fig. 1 Dynamic MRI of left trigeminal schwannoma (arrow) a before ab30 s c 60 s d 180 s e 210 s after administration of Gd- DTPA. The tumour enhanced gradually and homogeneously, with maximum intensity at 210 s e in three, other subtypes in four and unknown in four. The sites of the meningiomas were: parasagittal in six, convexity in four, sphenoid ridge in three, parasellar in three, intraorbital in two, petrous ridge in two, tentorium in two and clivus in one. The schwannoms originated from the acoustic nerve in ten cases, the trigeminal nerve in three and the hypoglossal in one. All patients were examined on a 1.5-T superconductive imager. They underwent dynamic studies after routine T1-weighted (600/ 15/1-TR/TE ms/excitation) and T2-weighted (2300/90/1) imaging. Dynamic MRI was performed as follows: 10 s after rapid intravenous injection of gadopentate dimeglumine (0.1 mmol/kg), images were obtained every 30 s for 210 s, with a spin-echo sequence: 200/ 15/1, section thickness 5 mm, acquisition matrix 128 256 and field of view 25 cm. Actual sampling time per set of images was 26 s, and 1 5 contiguous sections were imaged simultaneously. The sequential contrast enhancement pattern of the tumours was assessed qualitatively; homogeneity of enhancement were also assessed. The dynamics of contrast enhancement of the tumours were also analysed using the CER. This was obtained for each region of interest as follows: (Sn-So) 100/Smax-So, where Sn indi-

635 a b c d Fig. 2 Dynamic MRI of meningothelial meningioma of the left sphenoid ridge (arrow); timing as in Fig. 1. The tumour enhanced rapidly and homogeneously, with maximum intensity at 30 s e cates signal intensity at each dynamic phase; So signal intensity on the image obtained before administration of contrast medium; and Smax, maximum signal intensity of all dynamic images. The regions of interest within the tumour were placed so as to exclude cystic or necrotic areas. Their size varied between patients, because we chose homogeneously enhancing areas which were as large as possible. The CER was divided into three patterns: a sharp rise with a peak within 60 s (pattern A), relatively rapid increase with a peak between 60 and 210 s (B), and slow increase without a peak (C). The patterns were correlated with the histology of the tumours. We also analysed the signal intensity of the tumours on T2-weighted images and correlated these with the dynamic patterns. Signal intensity on T2-weighted images was described as low, isointense, or high relative to cerebral cortex. Results On visual assessment of the dynamic images, all schwannomas showed gradual, steady homogeneous contrast enhancement (Fig. 1). Meningiomas showed a variable dynamic pattern, with relatively homogeneous contrast enhancement. All meningothelial meningiomas showed rapid enhancement (Fig. 2), whereas all fibrous meningiomas enhanced gradually (Fig.3). Quantitative assessment confirmed that meningiomas had a wider range of dynamic patterns. Of the 23 tumours, 11 (48%) showed pattern A, and 8 (35%) pattern C. All schwannomas showed pattern C (Fig. 4). Pattern A was been in 4 of 6 meningothelial meningiomas (Fig. 5), whereas all fibrous meningiomas showed pattern C (Fig. 6) and various patterns were observed in transitional meningiomas (Fig. 7). On T2-weighted images, 10 (43%) of 23 meningiomas gave higher signal than cerebral cortex (Fig. 8), while 9 (39%) were isointense and 4 (17%) gave lower signal. All schwannomas gave higher signal than the cortex. Of the 15 meningiomas showing pattern A or B, a gave higher signal and the remaining 6 were isointense with cerebral cortex. Of the 8 showing pattern C, four

636 a b c Fig. 3 Dynamic MRI of left parasagittal fibrous meningioma (arrow); timing as in Fig. 1. The tumour enhanced gradually and homogeneously, without maximum intensity at 210 s d e gave low signal, three were isointense, and one gave high signal. Thus, only one meningioma could not be differentiated form the schwannomas by a combined reading of dynamic pattern and the signal intensities on T2-weighted images. All fibrous meningiomas gave lower signal than cerebral cortex (Fig. 9). Discussion Should one wish to differentiate schwannomas and meningiomas, morphological characteristics, such as extension along the course of cranial nerves in the case of schwannomas and linear enhancement along the dura emanating from the dural margin for meningiomas, are often reliable [8, 9]. However, sometimes it is difficult to differentiate these tumours by morphological characteristics alone [10, 11]. There are several reports on the usefulness of dynamic MRI for differentiating schwannomas and meningiomas, which frequently originate in similar sites, such as the cerebellopontine angle and the skull base. In 12 cases of meningioma, the typical pattern of CER was reported to be a rapid rise with a gradual decline [6]. In the current, almost half of meningiomas showed this pattern. A third showed a slow, steady increase, similar to schwannomas, all of which showed a slow increase without a peak within 210 s. Therefore, if a rapid increment and gradual decline is observed, a schwannoma is unlikely. However, if a slow increment without a peak is observed, a meningioma cannot be excluded. On T2-weighted images, meningiomas showed variable signal intensity, whereas all schwannomas gave high signal. Although ten (43%) meningiomas gave high signal, there was only one which showed both pattern C on dynamic MRI and high signal on T2-weighted images. Watabe et al. [12] measured of signal enhancement and relaxation rate increments with Gd-DTPA for schwannomas and meningiomas. The average T1 increment on contrast enhanced images was almost twice as high in schwannomas as in meningiomas, mainly deriving from longer intrinsic T1 values in schwannomas. The signal-enhancement increment was higher in schwannomas than in meningiomas, but this difference was poorly appreciated on contrast-enhanced images. In the dynamic study by Fujii et al. [4], using a short- TE FLASH sequence with measurement times of 10 s per image, the signal-enhancement increment in the early phase in meningiomas was approximately four times that in schwannomas. However, in the early phase, meningiomas had a wider range of signal enhancement, partially overlapping with schwannomas; the latter had consistently low signal enhancement. These results are roughly consistent with ours. With a fast snapshot- FLASH technique which allowed measurement times of

637 Fig. 4 Sequential contrast-enhancement ratio (CER) of schwannomas shows slow increase, without a peak within 210 s Fig. 6 Sequential CER of fibrous meningiomas shows slow increase, without a peak within 210 s Fig. 5 Sequential CER of meningothelial meningiomas shows a rapid rise with gradual decline. Peak enhancement was observed within 30 150 s Fig. 7 Sequential CER of transitional meningiomas shows variable dynamic patterns 1 s per image, Nagele et al. [5] thought that early enhancement is mainly influenced by the degree of vascularity. Maeda et al. [13] used dynamic susceptibilitycontrast imaging with a gradient-echo technique to assess the vascularity of meningiomas and schwannomas: meningiomas showed a wider range, while schwannomas showed constant low vascularity. Hypovascular meningiomas probably cannot be differentiated from schwannomas even with these rapid sequences, although additional information about the vascularity of the tumour may be obtained. We believe that for differentiating meningiomas from schwannomas, SE sequences with measurement times of 30 s can provide information almost equal to that of other sequences with higher temporal resolution. The SE sequence has superior contrast and lacks susceptibility artefact. Though the turbo-se sequence has the same advantages with high spatial and temporal resolution, the SE sequence is available on relatively old imagers. We classified the histologic subtypes of meningiomas according to the 1993 WHO histological typing of CNS tumours [7]. In this typing, meningiomas are divided into 11 subtypes; MRI characteristics of various subtypes of meningiomas have been described [14, 15]. Although fibrous and transitional meningiomas tend to show low signal on T2-weighted images, it is difficult to reach a histological diagnosis on signal characteristics alone. According to the dynamic patterns reported previously, angioblastic meningiomas are markedly hypervascular; meningothelial meningiomas are moderately hypervascular, while fibrous meningiomas are hypovascular [4, 10]. In our study all meningothelial meningiomas showed a rapid increment with gradual decline, while all fibrous meningiomas showed slow increment without a peak. The latter are composed of interlacing sheets and fascicles of markedly elongated spindle-shaped cells which sometimes exhibit a prominent storiform appearance; they also have abundant extracellular reticular and collagen fibres [16]. It is supposed that the extensive extracellular matrix and the spindle-shaped cells prevent rapid contrast enhancement. Meningothelial meningiomas are composed of meningothelial cells which aggre-

638 Fig. 8 T2-weighted image of meningothelial meningioma: the left sphenoid ridge meningioma gives slightly higher signal than cerebral cortex Fig. 9 T2-weighted image of fibrous meningioma: the parasagittal meningioma gives lower signal than cerebral cortex 8 9 gate into whorls and sheets with a poor extracellular matrix [16]; lack of abundant extracellular matrix may permit immediate contrast enhancement with a gradual decline. Transitional meningiomas exhibit a mixture of both features [16]. It may be because of differing proportions of both tissues, that transitional meningiomas showed variable dynamic patterns. Genesis of the signal of meningiomas on T2-weighted-images involves four important constituents: collagen content, cellularity, ferric iron and psammomatous calcification [14]. Collagen content is subtypespecific, and fibrous and collagen-dominated transitional meningiomas usually give low signal [14]. It is supposed that meningiomas which contain abundant intracellular collagen fibres tend to show the dynamic pattern of gradual enhancement, similar to that of schwannomas, and also to give low signal on T2-weighted images, unlike schwannomas. References 1. Sakamoto Y, Takahashi M, Korogi Y, Bussaka H, Ushio Y (1991) Normal and abnormal pituitary glands: gadopentate dimeglumine-enhanced MR imaging. Radiology 178: 441 445 2. Korogi Y, Takahashi M, Sakamoto Y, Shinzato J (1991) Cavernous sinus: correlation between anatomic and dynamic gadolinium-enhanced MR imaging findings. Radiology 180: 235 237 3. Takahashi M, Sakamoto Y, Korogi Y, Seto H, Ushio Y (1992) Dynamic MR imaging for localization of the normal pituitary tissue in macroadenoma. Radiology 185: 327 P 4. Fujii K, Fujita N, Hirabaki N et al (1992) Schwannomas and meningiomas: evaluation of early enhancement with dynamic MR imaging. AJNR 13: 1215 1220 5. Nagele T, Peterson D, Klose U et al (1993) Dynamic contrast enhancement of intracranial tumors with snapshot- Flash MR imaging. AJNR 14: 89 98 6. Joo YG, Korogi Y, Hirai T et al (1995) Differential diagnosis of extra-axial tumours by dynamic spin-echo MRI. Neuroradiology 37: 522 525 7. Kleihues P, Burger PC, Scheithauer BW (1993) Histologic typing of tumours of the central nervous system. In: WHO International classification of tumours. 2nd edn. Springer, Berlin Heidelberg New York 8. Tokumaru A, O uchi T, Eguchi T et al (1990) Prominent meningeal enhancement adjacent to meningioma on Gd- DTPA enhanced MR images: histopathologic correlation. Radiology 175: 431 433 9. Goldsher D, Litt AW, Pinto RS, Bannon KR, Kricheff II (1990) Dural tail associated with meningiomas on Gd- DTPA enhanced MR images: characteristics, differential diagnostic value, and possible implications for treatment. Radiology 176: 447 450 10. Kutcher TJ, Brown DC, Maurer PK, Ghaed VN (1991) Dural tail adjacent to acoustic neuroma: MR features. J Comput Assist Tomogr 15: 669 670 11. Brady AP, Stack JP (1994) Case report: magnetic resonance demonstration of haemorrhagic acoustic neuroma. Clin Radiol 49: 61 63 12. Watabe T, Azuma T (1989) T1 and T2 measurements of meningiomas and schwannomas before and after Gd- DTPA. AJNR 10: 463 470 13. Maeda M, Itoh S, Kimura H et al (1994) Vascularity of meningiomas and schwannomas: assessment with dynamic susceptibility-contrast MR imaging. AJR 163: 181 186 14. Demaerel P, Wilms G, Lammens M et al (1991) Intracranial meningiomas: correlation between MR imaging and histology in 50 patients. J Comput Assist Tomogr 15: 45 51 15. Kaplan RD, Coons S, Drayer BP, Bird CR, Johnson PC (1992) MR characteristics of meningioma subtypes at 1.5 tesla. J Comput Assist Tomogr 16: 366 371 16. Burger PC, Scheithauer BW, Vogel FS (1991) Surgical pathology of the nervous system and Its coverings, 3rd edn Churchill Livingstone, New York, pp 193 437