OBJECTIVE. The purpose of our study was to compare the imaging characteristics of atypical teratoid–rhabdoid tumor with medulloblastoma and seek distinguishing features that can aid in preoperative diagnosis.
MATERIALS AND METHODS. Preoperative MRI examinations of 55 patients (36 medulloblastomas and 19 atypical teratoid–rhabdoid tumors) were analyzed retrospectively. Imaging characteristics of atypical teratoid–rhabdoid tumor and medulloblastoma were assessed with conventional MRI and CT. Diffusion-weighted imaging (DWI) was available in 27 patients (19 medulloblastomas and eight atypical teratoid–rhabdoid tumors). Apparent diffusion coefficient (ADC) values were calculated for 14 medulloblastomas and six atypical teratoid–rhabdoid tumors.
RESULTS. Both atypical teratoid–rhabdoid tumors in general and infratentorial atypical teratoid–rhabdoid tumors presented at a younger age than medulloblastomas. Eleven of 19 atypical teratoid–rhabdoid tumors were infratentorial. Cerebellopontine angle (CPA) involvement was more frequent (8/11, 72.7%) in atypical teratoid–rhabdoid tumor than in medulloblastoma (4/36, 11.1%) (p < 0.001). Intratumoral hemorrhage was more common in atypical teratoid–rhabdoid tumor (9/19, 47.4%) than in medulloblastoma (2/36, 5.6%) (p < 0.0001). All atypical teratoid–rhabdoid tumors and all medulloblastomas for which DWI was available displayed increased signal intensity on DWI compared with normal brain parenchyma. The mean ADC values for tumor types were not significantly different.
CONCLUSION. Atypical teratoid–rhabdoid tumor presents at a younger age than medulloblastoma. Although atypical teratoid–rhabdoid tumor and medulloblastoma display similar imaging characteristics on conventional MRI, CPA involvement and intratumoral hemorrhage are more common in atypical teratoid–rhabdoid tumor. If a pediatric posterior fossa mass that displays restricted diffusion is involving the CPA, atypical teratoid–rhabdoid tumor is a more likely consideration than medulloblastoma.
Medulloblastoma and atypical teratoid–rhabdoid tumor are childhood CNS neoplasms that display similar characteristics on routine histologic analysis and on neuroimaging [1, 2]. On routine histopathologic examination, misdiagnosis of some atypical teratoid–rhabdoid tumors as medulloblastoma underscores this similarity. Distinction is made by immunohistochemical methods, including immunohistochemistry for BAF47 antibody against the hSNF5/INI1 protein .
The imaging characteristics of medulloblastoma are well known to radiologists and clinicians. Medulloblastoma is generally hyperdense on unenhanced CT examinations and less hypointense on T2-weighted MRI than many of the brain tumors, features that are attributed to the increased nuclear–cytoplasmic ratio of the tumor cells. Medulloblastoma is usually a midline tumor, arising in the region of the fourth ventricle. Spinal drop metastasis and leptomeningeal spread are characteristic of medulloblastoma [4, 5]. Despite its similar histologic appearance to medulloblastoma, atypical teratoid–rhabdoid tumor presents earlier, with a remarkably more aggressive course . The original case descriptions of atypical teratoid–rhabdoid tumor emphasized its distinguishing pathologic features and its dismal clinical outcome. Imaging characteristics of atypical teratoid–rhabdoid tumor have been described in multiple reports [6–10]; the largest two imaging series included 11 patients each [11, 12].
To our knowledge, a study comparing imaging characteristics of medulloblastoma and atypical teratoid–rhabdoid tumor has not yet been published. In this study, we retrospectively reviewed the preoperative conventional MRI examinations in 36 patients with medulloblastoma and 19 patients with atypical teratoid–rhabdoid tumor. If performed, diffusion-weighted imaging (DWI) characteristics of the tumors were analyzed. Apparent diffusion coefficient (ADC) values were measured in 14 medulloblastoma and six atypical teratoid–rhabdoid tumor patients. We sought to identify parameters that may be useful in distinguishing atypical teratoid–rhabdoid tumor from medulloblastoma preoperatively.
Materials and Methods
This study was reviewed for human subject protection and confidentiality and was approved under the expedited review process by our institutional review board in accordance with standards of the National Institutes of Health. Consecutive patients with a diagnosis of either an atypical teratoid–rhabdoid tumor or a medulloblastoma from January 1985 until July 2007 were eligible. Pathology slides from resected tumor specimens were reviewed by one of the authors to confirm the diagnosis of atypical teratoid–rhabdoid tumor or medulloblastoma according to the established criteria in the World Health Organization classification of brain tumors [13, 14].
MRI—All patients underwent preoperative MRI performed using 1.5-T units. The MRI protocols differed slightly, but all included T2- and T1-weighted images in different imaging planes, with and without IV gadopentetate dimeglumine. Screening of the spine was performed with gadolinium-enhanced sagittal T1-weighted sequences at the time of brain imaging. Two pediatric neuroradiologists evaluated the examinations. ADC value measurements were made by two of the authors. On MRI, tumor location, signal and enhancement characteristics, presence of hydrocephalus, and intratumoral hemorrhage (described as T1 shortening on unenhanced images) were reviewed.
Lesions were described as solid if no cystic component was seen. If the solid component constituted approximately more than half of the tumor, it was classified as predominantly solid. If approximately equal amounts of solid and cystic components were present, the lesion was classified as mixed. When the solid component was less than half of the tumor bulk, the lesion was classified as predominantly cystic.
On the T2-weighted images, the dominant signal intensity of the solid component of the tumor was assessed in comparison with the cortical or deep gray matter as isointense or hyperintense. For intensity comparison, visual inspection is used generally; when in doubt, intensity measurements were made at the PACS workstation for studies performed after 1999. The presence of extreme T2 shortening was noted; the nodular T2 hypointense foci were presumed to represent hemosiderin deposition or calcification and differed from linear signal voids that were representative of vascularity.
The most intensely enhancing components of the tumors were identified and ranked as follows: no enhancement, mild enhancement, moderate enhancement, or intense enhancement (intense enhancement described as isointensity with choroid plexus signal on gadolinium-enhanced images). The presence of noncontiguous intraparenchymal involvement and intracranial leptomeningeal spread was noted. Spinal drop metastases were recorded.
DWI and ADC analysis—DWI was performed in 19 medulloblastoma and eight atypical teratoid–rhabdoid tumor patients with b values of 0 and 1,000. The raw data for calculation of ADC values were available in 14 medulloblastoma and six atypical teratoid–rhabdoid tumor patients. All atypical teratoid–rhabdoid tumors with available DWI and ADC measurements were infratentorial. The solid components of the tumors were assessed on DWI images. ADC measurements were not made in all patients who had DWI because some data were corrupt, disallowing reliable ADC measurements in five medulloblastomas and two atypical teratoid–rhabdoid tumors. ADC maps were generated with a monoexponential fit on a voxel-to-voxel basis for all imaging planes using nordicICE software (NordicImagingLab). Circular regions of interest, 15 pixels in diameter, were placed on the solid portions of the tumors to note the tumoral ADC value. If the tumor was present in more than one image, the ADC values were measured on each image and were averaged. Particular attention was paid to avoid regions of T1 shortening and extreme T2 shortening. Normal ADC values were obtained from the supratentorial deep gray matter structures.
CT—Preoperative unenhanced CT was available in 10 medulloblastoma and seven atypical teratoid–rhabdoid tumor patients. On CT, tumor density was compared with adjacent brain parenchyma. Presence of hemorrhage or calcification was noted.
The clinical data were expressed as mean ± SD where appropriate. Clinical parameters were compared using chi-square analysis. Overall survival was analyzed using the Kaplan-Meier method, and comparison of study groups was performed using the log-rank test. The statistical packages SPSS version 10.0 and Advanced Statistics version 7.5 (SPSS) for Microsoft Windows were used to conduct statistical analysis. A p value < 0.05 was considered statistically significant, and all tests were two-tailed.
The mean age at diagnosis of the 19 patients with atypical teratoid–rhabdoid tumor was 1.32 ± 1.26 years (age range, 0.01–5.26 years). There were nine males (47.4%) and 10 females (52.6%) with atypical teratoid–rhabdoid tumor. The mean age at diagnosis of the 36 patients with medulloblastoma was 6.52 ± 4.46 years (age range, 0.21–14.33 years). There were 27 males (75%) and nine females (25%) with medulloblastoma. Patient data, including tumor locations comparing the two tumor types, are described in Table 1.
TABLE 1: Patient Data
Atypical Teratoid—Rhabdoid Tumor (n = 19)
Medulloblastoma (n = 36)
Diagnosis age (y) (mean ± SD) (range)
1.32 ± 1.26 (0.01-5.26)
6.52 ± 4.46 (0.21-14.33)
Predominantly solid and solid tumor
Kaplan-Meier overall survival (%)
Note—Except where indicated otherwise, data are number with percentage in parentheses.
The 55 patients with atypical teratoid–rhabdoid tumor and medulloblastoma had been followed for a mean of 2.35 ± 2.49 years (range, 0.02–8.83 years) after biopsy confirmation of tumor. At the time of this review, 14 patients with atypical teratoid–rhabdoid tumor and nine patients with medulloblastoma had died secondary to tumor progression. Kaplan-Meier overall survival rates for atypical teratoid–rhabdoid tumor (26.3%) and medulloblastoma (75.0%) were significantly different (p < 0.0001, Fig. 1). Survival at 1 year was 27.9% and 86.9% for atypical teratoid–rhabdoid tumor and medulloblastoma, respectively.
At presentation, 11 of 19 atypical teratoid–rhabdoid tumors (57.8%) were infratentorial. Five atypical teratoid–rhabdoid tumor patients had supratentorial tumor only. Three atypical teratoid–rhabdoid tumor patients had both supratentorial and infratentorial tumor. By definition, all medulloblastomas were infratentorial (Fig. 2A, 2B). Position relative to the tentorium was found to be a significant prognostic factor in overall survival for atypical teratoid–rhabdoid tumor. Infratentorial location was associated with a better prognostic factor for atypical teratoid–rhabdoid tumor (log-rank test, p = 0.04). Table 2 contrasts characteristics of infratentorial atypical teratoid–rhabdoid tumors and medulloblastomas.
Note—Except where indicated otherwise, data are number with percentage in parentheses.
Whereas eight of the 11 (72.7%) patients with infratentorial atypical teratoid–rhabdoid tumor had tumor at the cerebellopontine angle (CPA) (Fig. 3A, 3B), extension of tumor to the CPA in medulloblastoma was seen in four of the 36 (11.1%) patients (chi-square test, p < 0.001). There was no difference in intracranial leptomeningeal tumor dissemination (10.5% in atypical teratoid–rhabdoid tumor, 19.4% in medulloblastoma; p = 0.40) and spinal drop metastases (26.7% in atypical teratoid–rhabdoid tumor, 19.4% in medulloblastoma; p = 0.57).
Hydrocephalus was seen in 32 (88.9%) patients with medulloblastoma and 13 patients (68.4%) with atypical teratoid–rhabdoid tumor (chi-square, p = 0.06). Of the 11 infratentorial atypical teratoid–rhabdoid tumor patients, seven (63.6%) had hydrocephalus at presentation and, compared with medulloblastomas, hydrocephalus was less common in infratentorial atypical teratoid–rhabdoid tumors (chi-square, p = 0.05).
Intratumoral hemorrhage, described as T1 shortening, was significantly more common in atypical teratoid–rhabdoid tumor (9/19, 47.4%) than it was in medulloblastoma (2/36, 5.6%) (chi-square, p < 0.0001). Intratumoral hemorrhage was more common in infratentorial atypical teratoid–rhabdoid tumors (5/11, 45.5%) than in medulloblastomas (2/36, 5.6%) (chi square, p < 0.001).
More medulloblastomas were solid or predominantly solid (33/36, 91.7%) than atypical teratoid–rhabdoid tumors (12/19, 63.2%, Fig. 4A, 4B) (chi-square, p = 0.02). There was no difference in the intensity of the solid components between the tumor types. Of the 19 atypical teratoid–rhabdoid tumors, six (31.6%) had solid components that were isointense to gray matter on T2-weighted images, whereas 13 of the 36 (36.1%) medulloblastomas had similar solid components (chi-square, p = 0.74). The presence of extremely T2-weighted hypointense nonlinear (i.e., nonvascular) foci was not helpful in distinguishing the tumor types. Seven of 19 (46.8%) patients with atypical teratoid–rhabdoid tumor and eight of 36 (22.2%) patients with medulloblastoma had extremely T2-weighted hypointense foci (chi-square, p = 0.25).
DWI and ADC Values
All 27 tumors (atypical teratoid–rhabdoid tumors and medulloblastomas) for which DWI was available displayed restricted diffusion compared with the normal brain parenchyma. The mean ADC value for six atypical teratoid–rhabdoid tumors was 0.55 ± 0.06 × 10–3 mm2/s (range, 0.45–0.60 × 10–3 mm2/s). The mean ADC value for 14 medulloblastomas was 0.47 ± 0.16 × 10–3 mm2/s (range, 0.27–0.83 × 10–3 mm2/s). In normal brain parenchyma, the mean ADC values were 0.86 ± 0.06 × 10–3 mm2/s (range, 0.79–0.96 × 10–3 mm2/s) for atypical teratoid–rhabdoid tumor and 0.81 ± 0.07 × 10–3 mm2/s (range, 0.69–0.96 × 10–3 mm2/s) for medulloblastoma patients. The tumor ADC values were not statistically different for the two tumor types (Wilcoxon's rank sum test, p = 0.09).
Ten patients with medulloblastoma had preoperative CT available. The tumor was hyperdense to adjacent parenchyma in all patients. Hyperdense foci were present in two tumors; in one there were foci of T1 shortening on MRI corresponding to hemorrhage. Of the seven atypical teratoid–rhabdoid tumor patients who underwent preoperative CT, the tumor was hyperdense in six. In one patient, the tumor included cystic hypodense foci, but the solid component was hyperdense. Two atypical teratoid–rhabdoid tumors had punctate hyperdense foci without corresponding T1 and T2 shortening on MRI.
After the first publication describing atypical teratoid–rhabdoid tumor of the CNS as a new pathologic entity in 1987 , the radiologic descriptions of atypical teratoid–rhabdoid tumor appeared in the literature in the early 1990s [16, 17]. Although the tumor was originally described as primary rhabdoid tumor of brain, atypical teratoid–rhabdoid tumor has become the universally accepted name.
Medulloblastoma and atypical teratoid–rhabdoid tumor constitute the majority of hypercellular primary CNS neoplasms of childhood. Differentiation of atypical teratoid–rhabdoid tumor from medulloblastoma has important clinical and prognostic implications because atypical teratoid–rhabdoid tumor has a more malignant biologic behavior and is less sensitive to therapy [1, 18].
The imaging characteristics and pathologic appearances of atypical teratoid–rhabdoid tumor and medulloblastoma are similar to each other. In fact, on routine histopathologic examination, atypical teratoid–rhabdoid tumor may be misdiagnosed as medulloblastoma because areas with atypical teratoid–rhabdoid tumor features may be obscured by an extensive medulloblastoma component [19, 20]. For example, in their review of 127 cases that were originally diagnosed as medulloblastoma, Ho et al.  found that six of the tumors that were initially diagnosed as medulloblastoma were actually atypical teratoid–rhabdoid tumor. Distinction between medulloblastoma and atypical teratoid–rhabdoid tumor is made by the presence of nests or sheets of rhabdoid cells in the latter, which are similar to those cells seen in the rapidly fatal rhabdoid tumor of the kidney .
If the rhabdoid cells are few, immunohistochemical and molecular techniques are required to distinguish atypical teratoid–rhabdoid tumor from medulloblastoma . Moreover, deletion or mutation of the hSNF5/INI1 gene within the atypical teratoid–rhabdoid tumor has been associated in nearly 85% of patients . Immunohistochemistry using the BAF-47 antibody directed against the protein product of the hSNF5/INI1 gene appears to be even more sensitive and is very specific in discriminating atypical teratoid–rhabdoid tumor from medulloblastoma. Judkins et al.  found absence of BAF-47 staining in 20 of 20 atypical teratoid–rhabdoid tumors whereas they reported BAF-47 staining in 33 other CNS tumors including medulloblastoma and primitive neuroectodermal tumor.
Imaging characteristics of atypical teratoid–rhabdoid tumor were reported in numerous manuscripts [6–12]. In these reports, heterogeneity of the lesions, intratumoral calcification and hemorrhage, drop metastases, and leptomeningeal spread were described.
The mean age (1.32 years) of our atypical teratoid–rhabdoid tumor patients is lower than the mean ages (2.42–5.50 years) reported in large series [2, 11, 12]. The mean age at diagnosis for medulloblastoma in our study is similar to that published in other large series [21, 22]. In our series, the mean age of presentation of atypical teratoid–rhabdoid tumor was lower than the mean age of presentation of medulloblastoma: 1.32 years versus 6.52 years. Consistent with other reports, in our series, atypical teratoid–rhabdoid tumor patients had a worse prognosis compared with medulloblastoma patients.
Imaging characteristics of atypical teratoid–rhabdoid tumor and medulloblastoma are similar, but there are some differences. In our study, 11 of 19 atypical teratoid–rhabdoid tumors were infratentorial, whereas, by definition, all medulloblastomas are infratentorial. When only infratentorial atypical teratoid–rhabdoid tumors are taken into consideration, CPA involvement was significantly more common in atypical teratoid–rhabdoid tumor than in medulloblastoma (p < 0.001). In our study group, 11.1% CPA involvement in medulloblastoma was comparable to the frequency of 15% given in 13 patients . In a report of 11 atypical teratoid–rhabdoid tumors , the CPA involvement was much less common than in our series: Only one of the seven infratentorial atypical teratoid–rhabdoid tumors had CPA involvement. In the other large series of atypical teratoid–rhabdoid tumor, seven of the 11 tumors were infratentorial ; in this report no specific reference to the CPA involvement was made. In our series 72.7% of infratentorial tumors had CPA involvement. Intratumoral hemorrhage, described as T1 shortening, was more common in atypical teratoid–rhabdoid tumor than medulloblastoma when all atypical teratoid–rhabdoid tumors are assessed (p < 0.001) and when infratentorial atypical teratoid–rhabdoid tumors are analyzed separately (p = 0.004).
Medulloblastoma was more commonly a solid or predominantly solid tumor compared with all atypical teratoid–rhabdoid tumors (91.7% vs 63.2%; p = 0.02, chi-square); however, there was no statistically significant difference in the solid–cystic appearance of infratentorial atypical teratoid–rhabdoid tumors and medulloblastomas. No statistically significant difference was seen in the enhancement characteristics, T2 signal intensity of the solid components, presence of extremely T2 hypointense foci, and presence of leptomeningeal spread–spinal drop metastasis between medulloblastoma and atypical teratoid–rhabdoid tumor.
DWI is helpful in distinguishing common pediatric cerebellar tumors on the basis of these tumors' different ADC values. Low tumor ADC values compared with normal brain parenchyma have been linked to hypercellularity of the tumors. There is relative restriction to the random movement of the water molecules in the small cells of medulloblastoma. Medulloblastomas are uniformly hyperintense on DWI, with lower ADC values than those of ependymomas and juvenile pilocytic astrocytomas . All eight medulloblastomas in the series of Rumbolt et al.  showed increased signal on DWI, and they had a mean ADC value of 0.66 ± 0.15 × 10–3 mm2/s (range, 0.48–0.93 × 10–3 mm2/s). Chawla et al.  reported six medulloblastomas that were hyperintense on DWI with ADC values ranging from 0.53 to 0.64 × 10–3 mm2/s. In our series, the mean ADC value for 14 medulloblastomas was 0.47 ± 0.16 × 10–3 mm2/s (range, 0.27–0.83 × 10–3 mm2/s). This figure is similar to but slightly lower than published figures in the literature [24, 25].
In their article on the utility of the ADC values to distinguish pediatric cerebellar tumors, Rumbolt et al.  provided ADC measurements for two infratentorial atypical teratoid–rhabdoid tumors that were 0.55 and 0.60 × 10–3 mm2/s. In our study, DWI was available in eight atypical teratoid–rhabdoid tumors, and we were able to obtain ADC values in six of them (mean, 0.55 ± 0.06 × 10–3 mm2/s; range, 0.45–0.60 × 10–3 mm2/s). All atypical teratoid–rhabdoid tumors were hyperintense on DWI and their ADC values were not significantly different from those of the 14 medulloblastomas for which ADC measurements were available. Therefore, for a pediatric posterior fossa neoplasm that is hyperintense on DWI, atypical teratoid–rhabdoid tumor has to be a consideration in addition to medulloblastoma.
To our knowledge, this is the first report that compares the radiologic features of atypical teratoid–rhabdoid tumor and medulloblastoma. Atypical teratoid–rhabdoid tumor presents at a younger age than medulloblastoma. General imaging characteristics of atypical teratoid–rhabdoid tumor and med ullo blastoma reflect the histopathologic similarities of these two tumors. Both tumors are hyperintense on DWI, and ADC values do not help distinguish these tumors from each other. For pediatric posterior fossa tumors that show restricted diffusion, considerations are medulloblastoma and atypical teratoid–rhabdoid tumor. In the posterior fossa tumors of children, lower ADC values are found in medulloblastoma and atypical teratoid–rhabdoid tumor compared with the ADC values of juvenile pilocytic astrocytomas and ependymomas. If a posterior fossa tumor showing restricted diffusion is not in the midline and if there is involvement of the CPA, it is more likely to be atypical teratoid–rhabdoid tumor than medulloblastoma.
Ho DM, Hsu C, Wong T, Ting LT, Chiang H. Atypical teratoid/rhabdoid tumor of the central nervous system: a comparison with primitive neuroectodermal tumor/medulloblastoma. Acta Neuropathol 2000; 99:482 –488
Giangaspero F, Bigner SH, Kleihues P, Pietsch T, Trojanowski JQ. Medulloblastoma. In: Kleihues P, Cavenee WK, eds. Pathology and genetics of tumors of the nervous system: World Health Organization classification of tumors. Lyon, France: International Agency for Research on Cancer Press, 2000:129 –137
Rorke LB, Biegel JA. Atypical teratoid/rhabdoid tumor. In: Kleihues P, Cavenee WK, eds. Pathology and genetics of tumors of the nervous system: World Health Organization classification of tumors. Lyon, France: International Agency for Research on Cancer Press, 2000:145 –148