Recommendations for future research
Step One
As a first step, dozens of descriptive studies could be conducted at relatively little cost using data linkage. For example, recent reports of breast cancer clusters in California could be evaluated using incidence rates mapped to census tracts and other units smaller than the county level. In this manner, one could avoid invalid comparisons of small counties (e.g. Marin) with large, more heterogeneous counties (e.g. Los Angeles). New statistical applications, including Bayesian methods, need to be developed to make such comparisons, and could be developed under the sponsorship of the California Breast Cancer Research Program.
Step Two
As a second step, complementary data collection sponsored by the California Breast Cancer Research Program could be used to maximize the potential of existing infrastructures. For example, databases maintained by the California Cancer Registry could be supplemented with survey information. Residential history could be collected routinely on a subset of cancer patients, or even all patients with certain forms of cancer, and integrated with the Cancer Registry database. Existing databases containing information on pesticide use, water quality, and other environmental exposures could then be linked to Cancer Registry data to generate new hypotheses about potential etiologic agents. Survey sampling methods could be used to gather the information needed to examine correlations between breast cancer incidence and distributions of known breast cancer risk factors, screening, social class, and access to health care. Such investigations are necessary to address potential confounding variables in studies of the environment, as well as to identify inequalities in access to health care. Studies of migrant communities and cross-regional comparisons of breast cancer rates will further help to uncover modifiable risk factors that are amenable to population based interventions to reduce the burden of breast cancer in California.
Step Three
As a third step, researchers in academic institutions in California could be funded through the California Breast Cancer Research Program to conduct focused epidemiologic studies to better understand novel etiologic agents. Many of these institutions already have experience in epidemiologic studies and could be sponsored to expand the scope of existing studies or initiate new studies. Many investigators have already collected DNA from study participants and could be sponsored to study interactions between newly identified candidate environmental exposures and genetic susceptibility. Genetic markers may be quite useful in identifying low-dose environmental effects and helping to uncover biologic mechanisms, so genetics research and research on the environment can work hand in hand. Most agencies that fund breast cancer research list gene-environment interaction as a high priority. However, most of the emphasis and funding has focused on genes, not the environment. Better survey instruments, studies that link residential history to existing environmental databases, and new biomarkers of environmental exposure are areas that merit extensive research.
Examples
Specific examples of the types of studies that could be performed are listed below. In these proposed studies, environmental factors are not studied in isolation but within a greater context that includes social class and urban rural differences at the macro level, and genetic susceptibility and biological mechanisms at the micro level. Only by integrating research across all of these areas, a feat that has never been previously accomplished or proposed by any funding agency, is research on breast cancer and the environment likely to succeed.
a. Descriptive studies
Public health surveillance through improved monitoring of breast cancer rates was a primary recommendation of the summit on breast cancer and the environment held in Santa Cruz, California, in 2002. Studies are needed to develop methods for calculating breast cancer incidence and mortality rates at the census tract and other subcounty levels. Newly developed Bayesian estimation procedures could be applied, such as those developed by Tom Mack at the University of Southern California. Newer cluster methods developed by Kuldorff (1997) and others could be used to identify regions with higher and lower breast cancer rates. Such analyses would need to take into account differences in known risk factors, screening, and population demographics. Once regions of interest were identified, ecologic studies linking cancer rates to environmental databases could be conducted.
Cancer rates also need to be studied over time among migrants. Periodic surveys of immigrants could provide clues regarding the role of reproductive patterns, diet, and physical activity.
b. Urban rural differences
A comprehensive, statewide study could be conducted comparing breast cancer incidence rates in regions defined according to newly developed indices of urban/ rural status. Urban Influence Codes (UICs) (Baer, 1997) were developed by the US Department of Agriculture as a county level measurement that captures adjacency to large metropolitan areas. Using UICs, rural areas are not categorized according to population density alone, but by proximity to urban areas. Rural Urban Commuting Area Codes (RUCACS) were developed by the US Census Bureau (Comartie 1996; University of Washington School of Medicine, 2003). RUCACS are used at the census tract level and categorize communities according to traffic flow where people commute for employment. They are especially useful where there are pockets of urbanization within non-metropolitan areas (so-called micropolitan areas) (Hewitt, 1992).
c. Studies of social class and traditional risk factors
Ecologic studies could be performed linking incidence rates to distributions of known risk factors, screening practices, and social class. Group level proxies for physical activity (availability of bike trails, parks, etc.) could also be explored. Currently funded studies such as that of Margaret Wrensch at the University of California, San Francisco, examining risk factors for breast cancer in Marin County, including adolescent exposures, are particularly important. Several measures have been created and validated to capture information on social class that could be incorporated into cancer registry data collection (Krieger et al., 1997). Information on known risk factors is also important when evaluating regional differences in breast cancer, and may make important contributions to areas where rates appear to be elevated (Robbins et al., 1997).
d. Studies of environmental contamination
Once information is collected at the aggregate level on potential confounding variables, one could begin to explore the relationship between regional (group-level) environmental exposures and breast cancer. One could link residential history information to regional information regarding crop and pesticide use practices. For example, ecologic studies could be performed linking breast cancer incidence rates to Proposition 65-mandated reporting of pesticide use and related databases. Geo-coding and GIS methods could be used to link incidence rates to levels of environmental pollutants, toxins, and ambient levels of light at night. Residential history information or some proxy of migration frequency would need to be included. Analytic methods have been developed for measuring levels of endocrinedisrupting chemicals in ground water and other locations that could be used to characterize census tracts (Rudel et al. 1998).
One potential statistical technique that takes advantage of regional information on health outcomes is contextual analysis, in which group effects are estimated after controlling for individual-level covariates (Humphreys, 1991). Studies of pesticide exposure, for example, could be modeled at the group level using pesticide use databases and at the individual level using blood samples. One could collect residential history information (for example, for ten years prior to diagnosis of breast cancer), and link water quality, crop spraying, and other ecologic level information, while adjusting for individual level risk factors that are potential confounders as determined from questionnaires.
Surveys of wildlife, including reproductive problems, birth defects, and fetal malformations, could be used to characterize census tracts according to potential exposure to endocrine disruptors and other chemicals. Health outcomes in commercial and domestic animals might also prove useful in this regard. This information could be incorporated into ecologic studies of patterns of breast cancer incidence in women.
e. Incidence rates for subtypes of breast cancer
Incidence rates for different stages of breast cancer, specific histologic subtypes of breast cancer (in particular, inflammatory breast cancer), and subtypes according to hormone receptor status could be analyzed. Subtypes of breast cancer may have distinct etiologies and show up in specific geographic patterns, or according to social class or other characteristics of breast cancer patients. Incidence rates and distributions for multiple primary cancers (breast as well as other types of cancer) could be evaluated.
f. Interdisciplinary investigations
The summit on breast cancer and the environment held in Santa Cruz, California, in 2002 recommended that multi-disciplinary and inter-disciplinary approaches be used to study breast cancer and the environment. Toxicologists, epidemiologists, biologists, geneticists, social scientists, and others all need to be involved in studies where they act not as separate investigators, but as part of research teams. The interdisciplinary research team approach is the foundation for the proposed Centers for the Study of Breast Cancer and the Environment to be funded by the National Institute of Environmental Health Sciences.
Exposure assessment
One of the most important areas of investigation is improved exposure assessment. Direct measurements of persons (e.g., blood levels of compounds) and the environment (e.g. water sampling) need to be refined and correlated for a number of candidate compounds. But many chemical residues are non-persistent, and we need better ways to model exposure. Some of these exposures may leave traces in the body, measured as increased expression of specific genes in circulating lymphocytes, by modulation of single transduction and by changes in the immune system. Toxicologists working with epidemiologists could develop new biomarkers of exposure, as well as early disease, for studies of breast cancer and the environment.
Gene-environment interaction
It is important to collect more information on the environment while simultaneously investigating mechanisms of disease and susceptibility to exposure. One of the most promising areas of research is studies of geneenvironment interaction. Such studies were recommended as a promising area of research into the role of cancer and the environment in a recent report by the Institute of Medicine. To further such investigations in an efficient and timely fashion, it would be useful to provide supplemental funding for investigators with existing epidemiologic studies that collected DNA samples. Studies of gene-environment interaction would be an efficient use of resources, since only funds for genotyping would need to be provided. However, one needs to think beyond the paradigm of genes involved in hormone metabolism. Candidate genes include variants involved in DNA repair, oxidative stress, and carcinogen metabolism.
Few studies have focused on mutations in P53 and other somatic alterations in breast tumors. It would make sense, when trying to understand etiology, to look first at breast tumors and discern whether patterns of somatic alterations provide clues to etiology. Mutations in P53 and other genes can be classified according to frequency, location, and type of base change (transition, transversion, deletion, etc.), and correlations with exposure histories conducted (Conway, 2002). Novel studies using new technologies for somatic changes could also be funded. For example, comparisons of cDNA expression array profiles for cancers that appear to occur in space-time clusters. Do such tumors share patterns of somatic alteration, suggesting a common etiology?
Augmenting existing epidemiologic studies
The CBCRP could fund studies to expand the environmental portion of existing epidemiologic studies. If study participants have previously collected residence histories, this information can be geocoded and linked to environmental databases funded by the CBCRP. Study participants who provided residential histories could then be categorized according to potential for exposure to water contaminants and a variety of environmental exposures. This information could be studied in conjunction with previously collected information on reproductive histories and other traditional risk factors for breast cancer.
g. Breast biology and breast development
Until more is known about normal breast development and breast biology, studies of environmental risk factors for breast cancer rest on a weak foundation. Both the Institute of Medicine report (2002) and the NCI Progress Review Group on Breast Cancer (1998) listed breast biology and breast development as important areas for future research. The NCI/NIEHS funded Centers for Breast Cancer and the Environment include development of animal models for breast development and susceptibility, and will examine the effects of chemical carcinogens and endocrine disruptors on breast biology. But more studies are needed of breast metabolism in humans, including the distribution and metabolism of environmental chemicals in breast tissue. Studies such as that of Peggy Reynolds at the California Department of Health Services and Vicki Davis at Cedars-Sinai of xenoestrogens in breast tissue should be encouraged, and could be expanded to understand how these compounds affect breast development. The previously mentioned studies of EBV and BLV and breast cancer will also yield important information with regard to breast biology.
