Epidemiology of Brain Tumors

Published on 26/03/2015 by admin

Filed under Neurosurgery

Last modified 26/03/2015

Print this page

This article have been viewed 2616 times

CHAPTER 106 Epidemiology of Brain Tumors*

James L. Fisher, Margaret Wrensch, Joseph L. Wiemels, Judith A. Schwartzbaum

Descriptive Epidemiology

Brain tumor incidence rates have increased over time, with the rate of increase depending on histologic subtype, sex, age, race, and geographic area. For all central nervous system (CNS) tumors, of which brain tumors represent approximately 88%, the age-adjusted average annual (2000 to 2004) incidence rate for females (17.2 per 100,000) is greater than that for males (15.8 per 100,000).¹ Table 106-1 shows the age-adjusted average annual (2000 to 2004) incidence rates and median ages at diagnosis for the major histologic groupings and their selected common histologic subtypes of brain tumors.

TABLE 106-1 Number of Cases, Median Ages at Diagnosis, and Age-Adjusted, Average Annual (2000-2004) Incidence Rates* of Primary Brain Tumors (Major Histologic Groupings and Selected Histologic Subtypes) According to Sex

Glioma, which consists of a number of histologic subtypes, and meningioma have age-adjusted average annual incidence rates of 6.45 and 5.55 per 100,000 population, respectively.¹^,² As shown in Table 106-1, men have higher incidences rates of glioma, germ cell tumors, and cysts, whereas women have higher meningioma incidence rates. In the United States, the median age at diagnosis of a primary brain or CNS tumor between 2000 and 2004 was 57 years.¹ For all major histologic groups except germ cell tumors and cysts, incidence rates increase with age. Average annual incidence rates by age at diagnosis for selected histologic types common in adults and children/adolescents (aged 0 to 19 years) are shown in Figures 106-1 and 106-2, respectively. In adults (Fig. 106-1), incidence rates of meningioma and glioblastoma increase with advancing age, except for a decline in the incidence rate of glioblastoma in those 85 years and older. In children/adolescents (Fig. 106-2), incidence rates of all non–germ cell histologic types decrease throughout childhood and adolescence, whereas the incidence of germ cell tumors reaches a peak during the adolescent years.

FIGURE 106-1 Age-specific incidence rates of primary neuroepithelial brain tumors and meningioma, 2000 to 2004, Central Brain Tumor Registry of the United States (CBTRUS), reported in tabular form in CBTRUS (2008) Statistical Report: Primary Brain Tumors in the United States, 2000-2004.¹

FIGURE 106-2 Age-specific incidence rates of primary central nervous system histologic types more common in children, 2000 to 2004, Central Brain Tumor Registry of the United States (CBTRUS), reported in tabular form in CBTRUS (2008) Statistical Report: Primary Brain Tumors in the United States, 2000-2004.¹ NOS, not otherwise specified.

Survival Time from Diagnosis of Glioma and Meningioma

Survival time varies greatly by histologic type and age at diagnosis, as shown in Figure 106-3. For each age group, relative 2-year survival is lowest for patients with glioblastoma. In general, within histologic types survival time decreases with age. The mechanisms for the strong, consistent inverse association between age and survival are poorly understood and deserve further exploration. In patients in whom primary malignant brain tumors were diagnosed between 2000 and 2005, only 39.1% and 31.0% will be alive 2 and 5 years, respectively, from the time of diagnosis.³ Although the prognosis is poor for most patients with malignant brain tumors, 2-year survival rates increased from 28.6% in 1975 to 38.2% in 2004.³ The largest improvements in survival occurred in patients younger than 65 years in whom tumors other than anaplastic astrocytoma and glioblastoma were diagnosed. This may be attributed not necessarily to improvements in treatment but to improvements in imaging technology that allow earlier identification of tumors. There has been little change in the poor survival of patients with glioblastoma. Because glioblastoma appears to have only a brief preclinical period, improvements in survival will probably be attained only with technologies outside the current treatment paradigm, which includes surgery, radiation therapy, and alkylating chemotherapy. The concomitant addition of and maintenance with the chemotherapeutic agent temozolomide (Temodar) has improved the median survival time for glioblastoma patients by 2.5 months.⁴

FIGURE 106-3 Two-year relative survival probabilities of patients with primary malignant central nervous system tumors according to age at diagnosis and histologic type, based on follow-up of individuals with tumors diagnosed between 1973 and 2004 in the Surveillance, Epidemiology and End Results (SEER) program as compiled by the Central Brain Tumor Registry of the United States (CBTRUS) and reported in tabular form in CBTRUS (2008) Statistical Report: Primary Brain Tumors in the United States, 2000-2004. NOS, not otherwise specified.

Population-based data from Norway and Finland suggest that survival of patients with meningioma also improved between the 1950s and 1990s,⁵^,⁶ and it is possible that this improvement is likewise associated with advances in imaging technology. McCarthy and colleagues estimated a 69% 5-year survival rate for meningioma, with those younger at diagnosis having a more favorable prognosis (5-year survival probabilities were 81% versus 56% for patients in whom meningioma was diagnosed before versus after 65 years of age).⁷ Survival from meningioma is generally poorer for patients with malignant versus benign histology, with a 5-year survival probability of 55% versus 70%⁷; however, the vast majority (96%) of meningiomas are not malignant.

Environmental Risk Factors

Ionizing Radiation

Sources of exposure to ionizing radiation include therapeutic and diagnostic medical procedures, occupation, atmospheric testing of nuclear weapons, and proximity to atomic bomb explosions in Japan. Survivors of the bombing of Hiroshima have elevated incidence rates of meningioma that increase with the estimated dose of radiation to the brain.⁸ These survivors also have higher incidence rates of glioma, schwannoma, and pituitary tumors, although there is no increased risk for brain tumors in survivors who were exposed in utero.⁹ The use of ionizing radiation to treat tinea capitis and skin hemangioma in infants and children has been associated with significantly elevated relative risks for nerve sheath tumors, pituitary adenoma, meningioma, and glioma.⁹^,¹⁰ However, there are mixed results from studies involving other sources of exposure to diagnostic and therapeutic irradiation of the head and neck.¹¹^,¹² One study found that radiographs performed 15 to 40 years preceding diagnosis appeared to increase the risk for meningioma,¹³ and another study found meningioma risk to be associated with radiographs before the age of 20 years or taken before the year 1945.¹⁴ Further evidence possibly supporting an effect of ionizing radiation on brain tumor risk includes the observation that second primary brain tumors occur more frequently than expected in patients previously treated for brain tumors with radiation therapy; the standardized incidence ratio for second CNS tumors in brain tumor patients treated by surgery alone is 2.0 (95% confidence interval [CI], 1.2 to 3.2) versus 5.1 (95% CI, 2.5 to 9.4) for patients treated by radiotherapy with or without surgery or chemotherapy, or both.¹⁵ It is also possible that these results can be attributed to the fact that people with higher grade tumors are more likely to both receive radiation therapy and have a recurrence. Results from case-control studies of exposure to ionizing radiation and glioma risk are varied, perhaps because of underreporting of exposure, imprecise estimates of age at first exposure, or a low prevalence of exposure to high doses of ionizing radiation.¹⁶ The consistent and strong results from prospective studies of people exposed to ionizing radiation provide unquestionable evidence of a linear dose-response association between ionizing radiation exposure and glioma risk.¹⁶ Future studies should consider the potential for interaction between ionizing radiation and both age at exposure and genetic variation that may mediate the exposure. Regardless of the strong evidence for an association between ionizing radiation and brain tumors, therapeutic doses of ionizing radiation probably contribute to the development of only a small proportion of brain tumors because exposure to therapeutic levels of ionizing radiation is rare and the vast majority of glioma and meningioma patients report no such exposure; in one study, between 1% and 3% of glioma and meningioma patients, as well as controls, reported a history of at least one therapeutic dose of ionizing radiation before diagnosis of their brain tumor.¹⁷ Elucidating a possible role of more common radiation exposure, such as that resulting from dental radiographs, will require reliable assessment of exposure.

Cellular Telephone Use

Public concern over the potential health effects of cell phones has prompted studies focused on exposure to radiofrequency fields and brain tumor risk. In general, risk for meningioma does not appear to be increased as a result of cell phone use ¹⁸^–²⁰; however, there is some limited evidence that use of an analog cell phone,²¹ especially of longer duration (>10 years),²² may increase meningioma risk (although exposure to analog cell phones has greatly diminished because they have largely been replaced by digital cell phones). Mixed results have been reported for association between cell phone use and acoustic neuroma risk,²¹^–²⁶ yet relative risks from studies of more than one histologic type of brain tumor have been greater for acoustic neuroma than for meningioma and glioma, especially when cell phone use is on the same side of the head as the tumor (ipsilateral).²⁵

A relatively large number of epidemiologic studies of cell phone use and glioma risk have been conducted. Results from these studies have suggested that short-term cell phone use is probably not associated with risk for glioma.18–20,22,27–35 There are limited data and inconsistent results pertaining to long-term use and glioma risk,^{18–20,22,28–31}^,34 ^,36 with the most compelling results suggesting evidence of increased glioma risk as a result of ipsilateral cell phone use.^19,^22,^29,^31,³⁴ However, these results have probably been affected by small sample size and selection and recall bias.²⁰^,³⁷ Some of the increased risk resulting from ipsilateral cell phone use may be attributable to recall bias because contralateral cell phone use appears to reduce risk; in the absence of recall bias, one would not expect cell phone use to decrease risk. The largest population-based case-control study reported to date (1522 glioma cases and 3301 controls), conducted in five Nordic countries and the United Kingdom, found no consistent evidence overall of increased risk for glioma related to the use of cell phones, nor was increased glioma risk found in the most highly exposed group.²⁴ Thus far, no study has demonstrated irrefutable evidence of an association between long-term cell phone use and increased glioma risk. However, if the latency period is at least 5 years long, earlier studies lacked sufficient numbers of long-term cell phone users to adequately evaluate the relationship. The association of glioma risk with long-term cell phone use has not yet been convincingly demonstrated but will continue to be examined in the context of more refined studies with greater statistical power because of the increasing number of people who are long-term cell phone users and the potential release of individual records by cell phone companies for better assessment of exposure.

Risk and Preventive Factors for Which Evidence Is Inconclusive

Numerous dietary, experiential, and environmental factors studied in relation to brain tumor risk have shown inconsistent associations, that is, one or more positive studies but some with no association found. Such factors include head injury and trauma,³⁸^–⁴³ dietary calcium intake (for glioma),⁴⁴^,⁴⁵ dietary intake of N-nitroso compounds (for glioma and meningioma),⁴⁶^–⁴⁹ dietary antioxidant intake (for glioma),^44,^46,^47,⁴⁹ dietary maternal intake of N-nitroso compounds (for childhood brain tumors),³⁸^,⁴³ dietary maternal and early life intake of antioxidants (for childhood brain tumors), maternal folate supplementation (for primitive neuroectodermal tumors),³⁸^,⁵⁰ tobacco smoking (for glioma and meningioma),^38,^47,⁵¹ alcohol consumption (for glioma, meningioma, and childhood brain tumors),⁹^,⁵² and exposure to electromagnetic fields (for childhood and adult brain tumors).³⁸ In addition to small study sample sizes, possible explanations for the inconsistent findings include invalid or imprecise measurements of exposure (resulting from the use of self-reported or proxy-reported exposure and from lack of validation of the exposure), unfocused hypotheses (resulting from studies of large numbers of exposures without a specific rationale for examining many exposures or from studies in which relationships between exposure and brain tumor risk are examined without respect to known confounding or modifying factors), inherited or developmental variations in metabolic or DNA repair pathways that modify the effect of environmental factors on brain tumor risk, and unaccounted-for protective environmental exposures or conditions (e.g., allergy). The levels generally encountered for some examined exposures (e.g., chemical compounds) may also often be too low to have a measurable impact on brain tumor risk. Continued progress in understanding risk factors is dependent on the construction of large studies with better assessment of exposure, along with analysis of genetic factors that influence the effects of such exposure.

Reproductive and Menstrual Factors

In part because females have a lower risk for glioma and a greater risk for meningioma, investigators have examined factors associated with endogenous and exogenous female hormonal status as potentially being related to brain tumor risk, including menopausal status, ages at menarche and menopause, parity, and use of oral contraceptives and hormone replacement therapy (HRT). Estrogen and other female hormone levels are greater in general and without respect to exogenous estrogen sources in women between the ages of menarche and menopause; therefore, investigators have examined age- and sex-specific glioma and meningioma incidence rates to look for patterns in rates potentially related to these hormones. One study suggested that the gender differential in glioma rates occurred primarily from menarche to menopause and decreased in postmenopausal age groups,⁵³ whereas another showed that postmenopausal women had an increased risk for glioma and acoustic neuroma.⁵⁴ However, the latter notion is not fully supported by Figure 106-4, which shows that glioma incidence rates derived from the Surveillance, Epidemiology, and End Results (SEER) program are greater in males within each age group and that the ratio of male-to-female incidence rates increases with advancing age and reaches a plateau after the age of 40, with males having at least a 40% greater risk than females for all age groups 40 years and older. For meningioma, in general, there are consistent results suggesting greater risk in women during their reproductive years,^5,^54,⁵⁵ and these results are supported by Figure 106-5, which shows that SEER meningioma incidence rates are greater in females (indicated by lower male-to-female incidence rate ratios) and the greatest differential occurs during the approximate ages of 30 to 54 years.

FIGURE 106-4 Ratios of male to female average annual glioma (2000 to 2005) incidence rates according to age group based on the Surveillance, Epidemiology, and End Results (SEER) program SEER*Stat Database: Incidence—SEER 17 Regs Limited-Use + Hurricane Katrina Impacted Louisiana Cases, Nov 2007 submission.

FIGURE 106-5 Ratios of male to female average annual meningioma (2000 to 2005) incidence rates according to age group based on the Surveillance, Epidemiology, and End Results (SEER) program SEER*Stat Database: Incidence—SEER 17 Regs Limited-Use + Hurricane Katrina Impacted Louisiana Cases, Nov 2007 submission.

There is no consistent or convincing evidence that parity is associated with risk for either glioma ^54,⁵⁶^–⁵⁹ or meningioma.⁵⁴^,⁶⁰ Two recent studies have suggested a possible increase in glioma risk as a result of later (14 years or older versus younger than 12 years) age at menarche,⁵⁷^,⁵⁹ but more studies are needed to confirm this relationship. Furthermore, inconsistent results pertaining to oral contraceptive use and HRT have been reported for both glioma^55,^57,⁵⁹ and meningioma⁵⁵^,⁵⁷ risk. Results from a population-based case-control study conducted by Wigertz and coworkers revealed elevated meningioma risk in women who had used long-acting hormonal contraceptives (odds ratio [OR] for at least 10 years of use, 2.7; 95% CI, 0.9 to 7.5) and postmenopausal women who had ever received HRT (OR, 1.7; 95% CI, 1.0 to 2.8).⁵⁵ These authors also reported that the use of oral contraceptives and HRT had no impact on glioma risk.⁵⁵ Female hormones probably play a role in the development of some meningiomas, which may partially explain why women have lower glioma incidence rates, but our understanding of the mechanisms governing their role is limited, perhaps in part because the menstrual and reproductive factors that have been examined are insufficient to accurately characterize lifetime estrogen or other hormonal exposure.

Glioma, Allergic Conditions, Infections, and Associated Immunologic Factors

Since 1990, the results of 10 case-control and 1 of 2 cohort studies have shown that self-reported allergies are inversely related to glioma risk. Linos and associates conducted a formal meta-analysis of a subset of these studies and concluded that the strong inverse association between self-reported allergies and glioma (OR, 0.61; 95% CI, 0.55 to 0.67) is probably not attributable to methodologic bias alone.⁶¹ Further evidence of this inverse association was contributed by Wiemels and coworkers, who found that total IgE levels were lower in glioma patients than in controls.⁶²

Although mechanisms governing potential protection have not been identified, they may arise from the anti-inflammatory effects of interleukin-4 (IL- 4) and IL-13 cytokines involved in allergic and autoimmune disease ⁶³ or from increased tumor immunosurveillance in those with allergies and autoimmune disease.⁶⁴ It is also possible that the inverse association results from immune suppression by the preclinical tumor.

In addition to their role in allergic conditions, IL-4 and IL-13 cytokines also inhibit the growth of glioma cell lines. Although IL-4 is not expressed in the normal adult brain, it is strongly expressed during brain injury,⁶⁵ where invading T cells may be a source of this cytokine.⁶⁶ Barna and associates found that three normal astrocytic, two low-grade astrocytoma, and three of four glioblastoma cell lines that they evaluated expressed IL-4Rα receptors.⁶⁶ However, IL-4 suppresses DNA synthesis and cell proliferation only in the normal astrocytic and low-grade astrocytoma cell lines but not in the glioblastoma cell lines. Their results suggest that IL-4 could play a role in the inhibition of glioblastomas that arise from astrocytomas but may not be involved in glioblastomas that arise from other pathways.⁶⁷ In support of a role for IL-4 in glioma pathology, Faber and colleagues reported that IL-4 increases the number of T-cell precursors in glioblastoma patients.⁶⁸ Saleh and associates attributed the growth-inhibiting properties of mouse IL-4 on implanted C6 glioma cell lines to its ability to promote eosinophil infiltration and inhibit angiogenesis.⁶⁹ Furthermore, they observed that C6 cell gliomas implanted in rats that produce retroviral IL-4 are rapidly eradicated.⁷⁰

In view of a possible role of IL-4 and IL-13 in both allergic conditions and glioma, Schwartzbaum and coworkers identified polymorphisms of the IL-4Rα and IL-13 genes that increase risk for allergic conditions.⁷¹ Although these germline genetic variants are not sensitive indicators of the presence of allergic conditions, they do provide a measure of risk for these conditions free of recall bias. The working hypothesis was that in individuals with IL-4Rα or IL-13 polymorphisms that increase risk for allergic conditions, glioblastoma risk would be decreased. Using data from a small case-control study (111 glioblastoma cases, 422 controls), the authors found results consistent with their hypothesis. Each of the two IL-4Rα and IL-13 single nucleotide polymorphisms (SNPs) associated with increased risk for allergic conditions were also related to decreased glioblastoma risk. Wiemels and coauthors confirmed their finding for one of the IL-13 SNPs in a larger case-control study of glioma (456 glioma cases, 541 controls).⁷² Furthermore, they reported that this IL-13 SNP was inversely associated with IgE levels in controls (P = .04). However, they did not find associations between IL-4Rα SNPs and glioma as had Schwartzbaum and associates, but they did see a borderline association between an IL-4Rα haplotype (OR, 1.5; 95% CI, 1.0 to 2.3) and glioma. They also found that a rare IL-4 haplotype was associated with decreased glioma risk (OR, 0.23; 95% CI, 0.07 to 0.83).

A larger study of the original four IL-4Rα and IL-13 genetic variants by Schwartzbaum and colleagues did not provide strong support for their original observations. Nonetheless, they found an IL-4Rα haplotype associated with glioblastoma (OR, 2.26; 95% CI, 1.13 to 4.52) and inversely related to self-report of hayfever or asthma in controls (OR, 0.39; 95% CI, 0.16 to 0.98).⁷³ Although Wiemels and associates also found suggestive evidence of an association between an IL-4Rα haplotype and glioma, when they restricted their haplotype to the same IL-4Rα SNPs that Schwartzbaum and coworkers examined, they observed no evidence of an association with glioma (OR, 1.13; 95% CI, 0.83 to 1.53).

The tumor itself may also have mechanisms that inhibit the ability of the immune system to eradicate it. In recent in vitro studies of glioma, human glioma cell lines were found to secrete immunosuppressive cytokines that can selectively recruit regulatory T cells into the tumor microenvironment.⁷⁴ In addition, Chahlavi and associates demonstrated that glioma cell lines mediate immunosuppression by promoting T-cell death through tumor-associated antigens and gangliosides.⁷⁵ Two of the major immunosuppressive cytokines that are present in both the glioma microenvironment and peripheral blood of glioma patients, IL-10 and transforming growth factor-β, induce immune tolerance and thereby inhibit allergy and asthma.⁷⁶ Elevated IgE concentrations may therefore indicate low levels of immunosuppression and the resulting ability to conduct antitumor immunosurveillance against incipient glioma. Alternatively, the relative absence of allergies in glioma patients may show merely that these tumor-induced cytokines have suppressed the immune system.

Reduced glioma risk has also been attributed to a reported history of varicella-zoster virus (VZV) infections and positive IgG to VZV.⁷⁷^–⁷⁹ Papovaviruses including simian virus 40 [SV40], JC and BK viruses, adenoviruses, retroviruses, herpes and influenza viruses, and parasitic infections (Toxoplasma gondii) have also been investigated in relation to the genesis of glioma in experimental animals and in limited epidemiologic studies; however, the potential risk from these agents has generally been inadequately addressed in epidemiologic studies.⁸⁰^,⁸¹ With relative consistency, results from two case-control series suggest that previous clinical disease associated with VZV infection and anti-VZV IgG levels may be inversely associated with adult glioma risk.⁷⁷^–⁷⁹ It might be the specific nature of the immune system’s response to antigens and not exposure to the antigen per se that is responsible for this inverse association with glioma.⁶²

At present, there is no strong epidemiologic evidence suggesting that human cytomegalovirus (HCMV) plays a role in the development of glioma. However, HCMV nucleic acids and proteins have been found in the tumors of glioblastoma patients, HCMV DNA has also been found in the peripheral blood of glioblastoma patients,⁸² and Scheurer and coauthors reported detection of HCMV infection in 21 of 21 patients with glioblastoma.⁸³ However, Wrensch and colleagues⁷⁹ and Poltermann and associates⁸⁴ reported that antibody positivity to HCMV in glioma patients was not different from that in controls and the general population, respectively. The presence of HCMV gene products in blood or tumor tissue may result from reactivation of infection or from infected tumor cells shedding viral DNA.⁸² Renewed interest in HCMV may spark new epidemiologic studies, and these studies should consider the potential importance of low-level infection, which requires stringent technical conditions to adequately detect such infection.⁸³

With respect to SV40, between 1955 and 1963 an unknown proportion of all inactivated and live polio vaccines distributed were contaminated with SV40.⁸⁵ In Germany, where children were monitored over a 20-year period, those inoculated with the polio vaccine contaminated with SV40 had a higher incidence of glioblastoma, medulloblastoma, and some less common brain tumor types than did children not given the contaminated vaccine.⁸⁶ In the United States, no difference in brain tumor risk was found for glioma or meningioma between the two groups of children,⁸⁷ but one study reported that the incidence of ependymoma was 37% greater in children receiving the contaminated vaccine.⁸⁵ Results pertaining to infections should be validated by studies in which serologic measurement of viral or bacterial exposure is ascertained before the development of brain tumors and in which there is serologic or symptom-based confirmation of infection.

Results pertaining to human leukocyte antigens (HLAs)—cell surface molecules that modulate immune responses, in part by presenting antigenic peptides to T lymphocytes—also suggest the importance of immunologic responses in glioma development. Tang and colleagues showed that glioblastoma is positively associated with the HLA genotype B*13 and the HLA haplotype B*07-Cw*07 (P = .01 and P < .001, respectively) and is inversely associated with the genotype Cw*01.⁸⁸ Interestingly, if confirmed, these results could partially explain the increased glioblastoma incidence in whites because B*07 and B*07-Cw*07 are more common in whites than nonwhites. Guerini and associates compared a small group of glioma patients in northern Italy with control organ donors from the same region and demonstrated a positive association between HLA-DRB1*14 and the presence of symptomatic cerebral glioma.⁸⁹ Facoetti and coworkers found that HLA class I antigens were lost in approximately half of glioblastoma tumors but in only 20% of grade 2 astrocytoma tumors, selective HLA-A2 antigen loss was observed in approximately 80% of glioblastoma lesions and half of the grade 2 astrocytoma tumors, and HLA class I antigen loss was significantly (P < .025) correlated with tumor grade.⁹⁰ Studies of HLA may contribute to our understanding of the immune escape mechanisms used by glioma because HLA antigens mediate interactions of tumor cells with the host immune response; furthermore, HLA antigen defects in astrocytoma brain tumors may explain the relatively poor clinical response rates observed in the majority of the T-cell–based immunotherapy clinical trials.⁹⁰