The following are also group members of the Austrian Brain Tumor Registry: Amedeo A. Azizi, MD (Department of Pediatrics, Medical University of Vienna); Michael Scarpatetti, MD (Institute of Pathology, Medical University of Graz); Georg Ebetsberger, MD (Department of Pediatrics, Children's and Maternity Hospital Linz); Serge Weis, MD (Department of Pathology and Neuropathology, State Neuropsychiatric Hospital Wagner-Jauregg); Neil Jones, MD (Department of Pediatrics, Paracelsus Private Medical University Salzburg); Andreas Klein-Franke, MD (Department of Pediatrics, Medical University of Innsbruck); William Sterlacci, MD (Institute of Pathology, Medical University of Innsbruck); Barbara Jauk, MD (Department of Pediatrics, State Hospital Klagenfurt); Andreas Kiefer, MD (Institute of Pathology, State Hospital Klagenfurt); Guido Mueller, MD (Department of Pediatrics, State Hospital Feldkirch); Ulrike Gruber-Moesenbacher, MD (Institute of Pathology, State Hospital Feldkirch); Angelika Reiner-Concin, MD (Institute of Pathology, Danube Hospital Vienna); and Hans Feichtinger, MD (Department of Pathology, Krankenanstalt Rudolfstiftung Vienna).
Atypical teratoid/rhabdoid tumors are highly malignant embryonal central nervous system (CNS) tumors that were defined as an entity in 1996. As compared with other malignant CNS tumors, their biological behavior is particularly aggressive, but patients may benefit from an intensified treatment. Atypical teratoid/rhabdoid tumors display a complex histomorphology, which renders them prone to misdiagnosis. They occur predominantly in young children, with an estimated prevalence of 1% to 2% among all pediatric CNS tumors. However, population-based data on the incidence of these tumors are not yet available.
A nation-wide survey of malignant high-grade CNS tumors (World Health Organization grade III/IV), diagnosed in children (aged birth to 14 years) from 1996 to 2006 was conducted by the Austrian Brain Tumor Registry. A central histopathology review was performed including the assessment of SMARCB1 (INI1) protein status.
A total of 311 newly diagnosed, malignant CNS tumors were included. Atypical teratoid/rhabdoid tumors constituted the sixth most common entity (6.1%), referring to an age-standardized incidence rate of 1.38 per 1,000,000 person-years in children. Peak incidence was found in the birth to 2 years age group, where they were as common as CNS primitive neuroectodermal tumors and medulloblastomas. A total of 47.4% of atypical teratoid/rhabdoid tumors were initially diagnosed, whereas 52.6% were retrospectively detected by the central review. The 5-year survival of atypical teratoid/rhabdoid tumor patients was 39.5%, with 66.7% in the correctly diagnosed group versus 15.0% in the not recognized group (P = .0469).
Central nervous system (CNS) tumors represent >20% of all childhood malignancies (0-14 years) in developed countries1, 2 and constitute the most common cause of cancer-related death in this age group.2
Atypical teratoid/rhabdoid tumors are rare, highly malignant, embryonal CNS tumors, which occur sporadically, or in the context of a rhabdoid tumor predisposition syndrome.3 Malignant rhabdoid tumors were originally described in the kidney4 and subsequently observed in soft tissues and the CNS.5, 6 Atypical teratoid/rhabdoid tumors were defined as an entity in 1996.7 In 2000, atypical teratoid/rhabdoid tumors were introduced to the World Health Organization (WHO) brain tumor classification8 and the International Classification of Diseases for Oncology (ICD-O, third edition).9
Histopathologically, atypical teratoid/rhabdoid tumors are characterized by rhabdoid tumor cells and variable areas of primitive neuroectodermal, epithelial, and mesenchymal differentiation.3 This complex morphology may render the differential diagnosis to other malignant CNS tumors, including CNS primitive neuroectodermal tumors (PNETs)/medulloblastomas, choroid plexus carcinomas, germ cell tumors, or malignant gliomas difficult.10, 11 However, atypical teratoid/rhabdoid tumors display a distinct genetic profile characterized by the biallelic inactivation of the SMARCB1 (hSNF5/INI1) gene at chromosomal locus 22q11.23,12, 13 which causes the loss of SMARCB1 (INI1) protein expression. Since the recent introduction of a sensitive and specific monoclonal antibody against the SMARCB1 protein (Ab No. 612110, BD Transduction Laboratories, San Jose, Calif), diagnostic means have been significantly improved.11, 14
The biological behavior of atypical teratoid/rhabdoid tumors is highly aggressive, and the prognosis is exceedingly dismal compared with other malignant brain tumors. Reported survival times have ranged from 0.5 to 11 months, with a particularly poor outcome for infants.6, 7, 10, 11, 15, 16 However, recent studies provide evidence that patients benefit from intensified multimodal therapies,17-20 and 2-year overall survival rates of up to 70% have been reported.21
Atypical teratoid/rhabdoid tumors have been predominantly observed in children, particularly in very young children younger than 3 years.6, 7, 10, 13, 15, 17, 18 Only single cases have been reported in adults.22-32 Several large, hospital-based series established an atypical teratoid/rhabdoid tumor prevalence of 1% to 2% among pediatric brain tumors,8, 33-35 but population-based data on the incidence of atypical teratoid/rhabdoid tumors in children are not yet available. However, such data provide important insights into the burden of disease and might contribute to patient care. Hence, we present the first population-based study of histologically confirmed atypical teratoid/rhabdoid tumors in the pediatric population.
MATERIALS AND METHODS
Study Design and Source of Data
A retrospective nation-wide survey on malignant, high-grade CNS tumors (WHO grade III and IV tumors3) in children younger than 15 years was conducted by the Austrian Brain Tumor Registry.36 Since its establishment in 2005, the Austrian Brain Tumor Registry has evolved as a national neuro-oncology network, with active contributions from various disciplines (neurosurgeons, pathologists, neuro-oncologists) in all Austrian neuro-oncology units. The Austrian Brain Tumor Registry cooperates closely with the population-based Austrian National Cancer Registry. For this study, all newly diagnosed cases from 1996 (after the definition of atypical teratoid/rhabdoid tumor as an entity) through 2006 were included. To encompass a maximum number of patients, the Austrian Brain Tumor Registry network was used for retrospective case ascertainment. Active case reporting included all neuro-oncology units (n = 7) that are routinely engaged in the diagnosis and management of pediatric brain tumor patients in Austria. A permanent Austrian residence was considered mandatory. Tumors at any of the following sites were included9: brain (C71.0-C71.9), meninges (C70.0-C70.9), spinal cord, cranial nerves and other parts of the CNS (C72.0-C72.9), pituitary and pineal glands (C75.1-C75.3), and olfactory tumors of the nasal cavity (C30.0). All metastatic tumors were excluded. Eligible cases were identified through local pediatricians and pathologists. In a second step, data from these patients were matched with those from the population-based Austrian National Cancer Registry. The following parameters were abstracted for each case: personal identifiers, sex, date of birth, date of surgery, histopathological diagnosis. For patients with atypical teratoid/rhabdoid tumors, information on the extent of resection, metastatic stage, treatment, and follow-up (last update May 2010) was retrieved from the local pediatricians. Personal identifiers were pseudonymized, and all data were stored in the Austrian Brain Tumor Registry database.
Central Histopathology Review
In all available cases, paraffin blocks or alternatively 10 unstained sections were retrieved from the local pathology departments. For the central histopathology review, conventional histological stainings and the systematic analysis of SMARCB1 protein expression were performed, as previously described.11 All histopathological diagnoses were classified according to the WHO 2007 diagnostic consensus criteria.3 Two neuropathologists (A.W., J.A.H.) at the Institute of Neurology, Medical University of Vienna independently reviewed the hematoxylin & eosin and SMARCB1 staining of the total series. In all SMARCB1-negative tumors, a panel of additional immunohistochemical markers typically expressed in atypical teratoid/rhabdoid tumors (cytokeratin, epithelial membrane antigen, neurofilament protein, glial fibrillary acidic protein, muscle actin, vimentin) was performed. Atypical teratoid/rhabdoid tumor was diagnosed when divergent differentiation along epithelial, mesenchymal, neuronal, or glial lines was found in addition to complete loss of SMARCB1 protein expression in tumor cell nuclei, but expression was retained in pre-existing cells (eg, endothelial cells). In addition a third neuropathologist (C.H.), experienced in the diagnosis of pediatric brain tumors reviewed all atypical teratoid/rhabdoid tumors.
All statistical data analyses were performed with SAS version 9.2 (SAS Institute, Cary, NC), SPSS version 17.0 (SPSS Inc., Chicago, Ill), and Excel version 12.1.0 (Microsoft, Redmond, Wash). All incidence rates were age-adjusted to the WHO world standard population using direct methods and expressed per 1,000,000 person-years. National population estimates per single year were provided by Statistics Austria. Special analysis was performed for 5 different age groups: 0-2 years, 3-5 years, 6-8 years, 9-11 years, and 12-14 years.
Differences in frequency of misdiagnoses before and after the introduction of atypical teratoid/rhabdoid tumors to the WHO brain tumor classification in 2000, as well as metastatic disease, extent of resection (gross total, subtotal, biopsy), and localization (supratentorial, infratentorial, spinal) within subgroups of initially recognized versus not recognized cases, were tested by Fisher exact test. Differences in age were tested by Wilcoxon-Mann-Whitney test. The distribution of annually observed numbers of atypical teratoid/rhabdoid tumors was compared with a Poisson distribution (Kolmogorov-Smirnov test). Survival probabilities were estimated according to Kaplan-Meier. Univariate survival analyses were performed for age, localization, extent of resection, initial establishment of the diagnosis (recognized vs nonrecognized atypical teratoid/rhabdoid tumors), and metastatic disease. Differences in survival curves were tested with the log-rank test. To clarify the impact of an intensified treatment regimen (MUV ATRT dose-intense, multiagent chemotherapy including doxorubicin, cyclophosphamide, vincristine, cisplatin, ifosfamide, etoposide, and methotrexate, augmented by intrathecal therapy and high-dose chemotherapy [modified Finlay protocol] and followed by local radiotherapy; EU rhab [consensus therapy of the European Rhabdoid Registry consists of intensive induction CTX, early RTX, and if possible complete surgical resection]) on survival, a Cox model was estimated with intensified treatment regimen as time-dependent binary covariate and initial establishment of the diagnosis as fixed covariate.
Ethical approval was obtained from the Ethics Committee of the Medical University of Vienna. Reporting through the established Austrian Brain Tumor Registry warranted data confidentiality.
During the period 1996 to 2006, a total of 319 primary malignant, high-grade CNS tumors (WHO grade III and IV) were reported in the Austrian pediatric population (annual estimate: 1.4 million children). Of these 319, 225 (70.5%) were microscopically verified at diagnosis, whereas 29.5% were either diagnosed clinically (malignant brain stem gliomas, n = 19), were diagnosed by specific tumor markers (germ cell tumors, n = 3), or had a basis of diagnosis that was unknown (n = 72). In 202 (89.8%) of 225 microscopically verified cases, tissue was provided for a central histopathology review. On central review, another 8 cases were excluded, because of reclassification as low-grade gliomas. Thus, in total 311 tumors were included in the analysis. On central histopathology review, 19 cases displayed morphological and immunohistochemical features of atypical teratoid/rhabdoid tumor including complete loss of SMARCB1 protein expression in tumor cell nuclei. In addition, 2 tumors with morphological features of epithelioid sarcoma showed lack of SMARCB1 protein expression. As loss of SMARCB1 expression has been described in epithelioid sarcomas,37 neither tumor was classified as an atypical teratoid/rhabdoid tumor.
The median age at diagnosis of the total cohort was 7.5 years. Of 311 tumors, 173 were diagnosed in males (55.6%), 138 (44.4%) in females (male/female ratio, 1.25). Embryonal brain tumors constituted the largest group (n = 142, 45.7%), followed by astrocytic (n = 86, 27.7%) and ependymal tumors (n = 24, 7.7%). Incidence data of the 10 most common tumor entities are provided in Table 1. Atypical teratoid/rhabdoid tumor constituted the sixth most common tumor entity; the age-standardized incidence rate was 1.38 per 1,000,000 person-years (age-specific incidence rate, 1.28 per 1,000,000 person-years). As some variation in the incidence of atypical teratoid/rhabdoid tumors was noted (range, 0-4 cases per year), the distribution of annually observed tumors was analyzed and found to be consistent with the expected distribution for a rare event of 19 of 11 = 1.73 cases per year (P = .8531).
Table 1. Incidence Data of the 10 Most Common Malignant, High-Grade CNS Tumors
Classification of children into 5 age groups showed a peak incidence of atypical teratoid/rhabdoid tumors among children younger than 3 years (Fig. 1). In this age group, atypical teratoid/rhabdoid tumors were as common as CNS PNETs or medulloblastomas. In contrast, atypical teratoid/rhabdoid tumors were rare in all other age groups.
Nine (47.4%) of 19 atypical teratoid/rhabdoid tumors were initially diagnosed, whereas 10 (52.6%) of 19 cases were retrospectively detected by the central histopathology review. Initial diagnoses of those tumors were medulloblastoma (n = 6), ependymoblastoma (n = 1), anaplastic meningioma (n = 1), small cell sarcoma (n = 1), and Ewing sarcoma (n = 1). Misdiagnoses of atypical teratoid/rhabdoid tumors tended to be more common in the period 1996 to 2000 (before the introduction of atypical teratoid/rhabdoid tumor to the WHO brain tumor classification) than after 2000 (P = .0698). In contrast, none of the initially diagnosed atypical teratoid/rhabdoid tumors was reclassified on central review. Clinical characteristics and treatment regimens of atypical teratoid/rhabdoid tumor patients are listed in Table 2. The male/female ratio was 0.9. Median age at diagnosis was 1.4 years, with a range from 0.2 to 14.4 years. At the time of diagnosis, 68.4% of the patients were younger than 3 years. The groups of initially recognized and nonrecognized cases did not differ significantly in terms of age (P = .7801), metastatic disease (P = .6381), extent of resection (P = .4560), and localization (P = .2762). Figure 2 illustrates the Kaplan-Meier survival curves of atypical teratoid/rhabdoid tumor patients, stratified by initially recognized versus not recognized atypical teratoid/rhabdoid tumors. Overall survival of nonrecognized patients was significantly worse compared with initially recognized cases (P = .0469, log-rank test). Five-year survival of all atypical teratoid/rhabdoid tumor patients was 39.5% (66.7% in the initially diagnosed group vs 15.0% in the nonrecognized group). Extent of resection was significantly associated with survival (P = .0127), whereas no significant associations could be observed for age (P = .3565), localization (P = .4928), or metastatic disease (P = .1182). Analysis of treatment strategies revealed that the important factor for prolonged survival was an intensified treatment (MUV ATRT protocol; EU rhab protocol; P = .0241, Cox model).
Table 2. Clinical Characteristics and Treatment of AT/RT Patients
At first single-center studies on atypical teratoid/rhabdoid tumors documented their occurrence predominantly in very young children younger than 3 years.6, 7, 10, 11, 13, 15, 17, 18 This was confirmed in larger single-center series, based on continuous pediatric CNS tumor collectives including all types of brain tumors. These microscopically verified series established a low prevalence of atypical teratoid/rhabdoid tumors between 0.9% and 2.1%.33-35 A similar prevalence of 1.1% was found in a multicenter, hospital-based study of pediatric brain tumors in France.38 However, definite conclusions on the population-based incidence can be provided only by cancer registries, in particular specialized brain tumor registries such as the Central Brain Tumor Registry of the United States39 or projects such as Automated Childhood Cancer Information System, a European project on the incidence and survival of pediatric tumors.1, 40, 41 However, data published by these organizations are mostly presented as summary rates for groups of tumors (eg, embryonal tumors or gliomas); exact incidence rates of rare tumor entities (including atypical teratoid/rhabdoid tumors) are not provided. With the increasing use of the ICD-O third edition of 2000, reliable incidence data of atypical teratoid/rhabdoid tumors will be available in cancer registries in several years time. Therefore, our approach was to actively report cases from all Austrian neuro-oncology units from 1996 onward, when atypical teratoid/rhabdoid tumors were defined as a distinct tumor entity. We screened for all malignant, high-grade pediatric CNS tumors entering into the differential diagnosis of atypical teratoid/rhabdoid tumor. As others and we have previously shown that atypical teratoid/rhabdoid tumors are frequently misdiagnosed,10, 11 a central histopathology review including the assessment of SMARCB1 protein status was performed, which was feasible in the majority of all tumors (70.5%). By using this strategy, we found an atypical teratoid/rhabdoid tumor incidence rate of 1.38 per 1,000,000 person-years (95% confidence interval, 0.83-2.15) in children, in accordance with the low atypical teratoid/rhabdoid tumor prevalence of single-center studies.33-35 Atypical teratoid/rhabdoid tumor is a rare CNS tumor. However, in very young children (0-2 years), we found atypical teratoid/rhabdoid tumor to be as frequent as medulloblastoma and CNS PNET (17.3%, 16.0%, and 13.3%, respectively). To ensure a population-based dataset, we also included nonmicroscopically verified cases. Thus, the actual incidence rate of atypical teratoid/rhabdoid tumor might be slightly higher, as single nonmicroscopically verified cases might have been missed. Comparison of the incidence rates of the most frequent pediatric CNS tumors across Europe, England, and the United States shows highly similar rates in Austria, thereby confirming a high degree of case ascertainment and the validity of our findings.1, 39, 42
Half of the atypical teratoid/rhabdoid tumors in our series were initially not recognized. Such misdiagnoses tended to be more common in the first years after the definition of atypical teratoid/rhabdoid tumor as an entity and before the introduction of the anti-SMARCB1 antibody as diagnostic tool. This finding emphasizes the relevance of a systematic analysis of the SMARCB1 protein status in malignant pediatric brain tumors. Furthermore, it underlines the importance of a central histopathology review in epidemiological studies of tumors with a complex histopathology.
In the literature, single malignant rhabdoid tumors with retained SMARCB1 protein expression have been reported.43 We did not detect such tumors in our series. Loss of SMARCB1 protein expression is characteristic for malignant rhabdoid tumors, including atypical teratoid/rhabdoid tumors, but not absolutely specific. Alterations of the SMARCB1 gene have recently been implicated in several other tumors, such as epithelioid sarcomas, single cases of gangliogliomas, familial schwannomas, and cribriform neuroepithelial tumors.37, 44-49 These findings indicate that the morphology needs to be cautiously interpreted in conjunction with the immunohistochemical profile for the differential diagnosis of atypical teratoid/rhabdoid tumors. Indeed, 2 tumors in our series displayed features of epithelioid sarcoma and showed loss of SMARCB1 protein expression. Whether these tumors fall in the spectrum of rhabdoid tumors, or represent a separate tumor entity with alteration of the SMARCB1 gene, needs to be clarified in larger patient cohorts.
Notably, the most frequent misdiagnosis of atypical teratoid/rhabdoid tumor was medulloblastoma. This might indicate that the overall survival of patients with embryonal brain tumors, especially in the young medulloblastoma age group, improves because nonrecognized atypical teratoid/rhabdoid tumors are no longer included.
As previously reported,6, 7, 10, 11, 15, 16 the overall survival of atypical teratoid/rhabdoid tumor patients is poor; this was also found in our series. However, we could show that patients whose tumors were initially diagnosed as atypical teratoid/rhabdoid tumors and consequently treated according to an intensified protocol showed a better outcome compared with those who were initially not recognized. This is in line with our previous findings in a single-center study,11 and emphasizes the importance of a correct diagnosis and appropriate treatment of atypical teratoid/rhabdoid tumor patients.
In conclusion, we present the first population-based study of histologically confirmed atypical teratoid/rhabdoid tumors in children, showing a high incidence in the very young (0-2 years), where they constitute together with CNS PNETs and medulloblastomas the most frequent malignant, high-grade CNS tumors. An increasing awareness among clinicians and pathologists of this high disease occurrence will help to optimize the diagnostic and therapeutic management of atypical teratoid/rhabdoid tumor patients.
We thank Irene Leisser for excellent technical assistance, Andreas Jurkowitsch for information technology support, and Leo Karger, Reinhard Motz, Franz Wuertz, Agnes Gamper, Hans Maier, Christian Urban, Gabriele Pammer, and Selma Hoenigschnabel for assistance in case ascertainment.
CONFLICT OF INTEREST DISCLOSURES
Supported by the Anniversary Fund of the Österreichische Nationalbank (grant No. 12268).