Key Points

Incidence and Overall Survival

The overall incidence of childhood brain tumors is 5.67 per 100,000 person-years for children. Brain tumors are the second-most common childhood malignancy and the most common primary solid tumor. Pediatric brain tumors are a diverse group of tumors and outcomes are dependent on histologic subtype and treatment. Approximately 75% of children diagnosed with pediatric brain will experience long-term disease-free survival (DFS).

Etiology

At present, the etiology of most pediatric brain tumors is unknown. Genetic syndromes that may predispose children to certain subtypes of brain tumors include neurofibromatosis, Li-Fraumeni, Gorlin's syndrome, Turcot's syndrome, and ataxia-telangiectasia.

Histologic Subtypes

Brain tumors in children have traditionally been classified by histology, which is summarized as follows based on the World Health Organization (WHO) classification of tumors of the central nervous system 2016 update: embryonal tumors; medulloblastoma; central nervous system embryonal tumor, not otherwise specified; atypical teratoid rhabdoid tumor; embryonal tumor with multilayered rosettes; astrocytic tumors (pilocytic astrocytomas [WHO grade 1], astrocytoma [WHO grade 2], anaplastic astrocytoma [WHO grade 3], glioblastoma [WHO grade 4]); diffuse midline glioma H3 K27M-mutant (ependymal tumors, myxopapillary [WHO grade 1, usually of spine], classic/differentiated [WHO grade 2], anaplastic [WHO grade 3]); ependymoma, RELA fusion–positive (WHO grade 2 or 3) (craniopharyngioma, germ cell tumors, pure germinoma, nongerminomatous germ cell tumors, embryonal, yolk-sac, choriocarcinoma, immature teratoma, mixed germ cell tumors [NGGCT components ± germinoma], mature teratoma, choroid plexus tumors), pineal region tumors.

Key Molecular Features of Pediatric Brain Tumors

Although histology continues to play an important role in the classification of childhood brain tumors, the WHO has decided to make changes in the recent WHO 2016 classification manual to incorporate specific molecular information, to assist with the better classification of these tumors. In many cases, molecular subtyping also is prognostic and possibly predictive for the response to therapy. We review key molecular features and subtypes of selected pediatric brain tumors. Their relevance to the radiation oncologist is included in the subsequent sections on the clinical management of these tumors.

Molecular Classification of Pediatric Brain Tumors

Medulloblastoma

Historically, medulloblastoma has been classified by a diverse collection of pathologic, clinical, and demographic features. Both treatment decisions and prognosis were largely based on features such as age (older or younger than 3 years old), tumor location (infra- versus supra-tentorial), extent of resection (gross total resection [GTR] or not), and evidence of disseminated disease in the cerebrospinal fluid (leptomeningeal spread or not). These factors led to the assignment of medulloblastoma into one of two risk groups, standard- or high-risk disease, which dictated the difference between the utilization of craniospinal irradiation (CSI) doses of 23.4 versus 36 Gy, respectively. The 2007 World Health Organization (WHO) classification of medulloblastoma further divided the disease in to four groups based on histology: classic, large-cell/anaplastic variant, desmoplastic/nodular medulloblastoma, and medulloblastoma with extensive nodularity ; however, in 2016, the WHO classification markedly changed its classification by incorporating molecular features along with histology, leading to the four main subgroups: (1) Wingless (WNT), Sonic Hedgehog (SHH), Group 3, and Group 4. Emerging data suggest that these subgroups are powerful predictors of disease behavior and overall patient prognosis. Key features of each subgroup are presented in Table 78.1 , and we discuss each subgroup in greater detail. We also briefly discuss supratentorial primitive neuroectodermal tumors (PNETs) here.

TABLE 78.1
Medulloblastoma
Subgroup WNT SHH Group 3 Group
Cases 10% 30% 25% 35%
Histology Classic, rarely LCA Classic, DNMB, LCA, MBEN Classic, LCA Classic, rarely LCA
Key gene mutations CTNNB1 PTCH1, SMO, SUFU, TP53
Key molecular pathways WNT signaling SHH signaling
PI3K signaling
MYC signaling
TGFab signaling
NFkB signaling
Neuronal signaling
DNMB, Desmoplastic/nodular medulloblastoma; LCA, large cell/anaplastic; MBEN , medulloblastoma with extensive nodularity; SHH , Sonic Hedge Hog; WNT , Wingless.

WNT Tumors

These tumors are the least common of the four subgroups, occur most frequently in children older than 3 years old and almost always are located in the fourth ventricle or near the brainstem. They have the most favorable prognosis, with 5-year survival rates exceeding 95%. The link between WNT signaling and medulloblastoma was first made in patients with Turcot's syndrome, a hereditary disease caused by mutations in the adenomatous polyposis gene, which functions as a negative regulator of WNT signaling. This syndrome was found to be associated with a higher incidence of medulloblastoma. It was subsequently discovered that subsets of sporadic medulloblastoma harbor WNT signaling pathway mutations, most notably in the CTNNB1 gene, which encodes β-catenin. The WNT signaling pathway is highly conserved in metazoans, and is involved in a diverse range of cellular process, including cell-cycle control, stem cell renewal, and cell motility/polarity.

SHH Tumors

These are a heterogeneous group of tumors, which account for nearly one-third of all medulloblastoma. They are seen most frequently at either ends of the age spectrum (i.e., infants and adults), and they are associated with an intermediate prognosis. A link between the SHH signaling pathway a medulloblastoma was initially recognized in patients with Gorlin's syndrome, which is characterized by craniofacial abnormalities and predisposition to several cancers, including medulloblastoma. Mutations in PTCH1 have been described in Gorlin's patients, which is a negative regulator of the SHH pathway. Subsequent studies linked medulloblastoma to several other key gene mutations or other genetic alterations within the SHH signaling pathway, including SUFU, SMO, SHH, GLI2, and MYCN . The SHH signaling pathway plays a critical role in numerous cellular processes, including cell proliferation, differentiation, and vertebrate organogenesis.

Group 3 Tumors

These tumors account for a quarter of medulloblastomas, are found in younger patients, and are associated with the worst prognosis. Likely because of their location near the fourth ventricle at the midline of the brainstem, cerebrospinal fluid (CSF) metastases are common and frequently seen at the time of initial diagnosis. Although specific mutations are not common in this subgroup, focal somatic copy number aberrations have been described, including amplification of the MYC oncogene.

Group 4 Tumors

Group 4 tumors are the most common subgroup and they share many similarities to group 3 medulloblastoma, including close proximity to the fourth ventricle and frequent metastases at diagnosis. Overall, it is the most poorly understood tumor within the four subgroups from a molecular perspective. Somatic copy number aberrations have been described in this tumor, including duplication of the SNCAIP gene, which encodes a protein implicated in Parkinson's disease.

Central Nervous System Embryonal Tumor Not Otherwise Specified (Formerly Supratentorial Pnets)

These tumors resemble medulloblastoma but differ in their location outside of the posterior fossa in the brain. They are genetically heterogeneous tumors, with some reports of copy number alterations in MYCN , PDGFB and PDGFRA, as well as deletions in CDKN2A/B . More recently, a high-level amplicon involving the chr19q13.41 microRNA cluster (C19MC) was reported in a subset (~25%) of these PNETs. In the 2016 central nervous system (CNS) WHO classification, the presence of C19MC amplification results in the diagnosis of embryonal tumor with multilayered rosettes (ETMR) (PMID: 27157931 ). On the basis of DNA methylation profiling, it has been shown that a significant proportion of institutionally diagnosed CNS-PNET are in fact other well-defined CNS tumor entities (i.e., high-grade glioma), whereas four distinct tumor entities have emerged from the remaining histologically defined CNS-PNETs. This has resulted in removal of the term PNET and addition of the term CNS embryonal tumor, not otherwise specified, in the updated WHO classification.

Ependymoma

Like medulloblastoma, treatment decisions and prognostication for ependymoma have been defined by their classification into groups based on location (supra- versus infra-tentorial, or spinal), extent of resection (GTR or not), histology (presence of anaplasia), and age. The current WHO 2016 classification of these tumors is histology-based and consists of: subependymoma, myxopapillary ependymoma, classic ependymoma, and anaplastic ependymoma. A landmark paper in 2015 has radically changed the landscape of ependymoma, which used DNA methylation profiling to classify these tumors into nine unique molecular subtypes based on three anatomic locations. A notable feature of this new classification is the stark lack of concordance with histologic grading. Next we discuss key molecular features these subtypes, with a focus on supratentorial and posterior fossa tumors (summarized in Table 78.2 ).

TABLE 78.2
Ependymoma
SUPRATENTORIAL POSTERIOR FOSSA SPINE
Subgroup ST-SE ST-EPN-YAP1 ST-EPN-RELA PF-SE PF-EPN-A PF-EPN-B SP-SE SP-MPE SP-EPN
Age Distribution Adults Infants>children>adults Children>infants and adults Adults Infants>children>adults Adults>children Adults Adults>children Adults>children
Histology (WHO Grading) Subependymoma
(WHO I)
(Anaplastic)
Ependymoma
(WHO II/III)
(Anaplastic)
Ependymoma
(WHO II/III)
Subependymoma
(WHO I)
(Anaplastic)
Ependymoma
(WHO II/III)
(Anaplastic)
Ependymoma
(WHO II/III)
Subependymoma
(WHO I)
Myxopapillary
Ependymoma
(WHO I)
(Anaplastic)
Ependymoma
(WHO II/III)
Notable Genetic Aberrations Balanced Genoma Yap-1 fusions RELA fusions
Chromothripsis
Balanced genoma Balanced genoma Chromsomal instability 6q Deletion Chromsomal instability NF-2 Mutation
Chromsomal instability
Prognosis Good Good Poor Good Intermediate Good Good Good Good
ST-SE , supratentorial subependymoma; ST-EPN , supratentorial ependymoma; PF-SE , posterior fossa subependymoma; PF-EPN , posterior fossa-ependymoma; SP-SE , spinal cord subependymoma; SP-MPE , spinal cord myxopapillary ependymoma; SP-EPN , spinal cord ependymoma.

Supratentorial Tumors

Approximately two-thirds of supratentorial ependymoma harbor fusions between the RELA gene and C11orf95, termed ST-EPN-RELA, which is associated with an extremely poor prognosis. The former gene is involved in NF-κB signaling, and little is known about the function of latter gene. However, emerging preclinical data suggest that RELA fusions are key oncogenic drivers, promote tumor cell invasion, and are associated with resistance to cytotoxic chemotherapies. Of note, the most recent WHO 2016 classification for ependymoma now includes a subcategory for RELA fusion-positive ependymoma. The remaining one-third of supratentorial ependymoma lack RELA fusions, but instead harbor YAP1 fusions, a gene involved in the Hippo signaling pathway. These tumors are termed ST-EPN-YAP1, and emerging data suggest that these tumors are associated with an excellent prognosis.

Posterior Fossa Tumors

Before the aforementioned ependymoma DNA methylation study, two previous studies had already suggested two biologically distinct forms of posterior fossa ependymomas (PF-EPN-A and PF-EPN-B). These two groups demonstrated unique differences with regard to molecular characteristics, patient demographics, and clinical outcomes. PF-EPN-A tumors tend to occur in younger patients, are likely to recur locally, metastasize distantly, and overall are associated with a much poorer prognosis in comparison to PF-EPN-B tumors. At the molecular level, PF-EPN-A tumors are most notable for transcriptional silencing driven by a CpG-island methylator phenotype, which is associated with hyperactivity of the Polycomb Repressive Complex 2 (PRC2).

Glioma and Diffuse Intrinsic Pontine Glioma

Historically, pediatric gliomas have been classified using the conventional WHO grading system (I-IV). Clinical trials for pediatric high-grade glioma (HGG; i.e., WHO grades III–IV) were modeled based on adult GBM studies, with the assumption that they were histologically similar tumors. In the adult GBM realm, the Stupp trial defined the current standard of care for this disease consisting of radiotherapy (RT) with concurrent and adjuvant TMZ, and methylation of the O 6 -methylguanine-DNA methyltransferase (MGMT) gene emerged as a key biomarker for the response to TMZ. However, the addition of TMZ to RT failed to show activity in pediatric HGGs, compared with historical controls in patients treated with CCNU and vincristine. In addition, the prognostic and predictive value of MGMT promoter methylation in pediatric HGG has remained controversial.

In 2010, a landmark study by Verhaak et al. defined four unique molecular subtypes in adult GBM (proneural, neural, classical, and mesenchymal). Soon after, it was discovered that pediatric HGGs also could be classified into molecular subtypes, which largely were distinct from adult GBM (isocitrate dehydrogenase [IDH], K27, G34, Receptor Tyrosine Kinase I and II, and mesenchymal). In the past, diffuse intrinsic pontine glioma (DIPG) rarely was biopsied over concerns of morbidity, but two studies suggested these procedures were safer than previously thought, which, along with innovative postmortem autopsy studies opened the door for the comprehensive molecular profiling of these tumors as well. The results of these studies revealed that DIPGs harbor unique molecular characteristics, which suggest that these tumors indeed are distinct from both adult GBM and pediatric HGGs. Next we review the key molecular features of both low-grade gliomas (LGGs) and HGGs, as well as DIPG.

High-Grade Glioma

As noted previously, pediatric HGGs can be classified into six unique molecular subtypes. The K27 and G34 subgroups are composed of tumors with heterozygous mutations in two key histone genes, H3.3 and H3.1. (also referred to as H3F3A and HIST1H3B/C, respectively), resulting in amino acid substitutions at K27 or G34 . Histone H3 K27M mutations are found in approximately 80% of midline pediatric HGGs, including DIPG, and given the stereotypic location and uniformly poor outcome have been designated as diffuse midline glioma, H3 K27M-mutant tumors in the 2016 CNS WHO classification. This mutation disrupts the PRC2 complex via inhibition of its catalytic subunit, EZH2. This leads to reductions in the overall levels of trimethylated histone H3 marks at the K27 position (H3K27me3), which causes transcriptional de-repression .

In the adult setting, isocitrate IDH1/2 are common in LGGs (>70% tumors) and less frequent in glioblastoma. Early studies had suggested that IDH1/2 mutations were rare in pediatric gliomas (~10% or less), although these were based on the analysis of small case numbers. Because more comprehensive analyses emerged with larger case numbers, and using next-generation sequencing (NGS)-based mutation detection techniques, it has become apparent that IDH1/2 mutations are relatively common in the adolescent and young adult population. For example, a study by the Children's Oncology Group (COG) found IDH1/2 mutations in 35% of children ≥14 years of age, which was corroborated in two NGS-based studies of pediatric gliomas.

Low-Grade Glioma

Pediatric LGGs have been referred to as “genetically quiet,” with few mutations found in these tumors; however, alterations have been reported that appear to converge on the RAS/MAPK pathway. In particular, genetic aberrations in BRAF have been described in pediatric LGGs, including BRAF fusions with KIAA1549, as well as V600E point mutations.

DIPG

H3 K27M mutations are also common in DIPGs (in up to 80% of cases), and frequently co-occur with mutations in the Protein Phosphatase Mg 2+ /Mn 2+ Dependent 1D (PPM1D) gene. The PPM1D gene encodes a serine/threonine phosphatase that dephosphorylates numerous proteins primarily involved in the DNA damage response and cellular checkpoint pathways. Histone mutations in DIPG are also linked to activating mutations in the activing A receptor type 1 (ACVR1) gene. Interestingly, identical mutations have been reported in the hereditary disease, fibrodysplasia ossificans progressive.

You're Reading a Preview

Become a Clinical Tree membership for Full access and enjoy Unlimited articles

Become membership

If you are a member. Log in here