Introduction

Tumors of the central nervous system (CNS) are the leading cause of cancer-related morbidity and mortality, and are the most common solid malignancy in childhood. Past therapeutic approaches have relied on data from adult glioma biology and clinical trials; however, recent molecular studies of childhood gliomas reveal distinct differences that should translate into pediatric-specific glioma therapies. While there has been substantial improvement in our treatment approaches, including proton radiation therapy, minimally invasive surgical approaches, and molecularly guided targeted agents, the backbone of treatment remains surgery, radiotherapy and cytotoxic chemotherapy.

Low-grade gliomas (LGG) are the most common childhood brain tumor, representing over 30 % of all primary brain tumors in pediatric patients [1]. There is histologic heterogeneity even within the LGG subclass of gliomas. Most commonly seen in children are pilocytic astrocytomas (PAs) and diffuse (fibrillary) astrocytomas, but oligodendroglioma, ganglioglioma, pilomyxoid astrocytomas, and pleomorphic xanthoastrocytomas are also among the LGGs of childhood. These are currently treated in a similar fashion, though they have varying prognoses, and recently discovered molecular distinctions may soon have an impact on therapeutic decisions. Pilocytic astrocytomas are World Health Organization (WHO) grade I tumors that arise sporadically or in children with neurofibromatosis type 1 (NF1). In sporadic PAs, tandem duplications of the BRAF kinase gene have been identified as the most frequent genetic alteration, creating a novel fusion protein (KIAA1549:BRAF) with constitutively active BRAF activity [2, 3]. PAs arising in children with NF1 have allelic loss of the NF1 gene, resulting in loss of NF1 protein (neurofibromin) and hyperactivated signaling through the RAS pathway [4, 5]. The common end result of both mutations is increased mitogen-activated protein kinase (MAPK) pathway activity, promoting tumorigenesis and serving as tractable targets for anti-glioma therapy.

High-grade gliomas (HGG) comprise 8–12 % of pediatric brain tumors. Unfortunately, there has been little improvement in survival outcomes for this tumor in over 20 years of prospective randomized trials [6, 7•, 8]. Histologically, childhood anaplastic astrocytoma (AA, WHO grade III) and glioblastoma multiforme (GBM, WHO grade IV) appear similar to their adult counterparts, but increasingly, molecular distinctions between childhood and adult HGGs are being defined [9–13]. Recent analyses of DNA copy number, gene expression signatures, and sequencing of childhood HGGs revealed differences in expression of platelet-derived growth factor receptor (PDGFRA, overexpressed) and epidermal growth factor receptor (EGFR, repressed), as well as mutations in chromatin remodeling pathways regulating gene expression that clearly distinguish childhood from adult HGG [9, 13, 14] and have implications for future targeted therapies. Interestingly, infant HGGs (younger than 3 years old) have different genetic profiles, which may account for their better prognosis [15••, 16]. Expression of the DNA-repair enzyme, MGMT, a known prognostic factor in adult HGGs treated with temozolomide (TMZ), is also strongly correlated with outcome in pediatric HGG [7•, 17–19].

Brainstem gliomas account for 10–20 % of pediatric CNS tumors, and the majority are of the diffuse intrinsic pontine gliomas (DIPG) subtype. The latter have a peak age of onset in middle childhood (6–7 years) and are rare in adults. DIPG has a dismal prognosis, with a median survival of less than 1 year and fewer than 20 % of patients alive at 2 years. There have been minimal treatment advances over the past decades [20, Class III]. Improvements are hampered by the limited knowledge of DIPG biology, as historically this was a diagnosis made radiographically. Recently, there has been increased interest in performing stereotactic biopsy of DIPG [21]. This has been demonstrated to be safe and feasible by experienced pediatric neurosurgeons at high-volume academic centers (primarily in Europe) [22]. The ability to acquire tissue prior to therapy has yielded valuable data toward improving our understanding of the biology of DIPG and to help define genetic targets with therapeutic relevance [14, 15••, 21, 23, 24].

The ongoing challenges in treating childhood gliomas are currently being addressed by advances in diagnostic imaging, radiotherapy, and surgical techniques; however, future advances and improvements in outcomes are dependent on increasing participation in cooperative group studies in which tissue submission and biological correlates are incorporated into trial design.

Treatment

Diagnostic procedures

  • Magnetic resonance imaging (MRI) is the modality of choice for the diagnostic evaluation of childhood gliomas. It is important in disease staging, preoperative navigation, radiation planning, and in monitoring treatment response. LGGs are typically hypo-intense on T1-weighted, and hyper-intense on T2-weighted imaging. Pilocytic astrocytomas have a characteristic well-circumscribed cystic component with a contrast-enhancing mural nodule, while diffuse fibrillary astrocytomas have little enhancement and often have indistinct borders. HGGs tend to have a more heterogeneous appearance and are hypo-/iso-intense on T1 weighted imaging, hyper-intense on T2 sequences, and often show contrast enhancement with or without evidence of necrosis. Advanced imaging of gliomas can aid in preoperative diagnosis and may be prognostic. In comparison to LGGs, HGGs have increased choline peaks and elevated choline:N-acetylaspartate ratio on MR spectroscopy (MRS). They also show restricted diffusion and increased blood flow on diffusion/perfusion-weighted sequences [25, 26]. DIPGs have a classic MRI appearance of an expansile lesion centered in the pons that is hypo-/iso-intense on T1-weighted, hyper-intense on T2-weighted, and variably enhancing on post-contrast imaging. Specific MRI sequences, including multi-voxel MRS and dynamic susceptibility contrast (DSC) MRI of DIPG can predict short or long survival interval from diagnosis, though have yet to impact treatment decisions [27]. Metastatic disease is not uncommon for HGG; therefore, imaging of the entire neuraxis is indicated at diagnosis. In contrast, low grade and brainstem gliomas rarely spread, so spine imaging is not typically indicated unless multi-focal disease is seen in the brain [28].

Interventional procedures

Surgery

Surgery plays a critical role in the management of childhood gliomas by establishing a tissue diagnosis, decreasing compression of surrounding structures to alleviate symptoms and relieving obstruction of cerebrospinal fluid (CSF) flow. Surgery alone is essentially curative for LGGs when a complete resection can be achieved [29, Class III]. In addition to extent of resection, tumor grade is a prognostic factor for overall survival (OS) and progression free survival (PFS) [30, Class I]. Many LGGs, however, are located in deep midline structures and inaccessible for a complete resection. The surgical management of these tumors has greatly benefitted from new, minimally invasive neurosurgical techniques that include stereotactic biopsy for diagnosis and endoscopic third ventriculostomy for CSF diversion [31, 32]. Once diagnosis has been established appropriate adjuvant therapy can be delivered. The degree of surgical resection for HGGs is a critical component of therapy and predictive of improved event free survival. Specifically, patients achieving a gross total resection have a significantly better outcome than those with residual tumor [7•, Class III, 33, Class I]. The role for surgery in DIPGs is evolving. In the recent past, the diagnosis was made solely on imaging, and neuro-oncologists and radiation oncologists were comfortable treating without tissue. However, as survival has not improved despite multiple clinical trials, there is a push to biopsy tumors in order to provide tissue for biology, target identification, and treatment stratification [15••, 22].

Complications:

General neurosurgical complications include the risk of bleeding, infection, stroke, brain edema, and permanent neurological dysfunction. Depending on location, additional risks of vision loss, posterior fossa syndrome (mutism, oropharyngeal dyspraxia, emotional lability, ataxia, truncal hypotonia) [34], cerebral salt wasting, transient diabetes insipidus, and syndrome of inappropriate antidiuretic hormone secretion (SIADH) exist.

Special points:

The one exception in which surgery does not play much of a role is optic pathway/hypothalamic gliomas in patients with NF-1. These tumors have classic imaging features, a fairly reliable clinical course, and can be treated without a tissue diagnosis [35].

Radiation

Radiation therapy for childhood gliomas is a critical treatment modality for HGG and DIPG, but should be used judiciously in LGG, and requires a careful weighing of its risks to the developing brain. As long-term survival among children with CNS malignancies increases, the emergence of late morbidity related to radiotherapy is more apparent (see Survivorship and Surveillance) [36]. New developments in radiation treatment planning and techniques, including proton radiation therapy, should significantly decrease these late effects by limiting dose to developing normal tissues [37, 38]. In general, focal radiation with 54 Gy is standard dosing for glioma (boosted to 59.4 Gy for residual/unresectable HGG). The treatment margin beyond the resection cavity varies from 1 cm (for LGG) to 2 cm (for HGG) for treating marginal microscopic residual disease. The use of radiotherapy for LGG is discouraged upfront, and chemotherapy is often used to delay or avoid radiation, taking into account the age of the child, prior therapy, location of the tumor, and the presence of NF-1 [39–41, Class III]. When needed however, it is an effective therapy and results in long-term disease control [42]. Radiation therapy for HGG even after a complete surgical resection is standard practice and part of a multimodality treatment plan in children older than 3 years. The addition of chemotherapy to radiation has been accepted practice for many years; however, a “standard regimen” has not been identified and studies of new drug combinations are ongoing. In infants with HGG, chemotherapy alone is frequently employed to defer radiation until tumor progression or an older age is reached [43, Class III, 44]. For DIPG, radiation is the only therapeutic modality consistently demonstrating clinical and radiographic response, though without long-term disease control. Approximately 70 % of children have neurologic improvement and 40–60 % objective tumor response after conventional radiotherapy (54 Gy over 30 daily 180 cGy fractions) [20, Class III]. Higher dose, hyperfractionated (twice daily doses) RT to 64.8–78 Gy or hypofractionated (larger daily dose given over fewer days) has not demonstrated improved survival (reviewed in [45]). Additional radiation-based therapies, such as concurrent chemo-radiotherapy and radiation sensitizers, have no proven benefit [46–48, Class III], with the past 30 years of published clinical trials consistently reporting 2-year OS across studies of < 25 % [20, 45, Class III].

Complications:

Acute complications of radiation include nausea, vomiting, radiation dermatitis, alopecia, and fatigue/somnolence syndrome. Long-term late effects include radiation necrosis, hearing loss, bone hypoplasia, stroke, second malignant neoplasms, endocrine dysfunction and neurocognitive effects (highlighted in Survivorship and Surveillance section).

Special points:

The advantage of proton radiation therapy over standard techniques is based on the physical characteristics of the proton particle most notably the sharp dose fall-off beyond the target (tumor) along the particle path, substantially sparing normal adjacent tissue from unnecessary radiation exposure [37, 38].

Pharmacologic treatment

Chemotherapy

Chemotherapy is the first line of treatment for LGGs that are not amenable to surgical resection without high-risk of neurologic morbidity or for progressive or symptomatic tumors after sub-total resection or observation; especially in younger children or children with NF-1, who are at higher risk for radiation induced morbidity. Typical locations for such LGGs include the deep midline supratentorial region and optic pathway/hypothalamus. In a significant proportion of these cases, chemotherapy will delay or eliminate the need for radiation therapy. The combination of carboplatin and vincristine (CV) is widely used as front-line therapy for LGG, and has demonstrated disease control with objective responses and prolonged PFS [41, Class I]. An alternative regimen of thioguanine, procarbazine, lomustine, and vincristine (TPCV) [49] is also active against LGG, and when compared to CV in a randomized trial, may have superior long-term event-free survival [50••, Class I]. Cisplatin and etoposide has been used as front-line therapy in some European countries with very high response and long-term disease control [51, Class III]; however, many investigators are reluctant to use this combination because of the risk of high-frequency hearing loss, though this may be avoided with lower dosed regimens [52]. Single-agent weekly vinblastine, initially used as an alternative for patients with allergy to carboplatin [53, Class III], has reasonable response rates in patients who have progressed on first-line therapy and should be considered in advance of radiation for LGGs that have failed CV or TPCV [54, Class III]. Other retrieval regimens for recurrent LGG under study include bevacizumab in combination with irinotecan [55], and TMZ, though no randomized controlled trials have been performed to compare regimens [56]. In the treatment of HGG, early cooperative group trials demonstrated that the addition of chemotherapy (prednisone, lomustine, vincristine) to involved field radiation therapy improved survival over radiation alone [57, Class I]. Unfortunately, a “standard” chemotherapy regimen has not been clearly identified and studies comparing to historical controls may be inaccurate given advances in neuropathology [7•, Class III, 58]. There is a statistically significant improvement in survival with the addition of TMZ to radiation in adult HGGs [59, Class III]; however, the survival benefit of TMZ in childhood HGG was less clear in a recent, pediatric, non-randomized phase II trial [7•, Class III]. Temozolomide is still used as a backbone for HGG therapy; however, it is being studied in combination with targeted agents including bevacizumab, a VEGF inhibitor, and veliparib, a poly(ADP-ribose) polymerase (PARP) inhibitor; both protein biomarkers expressed in aggressive tumors and associated with resistance to therapy [24, 60]. Bevacizumab and irinotecan as a maintenance therapy with or without TMZ is also used frequently in pediatric HGG, though the only controlled studies have been in the relapsed setting, where it has had minimal efficacy compared to published trials in adults [61–64, Class III]. The use of high-dose chemotherapy with autologous stem cell support has been used in HGG to avoid radiation in the very young [43, Class III] and as a salvage regimen for recurrent HGG [65, Class III]. There is evidence of response, but it is unclear if this will translate into prolonged improved survival [44]. The current Children’s Oncology Group (COG) clinical trial for newly diagnosed HGG is a phase II randomized “pick the winner” approach, comparing two experimental chemoradiotherapy arms (suberoylanilide hydroxamic acid (SAHA) + radiation versus bevacizumab + radiation) to the standard arm of temozolomide given with local radiation. This is followed by maintenance chemotherapy with bevacizumab and temozolomide given to all subjects. This approach will evaluate whether we can improve upon TMZ+ radiation as the “de facto standard” chemoradiotherapy regimen for HGG. The use of conventional chemotherapy for DIPG is currently not indicated, and multiple clinical trials investigating dose, schedule, and delivery of agents to intensify or modulate the effects of local irradiation have not resulted in improved outcomes. Recent published efforts have focused on the use of radiosensitizers, such as motexafin-gadolinium [66], and rationally chosen molecularly targeted agents against PDGFR, RAS signaling, EGFR and vascular growth factors, but none of these clinical trials has demonstrated clinical benefit over radiation therapy alone [67–70, Class III]. A newer approach involves creating cancer vaccines to induce systemic immunity against tumor-specific or tumor-enriched antigens. Currently enrolling, early phase, pediatric DIPG trials utilizing this strategy use peptide-based vaccines against pooled glioma associated antigens [71] or against the mutated EGFR (EGFRvIII), known to be expressed in DIPG [72].

Complications:

Nausea, vomiting, hair loss, myelosuppression, immunosuppression with infection risk, allergic reactions (especially with carboplatin), ototoxicity (cisplatin), and risk of secondary malignancy with alkylators (procarbazine, lomustine, temozolomide), which may be higher in patients with NF-1.

Special points:

Given the dismal prognosis for both HGG and DIPG despite several decades of clinical trials, the importance of obtaining tissue specimens for biological studies to identify new targets and molecularly stratify patients is imperative.

Emerging therapies

Newer approaches to glioma treatment are needed for children with recurrent/refractory LGGs, or newly diagnosed HGG and DIPG, and rely on understanding and identifying biologically relevant targets to allow rationale, tumor-specific therapy. Tumor targets currently under investigation in the field of glioma biology include cell-signaling pathway receptors, effectors, and regulators of gene expression. In treating recurrent LGG, co-targeting mTOR and EGFR, both of which are highly expressed in LGG, resulted in prolonged stable disease in preliminary studies [73]. Oncogenic BRAF signaling, defined by the novel KIAA1549:BRAF fusion gene in LGG [3] provides multiple targets for inhibition, including BRAF and MEK (agents currently in clinical trial), Biologically relevant targets currently under pre-clinical or early clinical investigation in HGG and/or DIPG include EGFR (Erlotinib), PDGFRb/VEGFR/c-kit (sunitinib), c-MET (crizotinib), PDGFR (crenolanib) [15••, 24], and aurora kinase B [23].

Pediatric considerations

Survivorship and surveillance

Neurological deficits

  • Children with gliomas are at risk for neurological deficits as a consequence of multiple factors, including the presence of tumor in a specific location, prolonged hydrocephalus, surgical resection, and the effects of therapy. These may include sensory or motor deficits, vision loss, difficulty with speech and coordination. Physical, occupational, and speech therapies should be coordinated and tailored to individual needs.

Neurocognitive deficits

  • As treatment for pediatric gliomas advances and becomes more targeted and precise, the risk of damage to the developing normal brain should be lessened. However, studies of long term survivors of childhood glioma, including a population of children treated from 1985 to 1992 for HGG (median follow-up of 15.1 years), demonstrate impaired neuropsychological functioning, including low average mean intellectual function, and below average or impaired executive functioning, visual learning and memory, and psychomotor processing speed [74]. In this study, independent risk factors for lower neuropsychological functioning included midline or infratentorial tumor location, female sex, and young age at treatment [74]. Similarly, several studies of children with cerebral hemisphere and midline (optic pathway/hypothalamic) LGGs treated with radiotherapy demonstrate significant long-term intellectual impairment [75–77]. Patients who underwent radiotherapy at very young ages (median age 4 years in one study, mean age 5.7 years in another) had the most significant decline in cognitive function [75, 77, Class IV]. A recent study examining the intellectual outcomes in children with LGG (without NF-1) treated first-line with chemotherapy and subsequent radiotherapy at relapse, demonstrated normal range neurocognitive testing in those who avoided radiotherapy at a young age [78]; highlighting the important approach of deferring or delaying radiotherapy in the youngest children. Neuropsychologists are a critical part of the treatment team, and prospective longitudinal testing is essential in order to better understand the impact of disease and treatment on neurocognitive development, and to timely target remedial interventions to improve outcomes. Systematic longitudinal assessment of the neurocognitive functioning of survivors of childhood gliomas should be a standard part of long-term follow-up care [36].

Neurovascular events

  • Cranial radiation increases the risk of vascular injury including large-vessel stenosis, atherosclerosis, and vascular insufficiency. The rate of neurovascular events (stroke or transient ischemic attack) is 100-fold higher in children radiated for brain tumors compared with the general pediatric population [79]; the highest risk for stroke occurs in patients who receive radiation to the circle of Willis, common for inoperable midline gliomas [79]. The occurrence of stroke continues to increase over time among brain tumor survivors, with a mean time of 13.9 years from brain tumor diagnosis to “late occurring stroke” [80]. This wide range of events emphasizes the need for long-term follow-up in a multi-disciplinary survivorship program, and justifies the importance of continued neuro-imaging surveillance [81].

Endocrine dysfunction

  • Precocious or delayed growth or pubertal development may be a presenting sign of a hypothalamic glioma. Anterior pituitary dysfunction resulting in deficits in thyroid, growth, cortisol, and sex hormones are frequent after cranial radiation. Concerns that growth hormone replacement may increase the risk of tumor growth or recurrence are not borne out by large studies (reviewed in [82]).