Cancer Characteristics, Definitions, and Recent Investigations

Cancer is not one disease, but rather many different diseases with different causal mechanisms that share a similar characteristic: uncontrollable cell growth and division (3,4). Cancers, as a group, are very common. Cancers are the second leading cause of death in the United States, exceeded only by diseases of the heart and circulatory system (3,4). According to the American Cancer Society, one of every two men and one of every three women will be diagnosed with some form of cancer at some time in their lives (5,6). Different cancers also have varying latency periods, which can also present challenges in the evaluations of cancer patterns (7,8).

Given the frequency with which cancers are diagnosed, situations may arise where an unusual number of primary-site cancers (i.e., place in the body where the cancer started) are diagnosed among people in a particular location. It is possible this may be due to chance. These unusual patterns may also result from the following:

  • Differential recommended cancer screening practices
  • Access to health care, which may be more reflective of other social and economic factors (e.g., limited access to optimal healthcare services)
  • Genetic susceptibility to a particular cancer
  • Behavioral risks and social determinants of health, occupational exposures, and in some cases, exposures to environmental sources

Revised Definition of a Cancer Cluster and Consideration of Unusual Patterns

The intent of these guidelines is to enhance the historic approach for investigating community cancer and environmental concerns.

The 2013 Guidelines defined a cancer cluster as “a greater than expected number of cancer cases that occurs within a group of people in a geographic area over a defined period of time.” CDC/ATSDR revised the definition to recognize that some cancers may be similar etiologically (in terms of risk factors, causes, or origin), and present the revised definition of a cancer cluster as “a greater than expected number of the same or etiologically related cancer cases that occurs within a group of people in a geographic area over a defined period of time.”

This definition can be further understood as follows:

  • A greater than expected number: When the number of observed cases is greater than typically observed in a similar setting.
  • Of the same or etiologically related cancer cases: Cases are of the same type, are within a family of tumors (e.g., Ewing’s family of tumors), or have a known or suggested link to the same specific environmental or chemical exposures. It is possible to consider multiple cancer types when such a known exposure (e.g., radiation or a specific chemical) is linked to more than one cancer type or when more than one contaminant or exposure type has been identified.
  • Within a group of people: The population in which the cancer cases are occurring is defined by its demographic factors (e.g., race, ethnicity, age, and sex).
  • In a geographic area: The geographic area may be based upon pre-existing geopolitical boundaries (e.g., census tract, county, or ZIP code/ZIP code tabulation area). It may be defined according to the nature and extent of potential exposures that may cross multiple or partial boundaries. For example, air pollution from a hazardous waste incinerator which may cross multiple counties or census tracts. These geographic boundaries are used to determine the number of cancer cases as they relate to the total population in this predefined area. It is possible to create or obscure a cluster inadvertently by modifying the area of interest.
  • Over a period of time: The time frame used to establish the beginning and end dates for analysis. The time period chosen for analysis will affect both the total cases observed and the calculation of the expected incidence of cancer in the population.

It is possible that not every unusual pattern will meet the definition of a cluster as described above. However, unusual patterns that meet some of the criteria and also have plausible environmental concerns still warrant further evaluation or assessment. For example, many of the same cancer cases may be present but may be dependent upon a factor such as a water distribution system rather than a traditional boundary like a census tract or county.

With the advent of spatial statistics new approaches to determining if the probability of a spatial pattern is likely, locating and quantifying the strength of spatial associations, and examining possible relationships between a focused location and event occurrence are possible. Technical information about the application of spatial and temporal methods are provided in Appendix B. Unusual patterns and clusters are described in Phase 2.

Limitations Associated with the Investigation of Unusual Patterns of Cancer

Because cancer as a group of diseases is common, cases might appear to occur atypically within a community. As the U.S. population ages, and as cancer survival rates continue to improve, in any given community, many residents will have had some type of cancer, thus adding to the perception of excess cancer cases in a community. Multiple factors affect the likelihood of developing cancer, including age, genetic factors, and lifestyle behaviors (9). An excess of observed cancer cases within a given population may occur as a result of statistical fluctuation and random chance and may also occur without a discernible cause (10,11), given:

  • Many types of cancer vary in etiologies, predisposing factors, target organs, and rates of occurrence.
  • Cancers often are caused by a combination of factors that interact in ways that are not fully understood.
  • For the majority of adult cancers, the long latency period (i.e., the time between exposure to a causal agent and the first appearance of symptoms and signs) complicates attempts to associate cancers occurring at a given time in a community with local environmental contamination. Often decades intervene between the exposures that initiate and promote the cancer process and the development of clinically detectable disease (12).
  • Population mobility can also impact the ability to optimally determine the impact of environmental exposures on disease occurrence.

Although the causes of many cancers are unknown, some causal relationships have been shown between environmental exposures and development of cancer in specific organs (e.g., inhalation of asbestos and mesothelioma)(13). According to inputs received for these guidelines, health departments respond to inquiries that typically stem from concerns that are predominately environmental. These guidelines encourage the exploration of environmental concerns as further described in Phase 2.

Recent Cancer Cluster Inquiries and Investigations

The STLT survey asked health agencies to estimate the number of inquiries they receive about excess cancer in the last 7 years (2013–2019). Based on the survey results, 53% of STLT agencies received 1–5 inquiries each year, 23% received 6–10 inquiries, 15% received 11–25 inquiries, and 6% received more than 25 each year (note: Two agencies reported that they did not know how many inquiries they received). STLT agencies received inquiries most often from individual residents, physicians and healthcare providers, and community advocacy groups. More information from the STLT survey can be found in the report [PDF – 1 MB].

CDC/ATSDR also conducted a comprehensive literature review to identify published studies investigating non-occupational or non-disaster potential cancer clusters in the United States within the last 10 years (January 1, 2010−April 8, 2020). Manuscripts reviewed discussed epidemiologic investigations of cancer clusters and/or elevated cancer incidence in communities (14–23). Most investigations were prompted by a suspected environmental exposure. A brief summary of the articles is provided below.

Overall, the findings revealed the following:

  • The populations within these investigations varied in size, age group, and by cancer types studied.
  • Ecological study designs were most often implemented, followed by case-control and cross-sectional study designs.
  • Cancer registry data were used for cancer outcome data in the majority of studies; alternately, some studies used medical records as the source of cancer outcome data.
  • Potential exposures were ascertained using a variety of methods including surveys, environmental sampling, biological sampling, interviews, surrogate exposures (e.g., using the distance from a nuclear power plant to a ZIP Code as a proxy for individual exposure (14)), and pre-existing data from the Environmental Protection Agency and/or state health departments.
  • The majority of studies used geospatial methods to identify spatial patterns or to generate standardized incidence ratio maps.
  • Results of the studies varied. While some were unable to identify associations between cancer and environmental exposures (14–16), several demonstrated statistically significant associations, with noted limitations (17–23).

From this literature review, several articles reported a statistically significant association between the specified cancer and the exposure measure of interest but cited some limitations with interpreting the data. Of note, in epidemiology an association does not equate to causation. Many of these limitations are inherent to an ecological study design (see Phase 3 for more details). Limitations include the inability to measure the exact exposure biologically and instead relying on proximity to sites as an opportunity for exposure, the inability to measure lifestyle or behavioral risk factors that are associated with increased cancer risk, and challenges with modeling water or air movements that impact potential exposures.

Recognizing that the work of state health partners may be released as reports and not as published literature, a search of the gray literature was conducted to supplement the articles identified in the peer-reviewed literature. These reports aligned with feedback from the STLT and community focus groups and provided information relative to techniques and methodologies being used across public health agencies.