The objective of this report is to provide an overview of the descriptive epidemiology of brain tumors in the United States. Four data sources are used. Information on the incidence of all primary brain tumors, both malignant and benign, for the five-year time period 1990-1994, is presented from the CBTRUS database. Data for malignant tumors only, including incidence and mortality, are taken from the Annual Cancer Statistics Review of the Surveillance, Epidemiology, and End Results (SEER) Program of the National Cancer Institute¹and from the report of the North American Association of Central Cancer Registries (NAACCR).² Survival rates are estimated using data from the SEER program and the National Cancer Database (NCDB).

BACKGROUND

Until CBTRUS was incorporated in 1992, standard brain tumor data analysis in the United States was limited to primary malignant tumors. However, intracranial tumors, to an extent regardless of their histological behavior, invade and displace critical areas of the brain related to function.³ And, for this reason, benign brain tumors may be difficult to distinguish from malignant tumors either by symptoms or outcome. A histologically benign tumor may produce devastating effects in a relatively short period of time based on its location, while a malignant neoplasm may not produce overt symptoms as quickly.⁴ Benign tumors can and often do impose the same costs to patients in terms of medical care, lost productivity and case fatality as malignant tumors. In addition, newly discovered molecular markers may indicate that certain benign brain tumors become malignant over time. Therefore, in order to present a complete picture of this disease, public health surveillance systems in the United States must collect and report benign brain tumor data along with the malignant data. Without such population-based data, analytic epidemiologic studies to evaluate disease etiology and assess disease consequences are greatly hindered.³

Currently in the United States incidence of cancer is routinely collected by the SEER program, the NAACCR and the NCDB. SEER and NAACCR list malignant tumors of the brain and central nervous system in their standard incidence reports. CBTRUS has determined that by ignoring benign tumors, these reports may underestimate the occurrence of all brain tumors by at least 50%.⁵ The NCDB voluntarily collects hospital-based data on benign and malignant tumors from approximately 1,000 of 2,200 hospitals participating in the tumor registry program of the American College of Surgeons and the American Cancer Society. For the data presented here, it was estimated that the NCDB represents 39% of brain tumors in the United States ⁶ although this percent has increased in recent years.

The CBTRUS database attempts to fill a gap in the available information about brain tumors by providing accurate, population-based incidence rates by histology, age, gender, and race that were previously unavailable in a systematic way. These data are useful for surveillance and may serve as a baseline for comparison with regional rates. They are also important for the allocation and planning of specialty health care services, for planning programs of disease prevention and control, and in the development of research proposals including those that investigate etiology.

CBTRUS, therefore, encourages all registries to expand their primary brain tumor data collection to include tumors of benign and uncertain behavior. As of 1998, 17 U.S. state registries collect or have begun to collect benign brain tumor data on a voluntary basis (Fig. 1). Of these states, 12 participate in the CBTRUS (Fig. 2). CBTRUS welcomes collaboration with regions collecting these data which are not yet CBTRUS participants.

DEFINITION OF RATES

Incidence, mortality, and survival rates are cited in this report. Incidence rates measure the occurrence of newly diagnosed cases of disease. Mortality rates measure deaths from disease. When the prognosis for a disease is poor, mortality rates approximate incidence rates. Survival rates reflect the clinical outcome of disease and often are expressed as a percent. Observed survival rates are taken from life-table estimates and yield the probability of surviving two or five years after diagnosis. Relative survival rates are defined as the observed probability of survival adjusted for the expected survival rate of the U.S. population for that age, gender, and/or calendar year.

METHODS

Data Collection

CBTRUS obtained incidence data from 12 collaborating state cancer registries that include cases of benign and malignant brain tumors. Data were requested from each registry on all cases newly diagnosed between 1990 and 1994 with a primary tumor at any of the following sites (International Classification of Diseases for Oncology⁷ (ICDO) codes in parentheses): brain (C71.0-C71.9), meninges (C70.0-C70.9), spinal cord, cranial nerves, and other parts of the central nervous system (C72.0-C72.9), and pituitary and pineal glands (C75.1-C75.3). Data were received without personal identifiers. Population data for each state were obtained from the SEER program website¹, which receives yearly population estimates from the U.S. Bureau of the Census.

The Virginia registry provided 1994 data only, the first year for which benign brain tumor data were available in that state. Virginia data were not included in the incidence rate calculations. One registry (Missouri), which previously provided data to CBTRUS, did not participate in the 1997 CBTRUS call-for-data.

To edit state data, CBTRUS used a modified metafile from a computer editing program^8, which generates warnings when illogical or impossible site, behavior, and/or morphology codes are detected. Queries about possible errors were sent to each collaborating region for review and were corrected as needed. Connecticut and Massachusetts submitted data in a format incompatible with CBTRUS computer data editing. Corrected data were incorporated into the CBTRUS database and aggregate descriptive statistics calculated.

For the purposes of this report, SEER data were taken from the SEER website and annual publication.¹ The SEER program collects cancer data from selected population-based cancer registries which represent about 10% of the U.S. population. SEER registries routinely follow-up cases annually which allows estimation of survival rates. The SEER program also reports mortality rates by state using data from the National Center for Health Statistics.

Additional data were obtained from the NAACCR and the NCDB. NAACCR publishes incidence rates from participating population-based cancer registries in North America.² NAACCR reports are updated annually. The National Cancer Database (NCDB) is a large hospital-based reporting system.⁶ NCDB brain tumor data were provided by the NCDB Data Request Committee and Mr. Herman Menck at the American College of Surgeons and were available for 1985-1988 and 1990-1992.

Classification by Histology

CBTRUS histology groupings were developed in collaboration with the CBTRUS consulting neuropathologist, Dr. Janet Bruner, of The University of Texas M.D. Anderson Cancer Center. These groupings, which aim for improved clinical relevance, utilize ICDO-morphology codes and are broadly based on the new World Health Organization (WHO) categories for brain tumors⁹. The listing of ICDO^7-morphology codes included in each grouping is presented in Table 1. This classification scheme was revised from the one CBTRUS used previously in order to accommodate data from Massachusetts which did not include behavior code.

Estimation of Incidence and Mortality Rates

Incidence rates were estimated using SAS, the computer-based statistical analysis system¹⁰. Overall rates for benign and malignant tumors and rates for selected histology groupings by gender and age were estimated using data from eleven regions. Age-adjusted rates based on five-year age groupings were standardized to the 1970 U.S. population to allow comparison with NAACCR and SEER rates.

Incidence and mortality rates for malignant tumors were obtained from the SEER website and the SEER Cancer Statistics Review¹. State specific mortality rates from 1988-1994 from the NAACCR 1998 publication² were also reported. Both NAACCR and SEER estimates are standardized to the 1970 U.S. population.

Estimation of Survival Rates

The SEER* Stat for Windows 95/NT program¹¹ was used to estimate two-, five- and ten-year observed and relative survival rates for primary malignant brain tumor cases diagnosed between 1973 and 1994. This program utilizes life-table (actuarial) methods to compute survival estimates and accounts for current follow-up. The SEER grouping for brain and other nervous system cancer included ICDO-site codes C70.0-C72.9. Morphology codes 953_, 9590-9989 (lymphomas, meningiomas) were excluded. Survival estimates for tumors of the pituitary and pineal glands were determined by selecting site codes C75.1-C75.2 and C75.3 respectively. Excluded from SEER analyses were any second or later primary tumors, cases diagnosed at autopsy, cases in which race is other or unknown, and cases known to be alive but for whom follow-up time could not be calculated. Survival rates were estimated for selected histologies by age when sample sizes were adequate (>10 deaths).

Two- and five-year survival for meningiomas and neurilemomas were estimated previously using data from the NCDB.^{12, 13}

continue >