1.1 Introduction

Tumors of the central nervous system (CNS) comprise a broad and diverse collection of neoplasms within pediatric oncology. Yet when taken together pediatric brain and spine tumors represent the most common childhood cancer with an incidence of 5.57 per 100,000 annually and are a leading cause of cancer-related death in patients under 19 years of age (Ostrom et al. 2014; Siegel et al. 2015). Factors such as genetic predisposition, age, and sex play an increasingly significant role in understanding presentation, management, and etiology of childhood brain tumors. Although long-standing observations regarding general patterns of CNS tumors continue to be clinically useful, the introduction of molecular subtypes, such as in medulloblastoma and ependymoma, and the discovery of epigenetic regulators, such as in diffuse intrinsic pontine gliomas (DIPG) and other diffuse midline gliomas with H3K27M mutations, have repurposed epidemiological findings and reconceptualized CNS tumor classification (Louis et al. 2016). The elucidation of the molecular profile of pediatric CNS tumors has made it clear that epidemiology, viewed through a prism of genetics and epigenetics, can offer even greater insights into this incredibly challenging group of tumors. Epidemiology today considers not only environmental, parental, and birth factors that may increase the risk of pediatric CNS tumors, but also germline and molecular features that are causal or pathognomonic of tumor types and subtypes.

1.2 Astrocytomas and Other Gliomas

The gliomas are a heterogeneous group of tumors, comprised mostly of astrocytomas. Pediatric astrocytomas are divided into four grades by the World Health Organization (WHO), with pilocytic astrocytomas (WHO grade I) being the most common subtype of pediatric CNS tumor, comprising approximately 15% (Ostrom et al. 2014; Louis et al. 2007). The incidence of pilocytic astrocytomas in children in England and the USA is 0.75–0.97 per 100,000, and these tumors have an exceedingly low incidence of metastasis or malignant transformation (Ostrom et al. 2014; Stokland et al. 2010; Fisher et al. 2008; Arora et al. 2009). Although they may occur in any CNS location including the spine, they most commonly arise from the posterior fossa, optic pathway and hypothalamus, or brain stem (Fernandez et al. 2003; Gajjar et al. 1997; Hayostek et al. 1993; Khatib et al. 1994). Diffuse astrocytomas (WHO grade II), anaplastic astrocytomas (WHO grade III), and glioblastomas (WHO grade IV) have an incidence of 0.27, 0.08, and 0.15 per 100,000 children 0–14 years of age, respectively. Low-grade gliomas, which are comprised of WHO grade I and II astrocytomas as well as WHO grade I gangliogliomas, most commonly present with greater than 6 months of symptoms (Fisher et al. 2008). The incorporation of molecular characteristics in the 2016 WHO classification of tumors of the CNS will assist in a deeper epidemiological understanding by addressing distinct biologic entities, such as diffuse gliomas with IDH mutations and diffuse midline gliomas with H3K27M mutations (Louis et al. 2016).

Children with pilocytic astrocytomas have excellent outcomes of >96% overall survival (OS) at 10 years, and patients with subtotal resections do not do significantly worse than patients with gross total resections (Ostrom et al. 2014; Gajjar et al. 1997). Posterior fossa tumors are common in children, with pilocytic astrocytomas being the second most common tumor arising in that location, behind only medulloblastoma; mean age of occurrence is 7.1 years (Smoots et al. 1998). Up to 60% of pilocytic astrocytomas are associated with a KIAA1549:BRAF fusion, which is associated with a better outcome (Becker et al. 2015; Jones et al. 2008). Optic pathway and hypothalamic astrocytomas are most often pilocytic astrocytomas, but other subtypes of low-grade gliomas also account for a small number of cases (Hoffman et al. 1993; Laithier et al. 2003). Optic pathway gliomas (OPGs) occur in approximately 15% of patients with neurofibromatosis type 1 (NF1), though they most often occur sporadically (Listernick et al. 1989). OPGs are reported to have a broad median age between 4.3 and 8.8 years, and those occurring in patients with NF1 present at a significantly earlier age than sporadic cases (Listernick et al. 1989; Nicolin et al. 2009; Singhal et al. 2002; Ahn et al. 2006; Janss et al. 1995; Khafaga et al. 2003; Jahraus and Tarbell 2006; Avery et al. 2011). The variation in age of presentation may be secondary to the presence of a cancer predisposition syndrome in NF1 patients, as well as the practice of asymptomatic surveillance imaging in that group, while 90% of sporadic cases present with new neurologic symptoms. Subependymal giant cell astrocytomas (SEGAs) are another WHO grade I astrocytoma subtype that develop almost exclusively in patients with tuberous sclerosis (TS), which occurs in 1 in 5600 live births (O’Callaghan et al. 1998). Five to twenty percent of patients with TS develop SEGAs, often in adolescence, but congenital cases have also been reported (Adriaensen et al. 2009; O’Callaghan et al. 2008; Hahn et al. 1991).

Several WHO grade II subtypes can be distinguished by histology and presentation. Pilomyxoid astrocytomas (WHO grade II) have a more aggressive course than pilocytic astrocytomas (WHO grade I), a greater propensity for growing in the hypothalamochiasmatic region, and often present earlier with a mean age of 3.3 years (Bhargava et al. 2013). Pleomorphic xanthoastrocytomas (WHO grade II) are typically located in the superficial temporal lobe; they classically present with seizures and have a median age at diagnosis of 20.5 years and an approximately 75% overall survival (Gallo et al. 2013; Perkins et al. 2012). These can rarely transform into a high-grade glioma.

Low-grade gliomas of the brain stem can be pilocytic astrocytomas or gangliogliomas, which typically occur dorsally and have the possibility of long-term cure. WHO grade II, III, and IV gliomas of the brain stem have dismal outcomes and together comprise diffuse intrinsic pontine glioma (DIPG). The 2016 WHO classification has adjusted that nomenclature in favor of diffuse midline gliomas, as diffuse gliomas of the pons, thalamus, and spinal cord may form a more biologically distinct category when H3K27M mutations are present (Louis et al. 2016; Shankar et al. 2016).

DIPGs arise most commonly in the ventral pons and comprise 10–15% of all pediatric CNS tumors and 80% of brain stem gliomas, affecting roughly 300 children in the USA each year (Ostrom et al. 2014; Ramos et al. 2013; Smith et al. 1998). Males and females are affected equally and the median age of presentation is 7 years (Lassiter et al. 1971; Lober et al. 2014; Veldhuijzen van Zanten et al. 2014). Presentation usually consists of a classic triad of ataxia, cranial nerve palsies, and pyramidal tract signs developing over 1 month, although atypical cases can present more slowly over several months (Fisher et al. 2000). It is now recognized that approximately 17% of patients undergo both local and distant neuraxis dissemination by 15 months, which is not far beyond the median overall survival of patients with DIPG, as only 10% of patients survive beyond 2 years and only 2–3% are considered long-term survivors (Gururangan et al. 2006; Hargrave et al. 2006; Jackson et al. 2013). Recently, 80% of DIPGs have been found to harbor mutations in K27M of histone 3.1 or 3.3, which are associated with mutations in ACVR1 and p53, respectively (Taylor et al. 2014; Wu et al. 2012).

High-grade gliomas (HGGs) occur much more frequently in adults, with an increasing incidence with age to a peak between the ages of 75 and 85 years (Ostrom et al. 2014). The outcomes of patients with high-grade gliomas appear to be inverse to patient age, as 5-year overall survivals for children less than three and those 3–14 years of age are 31–66% and 19%, respectively (Mathew et al. 2014). Glioblastoma has been reported in classic CNS tumor predisposition syndromes, such as neurofibromatosis, Li–Fraumeni, and Turcot syndromes, as well as in several genitourinary syndromes, such as Turner and Mayer–Rokitansky–Küster–Hauser syndrome, though the majority of cases are believed to be sporadic (Hanaei et al. 2015; Jeong and Yee 2014; Macy et al. 2012; Gonzalez and Prayson 2013).

1.3 Embryonal Tumors

Embryonal brain tumors are a diverse group of aggressive neoplasms, including medulloblastoma, primary neuroectodermal tumors (PNET), atypical rhabdoid/teratoid tumors (ATRT), and pineoblastoma, which share high mitotic activity and a predilection for dissemination throughout the neuraxis, and are all WHO grade IV (Louis et al. 2007). They account for 15% of CNS tumors in patients 0–14 years of age and 12% in those 0–19 years of age, with incidences of 0.78 and 0.64 per 100,000, respectively; these incidences have remained unchanged since at least 1990 (Ostrom et al. 2014; Johnston et al. 2014). Embryonal CNS tumors rarely occur outside of childhood with the median age at presentation being 7.3 years, and 44% of them being diagnosed between the ages of 4 and 9 years (Ostrom et al. 2014; Kool et al. 2012). Medulloblastomas, the most common malignant brain tumor in pediatrics, histologically appear as PNETs specifically arising in the posterior fossa (Northcott et al. 2011). A minority of medulloblastoma cases have been reported in patients with genetic predisposition syndromes such as Gorlin, Turcot B, Li–Fraumeni, ataxia telangiectasia, Nijmegen breakage, Rubenstein–Taybi, and Coffin–Siris syndromes (Distel et al. 2003; Hart et al. 1987; Larsen et al. 2014; Skomorowski et al. 2012; Taylor et al. 2001; Rogers et al. 1988). Overall, there is a male predominance of 1.5:1, with females reported to have superior outcomes, although again this is likely subgroup dependent, as there are fewer females in the higher risk Group 3 and 4, while more young females have sonic hedgehog (SHH) driven tumors (Louis et al. 2007; Northcott et al. 2011). Historically, patients clinically classified as average-risk had a 5-year OS of roughly 85%, while high-risk patients suffered poorer outcomes with near 70% OS and patients with large-cell anaplastic histology had particularly dismal outcomes (Kool et al. 2012; Gajjar et al. 2006; Packer et al. 2006; Tarbell et al. 2013; Ramaswamy et al. 2013). Overall, long-term survival in patients with medulloblastoma is achieved in only 66% of patients, with 10% suffering from secondary malignancies, 32% of which are secondary brain tumors (Ning et al. 2015).

Although particular subsets of medulloblastoma have long been suspected to behave differently, it is now commonly accepted that there are four distinct molecular subgroups: WNT, SHH, Group 3, and Group 4, which account for 11%, 28%, 27%, and 34% of cases, respectively (Kool et al. 2012; Northcott et al. 2011; Badiali et al. 1991). Prodromes may vary among the groups, ranging from only 2 weeks in patients with SHH tumors, to 4 weeks in patients with Group 3 tumors and 8 weeks in patients with WNT or Group 4 tumors (Ramaswamy et al. 2014). Furthermore, age of presentation varies as the incidence of SHH medulloblastomas is bimodal, peaking under 3 years and again over 15 years of age (Northcott et al. 2011). WNT and Group 4 both peak around age 11, but WNT tumors are essentially absent in infancy (Kool et al. 2012; Northcott et al. 2011). WNT tumors have no gender predominance, are the least frequent subgroup, and experience the best outcomes with greater than 90% overall survival (Ellison et al. 2011, 2011). Outcomes in patients with SHH tumors are inferior, though strongly age-dependent as the 10-year OS is 77% and 51% in infants and children, respectively (Kool et al. 2012; Ramaswamy et al. 2013). Despite presenting with metastatic disease in 17% of infants and 22% of children, SHH tumors most often recur locally (Kool et al. 2012; Ramaswamy et al. 2013). Group 3 and Group 4 occur nearly twice as often in males, accounting for the male predominance in medulloblastoma as a whole. Forty-seven percent of Group 3 medulloblastomas present with metastases and, while they do not have significantly worse prognoses than those without metastases, this subgroup overall suffers the poorest outcomes with long-term survival in less than 50% of patients (Kool et al. 2012; Northcott et al. 2011). Group 4 patients, on the other hand, have significantly different outcomes associated with the presence of metastases, ranging from nearly 40% OS (metastases present) to greater than 70% (metastases absent) (Kool et al. 2012). In patients that experience recurrence, the molecular subgroup remains consistent, and although outcomes are uniformly poor, Group 4 patients have the longest survival following recurrence (Ramaswamy et al. 2013).

Atypical teratoid/rhabdoid tumors (ATRTs) are embryonal CNS tumors with rhabdoid features that were initially described in the 1990s (Zuccoli et al. 1999). Since their initial description their incidence has increased, while the incidence of other PNETs has declined, more likely representative of a change in classification than a change in biological patterns of disease (Ostrom et al. 2014). The incidence of ATRT in childhood is approximately 0.1 per 100,000 with a peak between 1 and 2 years of age and no gender predisposition observed in the USA (Ostrom et al. 2014, b; Hilden et al. 2004; von Hoff et al. 2011; Woehrer et al. 2010). They account for 10% of CNS tumors in patients less than 1 year of age, but only 1.6% of all childhood brain tumors (Ostrom et al. 2014). The wide range of reported OS, between 28 and 48%, may be affected by delays in appropriate diagnosis, as one report noted a 5-year OS of only 15% in patients who were initially misdiagnosed (Ostrom et al. 2014; Hilden et al. 2004; von Hoff et al. 2011; Woehrer et al. 2010; Athale et al. 2009; Lafay-Cousin et al. 2012). Most reports conclude that metastatic disease at presentation is not prognostic, while descriptions of the prognostic impact of age differ (Ostrom et al. 2014; Hilden et al. 2004; von Hoff et al. 2011; Woehrer et al. 2010; Athale et al. 2009; Lafay-Cousin et al. 2012). The location of ATRTs, however, does appear to change with age, as patients under 1 year of age most commonly have infratentorial disease and the incidence of supratentorial disease increases with age (Ostrom et al. 2014). The characteristic loss of INI1 in these tumors is most commonly somatic, although germline mutations have been reported and can result in a rhabdoid tumor predisposition syndrome (RTPS) (Sredni and Tomita 2015; Taylor et al. 2000). The development of ATRTs has also been associated with low birth weight and twin pregnancies (Heck et al. 2013).

Pineoblastomas are malignant tumors of the pineal gland that, like other PNETs, are histologically similar to medulloblastomas, but display a distinct biology (Li et al. 2005). While some pineal tumors, such as germ cell tumors, occur more commonly in males, reports suggest pineoblastoma may be more common in females (Villa et al. 2012; Fauchon et al. 2000). Although patients with bilateral retinoblastomas may develop a pineoblastoma, “trilateral retinoblastoma” occurs in only 1% of patients with bilateral retinoblastoma and only in the setting of germline mutations (Ramasubramanian et al. 2013). While the majority of pineoblastoma cases appear sporadic, cases also have been reported as part of Turcot syndrome and with germline DICER1 mutations (Ikeda et al. 1998; Gadish et al. 2005; Sabbaghian et al. 2012).

1.4 Ependymoma

Virchow initially described ependymomas in the nineteenth century as CNS tumors originating from the walls of the ventricular system (Virchow 1863–67). Though ependymomas likely consist of several discrete subgroups that can be distinguished by location and molecular profile, most reports evaluate ependymomas as a whole or by grade, leaving their epidemiologic understanding incomplete. Ependymoma incidence in the USA is 0.3 and 0.29 per 100,000 children aged 0–14 years and 0–19 years, respectively, and has not increased since 1973; nearly one-third of cases occur in children under the age of 4 years (Ostrom et al. 2014; McGuire et al. 2009). Although 46% of ependymomas in adults are spinal, location varies according to age in children (Vera-Bolanos et al. 2015). The mean age for spinal, supratentorial, and infratentorial ependymomas are 12.2, 7.8, and 5 years, respectively (McGuire et al. 2009). The gender incidence may be affected by age and location, as the overall male-to-female ratio is 1.3:1, though males are more commonly affected by supratentorial ependymomas (1.4:1) and less commonly affected by spinal ependymoma (0.7:1) than females (McGuire et al. 2009; Dohrmann and Farwell 1976). Presentation with metastatic disease is rare in pediatric ependymomas but is more common in infants, although reports vary on whether supratentorial or infratentorial tumors are more likely to metastasize (Zacharoulis et al. 2008; Allen et al. 1998).

Currently, the treatment of ependymoma primarily varies according to age, grade, and location. In 2015, a new molecular classification was proposed though it has yet to be validated. It divides ependymomas into anatomical compartments: supratentorial (ST), posterior fossa (PF), and spinal (SP); tumors in each compartment are then divided into one of three subgroups: a subependymoma group and two other genetic or epigenetic subgroups (Pajtler et al. 2015). Supratentorial ependymomas are distinguished by either RELA fusions (ST-EPN-RELA), which occur at a median age of 8 years and result in frequent disease progressions, or YAP1 fusions (ST-EPN-YAP1), which occur at a median age of 1.4 years (Pajtler et al. 2015). Posterior fossa ependymomas are subdivided into those with a CpG methylator phenotype (PF-EPN-A), which account for 48% of all pediatric ependymomas and experience poor outcomes, and those that are not hypermethylated (PF-EPN-B), which often occur in older patients (EPN-PFB) (Pajtler et al. 2015; Parker et al. 2014; Witt et al. 2011).

Although histologic classification of WHO grade II or III in pediatric ependymoma may not offer prognostic significance, several WHO grade I subsets are clearly less aggressive neoplasms (Perilongo et al. 1997; Ross and Rubinstein 1989; Robertson et al. 1998). Subependymomas represent less than 1% of CNS tumors in children, are designated WHO grade I, and have essentially no metastatic potential (Scheinker 1945; Ragel et al. 2006). Myxopapillary ependymomas, also WHO grade I, have a median age of presentation of 36 years, yet are not uncommon in children with reports of patients as young as 6 years old being affected (Barton et al. 2010; Woesler et al. 1998). Despite their WHO grade I designation, the pediatric variant may be more aggressive than that seen in adults with a suggestion of dissemination in as many as 58% of patients (Fassett et al. 2005). Neurofibromatosis type II (NF2) is the most common hereditary predisposition for ependymoma, most often causing intramedullary spinal tumors of the cervical spine (Bianchi et al. 1994; Plotkin et al. 2011). Pediatric ependymomas have also been reported in Turcot B, MEN1, and Li–Fraumeni syndromes (Chan et al. 1999; Metzger et al. 1991).

1.5 Germ Cell Tumors

Germ cell tumors (GCTs) are a heterogeneous group of cancers with variable classification and nomenclature depending on the particular organ involvement. In the CNS, they are divided into germinomas, non-germinomatous germ cell tumors (NGGCT), and teratomas. The most common locations for GCTs are the suprasellar and pineal regions. GCTs account for 4% of pediatric CNS tumors with an incidence of 0.2 and 0.22 per 100,000 in children aged 0–14 and 0–19 years, respectively (Ostrom et al. 2014). Males account for 76% of all CNS GCTs, 58% of pituitary GCTs, and a remarkable 93% of pineal GCTs (Goodwin et al. 2009). In both sexes there is a small spike at birth and a much greater spike in adolescence with incidences peaking at roughly age 15. Race also influences incidence patterns, as in the USA nearly 20% of patients were Asian or Pacific Islander with an incidence of 0.26 per 100,000, double the 0.13 per 100,000 in white children 0–15 years of age (Goodwin et al. 2009). CNS GCTs also account for a greater percentage of pediatric CNS tumors in Japan, Korea, Taiwan, and China at 7.8%, 11.2%, 14%, and 7.9%, respectively (Cho et al. 2002; Mori and Kurisaka 1986; Wong et al. 2005; Zhou et al. 2008). Klinefelter syndrome is associated with the development of pediatric germ cell tumors including intracranial germinomas (Arens et al. 1988). Down syndrome and NF1 have also been reported in patients with intracranial germinomas (Hashimoto et al. 1995; Wong et al. 1995).

1.6 Family History

Despite the increasing awareness of CNS tumor genetic predispositions, further discussed within another chapter, there is still little evidence of the development of CNS tumors in the parents or siblings of affected children. The studies reporting increased pediatric CNS tumor incidence among siblings have been plagued by small numbers and an inability to exclude genetic predisposition syndromes; however, a larger Nordic cohort of patients showed no association among siblings outside of genetic predisposition syndromes (Draper et al. 1977; Farwell and Flannery 1984; Miller 1971; Winther et al. 2001). There have been several reports regarding the association of parental age with pediatric CNS tumors: two studies identified increased parental age as a risk factor, while one found only advanced maternal age to be a significant risk (Hemminki et al. 1999; Johnson et al. 2009; Yip et al. 2006). A review of Sweden’s Family-Cancer Database, consisting of over 13,000 CNS tumor diagnoses, found that oldest siblings were at increased risk for several childhood malignancies and this risk increased with the number of younger siblings (Altieri et al. 2006). The existence of three or more younger siblings resulted in a relative risk of 1.34, 2.3, 2.61, and 3.71 of astrocytoma, medulloblastoma, ependymoma, and meningioma, respectively (Altieri et al. 2006).

1.7 Birth History

As early as 1968, Kobayashi had published a report of the association between congenital anomalies and childhood cancer (Kobayashi et al. 1968). A review of 90,400 children found patients with congenital anomalies had a risk ratio of 5.8 (CI 3.7–9.1) of developing cancer in their first year of life (Agha et al. 2005). The risk was also increased for central nervous system and sympathetic nervous system tumors individually at a risk ratio of 2.5 (CI 1.8–3.4) and 2.2 (CI 1.4–3.4), respectively. A Bjørge et al. study of 5.2 million children and their families in Norway and Sweden also found patients with congenital anomalies had an increased cancer risk that extended into early adulthood (Bjorge et al. 2008). Furthermore, patients with CNS malformations were also at the highest risk of developing CNS malignancies, with a standardized incidence rate (SIR) of 58 (CI 41–80) and 8.3 (Louis et al. 2007; Stokland et al. 2010; Fisher et al. 2008; Arora et al. 2009; Fernandez et al. 2003; Gajjar et al. 1997; Hayostek et al. 1993; Khatib et al. 1994; Smoots et al. 1998; Becker et al. 2015; Jones et al. 2008; Hoffman et al. 1993) in Norway and Sweden, respectively. To assess potential cancer risk associated with congenital anomalies even outside of the setting of chromosomal defects, a review of the California Cancer Registry (CCR) found that between 1988 and 2004, children with congenital anomalies without chromosomal defects had a 1.8-fold increased risk of CNS cancer (Fisher et al. 2012). A further examination found a particularly increased risk in medulloblastoma (OR 1.7, CI 1.1–2.6), PNET (OR 3.64, CI 1.5–8.6), and germ cell tumors (OR 6.4, CI 2.1–19.6), as well as an increased risk in mothers with greater than two fetal losses after 20 weeks of gestation (OR 3.13, CI 1.3–7.4) (Partap et al. 2011).

Many large studies have evaluated the impact of birth weight on the risk of developing CNS tumors, with several suggesting an increased birth weight carries a greater relative risk, although the most common specific tumors types varied among studies (Bjorge et al. 2013; Harder et al. 2008; MacLean et al. 2010; Milne et al. 2008; Schmidt et al. 2010). In an examination matching each case (17,698) to 10 controls, Bjørge found an increased childhood cancer risk for higher birth weight infants, and also infants with larger head circumferences (Bjorge et al. 2013). Additionally, in an evaluation of Nordic children, Schmidt found a gestational age-adjusted birth weight of greater than 4.5 kg increased the risk of all CNS tumors (OR 1.27, CI 1.03–1.6), with the greatest increase among embryonal tumors (Schmidt et al. 2010). When 3733 CNS tumors from the CCR were matched to controls, Maclean et al. found an increased birth weight of 4 kg associated with an increased risk of CNS tumors, especially HGGs (MacLean et al. 2010). A meta-analysis of eight studies found that increased birth weight was associated with increased incidence of astrocytomas and medulloblastomas, but not ependymomas (Harder et al. 2008). Conversely, a study of over 600,000 live births in Western Australia between 1980 and 2004 found no association between birth size and the development of CNS tumors prior to age 14 (Milne et al. 2008).

1.8 Immune System

Although allergic conditions have been consistently reported as inversely associated with adult gliomas, reports in children have varied (Chen et al. 2011). In pediatrics, an initial report from the United Kingdom found that maternal asthma resulted in a decreased relative risk of their children developing a CNS tumor, particularly PNETs (Harding et al. 2008). Another study evaluating 272 matched case–control pairs in Canada found asthma associated inversely with the development of CNS tumors, especially ependymomas, while the relationship with eczema was not significant (Roncarolo and Infante-Rivard 2012). Furthermore, the use of asthma controller medications was found to be associated with an increased risk. However, a study of 352 pediatric brain tumors in Denmark, Norway, Sweden, and Switzerland found no association with asthma or eczema (Shu et al. 2014).

Studies evaluating the influence of prior infectious history on the development of pediatric CNS tumors have been conflicting. Harding et al. found infants without social interaction with other infants in the first year of life had an increased risk (OR 1.37, CI 1.08–1.75) of CNS tumors, especially PNET, compared to those who had such interaction (Harding et al. 2009). Attendance in day care also appeared to show a protective benefit, though not statistically significant. A Canadian study also found a reduced risk in patients with day care attendance, and, unlike Harding’s study, breastfeeding was found to be protective against the development of brain tumors (Shaw et al. 2006; Harding et al. 2007). Conversely, Anderson et al. found no association with day care attendance but that patients with more frequent sick days in the first 6 years of life had an increased incidence of gliomas and embryonal tumors (Andersen et al. 2013).

1.9 Environmental Exposure

Radiation therapy (RT), used decades ago to treat tinea capitis and more recently to treat childhood acute lymphoblastic leukemia (ALL), is known to cause secondary CNS tumors, especially meningiomas, p53 mutated glioblastomas, and PNETs (Kleinerman 2006; Ohgaki and Kleihues 2005). Fifty-three percent of secondary neoplasms in survivors of childhood ALL occur in the CNS and 89% of those are associated with prior cranial irradiation (Mody et al. 2008; Schmiegelow et al. 2013). The timing and outcome are dependent on pathology, as non-meningioma CNS tumors occur between 6.5 and 9.8 years and meningiomas occurred between 12.3 and 18.3 years after treatment, with OS of 18% and 96%, respectively (Schmiegelow et al. 2013). Prenatal diagnostic imaging has been evaluated as a potential cancer risk, but studies from the United Kingdom, Sweden, and Denmark did not describe a significant increase in pediatric CNS tumors in patients exposed to prenatal X-rays compared to controls (Mellemkjaer et al. 2006; Rajaraman et al. 2011; Stalberg et al. 2007). Diagnostic head X-rays also have not been associated with the development of CNS tumors (Khan et al. 2010). However, CT scans contribute to a slightly elevated risk of CNS tumors, with risk decreasing with increasing age at first CT scan exposure (Pearce et al. 2012; Mathews et al. 2013).

Magnetic fields, radio waves, and mobile phone use have not been found to be associated with an increase in pediatric brain tumors (Aydin et al. 2011; Elliott et al. 2010; Ha et al. 2007; Kheifets et al. 2010).

Although many different maternal medications have been evaluated, none have been found to consistently increase the risk of pediatric CNS tumors in offspring. A German study found an association between maternal prenatal antibiotic use and an increased risk of medulloblastoma (OR 2.07, CI 1.03–4.17) and astrocytoma (OR 2.26, CI 1.09–4.69) (Kaatsch et al. 2010). Although the odds ratio was similarly elevated in a Canadian study, the results were not statistically significant (OR 1.7, CI 0.8–3.6) (Shaw et al. 2006). A 2010 Swedish study evaluating potential associations with prenatal medications and the development of pediatric CNS tumors in children 0–14 years of age found no association with antibiotics, antifungals, antacids, analgesics, antiasthmatics, antiemetics, antihistamines, diuretics, folic acid, iron, laxatives, or vitamins, but did find an association with antihypertensives (OR 2.7, CI 1.1–6.5), particularly β-blockers (OR 5.3, CI 1.2–24.8) (Stalberg et al. 2010). An association between prenatal antihypertensive use and the development of pediatric CNS tumors, however, was not found in a German study evaluating pediatric CNS tumors diagnosed between 1992 and 1997 (Schuz et al. 2007). Amide or amine-containing medications can potentially be carcinogenic after conversion to N-nitroso compounds (NOCs) in the stomach, though three studies have all found little or no support for an association between maternal exposure and central nervous system tumors in subsequent children (Cardy et al. 2006; Carozza et al. 1995).

Prenatal vitamins, especially iron and folic acid, consistently have been shown to decrease the risk of pediatric CNS tumors (Bunin et al. 2005, 2006; Ortega-Garcia et al. 2010; Milne et al. 2012).

Although prenatal alcohol exposure can have a variety of toxic effects on the developing child, there is no clear increased risk of pediatric CNS tumors (Infante-Rivard and El-Zein 2007; Milne et al. 2013). The role of maternal tobacco smoking during pregnancy is unclear, as several reports have found no association (Filippini et al. 2002; Huncharek et al. 2002; Norman et al. 1996), while a review of the Swedish Birth Register of births between 1983 and 1997 found a hazard ratio of 1.24 (CI 1.01–1.53) (Brooks et al. 2004).

Pesticide exposure may have an association with pediatric CNS tumors. A review of 4723 patients from the North of England found no significant relationship between occupational exposure to pesticides and risk of any childhood cancer (Pearce et al. 2006). In contrast, a study from the USA found that paternal pesticide exposure was associated with an increased risk of his child developing an astrocytoma (OR 1.8, CI 1.1–31), but not PNET (Shim et al. 2009). A separate study investigating paternal hobbies did identify exposure to pesticides as increasing the risk of medulloblastoma and PNET (Rosso et al. 2008). An Australian study also found preconception exposure to pesticides increased the risk of pediatric CNS tumors (Greenop et al. 2013). The effect of residential pesticides may be contingent on particular predispositions as polymorphisms in PON1, a gene responsible for organophosphorous metabolism, may increase the risk of pediatric CNS tumors in exposed patients (Searles Nielsen et al. 2010).

An investigation of the risk of pediatric CNS tumors among children of parents working in a wide variety of occupations found no clear associations (Mazumdar et al. 2008). However, a separate analysis found that brain tumors were more common in children of mothers working in electronic component manufacturing (OR 13.78, CI 1.45–129) and garment and textile workers (IR 7.25, CI 1.42–37) (Ali et al. 2004). There also appears to be an increased incidence of CNS tumors among children whose parents are exposed to diesel fuel, but not other exhausts (Peters et al. 2013). Paternal polycyclic aromatic hydrocarbon exposure has also been linked to a subsequent increase in pediatric CNS tumors (OR 1.4, CI 1.1–1.7) (Cordier et al. 2004).

In conclusion, pediatric neuro-oncology is a rapidly evolving field in which molecular investigations are fueling a restructuring of tumor subgroups. Although pediatric CNS tumors have historically been distinguished by histopathology and location, driving mutations and epigenetic profiles are proving to not only be attractive therapeutic targets but also epicenters for new classifications. The challenge will be to integrate former classification systems with the latter, and, perhaps just as importantly, to frame our historical data according to the new groupings so that the decades of lessons learned in epidemiology can continue to be applied in the pursuit of improving outcomes for children with CNS tumors.