Introduction

Medulloblastoma (MB) and central nervous system primitive neuroectodermal tumor (CNS-PNET) are primary grade 4 embryonal brain tumors that require multimodal therapies. Although MB and CNS-PNETs are treated with similar approaches and treatment regimens, they represent biologically different diseases. Recent studies have revealed molecular subgroups for both MB and CNS-PNET which correlate with distinct clinicopathologic features and treatment response. This chapter will summarize current clinical understanding of MB and CNS-PNET in the context of recent molecular studies and postulate on the direction of future therapeutic approaches for these diseases.

Medulloblastoma

Epidemiology

Medulloblastoma is the most common malignant brain tumor in childhood, representing up to 25 % of all pediatric brain tumors. First described by Bailey and Cushing in 1925, MB arises from the cerebellum and is classified as a World Health Organization (WHO) grade 4 tumor. MB predominantly arises in children, with peak incidence between the ages of 3–4 and 8–9. The annual incidence has been estimated at 1 in every 200,000 children under the age of 15 [1]. Consistent with the classification of an embryonal tumor, MB is rarely seen in adults; 70 % of patients present before the age of 20. There is a modest male preponderance in a ratio of 1.4:1 [2]. A small proportion of MBs (<5 %) have inherited disorders with germline mutations, these include: Gorlin’s syndrome (also called as nevoid basal cell carcinoma syndrome) associated with patched-1 (PTCH1) gene mutations, Turcot syndrome associated with mutations in the adenomatous polyposis coli (APC) gene, and Li-Fraumeni syndrome caused by TP53 mutations [1]. These familial cancer syndromes provided some of the first clues about biological pathways that underlie MB pathogenesis.

Clinical Presentation

Patients with MB frequently present with signs and symptoms related to hydrocephalus secondary to fourth ventricular obstruction, with predominant symptoms of vomiting, headache, and nausea. Due to the localizing posterior fossa mass, ataxia, dysmetria, and diplopia secondary to sixth nerve palsy are often accompanying symptoms. The most common symptoms, especially in younger children, are nonspecific and include morning headaches with vomiting, irritability, and lethargy – subtle clinical signs that can present diagnostic challenges [3].

The differential diagnosis for MB includes a range of tumors with a predilection for the cerebellum. These include pilocytic astrocytoma, ependymoma, and other rarer embryonal tumors including atypical teratoid/rhabdoid tumors (ATRT) and ETANTR/EMTR (embryonal neoplasm with abundant neuropil and true rosettes). It can be difficult to distinguish these tumor entities by clinical symptoms alone; however, patients with pilocytic astrocytoma or ependymoma tend to have a longer duration of symptoms.

Imaging Findings

The initial diagnosis of MB is usually made on non-contrast computed tomography (CT) followed by magnetic resonance imaging (MRI) with contrast which is the preferred modality, and later confirmed with histopathology [4]. The classical CT finding is a hyper-attenuating midline mass that markedly enhances with contrast medium. Foraminal extension of tumor and a predominantly cystic mass with a mural nodule would, respectively, favor ependymoma and pilocytic astrocytoma over MB (Fig. 12.1). The presence of leptomeningeal or nodular metastases would favor a diagnosis of MB or another related embryonal tumor such as ATRT. It is important to note that both CT and MRI characteristic of MB may overlap with that of other tumor types; hence, surgical histopathology is needed for definitive tumor diagnosis. Spinal MRI is part of standard preoperative investigations for posterior fossa tumors and can demonstrate evidence of nodular or leptomeningeal dissemination. Attention should be paid to kidney abnormalities when spinal MRI scan is performed as the presence of metastatic lesions at this level would favor ATRT rather than MB.

Fig. 12.1
figure 1

CT and MR imaging of medulloblastoma.(a) Axial contrast-enhanced CT showing a hyperdense mass compressing the 4th ventricle (white arrow). Note the hydrocephalus with accompanying trans-ependymal flow (white arrow). (b) Axial T1 gadolinium-enhanced MRI and (c) axial FLAIR showing an enhancing mass in the midline cerebellum. (d) Axial T1 and (e) accompanying axial diffusion coefficient map showing restricted diffusion. (f) Sagittal T1 gadolinium-enhanced MRI through the midline spine showing laminar leptomeningeal enhancement along the dorsal and ventral aspects of the spinal cord and ventral brainstem indicative of metastatic dissemination (arrows)

Full size image

Histopathology

Microscopically, MB is often referred to as a “small round blue cell tumor,” given the characteristics of densely packed cells with prominent nuclei surrounded by scant cytoplasm under H&E staining. The WHO classification of tumors of the central nervous system identifies five major histological variants in MB: classic, desmoplastic, large cell, anaplastic, and medulloblastoma with extensive nodularity (MBEN) [5]. Large cell and anaplastic variants have been correlated with poorer patient outcomes, while the best survival has been reported in patients with desmoplastic or MBEN histology; however, prognostic correlations may be influenced by age and other clinical features. Furthermore, as histologic categorization may be influenced by tumor heterogeneity, patient prognostication based on histopathological findings alone may not be accurate. The histologic differential diagnosis for MB includes atypical teratoid/rhabdoid tumor (ATRT) and rare embryonal tumors such as ETANTR [6]. Both entities which arise predominantly in younger children should be part of the differential work-up in all young patients with suspected MB. As characteristic rhabdoid cells may be present in variable amounts in ATRTs, immunostaining for the IN1/SNF gene product, which is nearly universally absent in ATRT but retained in MB tumors, should be included in the diagnostic work-up. Most but not all ETANTR exhibit classic histologic features of ependymoblastic rosettes and neuronal differentiation on a neuropil background and can be differentiated from MB by strong immuno-positivity for LIN28 and/or genomic amplification of the C19MC locus on 19q13.42 [7, 8].

Current MB Staging and Risk Stratification

Traditionally, MB patients are assigned into different treatment risk groups according to clinical features, which include age, the extent of resection, and the presence of metastasis at time of diagnosis. High-risk patients are defined as those <3 years of age, with more than 1.5 cm postsurgical residual tumor or evidence of metastasis at presentation [3]. Up to one third of MB patients present with leptomeningeal metastasis to the brain and/or spine, which may be detected on a preoperative MRI scan of the brain and spine. Postoperative MR imaging of the brain should be performed as soon as possible after surgery to avoid postoperative artifactual imaging changes to assess tumor residual. Staging is completed with cytological examination of the cerebrospinal fluid (CSF) for evidence of microscopic dissemination. Patients with nonmetastatic disease and no significant postoperative residual tumor, who are greater than 36 months of age, are stratified as standard risk.

Currently, all MB patients receive risk-adapted multimodality protocols based on clinical risk stratification schema. However, it is now increasingly clear that stratification based on clinical assessment alone is inadequate. It has been argued that current staging system fails to detect the true extent of disease and results in frequent over-/undertreatment of patients. Devastating acute and long-term treatment sequelae in survivors are major concerns for MB patients treated with intensive chemoradiotherapeutic treatment. Conversely, even with aggressive therapies, up to 30 % of MB patients will succumb to their disease. The lack of reliable clinical predictors of MB outcome has prompted substantial studies to identify biological predictors of clinical phenotype in MB.

Molecular Features of MB

MB was first linked to abnormalities in the Wingless (WNT) and Sonic Hedgehog (SHH) developmental signaling pathways based on observed association of MB with Turcot syndrome and Gorlin’s syndrome, and demonstration of alterations, respectively, in the APC and PTCH genes in some MB [9]. Early small cohort studies also suggested that specific genetic alterations, notably, MYCC gene amplification and CTNNB1 mutations, had prognostic correlations in MB (reviewed in ref [10]). Recent global gene expression and copy number studies of substantial cohorts of MB have helped to consolidate these early findings and establish a molecular classification system for MB that correlates with clinical phenotypes and patient outcomes. Specifically several global gene expression profiling studies of substantial MB cohorts have now demonstrated that MB is comprised of four molecular variants termed WNT, Sonic Hedgehog (SHH), and group 3 and group 4 subtypes which are associated with distinct developmental pathway signatures and/or cytogenetic abnormalities. The 4 MB subgroups also correlate with distinct tumor histology, patient demographics, and survival [10]. The WNT subgroup exhibits the best prognosis of any subgroup (greater than 95 % survival) and typically occurs in older children and exhibit classic histology. SHH MB represents an intermediate prognosis subgroup with overall survival ranging from 60 to 80 % and is predominantly seen in infants and young adults; MB with desmoplastic histology is almost exclusively restricted to this subgroup. Group 3 and 4 tumors which are not associated with any specific developmental signatures have the worst overall survival. MYCC gene amplification was observed only in group 3 tumors which also display frequently anaplastic histology, while group 4 tumors commonly (>30 %) exhibited isochromosome 17q (i17q, loss of chromosome 17p and gain of 17q). Characteristics of these subgroup variants are summarized in Fig. 12.2 [10]. These findings have helped to significantly advance our understanding of MB molecular biology and provided valuable diagnostic and prognostic tools that are currently in consideration for use in up-front risk stratification of patients in clinical trials across North America and Europe. A significant challenge is the development of robust diagnostic assays that can be used in clinical trials to reliably distinguish MB subgroups with high sensitivity and specificity. Currently, most consistent results to subtype MB have been reported with the use of nuclear CTNNB1 immuno-positivity and monosomy 6 which identifies WNT MB and MYCC amplifications which identifies group 3 MB; however, assays to reliably identify SHH and group 4 MB are lacking. Promising assays for MB subgrouping which include immunostains for SFRP1 in SHH, NPR3 in group 3, and KCNA1 in group 4 MB [11], and newer focused transcriptional and methylation assays remain to be validated.

Fig. 12.2
figure 2

Genetic, demographic, and clinicopathological features of the four molecular subgroups of medulloblastoma

Full size image

The availability of tools to segregate molecular subtypes of MB will profoundly alter the design of MB clinical trials [11]. In addition to enhanced risk stratification for current conventional treatment regimens, molecular subtyping of MB will enable concerted investigations of novel therapy tailored to subgroup-specific biology. The inclusion of molecular analyses with traditional histo-clinical examination will be the standard of care in establishing the diagnosis and treatment stratification of MB in the near future.

Therapeutic Approaches

Multimodal approach of maximal safe surgical resection, radiotherapy to the primary tumor site and craniospinal axis, and systemic adjuvant chemotherapy are the current standard of care. Patients with MB often present with significant obstructive hydrocephalus, and thus management of increased intracranial pressure is commonly the priority. Patients may be managed with corticosteroids to alleviate tumor edema or require CSF diversion prior to surgery. With current treatment strategies, an anticipated 5-year overall survival (OS) has reached up to 80 % for patients with localized disease. However, metastatic and recurrent MBs still result in a significant mortality.

Surgery and Radiotherapy

Maximal safe resection of the posterior fossa mass is a key component and goal for patients with MB. With modern surgical techniques, gross total resection can be achieved in a majority of patients. As residual disease is associated with poorer outcome, immediate reoperation or second look surgery after adjuvant chemo- or radiotherapy may be considered.

Radiotherapy remains a critical part of the multimodal approach and is delivered early, within a month postsurgery, as delayed radiation results in poorer outcome [12]. The goal of radiotherapy is to control for both residual microscopic tumor in the primary site and to treat or prevent leptomeningeal disease along the craniospinal axis. Due to the severe toxic effects of irradiation to the developing nervous system, craniospinal radiation is often avoided or delayed in patients under the age of 3. Children with both average- and high-risk MB patients receive the same dose of local tumor bed irradiation of 5,400–5,580 cGy, but receive risk-adapted craniospinal irradiation. Children without tumor residual or metastasis received a 2,340 cGy craniospinal irradiation, while high-risk patients receive at least 3,600 cGy to the neuraxis [13]. Further reduction of craniospinal irradiation to 1,800 cGy for average-risk patients is currently being investigated in a phase III randomized control study by the North American Children’s Oncology Trial Group (COG).

In European and North American trials, the standard of care for MB involves postsurgical radiation followed typically by adjuvant cisplatin-based or high-dose chemotherapy. With such regimens, a 5-year survival for average-risk MB patients has reached 75–85 %; however, survival of high-risk patients is significantly poorer with a 5-year OS of 30–65 %. Significant improvement in survival has been achieved with chemoradiotherapy combination regimens; however, high-dose craniospinal radiotherapy continues to be associated with a high incidence of treatment-related complications. In addition to cognitive impairment, ototoxicity, thyroid dysfunction, growth failure, and endocrine abnormalities are significant sequelae in MB survivors. Intensity-modulated and proton-based radiation therapies represent promising newer normal tissue-sparing radiotherapeutic approaches. In addition to the use of conventional concomitant chemotherapy, development of novel radiosensitizers would be an important step toward minimizing radiation-associated toxicity.

Chemotherapy

Adjuvant chemotherapy plays an important role in the management of both average- and high-risk MB patients and has been used with the intent to permit reduction in radiation doses for older children or to avoid or delay radiation in younger children.

Radiation-sparing approaches for younger children with MB have been varied both in terms of chemotherapeutic regimens and inclusion of age range. In addition, some groups have used focal radiation up-front. Global trial groups have used three general approaches in children <3–5 years of age with MB to achieve a 5-year OS ranging between 50 and 70 %. These regimens have generally differed in the use of methotrexate, intraventricular treatment, and use of high-dose chemotherapy for consolidation. While the SFOP [14] group has used chemotherapy-based regimens without methotrexate, the UKCCSG/SIOP [15] and the German trial group protocol [13] employed a methotrexate-based chemotherapy backbone with intraventricular methotrexate treatment in the German experience. The Head Start consortium, which pioneered the use of high-dose chemotherapy, currently employs a methotrexate-based induction regimen with a single high-dose/stem cell rescue as consolidation [16], while treatment in the COG protocol is consolidated with 3 cycles of high-dose therapy and stem cell rescue [17]. As favorable outcomes with desmoplastic MB in young children have been seen across different infant MB studies, planned trials are examining desmoplasia as a criterion for stratification. An ongoing COG high-risk infant MB/CNS-PNET protocol is examining the benefits of methotrexate during induction in high-dose chemotherapy-based regimens, while the relative merits of single versus three stem cell rescue in consolidation remain to be investigated.

Treatment approaches to older children with MB have been more consistent. With the exception of a stem cell-based protocol from St. Jude’s, most trial groups have employed a conventional chemotherapy-based regimen with similar drugs. For average-risk patients, over the age of 3, both the COG and SIOP trial groups have employed a chemotherapy-based regimens with CCNU, VCR, and cisplatin, followed by reduced dose 2,340 cGy craniospinal irradiation and reported similar 5-year EFS, respectively, of 81 % [18] and 77 % [19]. In the St. Jude Medulloblastoma-96 protocol where standard-dose craniospinal in radiation is followed by four cycles of cyclophosphamide-based, dose-intensive chemotherapy, 5-year OS of 85 % has been reported [20], suggesting that dose intensification may benefit some average-risk patients.

Overall survival of patients with high-risk disease, specifically those presenting with metastasis, has been less favorable with 60–65 % long-term survival observed across chemotherapy and high-dose-based regimens. However, improved results have been reported by the Milan strategy in which patients received postoperative methotrexate, etoposide, cyclophosphamide, and carboplatin in a 2-month schedule, followed by hyperfractionated accelerated radiotherapy (HART). This regimen resulted in a 5-year OS of 73 %. COG also reported better survival in a phase I/II trial for high-risk MB in which patients received 15–30 doses of carboplatin along vincristine as radiosensitizers, followed by cisplatin-based maintenance chemotherapy. This study reported the highest overall survival at 82 % in high-risk MB patients who received 6 months of maintenance chemotherapy with cyclophosphamide thus suggesting a role for biologic-based maintenance chemotherapy in MB therapy [21].

Due to the increased incidence of secondary malignancies from irradiation and chemotherapy, periodic surveillance with brain and spine MRIs for disease recurrence, as well as secondary malignancies, is performed. In addition, regular neuropsychological and medical surveillance for end-organ toxicity is indicated for MB survivors. Recurrent disease, which occurs in approximately 25 % of patients with MB, remains a significant clinical challenge. Most relapses tend to occur within the first 3 years post-diagnosis, and long-term survival in this population remains very poor, with no clear effective rescue regimens reported. Limited successes have been reported with high-dose chemotherapy/autologous stem cell rescue regimens in older children with recurrent MB [22]. Higher salvage rates have been reported in younger children, who have not received prior irradiation, with radiation-based rescue protocols [23].

Molecular Therapeutic Targets

Basic research has led to tremendous gains in biological knowledge regarding MB. Specifically, the establishment of molecular subgroups for MB and the development of multiple group specific animal models are poised to transform future clinical trials and treatments for MB patients. It is expected that in the near future, patients will be stratified and treated based on the biological subgroup-specific makeup of their disease, which will hopefully lead to improved outcomes with less adverse effects. One main goal is to reduce morbidity of current treatment regimens in children with favorable biology disease. Specifically reducing chemotherapy and craniospinal irradiation for the favorable WNT subgroup could be one tangible approach that will minimize treatment toxicity.

To date, multiple pharmacological inhibitors for SHH-driven MB subgroups have been designed and have shown promising antitumor effects in SHH MB mouse models, and some are currently under clinical trials (Table 12.1). Promising preclinical studies, however, have not correlated with sustained disease response due to development of drug resistance with monotherapies [25]. These observations suggest that a combination of targeted therapies with conventional chemotherapy regimens may be required for optimal efficacy. Additionally, the use of several biologic agents that target signaling cross talk between SHH signaling and other key molecular pathways, such as AKT, Notch, TGF-β(beta), may also offer novel approaches to tailored therapy [31]. The feasibility of employing combination signaling therapies has been assessed in MB preclinical models. For instance, retinoic acid, which induces apoptosis together with histone deacetylase inhibitors, exhibits synergistic effects in xenograft and transgenic models [32]. Furthermore, a combination of LDE225 with PI3K inhibitors also markedly delays development of resistance thus suggesting the importance and promise of multiple pathway inhibition for sustained tumor response [26].

Table 12.1 Examples of preclinical pharmaceutical agents for targeted therapeutics in medulloblastoma
Full size table

Recent studies implicate the PI3K/AKT pathway, an effector and downstream target of MYCN/C, respectively, in SHH and group 3 MB biology. Thus, PI3K/AKT inhibitors also represent attractive new drug approaches for these patients [33]. Indeed small molecule inhibitors of PI3K/AKT have been demonstrated to suppress MB tumorigenesis in in vitro cell culture conditions [34], as well as in MYC-driven mouse models [33]. As MYC overexpression confers aggressive and metastatic behavior in MB [35], there has been considerable interest in directly targeting MYC or altering MYC downstream effects. Intriguingly, recent studies in myeloma and lymphoma models have shown the effective use of BET bromo-domain inhibitors to suppress MYC expression [36]. More interestingly, recent focus on screening synthetic lethal targets for MYC-driven cancer also identified potential new targets such as the core SUMOylation machinery and eukaryotic initiation factor complex assembly that are required to support MYC oncogenic state [36], for therapeutic opportunities. It is anticipated that with further dissection of MB subgroups and development of more precise subtype models that treatment of MB will approach a truly tailored approach with regimens that incorporate a spectrum of biologic agents that will improve patient outcomes with minimal sequelae.

Central Nervous System Primitive Neuroectodermal Tumor/Pineoblastoma

Epidemiology

Central nervous system primitive neuroectodermal tumors (CNS-PNETs) encompass a collection of embryonal tumors which are poorly differentiated with varying degrees of neuronal, astrocytic, or ependymal differentiation. They represent ~2.5 % of all childhood brain tumors [5] and are predominantly hemispheric in location but can also arise in multiple other CNS locations including the posterior fossa. In the most recent WHO classification system, CNS-PNETs are subcategorized based on location and histology and include CNS-PNET-NOS or supratentorial-PNET which are hemispheric tumors without any distinctive histologic features that overlap with tumors previously labeled as SPNET. Other categories identified by specific histologic features include CNS-neuroblastoma, CNS-ganglioneuroblastoma, medulloepithelioma, and ependymoblastoma. Cerebral neuroblastomas and ganglioneuroblastoma, respectively, display only neuronal differentiation or with the presence of ganglion cells. Medulloepitheliomas are rare and diagnosed based on the presence of features resembling embryonic neural tube formation, while ependymoblastoma are characterized by distinct multilayered “ependymoblastic” rosettes; however, accuracy and specificity of this subclassification is under debate [5], and ependymoblastomas has been proposed to overlap with the recently described aggressive ETANTR/EMTR histologic entity [6]. These observations highlight the significant challenge in histological classification of CNS-PNET, particularly for subtypes which exhibit closely related variant histology. Emerging data suggest these may also represent closely related biological and molecular entities. CNS-PNETs diagnosis and classification have been challenging and remain in flux. Notably, pineal region PNET has been considered and treated as SPNET in clinical trials; however, their biologic relatedness to tumors currently classified under the CNS-PNET umbrella remains unclear. A full understanding of the molecular spectrum of this broad category of CNS-PNET and their relationship to tumors restricted to the pineal region is critical for development of more specific diagnostics and therapeutics.

Clinical/Imaging Findings

Clinical presentation of CNS-PNETs varies and can include a broad range of symptoms related to tumor location; however, signs and symptoms of increased intracranial pressure are most common. CNS-PNETs located in the cerebral hemispheres usually present as heterogeneously enhancing large, deep-seated lesions that may contain areas of calcification and may be difficult to distinguish from other malignant hemispheric lesions.

Molecular Features of CNS-PNETs

Although CNS-PNETs may share close histologic resemblance with MB, cumulative studies of small tumor cohorts show they lack genomic features commonly found in MB including isochromosome 17q and MYCC amplification, reviewed in [37]. Li et al. [8] first studied a substantial cohort of 40 hemispheric CNS-PNET using global gene expression and copy number profiling and reported the discovery of a novel oncogenic miRNA cluster that was amplified in 25 % of CNS-PNETs. They demonstrated that tumors with C19MC amplification were frequently CNS-PNETs with variant histologic features and included tumors labeled as ETANTRs, CNS-PNETs with ependymal differentiation, medulloepithelioma, and ependymoblastoma. In a subsequent study, Korshunov et al. [38] confirmed that the C19MC amplicon identified CNS-PNETs variants with histologic features of rosette formation that they called EMTR. In a more recent study, Picard et al. conducted global profiling and immunohistochemical analyses on 142 CNS-PNETs arising in the cerebral hemisphere and demonstrated that the C19MC-amplified tumors (called group 1) also expressed high levels of LIN28, a pluripotency marker [7]. Notably, C19MC amplification/LIN28-positive CNS-PNETs are not restricted to the cerebral hemispheres, indicating C19MC/LIN28 identifies a single molecular disease that may exhibit disparate histologies and location. Thus, C19MC amplification and LIN28 expression represent powerful molecular tools to identify and define the true incidence of group 1 CNS-PNETs, which may not be as uncommon as previously anticipated.

Picard et al. [7] also identified two other molecular classes of CNS-PNETs with oligoneural (group 2) and mesenchymal (group 3) gene expression signatures in their global study. Group 2 CNS-PNETs characterized by an oligoneural gene signature, arose most frequently in older children, and were predominantly localized (15 % were metastatic). The group 3 mesenchymal subgroup, which were identified by the lack of both LIN28 and OLIG2 expression, were found across age groups and had the highest incidence (65 %) of metastases [7]. Thus, OLIG2 expression may serve as an important marker to identify CNS-PNETs with low risk of metastasis. To date no specific histologic features which correlate with group 2 and 3 CNS-PNETs signatures have been identified. Interestingly, a recent study suggested that a specific mutation of the H3.3 gene (H3.3G34R), which is seen in about 25 % of glioblastoma, is also found in subset of CNS-PNETs [39]. These findings highlight long-standing challenges with histologic diagnosis of CNS-PNET and suggest that molecular features of some tumors diagnosed as CNS-PNETs and GBMs may overlap. Together with the identification of candidate immuno-markers for CNS-PNETs, these studies will now enable better categorization of the spectrum of undifferentiated malignant neuroepithelial tumors arising in the cerebrum.

Although certain copy number features were more commonly seen in group 2 versus group 3 tumors, no cytogenetic alterations exclusive to group 2 and 3 tumors were observed. N-Myc amplification and deletion of CDKN2A/B tumor suppressor locus (Chr9p21.3) which have been previously reported were observed in a proportion of group 2 and 3 tumors [40]. Future larger-scale studies will enable more comprehensive analysis of the clinical implications of these genetic lesions in CNS-PNETs.

Therapeutic Approaches for CNS-PNETs

Similar to MB, maximal safe surgery is also recommended for CNS-PNETs; however, due to the deep-seated nature of these tumors, safe complete tumor resection is often difficult to achieve. Currently, postsurgical therapy for CNS-PNET is similar to that for high-risk medulloblastoma [41, 42] with delivery of higher dose craniospinal irradiation in older children. However, survival rates of CNS-PNET are significantly poorer, with overall survival rates reported at <50 % regardless of therapy received [17, 41]. Determinants of treatment failures in CNS-PNETs are not clear; however, limited data from a number of small studies suggest that an extent of resection and treatment with irradiation are important prognostic factors [43]. Notably, predominant local failures reported in retrospective studies of CNS-PNETs together with pilot studies of risk-adapted radiation [44, 45] suggest that a proportion of older children with hemispheric CNS-PNETs may be cured with reduced dose or volume of irradiation. As group 2 OLIG2+ CNS-PNETs are predominantly localized, OLIG2 may represent a promising marker for treatment stratification of CNS-PNETs.

Poorest outcomes have been observed in infants with CNS-PNET and may reflect omission or reduction of radiotherapy as radiotherapy has been identified as a significant predictor of progression-free and overall survival [42, 44]. However, it may also reflect the distinct biology of CNS-PNETs arising in younger children. Data to date suggest a particularly aggressive course for the ETANTR/EMTR tumors which comprise group 1 CNS-PNETs. Gene expression signatures suggest activation of SHH and noncanonical WNT pathways in C19MC/LIN28+ group 1 tumors [7]; thus SHH pathway inhibitors, such as GDC-0449 or LDE225, represent promising novel therapeutics that warrants further investigation in this disease. Significantly, recent studies demonstrate that LIN28 activates the insulin-PI3K-mTOR pathway in these tumors and treatment with mTOR inhibitors abrogates growth of an ETANTR/EMTR cell line [46]. These studies suggest that drugs targeting the insulin-mTOR signaling warrant further evaluation in this disease.

Pineoblastoma

Pineoblastomas are grade 4 tumors arising in the pineal region that have traditionally been included as “supratentorial PNETs” in clinical trials, as they histologically resemble small round blue cell embryonal tumors that arise in other regions of the CNS. They comprise 40 % of tumors arising within the pineal parenchyma [5]. The differential diagnosis for pineoblastomas includes germ cell tumors and low-grade glial tumors that arise in the region of the tectal plate/pineal gland as well as benign pineal tumors and other nonmalignant masses. Radiologic features of an enhancing poorly circumscribed pineal mass would favor a malignant tumor diagnosis such as germ cell tumor or pineoblastoma. Histopathologic features of pineoblastoma predominantly resemble that of primitive pineal or retinal tissues. Only a small number of pineoblastomas have been examined at the molecular level [40, 47]; thus a comprehensive molecular picture has yet to emerge. Notably tumors with molecular features of ETANTR/EMTR have also been reported in the pineal region, suggesting that a proportion of tumor diagnosed as pineoblastomas in younger children may be C19MC/LIN28 group 1 CNS-PNETs.

Demographically pineoblastomas are most frequently found in younger children and may be associated with heritable RB1 alterations and present concurrently with retinoblastoma [48]. Thus, ophthalmological examination of young children presenting with pineoblastoma is recommended to rule out “trilateral” retinoblastoma. Children can present with hydrocephalus due to third ventricular obstruction as well as ocular signs of a pineal mass.

Younger children with pineoblastoma frequently present with or recur with metastatic disease thus making treatment pineoblastoma highly challenging [49]. Metastatic disease at diagnosis and young age has been identified as negative prognosticators and is likely to be age-related treatment approaches. Notably, although Fangusaro et al. [50] reported worse outcomes for pineoblastoma than hemispheric CNS-PNET treated in the Head Start protocol, outcomes for pineoblastomas in older children treated with chemoradiotherapy protocols have been reported to be significantly better than hemispheric CNS-PNETs [51, 52]. Whether these discrepant observations reflect age-related differences in pineoblastoma biology remains to be studied.

Prospects and Future Directions

Since the first publication on gene expression signatures of embryonal tumors [53], molecular studies of MB and rarer embryonal tumors, such as CNS-PNETs, have rapidly advanced. Discovery of clinically relevant molecular subclasses of MB and CNS-PNETs has highlighted the molecular heterogeneity of these tumors and provided new diagnostic and prognostic tools. With new molecular tools in hand, we are poised to quickly enter an era of trials where molecular information will significantly enhance clinical predictors for risk stratification and patient outcomes. Additionally these studies have uncovered novel targetable genes and pathways that necessitate the development of biology-based therapeutic trials for evaluation. For both MB and CNS-PNETs, the rapid pace of these significant achievements has only been possible with the establishment of large collaborative consortia which have provided the power to delineate molecular subgroups in relatively rare diseases. Similar global efforts will be necessary in clinical trials to enable robust translation of biological discoveries to precision in patient care. Additionally, continued global effort will be necessary to provide further refinement in molecular classification, particularly for rarer types of embryonal brain tumors such as CNS-PNETs and pineoblastoma.