he American Cancer Society estimates that 1.25 million Americans T will develop cancer this year.' Approximately 550,000 die annually of neoplastic diseases, the second leading cause of death in the United States. Rising age-adjusted cancer mortality rates, together with the decline in heart disease mortality, will make cancer the leading cause of death within a few decades2 The sheer burden of cancer on the population highlights the need to identify not only better therapies, but also effective prevention ~trategies.~ History reveals that many past scourges, such as polio, plague, smallpox, and diphtheria, were controlled or eliminated primarily through preventive intervention^.^ Prevention of disease requires knowledge of causation. Epidemiologic studies have made major contributions to identifying the carcinogenic role of cigarette smoking, alcohol consumption, ionizing radiation, chemical carcinogens, and viruses and other organism^."^^^ Despite important advances, etiology remains uncertain for many forms of cancer, including carcinomas of the pancreas, prostate, and kidney. The carcinogenic effects of diet, hormones, environmental pollutants, and other agents are still c o n t r o ~e r s i a l . ~~~-~ Incomplete knowledge of causation is due, in part, to the inherent complexities of the carcinogenic process and the methodologies for assessing disease causation in Cancer is a multi-factorial, multi-stage disorder.12 Individual causal influences that slightly elevate cancer risk are difficult to evaluate in epidemiologic ~t u d i e s . ' ~f ' ~
The heterogeneity and diverse lifestyles of the U.S. population further obscure the effects of weak carcinogens. For many agents that have been investigated as candidate human carcinogens, the relative risk is often in the range of 1.5 to 3. These weak associations tend to generate inconsistent results among studies that also differ in research design, sample size, case selection, and other feature^.'^ My research career began at the Epidemiology Branch of the National Cancer Institute, under the mentorship of Robert W. Miller and Joseph F. Fraumeni, Jr. They encouraged young physicians to use their clinical skills not only to diagnose and treat cancers, but also to investigate and determine their causes. This research strategy involved identifymg patients with unusual clinical histories, physical findings, and laboratory abnormalities that might indicate exceptionally high risk of cancer. The case studies can uncover etiologic factors involved in cancer development in the general pop~lation.'~,'~ Many patients who came to my attention had strong family histories of cancer, often in Mendelian patterns suggesting inherited s~sceptibility.'~~'~ The diversity of site-specific familial cancers suggested the existence of multiple predisposing genes. Associations of