In 1926, the Yale-New Haven Hospital set up the first cancer registry in the United States, and in 1956, the American College of Surgeons (ACoS) launched a program to encourage hospital-based cancer registries. However, based on individual card files, these data were of little use to physicians and researchers, and it would take until the advent of computerized registry systems that facilitated pooling and analyzing regional and national data for cancer statistics to become routine and useful. The purpose of cancer registries is to collect, manage, and analyze data on cancer patients in order to uncover ethnic and gender differences, and possible causal relationships to potentially hazardous agents or behaviors. There are two types of cancer registries: hospital-based and population-based, which can be administrative, research, or cancer control oriented. The Surveillance, Epidemiology and End Results (SEER), the first population-based registry, was established in 1973 by NCI. It began collecting cancer incidence and survival data on January 1, 1973 from the states of Connecticut, Iowa, New Mexico, Utah, and Hawaii and the metropolitan areas of Detroit and San Francisco-Oakland. Atlanta and the 13 Seattle-Puget Sound counties were added in 1974–1975, as were 10 predominantly black rural counties in Georgia (1978) and American Indian areas in Arizona (1980). In 1992, minority Hispanic populations living in Los Angeles County and the San Jose-Monterrey area were added. In 2001, coverage was expanded to Kentucky, New Jersey, and the previously uncovered portions of California, and in 2010, coverage was extended to the state of Georgia. Information on cancer cases is also collected by NCI from Alaska natives. Currently (2012), SEER collects and publishes cancer incidence and survival data from population-based cancer registries covering approximately 28 % of the US population. Although SEER does not cover the entire US population, validation studies based on the recorded cause of death for 17 cancer sites representing two thirds of cancer cases in the United States revealed a 90 % correlation [292]. Data collected by SEER include patient demographics, primary tumor site, morphology, stage at diagnosis, first course of treatment, and patient survival, which are essential for planning and monitoring cancer control strategies, for identifying priorities in public health, and for allocating resources for the prevention and treatment of cancer. In 1992, Congress passed the Cancer Registries Amendment Act, establishing the National Program of Cancer Registries (NPCR) to be administered by the CDC. Since 1994, “NPCR and SEER together collect cancer data for the entire U.S. population. CDC and NCI, in collaboration with the North American Association of Central Cancer Registries, have been publishing annual federal cancer statistics in the United States” [293]. Based on these databases, the ACS compiles yearly estimates of cancer incidence, mortality, and survival, standardized to the 2000 United States population. At the international level, the IARC, a non-governmental organization founded in 1966, is dedicated to fostering the aims and activities of cancer registries worldwide in cooperation with the WHO. It must be noted that some countries rely on regional rather than nationwide databases to estimate cancer incidence, which are reported to IARC via the WHO, somewhat compromising the accuracy of their cancer statistics. Nevertheless, IARC’s “Cancer Incidence in Five Continents” series of monographs, updated every 5 years, has become the reference source on the global incidence of cancer. In contrast to population-based surveillance data that reflect national rates and trends, hospital registries reflect the type of practice, catchment area, and other factors peculiar to each institution.

Cancer incidence and mortality both can be expressed as total number of cases in a population over a particular period of time and can be sorted by site, region, race, age, and other demographics. For example, 1,638,910 million Americans are expected to develop cancer in 2012; 848,170 males and 790,740 females [294]. However, total cancer cases vary with population size, age composition, and other factors, thus precluding detailed comparisons of cancer incidence or trends over time in the same country, or between countries with populations of different demographics. This problem is overcome by expressing the incidence, mortality, and other statistics for each 100,000 people in the total population surveyed, or in any segment thereof (males, whites, etc.), adjusted for age distribution in the overall population. The latter is necessary because cancer predominates in the elderly but certain types of cancer are age-dependent. For example, in the US, the median age of cancer patients at diagnosis in the 2005–2009 period was 66, but 33 for testicular cancer, 14 for ALL [295], and 80 % of inherited retinoblastomas are diagnosed before age 3. Hence, the incidence of prostate cancer in 2009, for instance, can be reported as 151.9 per 100,000 men of all-ages and all races, as 62.2 per 100,000 men of all races below age 65, or further broken down by other demographics or tumor stage [296]. These adjustments enable comparing cancer rates over time in the same country and between countries with different population size and demographics composition. There is, however, one caveat: In the US, rates per 100,000 are age-adjusted to the 2000 US population, whereas data reported by the IARC are standardized to the 1980 age-adjusted OECD population and other international reports use the Segi’s or WHO’s age-adjusted “world” populations¹ to remove cross-country age variations. Unless otherwise specified, in this book, all references to cancer incidence and mortality rates will be population- and age-adjusted.

The American Cancer Society publishes yearly estimates of the numbers of new cancer cases and cancer deaths expected in the United States, based on last available actual rates (usually 5 years in arrears) projected onto yearly estimates of the size and age distribution of the United States population. While these estimates are only projections, they have proved reasonably accurate when compared to actual data gathered and tabulated several years later, thus justifying their interim use. The American Cancer Society estimates that 1,660,290 (averaging 4,500 each day) Americans will develop cancer and 580,350 (or 1,600 each day) will die of the disease in 2013 (Table 5.1) [297]. There are well over 200 different types of cancer and the ACS reports on approximately 50. However, their relative incidence and deaths rates within a population vary greatly. Indeed, five cancers account for nearly 2/3 of all new male and female cancers expected in 2013 and approximately 60 % of all cancer deaths in American men and women (Fig. 5.1), two of which (lung and colorectal) are lifestyle related. Interestingly, the projected two leading cancer deaths in men (prostate and lung) and in women (breast and lung) accounted for approximately the same fractions of new cases (42 and 43 %) and deaths (38 and 40 %), respectively, in 2013 and 1995 [298].

The cumulative life-long risk of developing any invasive cancer was nearly 1 in 2 (45 %) for an American male and more than 1 in 3 (38 %) for an American female in the 2007–2009 period (Table 5.2). Likewise, in the same timeframe the cumulative life-long risk of dying from cancer was nearly 1 in 4 for an American male (23 %) and 1 in 5 for an American female (19 %). However, the risk of developing and dying of cancer from any particular type of cancer is both gender and age-dependent. For example, while the male cumulative life-long risk of dying of prostate cancer (1 in 36) was identical to a woman’s cumulative risk of dying of breast cancer (1 in 36) during the 2007–2009 period, only 1 in 7,964 men developed prostate cancer before age 40 whereas 1 woman in 202 developed breast cancer by the same age [299]. In the same timeframe, the leading cause of cancer death in men through age 40 was leukemia, whereas it was lung cancer after age 40. In contrast, the leading type of cancer death in women was brain before age 20, breast between ages 20 and 59, and lung after age 60 [300]. However, these statistics can change over time. For instance, as increasing numbers of smoking adolescent females come to age and breast cancer is increasingly detected in surgically curable early stages, mortality rates from lung cancer will likely shift to younger females, eventually replacing breast cancer after age 40.

Cancer prevalence refers to the number of individuals with any type of cancer alive at survey time regardless of when the diagnosis was established and whether they are cured, dying of the disease, or somewhere in between. Non-melanoma skin cancers are usually excluded. In essence, prevalence includes all cases of new and preexisting cancers that are alive at a particular time regardless of cancer status. Collection of such data requires a sufficient period of time to capture all previously diagnosed cases. In the US, the Connecticut Registry is the only registry with sufficient follow-up data (cancers diagnosed after 1935) enabling calculation of cancer prevalence. Prevalence data from this regional registry is extrapolated nationwide based on the total US population. The major interest of cancer prevalence data is to policy-makers, for it identifies the level of human and financial burden imposed by cancer on the health care system and the level of support required from public and private sources. As of January 1st, 2009, 12,549,000 Americans were alive with cancer: 5,809,000 (46 %) of these were men and 6,740,000 (54 %) were women [301]. Prevalence of most cancers was gender-unrelated, fairly evenly distributed between the genders. However, some were gender-impacted, including 65 % of oropharynx, 74 % of urinary bladder, 78 % of esophageal, and 81 % of larynx, and 100 % of prostate cancers in men, and 100 % of uterine, 98 % of breast, and 78 % of thyroid cancers in women. The four most prevalent cancers accounted for 61 % of all cancer patients alive in 2009. They were: female breast (22 % of the total), prostate (19 %), colorectal (11 %), and gynecologic (9 %) cancer. This is not surprising because the most prevalent cancers are those with high incidence rates and long survivals, which in turn hinges on being more amenable to early stage diagnosis and on exhibiting relatively slow tumor growth than most. For instance, in the 2005–2008 period, female breast cancer was diagnosed while confined to the primary site in 60 % of the cases and 98 % survived 5-years [302]. In the same timeframe, 81 % of prostate cancer cases were diagnosed as localized disease and exhibited a 100 % 5-year survival [303]. In fact, “prostate cancer that is present in the prostate gland but never detected or diagnosed during a patient’s life…is greater than the number of men with clinically detected disease” [304]. These usually elderly men harboring unsuspected prostate cancer die of old age or of unrelated causes. On the other hand, aggressive cancers that are less frequently diagnosed in early stages, inoperable, or unresponsive to chemo- or radiation therapy represent a very small fraction of cancers included in any prevalence report. Such is the case of pancreatic cancer that even in its earliest stage (IA) exhibits a meager 2 % 5-year survival.