Epidemiology of Childhood Brain Tumors
Little is known about the etiology of brain tumors in children. Epidemiologists have applied their methods to uncover risk factors, but small sample sizes and consideration of all brain tumors as a single disease limited early studies. More recent studies investigated histological groupings of brain tumors and accrued larger sample sizes through national and international collaborations. Although these studies have not yet found conclusive evidence for new risk factors, they have provided clues leading us toward such factors.
In the absence of evidence of a critical time period, researchers have hypothesized that exposures prior to a child′s conception, during gestation, or during childhood could influence risk. Genetic predisposition factors could act alone or could interact with exposure to increase risk. Although preconception, gestation, and childhood are distinct time periods, individuals’ exposures are not neatly delineated. For example, the parents’ diet, jobs, and smoking habits may stay constant from before the child′s conception until the brain tumor diagnosis, making determination of the critical time period difficult.
Known Risk Factors
Ionizing radiation is known to increase the risk of brain tumors. Children treated with radiation for leukemia or benign conditions such as tinea capitis experience a higher incidence of brain tumors as children and as adults.1,2 In utero diagnostic X-rays in the 1940s to 1960s increased the risk of all cancers including brain tumors during childhood3,4; the increase in risk for all cancers was small, approximately 40%. A more recent study investigated brain tumors in children born in 1975 to 1984 in Sweden and used medical records to find information on X-ray examinations.5 About 20% of the mothers of children who later developed brain tumors and of the mothers of control children had exposure to an imaging procedure that used radiation. About half of the imaging procedures were of the abdomen, mostly pelvimetry and X-ray for fetal position, almost all in the last trimester. Abdominal exposure did not increase the risk of all brain tumors combined or of astrocytoma, but a non-significantly increased risk was observed for medulloblastoma/primitive neuroectodermal tumor (MB/PNET). Both the frequency of X-rays and the dose per film have declined greatly since the well-documented increased risk in the mid-20th century. It seems unlikely that prenatal X-rays are a major cause of brain tumors today, but more data would be helpful, especially for MB/PNET. Individuals with some genetic disorders develop brain tumors during childhood more often compared with the general population. Children with neurofibromatosis type 1, tuberous sclerosis, basal cell nevus syndrome, adenomatous polyposis of the colon, mismatch repair cancer syndrome, and Li-Fraumeni syndrome experience increased incidence of brain tumors6 ( Table 2.1 ).
Many other possible risk factors have been studied. Some are discussed in the following subsections.
High birth weight appears to increase the risk of leukemia, with recent studies suggesting that the risk extends to large-forgestational-age babies in the normal birth weight range.7,8 The mechanism is unknown, but some have suggested a role for in utero levels of insulin-like growth factor-1, estrogen, leptin, and other growth factors and hormones.9 The data on the relationship between high birth weight and childhood brain tumors conflict. For example, a Norwegian study observed an increased risk of medulloblastoma, but not astrocytoma, with birth weight over 4,000 g,10 whereas a California study found the opposite, that is, an increased risk of astrocytoma but not MB/PNET.11 Thus, no clear association between high birth weight and childhood brain tumors exists.
N-nitroso compounds (NOCs) include nitrosamines and nitrosamides, some of which are potent animal carcinogens.12 In addition, some nitrosamides induce nervous system cancer in animals when exposure is transplacental. The transplacental dose required for a given level of tumorigenesis is much lower compared with postnatal exposure.12 NOC precursors also induce cancer, indicating the formation of NOC in the body. When substances such as vitamin C that inhibit NOC formation from becoming precursors are fed to animals, cancer incidence can be reduced and even eliminated completely.13 NOCs and their precursors occur ubiquitously in the human environment in tobacco smoke, rubber products, medications, food, cosmetics, and other products.12
Based on the animal data, Preston-Martin et al14 first studied NOC exposure in relation to childhood brain tumors in 1982. The researchers observed increased risk associated with maternal exposure to burning incense, passive smoking, diuretics, antihistamines, and cured meats. Subsequent studies have not confirmed the associations with the nondietary sources of NOCs, but evidence of a role for cured meats has accumulated. In the largest study with sufficient numbers for analyses of specific histologies, frequent eating of cured meats by the mother during pregnancy appeared to increase the risk of glioma and the subcategories of glioma that were investigated, including pilocytic astrocytoma, anaplastic astrocytoma, other astrocytoma, malignant glioma, and ependymoma.15 In this study and in other studies, cured meat intake did not appear to increase the risk of MB/PNET.15–17 Despite the accumulating evidence on cured meats and gliomas, researchers do not consider the evidence conclusive. Although several studies provided consistent results, other explanations for the association have not been ruled out. For example, nearly all previous studies investigated only aspects of diet related to NOCs. No study of glioma in children has collected comprehensive data on the mother′s diet that would permit the estimation of intake of macro- and micro nutrients. Thus, it is possible that correlated dietary factors such as high fat or low vitamin C intake explain the cured meat association. The cured meat–glioma association is consistent and biologically plausible, but is not yet sufficient to prove an etiologic role of NOCs.
Basal cell nevus syndrome
Adenomatous polyposis (colon)
Mismatch repair cancer syndrome
Other Aspects of Diet
In several studies, use of prenatal multivitamins during pregnancy appeared to decrease the risk of brain tumors in children. The studies observed an effect for the histological categories astrocytoma, MB/PNET, and all other types combined.16–19 The critical time period(s) and nutrient(s) are unknown. In an international study, longer duration of use was associated with a larger decrease in risk for all histological categories.20 However, studies of MB/PNET suggested a protective effect of use close to conception rather than an effect of duration.16,17 The observed protective effect of multivitamins is not considered conclusive. The results are all derived from case–control studies in which parents are interviewed, and different participation rates among those with and without an exposure can bias the study. If health-conscious mothers were more likely to participate as controls and more likely to take prenatal multivitamins, this selection bias could produce the observed results when there is no true effect. Studies that assessed prenatal multivitamin use through medical records would strengthen or refute the evidence. Although the evidence is suggestive, limitations of the studies prevent us from concluding that multivitamins taken during pregnancy protect against the development of brain tumors in children.
As discussed above, researchers have studied maternal diet in the context of the hypothesis about NOCs. Although most studies included some fruits, vegetables, and other foods, the foods studied, the food groupings used, and the results varied, making interpretation difficult.14–16,21,22
Many studies have collected data to assess the relationship between maternal smoking and risk of childhood cancers including those of the central nervous system (CNS). The largest cohort study, based on record linkage in Sweden, found no increased risk of CNS cancers before age 6, based on almost 500,000 individuals and 81 cancers. 23 The meta-analysis of the 12 case–control studies of nervous system cancer did not show an association with maternal tobacco smoking during pregnancy.24 However, the meta-analysis of paternal smoking showed a slightly and statistically significant increased risk (relative risk = 1.2) for CNS tumors based on 10 studies.24 A male-mediated preconception effect of tobacco could explain the findings, but confounding and bias are alternative explanations of the weak association.
Use during pregnancy of nitrosatable drugs, that is, those that could contribute to NOC formation, was not associated with brain tumor risk in children.25 A Swedish study using information extracted from prenatal medical records observed a protective effect of borderline statistical significance for folic acid. In addition, β-blocking agents were associated with increased risk, but because very few mothers took these drugs, chance may explain the finding.26