Eur. Radiol. (2001) 11: 131±136 Ó Springer-Verlag 2001
F. Lafitte S. Morel-Precetti N. Martin-Duverneuil A. Guermazi E. Brunet F. Heran J. Chiras
Received: 10 March 2000 Revised: 18 May 2000 Accepted: 22 May 2000
)
F. Lafitte ( ) ´ S. Morel-Precetti ´ N. Martin-Duverneuil ´ A. Guermazi ´ E. Brunet ´ J. Chiras Service de Neuroradiologie Charcot, Hôpital de La SalpØtrire, 47, Boulevard de l`Hôpital, 75013 Paris, France F. Heran Service de Radiologie, Fondation Rothschild, 25±29, rue Manin, 75019 Paris, France
NE UR OR A DI O LOG Y
Multiple glioblastomas: CT and MR features
Abstract The aim of this study was to analyze the CT and MR features of multiple glioblastomas, and to determine the best imaging modality for the initial diagnosis. The CT (four exams) and MR imaging (eight exams) of eight patients with proven multiple glioblastomas were reviewed by two neuroradiologists. The lesions were always hypo- or isodense on CT and hyperintense on T2-weighted images (100 %). They were usually hypo- or isointense on T1-weighted images (90 %). Edema and mass effect were very variable. After contrast media administration, the enhancement was mostly strong (71 % on CT and 70 % on MR), often either heterogeneous or ring-like. The different lesions of a patient often had a different pattern
Introduction Multiple gliomas are rare and poorly documented. Their incidence is variable, ranging from 0.5 to 20 % [1, 2, 3] of all glioma's diagnosed; however, in most studies, the incidence at the time of diagnosis ranges from 2.5 to 5 % [1, 4, 5, 6]. The multiple gliomas have been classified as multicentric if they arise independently in more than one site, and as multifocal if they spread from a primary focus to other cerebral parenchymal areas [1, 7, 8]. The lesions can also be classified as early (synchronous), diagnosed during the initial presentation, or late (metachronous), appearing multifocally some time after the initial diagnosis [5, 9]. The MR and CT features of synchronous multifocal glioblastomas have not been previously assessed or
on MR (75 % of cases). Meningeal or ventricular enhancement, suggestive of a possible way of dissemination, was rare. In case of multiple cerebral masses, multiple glioblastomas should be considered as a possible diagnosis in addition to the better known diagnosis of brain metastases, abscesses, or multifocal lymphomas. Moderate edema and mass effect on MR associated with strong and heterogeneous enhancement are suggestive of feature of multiple glioblastomas. Magnetic resonance allows rarely the visualization of a dissemination route. Key words Brain neoplasm ´ CT brain neoplasm ´ MR ´ Multiple glioma ´ Glioblastoma
compared in large studies. Radiological diagnosis is difficult, and often the initial diagnosis is metastatic disease. The purpose of the present study was to analyze the CT and MR imaging features of eight patients with multiple foci of glioma, and to determine which imaging modality is the most efficient in determining the initial diagnosis.
Materials and methods This study concerned the CT (four exams) and MR imaging (eight exams) of eight patients with synchronous multiple glioblastoma admitted to our institutions from October 1994 to May 1998. The patients consisted of four men and four women, with a mean age of 54 years. The major clinical manifestations included sensitivo-mo-
132
Table 1 The CT data of the different locations for each patient. M/V meningeal or ventricular. 0 absent or insignificant; + moderate + + strong Patient no.
No. of lesions
Location
Density
Edema
Mass effect
Parenchymal enhancement
M/V enhancement
2
2
Right temporal Right temporal posterior
Hypointense Isointense
++ ++
++ ++
+ + , Heterogeneous + + , Ring shaped
0 0
3
2
Ponto-mesencephalic Left frontal
Hypointense
+
+
+ + , Homogeneous
0
Isointense
0
0
+ , Homogeneous
0
4
1
Right hemisphere
Isointense
+
++
0
0
5
2
Left temporal Left parietal
Isointense Isointense
++ ++
++ ++
+ + , Heterogeneous + + , Ring shaped
0 0
tor loss (75 %), aphasia (25 %), seizures (12.5 %), and headache (12.5 %). The images were reviewed by two experienced neuroradiologists. All patients had a tissue-proven diagnosis. The histological examination diagnosed glioblastoma multiforme in all cases.
The results of clinical data and presumed diagnosis are expressed as the percentage of the number of patients. The results of imaging exams are expressed as percentage of the number of lesions.
Imaging modalities
Results
All patients were examined with CT and/or MRI. The CT was performed in the axial plane, before and after intravenous injection of 100 cc of iodinated contrast medium. The slice thickness was 5 mm in the posterior fossa and 10 mm at supratentorial level. The MRI was performed with a head coil at a field strength of 1.5 T in three patients, 0.5 T in three patients, and 1 T in two patients. T2-weighted (TR/TE: 2300/30±90 ms) conventional spinecho (SE) or T2-weighted turbo-spin-echo (TSE; TR/TE: 3000/ 60±120 ms) sequences in the axial plane, and a T1-weighted (TR/ TE: 640/11 ms) conventional SE sequence in sagittal plane, were performed for all patients (slice thickness was 6 mm). All patients underwent at least one T1-weighted SE sequence after injection of 0.1 mmol/kg of gadopentate dimeglumine, performed in the axial plane only for three patients, and in several planes for the other five patients. The field of view was 22 cm. The matrix size was 256 256 on T2-weighted sequence and 256 192 on T1-weighted sequence. Data analysis The CT and MR were analyzed separately CT images Number of locations, site of locations, and tumoral density (hypo-, iso-, or hyperdensity) were analyzed. Peritumoral edema, mass effect, and parenchymal enhancement were evaluated with a threelevel scale (0: absent or insignificant; + : moderate; + + : strong). When moderate or strong, the enhancement was evaluated as homogeneous, heterogeneous, or ring-like. Meningeal or ventricular enhancement close to the lesion was evaluated as absent (0) or present (+). The MR images were analyzed for the same items as CT. On MR, the signal intensity on T1- and T2-weighted sequences was assessed instead of tumor density. Parenchymal and meningeal enhancement on T1-weighted sequences after gadolinium injection were also evaluated with the same scale as CT.
CT imaging Results of CT are presented in Table 1. Seven different localizations in four patients were detected on CT images. Before injection, these lesions were iso (71 %) or hypodense (29 %). One patient presented a diffuse brain swelling of the right hemisphere. After injection, the parenchymal enhancement of the lesions was considered as strong (71 %), moderate (14 %), or absent (14 %). Meningeal or ventricular enhancement was never observed. Edema and mass effect were considered as strong in 57 and 71 % of the lesions, respectively. The initial presumed diagnosis with CT was metastasis in two patients (50 %). Computed tomography was considered normal in one patient. In one patient no specific diagnoses was suggested. MR imaging Results of MR imaging are presented in Table 2. Magnetic resonance allowed detection of 20 different localizations in eight patients. All but one patient had unilateral lesions. All the lesions were supratentorial, except for patient 3. On T1-weighted SE images, ten lesions were hypointense (50 %), eight were isointense (40 %), and two hyperintense (10 %). On T2-weighted SE images all lesions were hyperintense (100 %). After injection of gadopentate dimeglumine, 14 lesions demonstrated a strong enhancement (70 %), three lesions moderate enhancement (15 %), and three lesions no enhancement (15 %). When moderate or
133
Table 2 The MR characteristics of the lesions for each patient. 0 absent or insignificant; + moderate; + + strong Case no.
No. of lesions
Topography
Signal intensity
Edema
Mass effect
Parenchymal enhancement
M/V enhancement
1
2
Right frontal
Hypointense T1, hyperintense T2 Ditto
++
+
+ + , Ring shaped
0
+
0
+ + , Ring shaped
0
2
2
Right temporal
++
++
+ + , Heterogeneous
0
++
++
+ + , Heterogeneous
0
3
2
Right temporal posterior Pons and mesencephalic Left frontal
Isointense T1, hyperintense T2 Hyperintense T1, hyperintense T2 Hypointense T1, hyper T2 Ditto
+
+
+ , Homogeneous
0
4
4
Right temporal
3
Right thalamic Right occipital Right parietal Left temporal
Right parietal
5
Left parietal
6
2
Left capsularlenticular Left parietal
7
2
Left lenticular Right frontal
8
3
Left frontal Right frontal Right temporal Right temporaloccipital
0
0
0
0
Isointense T1, hyper T2 Ditto Ditto Ditto Hypointense T1, hyper T2 Isointense T1, hyper T2 Hyperintense T1, hyperintense T2 Isointense T1, hyperintense T2 Ditto Hypointense T1, hyperintense T2 Ditto
0
++
0
0
0 0 0 +
++ ++ ++ +
0 + + , Homogeneous + + , Ring shaped + + , Heterogeneous
0 + 0 +
+
0
+ + , Heterogeneous
0
+
+
+ + , Ring shaped
0
+
+
+ + , Ring shaped
0
+ 0
+ 0
+ + , Ring shaped + + , Heterogeneous
0 0
++
0
+ , Heterogeneous
0
Hypointense T1, hyperintense T2 Ditto Ditto
++
+
+ + , Heterogeneous
0
++ ++
++ +
+ , Heterogeneous + + , Heterogeneous
0 +
strong, this enhancement was heterogeneous (45 %), ring-shaped (30 %), or homogeneous (10 %). A vascular peritumoral network was observed in one lesion (patient 3). Eleven lesions were located close to meningeal structures (55 %) and three lesions to ventricular structures (15 %). Meningeal enhancement was observed in two cases (10 %). Ventricular enhancement was observed in one case (5 %). Edema was considered as important in seven lesions (35 %), moderate in seven lesions (35 %), and absent in six lesions (30 %). Mass effect was considered as important in seven lesions (35 %), absent in seven lesions (35 %), and moderate in six lesions (30 %). The MR features of the different lesions in the same patient (signal and enhancement of the lesions) were different for six patients (75 %). The initial suggested diagnosis on MR images was metastasis in two
patients (25 %), glioma in one patient (13 %), and encephalitis in one patient (13 %). In the four other patients (50 %), several or no specific diagnoses were suggested.
Discussion Multiple gliomas are rarely observed, but increase probably in frequency. All but one of our cases were supratentorial, according to literature data [3, 7, 10]; however, in a previous study of posterior fossa glioblastoma multiforme [11], multiple foci were identified in three of nine patients. The multicentricity is most frequently found in glioblastoma multiforme but can be found in well-differentiated and mixed glioma as well [3, 12]. The distinction between multicentric and multifocal glioma is difficult. Considering multifocal gliomas, the
134
a
b
Fig. 1 a±c Patient 8. A 65-year-old man presenting with left hemiplegia. a T1-weighted spin-echo (TR/TE: 400/40 ms) MR right parasagittal slice after gadolinium injection. Huge right frontal lesion with marked and heterogeneous enhancement, and peripheral edema. A second location with strong enhancement involves the splenium of the corpus callosum. b T1-weighted spin-echo MR coronal slice after gadolinium injection. The right frontal lesion is well demonstrated, close to the meningeal structures, but without pathological meningeal enhancement. Moderate mass effect on the right frontal ventricular horn. A third location is visible in the right temporal lobe, with moderate enhancement. c T2-weighted turbo spin echo (TR/TE: 3000/120 ms) MR axial slice at the level of basal ganglia. Both right temporal and corpus callosum lesions appear strongly hyperintense. Between these two lesions, the brain parenchyma demonstrates normal signal intensity
Table 3 Comparative imaging features (CT and MR) for brain metastasis and multiple glioma Metastasis
Multiple glioma
Location
Infra- and supraten- Mostly supratentorial torial cortex/white spread by hemisphermatter junction ic pathways
Size and shape Peritumoral edema
Often comparable Important
Hemorrhage
Frequent
Unusual
Enhancement
Strong (+ + , ring shaped)
Strong (+ + , heterogeneous)
Margins (after injection)
Variable
Often indistinct
Often different Mild or weak
The presented results are based on the results of our study and on the literature (especially for the data concerning metastasis): [10, 11, 12, 15, 16, 17, 18]
appearance of multiple foci results most frequently from dissemination along cerebrospinal fluid pathway for the lesions in contact with subarachnoid or ventricular spaces [1, 8]. It could also result from direct spread across brain parenchyma, especially the corpus callo-
c sum; thus, the multiple locations often occur at the same time, are located in the same hemisphere or close to dissemination pathways, and are histologically aggressive [13]. In a recent report [12], considering early glioma with multiple foci, various patterns of spread were identified in 65 % of cases, and most of the lesions were classified as multifocal. However, in our report, if lesions close to meningeal or ventricular structures were frequently observed, meningeal enhancement suggestive of a possible way of dissemination was rare; therefore, most of the lesions were presumed to be truly multicentric. The origin of these true multicentric gliomas is unclear. Neoplastic transformation may occur in a wide area of tissue, but the different rate of tumor proliferation could explain the appearance of distinct areas of malignant lesions. Distinction between multicentric and multifocal glioma, as well as differential diagnosis with brain metastasis, are important because they could have therapeutic consequences: in cases of multifocal glioblastoma, if the subarachnoidian spaces are characterized as a dissemination route, adjuvant treatment could be associated. According to our results and to literature data [12], the majority of our lesions demonstrate spontaneous hypo- or isodensity on CT, hypo- or isointense signal on T1-weighted sequence, and hyperintense signal intensity on T2-weighted sequence (Figs. 1, 2). Mass effect and edema are absent or moderate on MR in a majority of cases (65 %). Enhancement is usually strong and rarely homogeneous. This is probably related to the fact that the lesions are usually highly malignant with areas of necrosis [14]. Spontaneous T1 hyperintense or heterogeneous signal is rarely observed and is related to hemorrhagic areas. It is important to notice that in the same patient the multiple locations can have different patterns, probably according to the different degrees of malignant transformation. Despite the lack of histologic specificity, MR
135
a
d
b
e
should be performed in place of or in addition to CT, because of the better detection of the lesions, and the possible detection of dissemination pathways. In our series CT seems to overestimate the importance of edema and mass effect, in comparison with MR. Metastatic disease and, since recently, multifocal lymphomas represent the most important differential diagnosis. Some criteria can be useful in distinguishing primary brain tumors from metastases (Table 3) [10, 11, 12, 15, 16, 17, 18]. It is remarkable that metastases are usually surrounded by important peripheral edema, and
c Fig. 2 a±e Patient 4. A 47-yearold man with clinical presentation of meningoencephalitis. a Computed tomography in axial plane after contrast infusion: swelling of the right cerebral hemisphere, with decreased conspicuity of right cerebral sulci and lateral ventricle. b T1weighted spin echo (TR/TE: 640/11 ms) axial slice demonstrates several areas of hypointense signal in the right hemisphere (arrows). c After gadolinium injection, the right occipital lesion enhances. d The parietal location demonstrates ring-like enhancement. e On T2-weighted spin-echo (TR/ TE: 2300/90ms) axial slice, these lesions show hyperintense signal, with indistinct margins. Diffuse brain swelling of the right cerebral hemisphere
occur preferentially at the junction between cortex and white matter. Multifocal lymphomas demonstrate usually moderate spontaneous hyperdensity on CT. On CT and MR they usually enhance strongly after contrast media injection. Peritumoral edema is mostly weak or absent. When several enhancing lesions with mass effect are identified, an infectious cause should also be discussed (bacterial, fungal, or parasital abscess), especially in case of fever or immunodepression (toxoplasmosis). The abscesses are frequently surrounded by a peripher-
136
al ring which appears hypointense on T2-weighted sequences. Diffusion-weighted imaging can usually help in discriminating tumoral necrosis (hyposignal) from an abscess (hypersignal). Stereotaxic biopsy is rarely needed [13]. Sometimes, neuroimaging abnormalities are atypical for brain neoplasm, and can be mistaken for white matter diseases. In one of our cases, diffuse hemispheric infiltration on CT and MR with atypical enhancement suggested the diagnosis of encephalitis (Fig. 2). Rogers et al. [19] report eight cases of diffuse cerebral bilateral astrocytoma mimicking leukoencephalopathy, vasogenic edema, or multiple sclerosis. These features are also observed in gliomatosis cerebri, which is a progressive infiltration of the central nervous system by neoplastic glial elements [20]. Although a highly dense tumoral cell infiltration is observed, a well-defined mass does not usually form, and radiological diagnosis is often difficult. The distinction between gliomatosis cerebri and multiple glioma is not clear,
since foci of true malignant glioma can occur in gliomatosis [21]. Magnetic resonance imaging is superior to CT in defining the extent of the lesions, but the features are aspecific and diagnosis is usually made by brain biopsy or autopsy [22]. In conclusion, multiple gliomas are rare and usually radiologically reported as metastasis or sometimes as lymphomas or white matter diseases. If moderate edema and/or mass effect are observed around the lesions, despite an important enhancement, diagnosis of multiple glioma should be discussed and cerebral biopsy widely performed. Most often no radiological evidence of a dissemination route is observed. Magnetic resonance imaging often allows detection of one or several lesions overlooked with CT. Acknowledgements The authors thank G. Podevins for the iconography.
References 1. Barzdorf V, Malamud M (1963) The problem of multricentric glioma. J Neurosurg 20: 122±136 2. Schiefer W, Hasenbein B, Schmidt H (1978) Multricentric glioblastoma. Methods of diagnosis and treatment. Acta Neurochir (Wien) 42: 89±95 3. Barnard RO, Geddes JF (1987) The incidence of multifocal cerebral gliomas. Cancer 60: 1519±1531 4. Russel DS, Rubinstein LS (1989) Pathology of tumours of the nervous system, 5th edn. Williams and Wilkins, Baltimore, pp 188±189 5. Kyritsis AP, Alfred Yung WK, Leeds NE, Bruner J, Gleason MJ, Levin VA (1992) Multifocal cerebral gliomas associated with secondary malignancies. Lancet 339: 1229±1230 6. Solomon A, Perret GE, McCormick WF (1969) Multicentric gliomas of the cerebral and cerebellar hemispheres. Case report. J Neurosurg 31: 87±93 7. Budka H, Podreka I, Reisner T, Zeiler K (1980) Diagnosis and pathomorphological aspects of glioma multiplicity. Neurosurg Rev 3: 233±241 8. Salvati M, Oppido PA, Artizzu S, Fiorenza F, Puzzili F, Orlando ER (1991) Multicentric gliomas. Report of seven cases. Tumori 77: 518±522
9. Pell MF, Revezs T, Thomas DGT (1991) Multicentric malignant glioma. Br J Neurosurg 5: 631±634 10. Misra BK, Steers AJW, Miller JD, Gordon A (1988) Multiple glioma presenting with hemorrhage. Surg Neurol 29: 73±76 11. Kuroiwa T, Numaguchi Y, Rothman MI, Zoarski GH, Morikawa M, Zagarto MT, Kristt DA (1995) Posterior fossa glioblastoma multiforme. Am J Neuroradiol 16: 583±589 12. Kyritsis AP, Levin VA, Alfred Yung WK, Leeds NE (1993) Imaging patterns of multifocal gliomas. Eur J Radiol 16: 163±170 13. Jomin M, Lesoin F, Lozes G, Delandsheer JM, Biondi A, Krivosic I (1983) Les gliomes à foyers multiples : à propos de 10 observations. Neurochirurgie 29: 411±416 14. Pierallini A, Bonamini M, Bozzao A, Pantano P, Stephano D di, Ferone E, Raguso M, Bosman C, Bozzao L (1997) Supratentorial diffuse astrocytic tumours: proposal of an MRI classification. Eur Radiol 7: 395±399 15. Mandybur TI (1977) Intracranial hemorrhage caused by metastasis tumours. Neurology 27: 650±655
16. Atlas SW, Grossman RI, Gomori JM, Hackney DB, Goldberg HI, Zimmermann RA, Bilaniuk LT (1987) Hemorrhagic intracranial malignant neoplasms: spin-echo MR imaging. Radiology164: 71±77 17. Egelhoff JC, Ross JR, Modic MT, Masaryk TJ, Estes M (1992) MR imaging of metastatic GI adenocarcinoma in brain. Am J Neuroradiol 13: 1221±1224 18. Leeds NE, Elkim CM, Zimmerman RD (1984) Gliomas of the brain. Semin Roentgenol 19: 27±43 19. Rogers LR, Weinstein MA, Estes ML, Cairncross JG, Strachan T (1994) Diffuse bilateral cerebral astrocytomas with atypical neuroimaging studies. J Neurosurg 81: 817±821 20. Nevin S (1938) Gliomatosis cerebri. Brain 61: 170±191 21. Romero FJ, Ortega A, Titus F, Ibarra B, Navarro C, Rovira M (1988) Gliomatosis cerebri with formation of a glioblastoma multiforme. Study and follow-up by magnetic resonance and computed tomography. J Comput Assist Tomogr 12: 253±257 22. Kandler RH, Smith CML, Broome JC, Davies-Jones GAB (1991) Gliomatosis cerebri: a clinical, radiological and pathological report of seven cases. Br J Neurosurg 5: 187±196