Risk factors for brain tumors

INTRODUCTION — Primary brain tumors are a diverse group of neoplasms arising from different cells of the central nervous system (CNS). (See "Classification and pathologic diagnosis of gliomas, glioneuronal tumors, and neuronal tumors", section on 'Histopathologic and molecular classification'.)

Although incidence rates for primary brain and CNS cancers were increasing through the late 1980s, the rates have been decreasing by approximately 0.2 percent annually over the past 15 years [1]. Among the factors that contributed to the early increase in rates were the introduction of noninvasive diagnostic technology including computed tomography (CT) in the 1970s and magnetic resonance imaging (MRI) in the 1980s, as well as better health care access for older adults [2].

These data, in conjunction with evidence suggesting the increase may have been occurring for many decades, leave open the possibility that environmental exposures may account for the increasing incidence of brain tumors.

General risk factors that have been associated with brain tumors are discussed here. A number of putative risk factors for brain tumors have been examined. A shortcoming in many reports is the tendency to group all brain tumors together; this approach may miss important exposures for specific histopathologic types of tumors [3].

The epidemiology of meningioma and its associated risk factors are discussed separately. (See "Epidemiology, pathology, clinical features, and diagnosis of meningioma".)

GENETIC PREDISPOSITION SYNDROMES — Approximately 1 to 5 percent of brain tumors are due to genetic syndromes that confer an increased risk of developing tumors of the nervous system [4-6]. Some of these tumors are associated with neurofibromatosis and several other inherited syndromes.

Neurofibromatosis type 1 — Neurofibromatosis type 1 (NF1) occurs in 1 of 3000 persons and is linked to a gene on chromosome 17. The NF1 gene encodes a protein called neurofibromin that restricts cell proliferation by activating guanosine triphosphate (GTP) hydrolysis on Ras proteins. (See "Neurofibromatosis type 1 (NF1): Pathogenesis, clinical features, and diagnosis".)

Multiple neurofibromas are seen, and some undergo malignant change to neurofibrosarcoma. (See "Pathogenetic factors in soft tissue and bone sarcomas".)

Other malignancies that develop in up to 5 to 10 percent of patients with NF1 include other malignant nerve sheath tumors such as malignant schwannomas and astrocytomas. The astrocytomas are usually low grade and frequently have a pilocytic histology. These lesions have a predilection for the optic pathways, hypothalamus, and cerebellum.

It has been proposed that malignant degeneration in NF1 reflects the two-hit hypothesis in which one allele is constitutionally inactivated in the germline while the other allele undergoes somatic inactivation (the second hit) [7]. Animal models are consistent with this hypothesis but suggest that the second hit can be a mutation in the p53 gene [8,9].

NF2-related schwannomatosis — NF2-related schwannomatosis (NF2) is an autosomal dominant disorder predisposing to multiple neoplastic lesions. This disorder is due to a mutation in the NF2 gene, a tumor suppressor gene on chromosome 22 that encodes a membrane cytoskeletal protein called merlin or schwannomin [10] that appears to be involved in actin-cytoskeleton organization [11]. Other modifier genes may also be involved [12]. (See "NF2-related schwannomatosis (formerly neurofibromatosis type 2)", section on 'Molecular pathogenesis'.)

The pathognomonic findings are bilateral vestibular schwannomas (acoustic neuromas). Vestibular schwannomas are seen in 90 to 95 percent of patients with NF2 and generally develop by 30 years of age. (See "Vestibular schwannoma (acoustic neuroma)" and "NF2-related schwannomatosis (formerly neurofibromatosis type 2)", section on 'Vestibular schwannomas'.)

Other types of brain tumors are also seen, the most frequent of which are meningiomas. Approximately one-half of individuals with NF2 have meningiomas, and multiple meningiomas are often present [13]. The incidence of meningioma increases with age, and the lifetime risk may be as high as 75 percent [14]. Patients with NF2 tend to develop meningiomas at an earlier age than those with sporadic meningiomas. The meningiomas seen in patients with NF2 are more frequently atypical or anaplastic compared with sporadic tumors [15,16]. (See "NF2-related schwannomatosis (formerly neurofibromatosis type 2)", section on 'Vestibular schwannomas'.)

von Hippel-Lindau syndrome — The von Hippel-Lindau syndrome is an autosomal dominant disorder associated with hemangioblastomas, pancreatic cysts and neuroendocrine tumors, renal tumors, and pheochromocytomas. The gene on chromosome 3p25 normally functions as a tumor suppressor gene. (See "Clinical features, diagnosis, and management of von Hippel-Lindau disease" and "Molecular biology and pathogenesis of von Hippel-Lindau disease".)

Li-Fraumeni syndrome — The Li-Fraumeni syndrome (LFS) is inherited as an autosomal dominant trait and is usually associated with a germline mutation in the tumor protein 53 (TP53) gene. (See "Li-Fraumeni syndrome" and "Gene test interpretation: TP53".)

LFS is primarily characterized by sarcomas, breast cancer, leukemia, adrenocortical cancer, and brain tumors occurring before the age of 45 years. A variety of brain tumors are seen in LFS, most commonly choroid plexus tumors in infants, medulloblastomas in children, and astrocytomas in young adults [17]. Of particular note, a high percentage of choroid plexus carcinomas are associated with germline mutations in TP53 even in the absence of another cancer or a positive family history. (See "Uncommon brain tumors", section on 'Choroid plexus carcinoma'.)

Outside of choroid plexus carcinoma, which is included in LFS criteria, recommendations for TP53 germline testing in patients with brain tumors have not been established. Some experts have proposed that several other brain tumors prompt consideration of germline testing, based on findings suggesting high prevalence of germline TP53 mutations in specific histologic and molecular subgroups, independent of personal and family history of other tumors [17]. These include:

●Lower-grade choroid plexus tumors in children with a mutant p53 pattern in tumor by immunohistochemistry (IHC). (See "Uncommon brain tumors", section on 'Choroid plexus tumors'.)

●Sonic hedgehog (SHH) medulloblastoma in children – Large sequencing efforts have found that germline TP53 mutations are present in 20 percent (13 out of 63) of SHH medulloblastomas in children aged 5 to 16 years and 8 percent (13 out of 170) of all pediatric SHH medulloblastomas [18]. Genetic testing should be considered for pediatric SHH medulloblastomas (table 1). (See "Histopathology, genetics, and molecular groups of medulloblastoma", section on 'Genetic predisposition'.)

●Astrocytomas with rare isocitrate dehydrogenase type 1 (IDH1) R132C mutation – A strong association between germline TP53 and a rare IDH1 mutation (R132C) has been identified in LFS families [19] and in two germline TP53 mutations identified in The Cancer Genome Atlas (TCGA) database [20], suggesting that patients with this mutation might be recommended for germline TP53 testing. (See "Classification and pathologic diagnosis of gliomas, glioneuronal tumors, and neuronal tumors", section on 'IDH1/IDH2 mutation'.)

Clinical criteria for LFS, testing recommendations, and cancer surveillance strategies are reviewed in detail separately. (See "Li-Fraumeni syndrome".)

Familial adenomatous polyposis — Familial adenomatous polyposis (FAP) is an autosomal dominant condition caused by a mutation to the adenomatous polyposis coli gene on chromosome 5. The majority of FAP-associated brain tumors are medulloblastomas, but gliomas have also been described. (See "Clinical manifestations and diagnosis of familial adenomatous polyposis".)

Mismatch repair deficiency

●Lynch syndrome – Lynch syndrome, previously called hereditary nonpolyposis colorectal cancer (HNPCC), is a cancer predisposition syndrome caused by a germline mutation that impairs deoxyribonucleic acid (DNA) nucleotide mismatch repair (MMR). Lynch syndrome genes include MLH1, MSH2, MSH6, PMS2, and EPCAM. Patients with Lynch syndrome are at increased risk for high-grade gliomas, in addition to colorectal cancer and other solid tumors (table 2) [21]. (See "Lynch syndrome (hereditary nonpolyposis colorectal cancer): Clinical manifestations and diagnosis".)

Although MMR deficiency and Lynch syndrome are rare among patients with glioma as a whole, the prevalence is higher in younger patients with IDH-wildtype glioblastoma in particular. In a study that included 1225 adult gliomas referred to a single neuropathology department for diagnosis and next-generation sequencing, nine gliomas were MMR deficient (0.73 percent), including eight IDH-wildtype glioblastomas and one IDH-mutant astrocytoma [22]. Five of nine patients had Lynch syndrome confirmed by germline testing (overall prevalence of Lynch syndrome, 0.41 percent). In a complementary analysis including 257 additional IDH-wildtype glioblastomas enriched for early-onset cases, MMR deficiency was present in 12.5 percent of tumors in patients between 18 and 39 years of age and 2.6 percent of tumors in patients between 40 and 49 years of age. These data support neuropathologic testing for MMR deficiency in IDH-wildtype glioblastomas in patients younger than 50 years of age at diagnosis.

All patients with MMR deficiency identified on somatic (tumor) testing should be offered genetic counseling and germline testing for Lynch syndrome (see "Lynch syndrome (hereditary nonpolyposis colorectal cancer): Clinical manifestations and diagnosis", section on 'When to suspect Lynch syndrome'). A diagnosis of Lynch syndrome has important implications for health maintenance, reproductive counseling, and cancer screening (table 3), as reviewed separately. (See "Lynch syndrome (hereditary nonpolyposis colorectal cancer): Cancer screening and management".)

●Constitutional mismatch repair-deficiency (CMMR-D) – Biallelic mutations in mismatch repair genes (eg, MLH1, MSH2, MSH6, PMS2) cause the CMMR-D syndrome. Patients with CMMR-D are at risk for brain tumors in late childhood (primarily glioblastoma) and may have an NF1-like phenotype, with café-au-lait macules and axillary freckling [23,24]. (See "Lynch syndrome (hereditary nonpolyposis colorectal cancer): Clinical manifestations and diagnosis" and "Neurofibromatosis type 1 (NF1): Pathogenesis, clinical features, and diagnosis", section on 'Constitutional mismatch repair-deficiency syndrome'.)

Glioblastomas and other tumors associated with CMMR-D have a high mutational load compared with sporadic tumors, predicting that they may be responsive to immune checkpoint inhibition, and at least one case report describes a favorable response to nivolumab in two children with recurrent glioblastoma related to CMMR-D [25]. (See "Tissue-agnostic cancer therapy: DNA mismatch repair deficiency, tumor mutational burden, and response to immune checkpoint blockade in solid tumors".)

Basal cell nevus syndrome — Persons with the basal cell nevus syndrome (Gorlin syndrome, nevoid basal cell cancer syndrome) have an increased risk of medulloblastoma. The syndrome is caused by germline mutations of the patched 1 (PTCH1) gene, a tumor suppressor gene. (See "Nevoid basal cell carcinoma syndrome (Gorlin syndrome)".)

OTHER GENETIC FACTORS

Familial glioma — The above inherited syndromes explain only a fraction of families with an aggregation of gliomas [26-28]. Little is known about the genetic factors underlying the apparent susceptibility to glioma in families without one of the recognized genetic syndromes. Shared environmental exposures may contribute to risk in some families as well.

In one population-based study that included 2141 first-degree relatives of 297 individuals with astrocytoma diagnosed over an eight-year period in Sweden, 5 percent had a familial aggregation of glioma [29]. Segregation analysis of 14 families with at least two affected first-degree relatives suggested an autosomal recessive pattern of inheritance in 65 percent (nine families), while an autosomal dominant inheritance pattern was suspected in only 21 percent (three families).

In a multinational study that included 376 families with two or more verified gliomas, the majority of families (83 percent) had only two gliomas, of which 57 percent were in first-degree relatives and 32 percent were in second-degree relatives [30]. The age, sex, and grade distribution of familial cases was similar to that of sporadic cases in the literature. An autosomal dominant pattern of inheritance with low penetrance was suggested by the large number of families with only two glioma cases in consecutive generations, but a clear mode of inheritance was not identified.

Large-scale sequencing efforts will likely identify causative germline mutations in some of these families. As an example, whole-exome sequencing of 90 affected individuals in 55 glioma families identified protein-changing variants in the protection of telomeres 1 (POT1) gene in three families with mostly oligodendrogliomas [31]. POT1 encodes a component of the shelterin complex, which participates in telomere maintenance and response to DNA damage. Incomplete penetrance was demonstrated in the two larger families, with approximately half of individuals known to carry the mutation developing a glioma. Additional cancers in these families included lung cancer (n = 3), leukemia, colon cancer, and kidney cancer (one each). Rare loss-of-function mutations in POT1 have also been described in familial melanoma [32,33].

Single nucleotide polymorphisms — Several genome-wide association studies (GWAS) have examined the risk of brain tumors and identified genetic polymorphisms associated with glioma, including TERT, CCDC26, CDKN2A/CDKN2B, RTEL1, PHLDB1, and EGFR [34-37]. A large meta-analysis identified 13 new susceptibility loci for glioma; while some susceptibility loci are common to all glioma subtypes (eg, TP53), others appear to be specific to either glioblastoma (eg, EGFR) or non-glioblastoma glioma (eg, PHLDB1) [38]. A single variant in the 8q24.21 region has been strongly associated with risk of IDH-mutant oligodendrogliomas and astrocytomas [39], with a magnitude of risk comparable to BRCA1 gene mutations and breast cancer. This single nucleotide polymorphism (rs55705857) has been shown to be causally linked to IDH-mutant low-grade gliomas through the organic cation transporter (OCT)-mediated regulation of MYC expression [40].

Two susceptibility loci have been identified for meningioma (10q12.31 and 11p15.5) [41,42].

Genetic polymorphisms of interest have also been identified using candidate gene approaches. For example, individuals with variants in genes involved in carcinogen metabolism [43], DNA repair [43,44], inflammation/allergies [45-48], and other immune responses [49] may be at higher risk of brain tumors. While collectively these gene variants may not provide strong enough associations to identify high-risk groups (for screening), they may provide new insights into the pathways involved in carcinogenesis.

IONIZING RADIATION — Exposure to ionizing radiation, as occurs from therapeutic radiation therapy or among atomic bomb survivors, is an established cause of brain tumors, including meningiomas, gliomas, and nerve sheath tumors [50]. The latency between irradiation and the development of brain tumors may be as short as five years or as long as many decades. Higher doses of irradiation are associated with both an increased risk of developing a secondary brain tumor and a shorter latency period.

Therapeutic brain radiation used for the treatment of primary brain tumors, acute leukemia, and other tumors has allowed patients to survive for extended periods, thereby placing them at risk for secondary malignancies. This association has been demonstrated for both meningiomas and gliomas. The risk is higher for meningioma than glioma and does not appear to plateau over time [51].

In the Childhood Cancer Survivor Study, glioma was diagnosed in 40 patients (0.3 percent), with the diagnosis being made at a median follow-up of nine years [52]. Gliomas occurred in 24 patients who had leukemia and eight with a previous brain tumor. The primary risk factor was antecedent radiation (odds ratio [OR] 6.8, compared with survivors who had not received cranial radiation). The risk of developing glioma was greatest in children who were irradiated before age five years, and was not related to other forms of treatment or the underlying diagnosis.

The combination of prophylactic cranial irradiation with antimetabolite therapy has resulted in a much higher incidence of second tumors in survivors of acute lymphocytic leukemia. As an example, brain tumors occurred in 6 of 52 children (13 percent) treated with combined therapy, compared with no brain tumors in 101 children who did not receive radiotherapy [53]. Three of the six children had a genetic defect in thiopurine metabolism that resulted in higher levels of the active metabolite of 6-mercaptopurine.

Lower doses of radiation are also important. The relationship of exposure of the brain and skull to lower doses of irradiation (eg, treatment of tinea capitis) is clearly linked to the subsequent development of meningiomas and other brain tumors. The relationship of irradiation and meningioma is discussed separately. (See "Epidemiology, pathology, clinical features, and diagnosis of meningioma", section on 'Ionizing radiation'.)

Childhood exposure to diagnostic head computed tomography (CT) may also be associated with an increased risk of brain tumors [54,55]. (See "Radiation-related risks of imaging", section on 'Children and adolescents'.)

NONIONIZING RADIATION

Low-frequency electromagnetic fields — Low-frequency electromagnetic field exposure has been postulated as a potential risk factor for brain tumors, and several positive associations were reported in historical literature. Biologic plausibility has not been established, however, and studies using more rigorous methodology, including direct, in-home measurement of electromagnetic fields, have concluded that a large effect of this type of radiation on the risk of adult brain tumors can be excluded [3,56]. The International Agency for Research on Cancer (IARC) has concluded that data are insufficient to classify low-frequency electromagnetic fields as a risk factor for brain tumors [57].

Similarly, there is no evidence that exposure of children or pregnant women to magnetic fields from high-current power lines, electric heating sources, or electric appliances is associated with the subsequent occurrence of brain tumors in children [56,58-61]. Analyses of occupational exposure to magnetic fields have also failed to establish a relationship to the risk of brain tumors [62,63].

Cellular phones and radiofrequency fields — Cellular telephones are a source of radiofrequency fields that have received coverage in the popular media as potential risk factors for brain tumors, presumably due to exposure of the head of the user to radiofrequency energy. Other sources of radiofrequency field exposure include microwave and radar equipment and occupational exposures (sealers, plastic welders, amateur radio operators, medical personnel, and telecommunications workers). Exposures to radiofrequency energy are difficult to quantify, even under laboratory conditions [64].

Numerous epidemiologic studies have looked for a possible relationship between cellular telephone use and the development of brain tumors [65,66]. A meta-analysis that included data from 22 case control series concluded that there was a slight increase in risk associated with cell phone use in those studies where investigators were blinded to whether the participant was a case or a control [65]. Furthermore, the risk appeared to be associated with an induction period of 10 years or longer. Experimental studies suggest that increased levels of oxidation with radiofrequency electromagnetic exposures may explain the association [67].

However, many of the retrospective studies could have biases, such as recall bias, that could explain the positive findings. A prospective cohort study that included over 775,000 females and 3268 brain tumors reported no increase in the incidence of brain tumors in cellular telephone users compared with nonusers, overall or by tumor subtype (including glioma and vestibular schwannoma), over 14 years of follow-up [68]. Similarly, analyses considering changes in incidence rates over time, prevalence of mobile phone use, and latency period provide no support for causal associations [69-71].

Due to inconsistencies observed in studies and potential biases in case-control studies, the World Health Organization (WHO)/IARC classified radiofrequency electromagnetic fields as possibly carcinogenic to humans (Group 2B) in 2011 [72].

OCCUPATIONAL EXPOSURES — A large number of studies have been conducted to examine whether occupational exposures are associated with the risk of brain tumors. Although some positive associations have been reported, there are many inconsistencies in the literature and results are often difficult to interpret given numerous methodologic issues.

Evidence suggests that professionals tend to have a higher risk of brain tumors, although the elevated risk may be a consequence of detection bias due to better access to care [73,74]. Agriculture workers, especially those exposed to herbicides and pesticides, may also have elevated risk of brain tumors; farming as an occupation and place of residence have been associated with a 1.3- to 3.6-fold increased risk of brain tumors [75-80].

Many studies have focused on electrical, rubber, and petroleum workers, where earlier studies had detected strong positive associations; however, the overall findings from these studies do not provide strong support for positive associations. No excess mortality from brain cancer was detectable in a meta-analysis summarizing risk estimates from 20 studies of rubber workers (relative risk [RR] 0.90, 95% CI 0.79-1.02) [81]. Similarly, at least two meta-analyses have found no significant overall increase in brain cancer mortality in petroleum workers [82,83]. Finally, for electrical workers, while a number of studies suggested that individuals in electrical occupations were at increased risk of brain tumors, with relative risks ranging between 1.5 and 2.5 [84-90], other studies did not confirmed these findings [2,4,91].

HEAD TRAUMA — Anecdotal descriptions of brain tumors arising after head trauma date back to the reports of Harvey Cushing in 1922 [92]. A number of other observations have implicated head trauma as a potential risk factor for brain tumors. The evidence is strongest for meningiomas and less convincing for gliomas. (See "Epidemiology, pathology, clinical features, and diagnosis of meningioma", section on 'Others'.)

Evidence for other types of brain tumors includes the following:

●In a cohort study of 228,055 Danish residents hospitalized for head injury and followed for an average of eight years, there were increases in the risk of brain tumor with only hemangioblastoma and hemangioma reaching significance (standardized incidence ratio [SIR] = 2.6) [93].

●Childhood brain tumors may be more common in firstborn children (higher risk of birth trauma) and in children with documented history of birth trauma (forceps delivery, prolonged labor, cesarean section) [94]. However, other studies of childhood brain tumors have not confirmed an increased risk among these groups [3].

Studies of head trauma and brain tumors may be confounded by ionizing radiation since individuals with a history of head trauma are more likely to have had exposure to ionizing radiation in the form of skull x-rays. Recall bias is another factor complicating interpretation of case-control studies of head trauma and risk of brain tumors [95]. Persons with brain tumors may be more likely to recall minor and major episodes of head injury than controls. As an example, one study reporting an overall positive association between head trauma and the risk of brain tumors found no association when the definition of head trauma was restricted to episodes that required medical attention [96].

ALLERGIES — There is increasing consensus for the inverse association between allergies (eg, asthma, eczema, hay fever) and glioma risk based on consistent findings across different populations and study designs [6,97]. Increased immune surveillance in patients with allergies has been postulated as a possible explanation of these observations, and more research is needed to understand the underlying mechanisms.

This relationship was initially observed in large observational studies [98-101]. A systematic review of the literature and meta-analysis based upon 3450 patients diagnosed with glioma provided further support for this relationship [102]. There was a decrease in the incidence of glioma in patients with a history of any form of allergy (relative risk [RR] 0.61, 95% CI 0.55-0.67). Similar, statistically significant risk reductions were observed when the risk of glioma was analyzed for individuals with either asthma or eczema. In comparison, significant relationship to allergy was not observed among 1070 patients with meningioma.

Several large epidemiologic studies have examined the relationship between serum levels of immunoglobulin E (IgE) and the risk of malignant brain tumors [103-106]. In the largest prospective study to date, samples from 594 blood donors who subsequently developed a glioma (including 374 with glioblastoma) were compared with 1177 paired controls [105]. An inverse relationship was present for total IgE and risk of glioma among males and females; elevated total IgE was associated with a 25 percent decrease in risk of glioma, compared with low total IgE level (odds ratio [OR] 0.75, 95% CI 0.56-0.99, comparing >100 to <100 international units/mL total IgE). Furthermore, the association was present for at least 20 years before the diagnosis of glioma. These findings are consistent with previous studies [103,104,106] and provide strong support for the allergy hypothesis.

DIET — Dietary constituents associated with an increased risk of brain tumors include N-nitroso compounds and possibly dietary fat intake. Aspartame ingestion has been suggested as a possible risk, but there are at present insufficient data to support this claim. Ingestion of antioxidants, fruits, and vegetables may reduce the risk of brain tumors.

N-nitroso compounds — N-nitroso compounds are potent neurocarcinogens in animal models [107]. Human exposure to these agents occurs from both endogenous and exogenous sources.

●Exogenous – The major exogenous sources of population exposures to N-nitroso compounds include tobacco smoke, cosmetics, automobile interiors, and cured meats [4]. Other sources include rubber products (baby pacifiers, bottle nipples) and certain drugs including antihistamines, diuretics, oral hypoglycemic agents, antibiotics, tranquilizers, and opiates. N-nitrosodiethanolamine, a carcinogen in animal models, occurs mainly as a contaminant in cosmetic products, soaps, shampoos, and hand lotions.

●Endogenous – Endogenous formation of N-nitroso compounds is a complex process that occurs in the stomach and is dependent upon the presence of N-nitroso compound precursors, gastric pH, the presence of bacteria, and other physiologic parameters [3]. Thus, measurement of exposure to endogenous N-nitroso compounds is extremely difficult.

The many epidemiologic studies that have examined the relationship between meat consumption and brain tumors in adults have produced inconsistent results.

In the four largest case-control studies, all of which included over 200 patients diagnosed with glioma and appropriate controls, and some assessment of meat intake [108-111], two reported a significant two- to threefold increased risk of glioma for high consumers of cured meat or bacon, as compared with those with a low intake [109,110]. However, excess risks were only observed among men and, in one, the relative risks were for high intake of cured meat in combination with low fruit and vegetable intake [109]. A meta-analysis that included nine observational studies (primarily case-control studies) reported a relative risk of 1.48 (95% CI 1.20-1.83) for adult glioma among individuals with a high intake of cured meat [112].

Two more recent publications using data from prospective cohort studies found no associations with meat intake or dietary N-nitroso compounds [113,114]. Both studies had over 300 glioma cases and detailed dietary assessment to examine these exposures and their potential relation to glioma risk. The lack of association in these two large prospective studies cast doubt on the N-nitroso compound hypothesis, at least in relation to adult glioma risk.

Antioxidants, fruits, and vegetables — Indirect support for the N-nitroso compound hypothesis includes the observation that certain inhibitors of the nitrosation process, vitamins C and E, appear to reduce brain tumor risk in adults and children [4,96,115]. Dietary studies have demonstrated a reduced risk of brain tumors in children who consume increased amounts of fruits and fruit juices [4]. Prenatal vitamin supplementation (including vitamins A and C and folate) and increased maternal intake of vegetables have been associated with a lowered brain tumor risk in the offspring [115-117].

However, most epidemiologic studies on adult brain tumors contain limited dietary assessment and few questions to explore the relationship between intake of fruits, vegetables, and vitamins and the risk of glioma. Of the nine case-control studies that include data on vitamin C intake from dietary sources and/or supplements, two reported a statistically significant inverse association between intake of supplemental vitamin C and glioma risk [111,118]. In a third report, patients who reported ever using supplemental vitamin C had a relative risk of glioma of 0.2, but this association did not reach the level of statistical significance, and vitamin C intake from foods was unrelated to risk [119]. Others report an interaction between intake of vitamin C and cured foods, such that men with high intake of cured meat who had a low intake of foods rich in vitamin C had a significant twofold increase in glioma risk relative to those with low cured meat intake and high intake of foods rich in vitamin C [109]. Findings were similar in women, although the increased risk (1.5) did not reach statistical significance.

Studies examining vitamin E intake and glioma risk have been inconsistent. A meta-analysis summarizing the evidence from 12 observational studies with a total of 3180 glioma cases reported a 12 percent decrease in risk, but the reduction was not statistically significant (relative risk [RR] 0.88, 95% CI 0.69-1.12) [120]. Further, combined results from two cohort studies with 920 glioma cases showed no association (RR 1.00, 95% CI 0.77-1.31), highlighting potential biases in case-control studies.

As with vitamin C and E intake, data regarding glioma risk and intake of fruits and vegetables are inconsistent. In some studies, a significant inverse association was observed between glioma risk and total fruit and/or vegetable intake [108,121] (which in one study was limited to women only [110]), but three others report no association [119,122,123]. No associations were observed for intake of total or individual fruits and vegetables and the risk of glioma in an analysis of three prospective cohort studies, which included 296 incident glioma cases [124]. In another large prospective study, a positive association was observed for fruit and vegetable intake and glioma risk [113]. As prospective studies are the least prone to recall and selection bias, these findings suggest that fruits and vegetables are unlikely to reduce the risk of glioma.

TOBACCO — Although the presence of nitrosamines in tobacco smoke has stimulated interest in tobacco exposure as a potential risk factor for brain tumors, there is little evidence that either active or passive smoking is a significant risk factor for brain tumors.

Although case control [4,94,125,126] and prospective cohort [127,128] studies have yielded some conflicting results, the majority of studies do not support an association between tobacco exposure and subsequent development of gliomas or meningiomas. A large, contemporary prospective study examining smoking history in detail, including dose, duration, and latency, found no association between smoking and risk of glioma.

Studies looking at the role of active smoking or passive smoking in mothers and the risk of childhood brain tumors have yielded conflicting results [4,115,129].

ALCOHOL — There has been speculation that consumption of alcoholic beverages may increase the risk of brain tumors since both beer and liquor contain nitrosamines. However, no consistent association between consumption of different types of alcoholic beverages and the risk of gliomas or meningiomas in childhood (maternal consumption) or adulthood has been demonstrated [125,130,131].

INFECTION — Most reported associations of infection with brain tumor have been inconsistent. As an example, prior infection with tuberculosis was suggested as a possible risk factor for glioma in one study [132] but not in another [133]. In one large study, subjects who reported a history of infectious diseases (eg, colds, flu), compared with those with none, had a 30 percent reduction in risk (relative risk [RR] = 0.72, 95% CI 0.61-0.85) [98]. Other proposed infectious risk factors have included the polyoma virus, simian virus 40 (SV40), neonatal viral infections, and infection with Toxoplasma gondii. At present, however, there is no compelling epidemiologic evidence establishing infectious agents as important factors in the etiology of brain tumors.

Viral infection — The relationship between antecedent viral infections and subsequent development of brain tumors appears to be complex. Evidence for possible interactions comes from the identification of viruses and virus-like particles in brain tumor specimens, as well as from epidemiologic studies.

●Polyomaviruses – Interest in simian virus 40 (SV40), a polyomavirus, was stimulated by animal studies documenting brain tumor development after intracerebral inoculation with SV40 and by human studies that isolated SV40 from brain tumor tissue [134]. It was unclear, however, if SV40 contributed significantly to malignant transformation or whether certain tumors provided a microenvironment that favored replication in patients with latent SV40 infection.

Poliomyelitis vaccine administered between 1955 and 1962 was contaminated with SV40, and vaccination cohorts have been the subject of study over subsequent decades [135-137]. However, elevated brain tumor rates have not been observed in these cohorts. In a nested case-control study, no significant association was reported between antibodies to SV40 (or two other polyomaviruses, JC virus and BK virus) as measured in prediagnostic serum and incident primary malignant brain tumors [138]. Similar null findings were reported for JC and BK viruses in two prospective cohorts, but a different polyomavirus, the Merkel cell polyomavirus, was associated with a higher risk of glioma [139].

●CMV – Data concerning a possible etiologic role for cytomegalovirus (CMV) are conflicting and controversial. While several studies have reported that a high percentage of gliomas are infected with CMV [140,141], other groups have not been able to replicate these findings [142-146], or have reported inverse associations for CMV and other herpesviruses (Epstein-Barr virus and herpes simplex virus) [147].

●In utero viral exposure – Whether exposure to maternal viral infection while in utero is a risk factor for brain tumors is unclear. A large case-control study found an increased risk (RR = 2.4) for all types of brain tumors after different maternal and perinatal infections [148]. In addition, an association between influenza infection in pregnant women and childhood brain tumors (odds ratio [OR] = 3.15) was suggested in a study in which mothers of 94 children with brain tumors or neuroblastomas and 210 controls were interviewed [146]. However, others have failed to confirm an increased risk of brain tumors in the offspring of mothers infected with varicella, rubella, or mumps during pregnancy [149].

●Varicella-zoster – By contrast, a protective role for antecedent infection with varicella-zoster was suggested by an analysis of 229 adults with glioma and 289 controls from the San Francisco Bay Area Adult Glioma Study [150]. Individuals with gliomas were significantly less likely than controls to have a self-reported history of chickenpox (OR 0.59), and they also had lower levels of immunoglobulins directed against varicella-zoster. A similar inverse association was observed for self-report of history of chickenpox in a case-control study of 325 adults with glioma and 600 controls (OR = 0.52) [147]. In the Glioma International Case-Control Study, a history of varicella-zoster virus infection was associated with a 21 percent reduction in risk of glioma, and the association was slightly stronger among high-grade gliomas [97]. Based upon these observations and those for allergies, immune modulation has been postulated to have a role brain tumorigenesis. (See 'Allergies' above.)

Toxoplasma infection — Infection with T. gondii has been associated with an increased risk of astrocytoma and meningioma in two case-control studies [151,152]. In one, a significantly increased risk of meningioma (OR = 2.1), but not glioma, was associated with the presence of immunoglobulin G (IgG) antibodies to T. gondii as measured by ELISA [152]. Although this parasite has a propensity to infect the nervous system, it has not been established as a major risk factor for brain tumors.

SUMMARY

●Genetic predisposition syndromes – A small proportion of brain tumors are due to genetic syndromes that confer an increased risk of developing tumors of the nervous system. These include neurofibromatosis type 1 (NF1), NF2-related schwannomatosis (NF2), von Hippel-Lindau syndrome, Li-Fraumeni syndrome (LFS), familial adenomatous polyposis, Lynch syndrome, and the basal cell nevus syndrome. (See 'Genetic predisposition syndromes' above.)

●Other genetic factors – Genetic susceptibility has also been noted to play a role in determining risk of brain tumors; a single variant in chromosomal region 8q24.21 has been causally linked to isocitrate dehydrogenase (IDH)-mutant low-grade glioma. (See 'Single nucleotide polymorphisms' above.)

●Ionizing radiation – Ionizing radiation is the only firmly established environmental risk factor for brain tumors. Cohort studies of atomic bomb survivors and childhood cancer survivors have demonstrated that cranial radiation is associated with increased risk for a variety of brain tumors, including meningiomas, gliomas, and nerve sheath tumors. (See 'Ionizing radiation' above.)

●Nonionizing radiation – The association between forms of nonionizing radiation, such as low-frequency electromagnetic fields and radiofrequency fields, and cancer is less clear, and the data do not support an important role for these types of radiation as risk factors for brain tumors. (See 'Low-frequency electromagnetic fields' above and 'Cellular phones and radiofrequency fields' above.)

●Other possible factors – Possible causative factors that require further investigation include allergies, physical and acoustic trauma, and certain infections. (See 'Allergies' above and 'Head trauma' above and 'Infection' above.)