Multiple Cancer Prevalence: A Growing Challenge in Long-term Survivorship

In the last three decades, cancer has been transformed from a fatal disease to one in which the majority of people diagnosed with cancer receive highly effective treatments that result in either cure or long-term survivorship. The number of people in the United States living with a cancer diagnosis has more than tripled from 1970 (1), and at January 1, 2003, it is estimated to include 10.5 million individuals (2). Factors that will continue to drive the increase in the number of survivors include the aging of the baby-boom generation, longer life-expectancy, increased use of effective cancer screening, and improvements in treatment. Eighteen percent (6.3 million) of the U.S. population, ages ≥65 years, have had a cancer (3). Although most cancer survivors have ages ≥65 years, more than one third have ages of <65 years. It is known that individuals with cancer are at increased risk for subsequent tumors (4-8) and further that, the reported prevalence estimates include people diagnosed with more than one primary malignant tumor. Risk may be due to several reasons, including genetic predisposition, environmental exposure, exposure to carcinogenic cancer therapies (i.e., iatrogenic risks associated with chemotherapy, radiotherapy and/or hormonal therapy), life-style factors, such as smoking, and, perhaps most importantly, aging. Demographic trends make this last risk factor particularly salient. Given that cancer is a disease associated with aging, as the U.S. population ages and cancer survival lengthens, the growing number of survivors with multiple malignant primary tumors will become an increasingly important medical issue, particularly among the elderly.

The aim of this work is to estimate the number of and describe the pattern of disease among cancer survivors living with a history of multiple malignant tumors in the United States. Frequency of multiple primary cancers of the same sites, as well as different sites, are examined, in relation to survivors' sex, age, race, time since diagnosis, and site of first tumor. Most previous analyses of multiple tumors, including a highly cited monograph using the Connecticut Tumor Registry (4), many subsequent papers (5-8), and a recent monograph using data from the Surveillance, Epidemiology, and End Results (SEER) program (9), focus on estimating the risk and causes of developing second or later primaries among specific groups of cancer survivors. These analyses compare the risk of developing a subsequent cancer relative to the general population risk. This paper is distinguished from these prior efforts, in that here, we estimate the number of persons alive diagnosed with multiple cancers rather than the excess risk. This measure represents the current population burden of people affected by cancer more than once and serves as an estimate of the potential financial, emotional, and social effects it creates. The motivation for this effort is the increasing public and research attention being focused on survivorship research and cancer care (10). Although we hear or know of people who have had more than one diagnosis of cancer, to our knowledge, this is the first population-based estimate of the number of U.S. cancer survivors who had more than one diagnosis of cancer. Understanding and knowing the U.S. cancer survivor population with respect to all its dimensionality, as for example, the occurrence of multiple tumors, will help to inform policy makers, researchers, and clinicians on how to better allocate resources and refine research questions geared to improving cancer survivorship outcome.

This report uses incidence and follow-up data from the SEER program registries in five states (Connecticut, Hawaii, Iowa, New Mexico, and Utah) and four metropolitan areas (Atlanta, Detroit, San Francisco-Oakland, and Seattle; ref. 11). Only malignant primary cancers diagnosed between 1975 and 2001 were considered and prevalence is calculated as of January 1, 2002. Each person was classified as having (a) only one primary cancer, (b) more than one primary in the same cancer site, or (c) more than one primary in at least two different cancer sites. The list of main sites used to classify tumors as belonging to the same or different site is given in Appendix A. First site is defined as the first primary cancer reported in the 1975 to 2001 period, although an individual could have had a primary cancer before 1975. This definition is used because it permits data consistency (e.g., Seattle joined SEER in 1974 and Atlanta in 1975) and also because there is a lack of information on cancers diagnosed before 1973. SEER collects the number (but not the site or behavior) of cancers that occur before the start of the registry. Age at prevalence is categorized in ranges (0-19, 20-49, 50-59, 60-69, 70-79, and 80+ years). Prevalence statistics by site of first primary in the 1975 to 2001 period are calculated by counting the person just once for her/his first primary cancer site. Thus, if a woman was diagnosed with a cancer (site unknown) before SEER at age 46 in 1970 and later diagnosed with breast cancer at age of 61 years in 1985 and subsequently developed colorectal cancer at age of 71 years in 1995, she would be classified as follows: breast (first primary), more than one primary mixed site, ages 70 to 79 years (at prevalence). Because all statistics are based on 1975 to 2001 data, her first cancer is ignored in the statistics. For each individual lost to follow-up, his or her probability of being alive at the prevalence time is estimated from an appropriate (age and year at diagnosis, race, sex, and first primary site) survival function, conditional on the time lost to follow-up (12). U.S. prevalence counts are calculated by multiplying the SEER age, sex, and race-specific prevalence proportions (p̂_race,sex,age) by the corresponding January 1, 2002 U.S. population, (N_race,sex,age). Combined U.S. cancer prevalence counts for all races are estimated by summing the counts for Whites/unknown, Blacks, and other races as follows:

Similarly, combined U.S. cancer prevalence counts for all ages are estimated by summing the counts for all ages and so on. To adjust for age differences in different subgroups (sex and/or race), the age-adjusted prevalence percentages are calculated by averaging age-specific prevalence percentages using as weights the 2000 U.S. standard population. Variance for the U.S. prevalence estimates are calculated as

However, they represent a lower bound because they only include the Poisson variability of SEER prevalence estimates (13) and not the variability in extrapolating SEER rates to the U.S. population.

The SEER rules (14) used to determine if the cancer is a later primary cancer consider the cancer site of origin, date of diagnosis, histology, behavior (i.e., in situ versus malignant), and laterality of paired organs. In general terms, tumors in different sites of different histologic types are considered separate primaries. Multiple lesions of different histologic types occurring at the same site are considered separate primaries regardless or whether they occur simultaneously or at different times. Further, a new cancer with the same histology and same site as an earlier one is considered to be the same primary cancer if diagnosed within 2 months or a separate primary tumor if diagnosed after 2 months, unless stated in the medical record to be a recurrent or metastatic disease. If an organ is paired, each member of the pair, generally, is considered a separate site. The SEER rules include several exceptions to the standard rules that pertain to histology, laterality, and dates of diagnosis. Exceptions worth mentioning are multiple invasive adenocarcinomas of the prostate and multiple invasive bladder cancers with histologic codes 8120 to 8130, reported as a single primary with the date of the first invasive lesion. A different set of rules are used to determine multiple primaries of the lymphatic and hematopoietic diseases. Metastatic and/or recurrent tumors, which are tumors that contain cells that are like those in the original (primary) tumor, are excluded from the SEER definition and not reported.

We estimate that of the estimated 9.6 million survivors diagnosed between 1975 and 2001, 7.9% were diagnosed with more than one primary malignant tumor. This represents 756,467 people living with a history of multiple primary cancers (Table 1). Of these individuals, 74% had two or more cancers of different primary sites, whereas 26% were diagnosed with two or more tumors of the same site. Although it seems there is a higher prevalence of multiple tumors among White males and females compared with Blacks, for males this difference is due to an older White population compared with the Black male population. After adjusting to the 2000 U.S. standard population, the differences in cancer prevalence among White and Black males disappear but not for women. Cancer prevalence for White females is higher than for Black females.

Table 2 shows the number of cancer survivors who had at least one primary in the last 27 years, by site of first primary, type of multiple primary, and age at prevalence date. Female breast is by far the site with the highest percentage of individuals with more than one primary of the same site (105,614), followed by colorectal cancer (36,082) and melanoma (15,601). The highest prevalence of multiple tumors, in order, occurs respectively when the first cancer is prostate cancer (94,477), female breast cancer (87,093), colorectal cancer (78,923), and urinary bladder (51,856). It is noted that people, ages ≥50 years, at January 1, 2002, who may have been diagnosed with leukemia during childhood (at ≤19 years), would have received this diagnosis before 1975 and, thus, before SEER data collection. This accounts for the zeros in the cells of Table 2 for people, ages ≥50 years, diagnosed with leukemia as children or adolescents. The cancer sites which show the highest prevalence of multiple tumors among its survivors (ages at diagnosis, >19 years), in order, are leukemia (14%), followed by urinary bladder (11.9%), kidney and renal pelvis (11.3%), and colorectal cancer (11.3%).

Table 3 shows sites of second or later primaries in percentages by site of first primary among survivors with two or more primary cancers. Among males, the most frequent site for a second or later primary is prostate, followed by colon and rectum, urinary bladder, and lung and bronchus. For female survivors, the most frequent second or later sites are breast, female genital system, colon and rectum, and lung and bronchus. For both male and female survivors, there is higher frequency of having later tumors of the same site as the first tumor, as well as tumors in sites which are the most frequent sites of cancer, such as prostate cancer and breast cancer among males and females, respectively. In addition, there is a correlation between multiple cancers across some of the sites, as for example, bladder, kidney and renal pelvis on the one hand, and oral cavity, pharynx, and lung and bronchus for males on the other hand. The latter representing cancer sites associated to a common risk factor, such as smoking (9).

Whereas cancers of the same site seem most frequently within 5 years of the original diagnosis for men and women combined, the most likely time for a second or later tumor of a different site to occur is between 5 and 10 years from the diagnosis of the first primary tumor (Table 4).

Advances in early detection, treatment, and supportive care have led to a record number of cancer survivors or individuals living with a cancer history in the United States. These survivors are teaching us, however, that, for some, cure comes with a significant cost. With the increased survival of cancer patients, the number of people who will go on to have additional tumors is also rising. In the majority of cases, this will be due to simply living longer to ages at which the risk of developing cancer is higher. In 2002, from a total of 125,623 tumors reported to the SEER-9 areas, 20,469 (16%) represented second or later primary cancers. One important area of research has been to monitor and identify the risk factors for development of multiple tumors as a complication of cancer therapy, including second cancers induced by radiotherapy or chemotherapeutic and hormonal agents (7, 8, 15-18). Investigators also have used SEER data to explore hypotheses on the environmental, genetic, life-style, and other causes of increased second cancer risk (4, 9, 19).

To our knowledge, this is the first time that the number of people living with more than one primary cancer in the United States is estimated. We estimate that there are 756,467 people diagnosed with more than one primary cancer in the United States. White women experience the highest burden of both multiple tumors, as well as single tumors. Breast, colon and rectum, and prostate are the sites with the highest prevalence of multiple tumors. We estimate that 192,708, 115,005, and 94,606 people with a first diagnosis of female breast, colorectal, and prostate cancer between 1975 and 2001, respectively, have a second or later primary cancer.

The estimates of the prevalence of multiple tumors provided here are subject to some limitations. First, these estimates reflect cancers diagnosed between 1975 and 2001 and not lifetime or complete prevalence, which represents all previous diagnosis. Lifetime prevalence can be calculated for first primaries only by using a statistical method that estimates the proportion of tumors diagnosed before years with data (20). However, survivors diagnosed between 1975 and 2001 represent a large proportion (95%) of the 10.1 million lifetime cancer survivors at January 1, 2002 (21). Using a more recent database, the lifetime cancer survivors at January 1, 2003, is estimated to be 10.5 million (2). Second, the estimates reflect the current SEER coding rules for multiple primaries, which have been used for over 25 years and are consistent with prior publications (5-9). The current rules may overcount multiple tumors, because some of the tumors with the same histology, diagnosed more than 2 months apart from a previous one and coded as later primary, actually, may represent a recurrence. The current rules have been revised. And beginning in 2007, site-specific rules with different time intervals to distinguish between multiple primaries and recurrences will be used (22). The 2007 rules specify site-specific cut points to distinguish between recurrence and primary tumors. For breast, the cut point is 5 years, whereas, for colon, it is 1 year. The effect of the different cut points for colon and breast are small but not negligible. For example, of the 20,651 second breast tumors reported in SEER from 1975 to 2001, only 357 (1.7%) had the same histology and were diagnosed within 5 years and in the same breast as the first breast cancer. Of the 6784 second colon tumors reported, 358 (4.0%) had the same histology and were diagnosed between 2 months and 1 year from the first colon cancer. The international coding rules (23) are less flexible and report only the first tumor ever. Third, the projection of prevalence proportions from SEER to the U.S. population account for differences in age and race distribution in the United States and SEER but not other factors, such as socio-economic status. Fourth, SEs are based on SEs from the SEER prevalent proportions. They are an underestimate because they do not account for random variation in extrapolating SEER proportions to the United States. The error in the SEER proportion estimates might be much smaller than the error in sampling from the United States.

A final caveat to interpretation of these findings is that a higher prevalence of multiple tumors might be related to a higher overall survival, both high cancer specific survival as well as greater life expectancy (i.e., lower all cause mortality) or higher risk of developing second or later primaries. Thus, the observation that prevalence of multiple tumors was highest among the White female population compared with the Black female population might be explained by higher overall survival for White women, as well as a higher screening rate among the White population (24), which may lead to more diagnoses and longer survival if these cancers are diagnosed at early stages.

With the exception of breast cancer, it is not surprising that the prevalence of multiple cancers where the sites are different is higher than the prevalence of multiple cancers where the sites are the same (Table 2). This is true because although the risk of developing a second cancer of the same site is sometimes substantially elevated above the risk in the general population, the combined likelihood of developing one of the many other types of cancer usually exceeds that of a second cancer of the same site. Female breast cancer is an exception, probably due to the high risk of developing a second breast cancer. The prevalence of multiple cancers of the same site for men diagnosed with prostate cancer is very small because only the first invasive adenocarcinoma of the prostate is counted and recorded in SEER even if a second one is diagnosed 2 or more months after the first one.

In this report, the cancer-survivor population is qualified in terms of being affected more than once by cancer. In terms of quantifying other segments of the cancer survivor population, research is being done to develop methods that estimate the proportions of survivors who will be cured from their cancer (cure prevalence; ref. 25), remain in active therapy (26), are living with a chronic illness or disability, or who will eventually die of their cancer by recurrences. Characterizing the cancer-survivor population is important to better target care and health promotion.

The findings in this report have important implications for public health practice. There is a growing need to promote health and ensure the social, psychological, and economic well-being of long-term cancer survivors and their families (10). People with multiple tumors represent a segment of the cancer survivor population affected more than once by cancer. Despite this, little has been written about the experience of or health-related quality of life outcomes for this vulnerable subpopulation (27), and more research is needed. Attention in particular to the effect of survivors' health behaviors and medical follow-up care on their overall morbidity and mortality is a growing area of clinical and research focus (28-30). Another area in which more information is needed is treatment of second and later cancers, because treatment for the first cancer might limit what can be given when a new cancer is diagnosed. As we learn more about who is at risk for a second or multiple malignancies, it is clear that survivors and their health care providers will need to work in partnership to reduce risk and enhance both survival and the survivorship experience.

Oral cavity and pharynx, esophagus, stomach, small intestine, colon and rectum, anus/anal canal/anorectum, liver and intrahepatic bile duct, gallbladder, other biliary, pancreas, retroperitoneum, peritoneum/omentum/mesentery, other digestive organs, nose/nasal cavity/middle ear, larynx, lung and bronchus, pleura, trachea/mediastinum/other respiratory organs, bones and joints, soft tissue including heart, melanoma of the skin, other nonepithelial skin, breast, female genital system, prostate, testis, penis, other male genital organs, urinary bladder, kidney and renal pelvis, ureter, other urinary organs, eye and orbit, brain and other nervous system, thyroid, other endocrine including thymus, Hodgkin lymphoma, non-Hodgkin lymphoma, myeloma, lymphocytic leukemia, myeloid and monocytic leukemia, other leukemia, mesothelioma, kaposi sarcoma, and miscellaneous.