Chapter 52A Cancer and the Nervous System
Epidemiology of Brain Tumors
Primary brain tumors are a diverse group of neoplasms arising from different cells of the central nervous system (CNS). Light microscopy classifies these tumors according to predominant cell type and grades them for malignancy based on the presence or absence of standard histopathological features. Gliomas are the most common malignant primary brain tumors (80%) and constitute 32% of all brain and CNS tumors; these tumors arise from astrocytes (astrocytomas), oligodendrocytes (oligodendrogliomas), or ependymal cells (ependymomas). Glioblastoma accounts for approximately half of gliomas (54%), and astrocytomas (including glioblastoma) account for approximately three-quarters of gliomas (76%) (CBTRUS, 2010). Meningiomas, which arise in the meninges, are also common brain tumors (33%), but unlike gliomas, they demonstrate mostly nonmalignant behavior (CBTRUS, 2010). The etiology of brain tumors remains unknown despite a large number of epidemiological studies. This type of cancer is associated with a unique set of challenges for observational study designs (Box 52A.1). Despite a slow start, observational brain tumor studies are likely to yield important clues into the pathogenesis of this cancer.
Classification
The different cellular origins of brain tumors contribute to the difficulty in achieving a single, widely accepted classification system. Historical attempts at developing a classification system for brain tumors date back to the 1830s. The most widely accepted classification for primary CNS tumors is the World Health Organization (WHO) system, developed in 1979 and subsequently revised in 1993, 2000, and 2007 (Kleihues et al., 2000; Louis et al., 2007). In this system, all CNS tumors are assigned different grades (I to IV) representing an estimate of malignancy. Molecular subclassification of specific primary CNS tumors has become routine in practice and may better predict prognosis in some cases. Chapter 52B contains a comprehensive discussion and history of brain tumor classification.
Descriptive Epidemiology
Incidence
The traditional source of descriptive data on brain tumors has been the Surveillance, Epidemiology, and End Results (SEER) program sponsored by the National Cancer Institute (NCI). This program collects population-based cancer data on approximately 26% of the U.S. population (including cancer registries for nine states) to gauge national trends in cancer incidence and survival (National Cancer Institute, 2010). Until recently these data encompassed only malignant tumors; however, effective January 2004, all cancer surveillance registries are required to expand their primary brain tumor data collection to include tumors of benign or uncertain behavior (Benign Brain Tumor Cancer Registries Amendment Act; Public Law 107-260). The Central Brain Tumor Registry of the United States (CBTRUS) established in 1992 is the nation’s largest population-based registry of primary brain and central nervous system tumors, compiling information from 47 population-based cancer registries (CBTRUS, 2010).
Incidence estimates differ according to the inclusion or exclusion of nonmalignant brain tumors. The American Cancer Society estimated the diagnosis of approximately 22,020 new cases of malignant primary CNS tumors in the United States in 2010 (American Cancer Society, 2010). For 2010, CBTRUS estimates approximately 23,720 new cases of malignant and 39,210 nonmalignant primary CNS tumors in the United States (CBTRUS, 2010). The annual incidence rate for malignant brain cancer for all races from 2004 to 2006 was 7.2 per 100,000 person-years and 11.5 per 100,000 person-years for primary nonmalignant brain tumors (CBTRUS, 2010).
Mortality and Prognostic Factors
Although primary malignant brain tumors account for only 2% of all cancers and are one-fifth as common as breast or lung cancer, they contribute to substantial morbidity, and prognosis is poor. The 5-year survival rate for primary malignant brain tumors (36% between 1999 and 2005) is the sixth lowest among all types of cancer (following pancreas, liver, esophagus, lung, stomach, respectively) despite having increased over the past 30 years (from 24%) (American Cancer Society, 2010). The 5-year survival rates vary substantially by histological subtypes, 79.1% for oligodendroglioma, 27.4% for anaplastic astrocytoma, and 4.5% for glioblastoma (CBTRUS, 2010). Malignant CNS tumors accounted for approximately 13,140 deaths in 2010 (CBTRUS, 2010). Age-specific mortality rates from brain tumors among all races demonstrate gradual increases with each decade until age 55, after which the rate increases dramatically (Fig. 52A.1).
Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Mortality—All COD, Public-Use With State, Total U.S. for Expanded Races/Hispanics (1990-2006), National Cancer Institute, DCCPS, Surveillance Research Program, Cancer Statistics Branch, released April 2009. Underlying mortality data provided by NCHS (www.cdc.gov/nchs).
Young age and lower pathological grade are favorable prognostic factors for primary brain tumors (CBTRUS, 2010). Less significant predictors of favorable prognosis include long duration of symptoms, absence of mental changes at the time of diagnosis, cerebellar location of the tumor, small preoperative tumor size, and completeness of surgical resection.
Gender and Race
A slight male predominance exists in the incidence of malignant CNS tumors (7.5 per 100,000 person-years for men versus 5.2 per 100,000 person-years for women) (SEERS, 2010). However, when assessing nonmalignant and malignant CNS tumor types together, the incidence rate in men is lower than in women (14.4 per 100,000 person-years for men versus 16.9 per 100,000 person-years for women; SEERS, 2010). A partial explanation of this gender disparity is the predominance of meningiomas in women (8.2 per 100,000 person-years) compared to men (3.7 per 100,000 person-years) (CBTRUS, 2010). Fig. 52A.2 compares the incidence rates of malignant brain tumors among whites and blacks. Compared to blacks, whites have a higher incidence of malignant brain tumors for both sexes; white males have an incidence rate of 8.4 per 100,000 person-years compared to black males at 3.7 per 100,000 person-years, whereas white females have an annual incidence rate of 6.5 per 100,000 person-years versus an annual incidence rate of 3.3 per 100,000 person-years among black females (Ries et al., 2006). Mortality rates among whites is higher (5.0 per 100,000 persons; 2002-2006) than among blacks (2.0 per 100,000 persons; 2002-2006) (SEERS, 2009), and 5-year relative survival rates are lower among whites (35%) than blacks (41%) (American Cancer Society, 2010). Comparisons between other racial groups within the United States are difficult because of the small numbers of cases, but incidence rates of malignant brain tumors among Asian Americans and Native Americans are lower than for either white or black Americans.
Fig. 52A.2 United States incidence rates for malignant brain cancer for 2000-2007, by race and sex.
Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Incidence—SEER 9 Regs Public-Use, Nov 2009 Sub (1973-2007), National Cancer Institute, DCCPS, Surveillance Research Program, Cancer Statistics Branch, released April 2010, based on the November 2009 submission.
Temporal Trends
Several studies have documented rising incidence rates for brain tumors in a number of industrialized countries. These increases appear confined mainly to the elderly population, with no clear ethnic, gender, or geographical differences. Overall, incidence rates among whites in the United States increased by 1.5% annually between 1975 and 1988 and leveled off between 1988 and 2001 (SEERS, 2010).
The increases in incidence observed in the United States are in accordance with observations in other parts of the world. The responsible factors remain unclear, but at least part of the increase is the result of more complete case ascertainment because of improved diagnostic technology and better access to health care and clinical specialization. The increase in brain tumor incidence correlates with the introduction of noninvasive diagnostic technology including computed tomography (CT) in the 1970s and magnetic resonance imaging (MRI) in the 1980s. In a study using CBTRUS data, the authors reported increases in the incidence of most glioma subgroups but decreases in the incidence of “not otherwise specified” (NOS) subgroups, suggesting that changes in classification and coding of brain tumors are likely to be responsible for some of the observed time trends (Jukich et al., 2001). However, the same study identified increases in the incidence of ependymomas and nerve sheath tumors, which were less likely to be artifacts of improvements in diagnosis.
Geographical Trends and Migrant Studies
Many other countries also experienced increased brain tumor incidence rates over the past 30 years. The highest rates are in industrialized nations such as the United States, Canada, Australia, and the United Kingdom; developing nations have lower incidence rates (Parkin et al., 2005). Generally, brain cancer incidence appears associated with level of economic development; concomitant differences in the availability of diagnostic technology (CT, MRI, neurosurgical technology) and access to health care may account for some of the observed disparities. However, even within the United States, a significant variation in the incidence of primary brain tumors exists among the states. Hawaii has the lowest rate (3.9 per 100,000 person-years), and Maine has the highest rate (8.0 per 100,000 person-years) (State Cancer Profile, 2010) (Fig. 52A.3). In contrast to international trends, a clear relationship between these rates and economic conditions does not exist.
In a study of Italians who migrated to Argentina, brain tumor mortality rates were lower among the migrants than in the host country (Roman, 1998). This suggests environmental factors may influence the development of brain tumors. However, disparities in completeness of case ascertainment in the countries under study complicate the interpretation of this study.
Primary Central Nervous System Lymphoma
The epidemiology of primary central nervous system lymphoma (PCNSL), a type of non-Hodgkin lymphoma, deserves special comment. A dramatic increase in the incidence of PCNSL occurred over the past few decades. This parallels a doubling of the incidence rate of systemic non-Hodgkin lymphoma over the past 4 decades. The incidence peaked in the early 1990s and has since decreased. The primary factor behind the rising incidence is the acquired immune deficiency syndrome (AIDS) epidemic. From 2% to 6% of individuals with AIDS develop PCNSL at some point in their disease course (Fine and Loeffler, 1999). The incidence of AIDS has decreased along with the rate of PCNSL since the beginning of the 1990s. However, AIDS does not fully account for the total increase in the incidence of PCNSL (Rubenstein et al., 2008). Recent studies suggest a continuing increased PCNSL incidence among non-AIDS populations (Haldorsen et al., 2007; Makino et al., 2006).
Analytical Epidemiology
Radiation
Ionizing Radiation
Ionizing radiation exposure from radiation therapy or among atomic bomb survivors is an established cause of certain types of brain tumors, especially meningiomas and nerve sheath tumors. The latency between irradiation and the development of brain tumors may be as short as 5 years or as long as many decades. In a cohort study of 10,834 children treated with cranial radiation for tinea capitis in the 1950s (mean estimated radiation dose, 1.5 Gy), elevated risks were found for nerve sheath tumors (relative risk [RR] 18.8), meningiomas (RR 9.5), and malignant gliomas (RR 2.6) (Ron et al., 1988). A strong dose-response relationship exists such that those who received an estimated dose of 2.5 Gy had relative risks approaching 20 for all brain tumors. With an additional 16 years of follow-up, the relative risk for meningioma remained high (overall, RR 4.63; 95% CI, 2.43-9.12) and was elevated regardless of age at exposure or latency period (Sadetzki et al., 2005). In the same study, the highest risk of malignant glioma was observed among children who were exposed when they were less than 5 years old (RR 3.56; 95% CI, 0.96-9.91). Reports of less dramatic but still significant elevations of risk for meningioma and nerve sheath tumors exist among individuals treated with variable doses of ionizing radiation for thymic enlargement, enlarged tonsils and adenoids, and thyroid and nasopharyngeal conditions. In a mutagen sensitivity assay, peripheral blood lymphocytes from glioma subjects were more prone to chromosomal damage when exposed to gamma radiation. In this study, mutagen sensitivity of lymphocytes was associated with an increased risk of glioma (OR 2.09; 95% CI, 1.43-3.06) (Bondy et al., 2001).