Skip to Content
Discovering the causes of cancer and the means of prevention

Accelerating Innovation in Cancer Research Through Epidemiology

An Invited Presentation to the President's Cancer Panel, October 2010

March 2011 - Linkage Newsletter

by Patricia Hartge, Sc.D.

Back to issue home page

Epidemiology continues to shape the research agenda for preventing cancer and understanding its causes. New epidemiological research will differ dramatically from previous research in its sheer size and scale, the technologies used, and the strong connections to other biomedical disciplines. By focusing on the most informative study populations, using new techniques for measuring genetic and environmental influences, and exploiting the emerging markers of cancer, including early disease and premalignant changes, we can accelerate the pace of discovery.

Study Populations

The most pressing requirement currently in the United States concerns the specific populations to study. Ideally, we want prospective data and samples of the general population because collecting them well before diagnosis eliminates many biases. We need simpler and less expensive ways to gather data about study participants over the course of their lives. We need access to their medical histories, routine medical visits, and tests as well as opportunities to take and store biological samples for use in specific nested studies. Assembling a large general population cohort with these key features from scratch may be impractical, but we may be able to create one within an existing, comprehensive medical care delivery system.

In a few countries, medical care data systems allow researchers to link outcomes, such as cancer occurrence, precancerous lesions, and survival, quickly and efficiently to the wide range of potential influences for all residents. In the United States, prepaid health plans, including those of Kaiser, Mayo, Geisinger, and the Department of Veterans Affairs, have conducted studies of cancer etiology and prognosis. To build a study population that is large and diverse enough to support prospective research in the United States, however, we likely need to include participants from more than one health care plan.

In addition to conducting prospective studies set in the general population, we will still need to study unusual populations because they can illuminate the causes of cancer and other major diseases. We will accomplish this type of research by identifying, enrolling, and following groups of people with specific exposures or genetic influences that may affect carcinogenesis: workers who clean up after man-made and natural disasters, families with distinctive genetic abnormalities (e.g., Li-Fraumeni syndrome or BRCA mutations), and patients exposed to a new drug or regimen.

Studying these special populations provides insights into long-term effects that are present but harder to observe among the general population because of lower doses and risks, lower frequency of genetic variants, or simply lack of measurement or documentation. For cohort studies of special populations, we not only can track health events, but we also can collect and store biological and environmental samples for focused studies.

Similarly, we must continue to conduct case-control studies, those in-depth examinations of patients with a particular cancer compared to their counterparts from the community at risk who did not develop the malignancy. Many new approaches are evolving within the basic case-control design, principally with vastly increased use of biological specimens, some collected within moments of cancer diagnosis. Case-control studies have only begun to exploit the mapping techniques and databases that can link participants' histories of homes and jobs to hundreds of associated environmental exposures.

Genetic and Environmental Interaction

Although genetic influences on cancer risk have been recognized for many years, high-throughput genotyping involving genome-wide association studies has transformed research on the etiology of cancer and many other diseases. This technological breakthrough and the agnostic search for signals have opened up avenues for research that otherwise would have remained hidden for years. For instance, we may not know why a variant in a "gene desert" (a stretch of DNA devoid of protein-encoding genes) is consistently associated with risks of breast, prostate, and several other cancers, but we must infer that an important model of genes and cancer induction needs major revision.

Metabolomics (the study of small-molecule metabolites in cells, tissues, and organisms) and proteomics (the study of all the proteins in cells, tissues, and organisms) may turn out to be even more revolutionary than genomics, but that is not clear yet. The mutable measures of metabolism may be more sensitive and accurate in tracking the triggers and stages of cancer development, but their dynamic nature will also make it harder to establish the correct time to perform sampling and to do it reliably. With that caution in mind, metabolomics and proteomics remain two of the most important avenues to pursue in the near term.

"Exposome" is a general term for broadly capturing the wide variety of exposures—radiation related, infectious, chemical, and physical—that a person experiences in the environment. It is too soon to know how this approach will evolve in practice, but it offers great conceptual appeal.

In the meantime, we are experiencing many technical changes that are accelerating the pace of research through the use of small, portable devices. Mobile phones and other handheld units are just beginning to revolutionize the collection of epidemiologic data. For instance, built-in GPS (global positioning system) locators yield data that we can link to geographic databases. Snapshots of plates of food link to nutrition databanks. Surveys of daily diet and exercise can be smartphone applications. Self-sample kits can be used in remote locales. Accelerating such simple technical adaptations offers great promise, in parallel with the hoped-for big-science breakthroughs to sample the entire exposome.

Cancer Outcomes: Precancer, Diagnosis, Prognosis

In recent years, we have experienced important changes in how we assess cancer, how we classify it, and how we view its natural history. First, as a result of refinements in classifying tumor pathology, we are distinguishing among cancers of the same organ and classifying them by etiology as well as prognosis. For example, we recognize the distinct etiology and behavior of estrogen receptor–positive and –negative breast cancers. Second, we are working with a gradually expanding list of recognized precursors, from colon polyps to Barrett's esophagus, that may suggest ways to screen populations and prevent cancer. Third, we often can accelerate epidemiologic study because the precursor may arise years before the cancer does.

To explore the outcome side of the equation, many etiological studies are expanding to include the evaluation of survival. Randomized clinical trials (RCTs) have rigorously tested the effects of an intervention on survival, but they are now gradually incorporating observational (nonrandomized) components, including assessment of genotypes, smoking, obesity, and other factors. For years, many observers have advocated the leveraging of RCTs despite the major logistical challenges that this would pose. With newer methods and advances in information technology, more RCTs will include elements of observational research that are focused on both etiology and prognosis.

The Organizational Impact of Future Research

Big science, team science, and consortia dot the cancer research landscape. The questions we need to ask often require numerous study participants, expertise from multiple scientific disciplines, and efforts that are too large for one institution to conduct alone. NCI has led the effort to create consortia for cohort, case-control, and family studies, enterprises that have been and will continue to be critical for accelerating research to discover the causes of cancer.

A Note on the President's Cancer Panel

The President's Cancer Panel was established by law to monitor the development and execution of the activities of the National Cancer Program and to report on its progress directly to the President of the United States. The panel, which comprises three people, two of whom are distinguished scientists or physicians, meets four times per year. Over the years, DCEG scientists have presented to the panel on a variety of subjects related to the causes of cancer and the means of prevention. Read more information on meetings of the President's Cancer Panel.

Back to the March 2011 issue of Linkage