Cancer Incidence 1992-1996

Index for Tables

Age-Standardized Incidence Rates

Table 2.1 provides a detailed profile of the statewide incidence of cancer in Minnesota for the five-year period, 1992-1996. (County-specific cancer data are provided in Appendix A PDF (274kb\96 pages).) Table 2.1 contains the number of cancers diagnosed in Minnesota and the corresponding age-standardized rate. Age-standardization adjusts for the distribution of the Minnesota population so that comparisons can be made to national age-standardized rates. (See Appendix C and Appendix D for a more complete definition of age-standardized incidence rates.) The national data are for the period 1991-1995 from nine regions of the SEER program of the National Cancer Institute. SEER refers to "Surveillance, Epidemiology and End Results," and represents separate population-based cancer reporting systems in various states and metropolitan areas. The nine regions encompass about 10 percent of the U.S. population. Although these rates are usually considered "national" rates and will be referred to as such in this report, it should be understood the SEER rates are not necessarily representative of the whole U.S. or the Midwest, and that rates vary, often significantly, from one SEER region to another. In addition, rates in Table 2.1 are for all races combined, and since nonwhite people are a greater percentage of the U.S. population than in Minnesota, the comparability of the two rates is decreased. (A complete discussion of the SEER program's comparability to the MCSS and the importance of racial and other data to interpreting cancer surveillance data is given in Section 2 of the MCSS 1993 report.) A description of the data presented in Table 2.1 using the category of "all cancers" as an example follows. During the period 1992-1996, a total of 52,803 cancers were diagnosed in Minnesota males, an average of 10,561 new cancers per year. This translates into an estimated 464.4 cancers per 100,000 males per year, an average annual rate that accounts for the age distribution of Minnesota males. The U.S. estimate was 501.0 cancers per 100,000 males per year. A total of 47,074 cancers were diagnosed among females during 1992-1996 for an average of 9,415 new cancers per year. The average annual rate among Minnesota females was 334.9 cancers per 100,000 per year, and the U.S. estimate was 347.5 per 100,000 per year. Interpretation of average annual 1992-1996 Minnesota rates, especially in comparison with national data, requires consideration of several factors in addition to the issue of racial composition discussed above. First, it is important to recognize that at the time of the data analysis for this report the most recent five-year period for which national data was available was 1991-1995. (Several years of cancer data are usually combined to provide a more precise estimate of the cancer risk.) The slight difference in time periods (1991-1995 vs. 1992-1996) has modest significance for the many cancers that have shown little if any change in rates during recent years. However, this time difference is important to consider for the cancers that have shown significant changes (increases or decreases) in rates over these years, e.g. prostate cancer. Second, pancreatic cancers had a lower incidence rate in Minnesota than nationally: 8.3 versus 10.3 per 100,000 per year in males and 5.7 versus 7.8 per 100,000 per year in females. This difference is an artifact of the pathology-based nature of the MCSS. Pancreatic cancer is generally a rapidly fatal disease, and sometimes a decision is made not to put the patient through the trauma of obtaining a biopsy. This is especially apparent in the oldest age groups, where the discrepancy between MCSS and national SEER rates is largest. The fact that the pancreatic cancer mortality rates in Minnesota were higher than the pancreatic cancer incidence rates is further evidence of the artifactual nature of the low Minnesota pancreatic cancer "incidence" (Tables 2.1 and 3.1). This is also true for liver cancer and to a lesser extent for lung cancer. The cost of obtaining all of the nonmicroscopically-confirmed cancers has not justified their ascertainment to date. Third, whenever a large number of comparisons are made, some differences will occur by chance alone. It is very difficult to sort out which differences between Minnesota and national data represent real differences in the rate of cancer occurrence and which are the result of the variable nature of these data. The methodological and demographic differences between MCSS and SEER make this task even more difficult. Because of these considerations, Minnesota and SEER rates have not been compared in a statistical sense to identify "significant" differences.

Several important conclusions can be drawn from Table 2.1. First, cancer rates in Minnesota are slightly lower than national rates for many types of cancer, which results in lower all cancer combined rates. The lower rate of lung cancer in Minnesota that has been associated with differences in smoking experiences (MCSS 1993, Section 3) accounts for some of this difference. Cancer rates for females in Minnesota are more similar to national rates though again lower lung cancer rates contribute to slightly lower all cancers combined rates for Minnesotans compared to national rates. Second, cancer rates are substantially higher (about 40 percent overall) in men than women. In fact, for those cancers that affect both sexes, cancer rates are higher in men for almost every type of cancer. Two major exceptions are cancers of the thyroid gland and gallbladder. Third, rates of specific types of cancer vary enormously. (See Section 4 for a more complete demonstration of year-to-year variability.) This is consistent with the fact that cancer is really many different diseases with different causes, natural histories, etc. Fourth, relatively few types of cancer account for the majority of cancers in men and women. By far, the most common cancer among men was prostate cancer (3,536 per year), while the most common cancer among women was breast cancer (3,043 per year). Lung cancer and colon cancer were the second and third most common cancers in both sexes. In men, cancers of the prostate, lung, colon, and rectum accounted for 58 percent of all cancers diagnosed between 1992 and 1996 in Minnesota. In women, cancers of the breast, lung, colon, and rectum accounted for 55 percent of all cancers, while cancers of the uterus and ovary accounted for another 11 percent.

County Cancer Incidence

As with previous MCSS reports, this report provides profiles of cancer occurrences for each of the 87 Minnesota counties (Appendix A). Cancer information at the county level has proven to be useful in the planning and development of health care resources and in a variety of educational programs. The information provided consists of the actual number of newly-diagnosed cancers that occurred among county residents during the five-year period 1992-1996. For perspective, a second number is also provided showing the estimated number of cancers that would have been expected in the county based on statewide rates and the population of the county. This information is presented for each major type of cancer and for all cancers combined. The practice of publishing county profiles has intentionally excluded statistical comparisons of individual rates among the counties. There are several reasons why such comparisons are not attempted. These reasons include the instability of cancer incidence rates due to relatively small numbers, even for the more common types of cancer; county differences in major risk factors (such as smoking, racial composition, diet) that are not accounted for in MCSS data; and the likelihood of many apparent rate differences that would be expected from chance alone.

Age-Specific Cancer Incidence Rates

Tables 2.2 and 2.3 contain cancer incidence rates for males and females for 18 separate age categories. These are referred to as age-specific rates. Although a few cancers, such as Hodgkin's Disease, show relatively modest changes over broad age ranges, the rates of most cancers vary significantly with age, generally showing sharp increases with age. Figure 2.1a shows graphically how cancer rates for all cancers combined vary with age for both males and females. Because there is over a 200-fold variation in overall rates depending on age, it is difficult to depict age-specific rates in a single graph. Therefore, Figure 2.1b shows the same data using a different scale. The first graph shows the rates on a conventional "linear" scale. On this scale, rates during childhood and young adulthood appear to be virtually zero and cannot be distinguished from each other. Figure 2.1a does illustrate, however, that most of our cancer risk occurs after middle age. The second graph shows the rates on a "log" scale (factors of 10) which can represent a much larger range of values. The scale of this graph spreads out the increases over a larger range and allows for detail of the younger age groups to be expressed

Median Age at Cancer Diagnosis

Table 2.4 contains the median age at diagnosis for selected cancers. The median is the 50th percentile: half of the patients were younger and half were older when diagnosed with the cancer. The median age is similar to an average age in that it is a measure of central tendency of the age distribution. Unlike an average, the median is not affected by extreme (younger or older) ages. Table 2.4 demonstrates two important aspects of cancer incidence. First, most cancers are diagnosed in older Minnesotans. The median age at diagnosis for all cancers was 69 years in males and 67 years in females. In other words, one-half of the male cancer diagnoses were in individuals older than 69 years and one-half of the females were older than 67 years. Second, there is a great range in the median age at diagnosis from 31 years for cancer of the bones and joints in males to 76 years of age for cancer of the stomach in females. The large range in age at diagnosis reflects the fact that the natural history varies greatly between different types of cancer. That the risk factors and natural history varies between different cancers is also exhibited in the many differences between median ages at diagnosis for males and females.

Lifetime Cancer Risk

In this section another measure of cancer occurrence is given, one that provides a quite different perspective on cancer occurrence. This statistic is the lifetime risk of cancer among Minnesota residents. Statements of lifetime risks are frequently reported in the media, and this may well be the type of cancer statistic that people have heard about most often. Because of the many important uses and implications of such projections to the residents of Minnesota, the MCSS has developed estimates of lifetime risks specific to Minnesota.

There are several major difficulties in estimating and interpreting lifetime cancer risks. First, there is the issue of what is meant by a "lifetime." This is important to consider because cancer rates increase dramatically with age (Figure 2.1a). For example, men in their early 70s have more than twice the overall cancer incidence of men in their early 60s (3,083 vs. 1,503 cancers per 100,000 per year). The overall rate increases by 53 percent (from 1,042 to 1,598 per 100,000) in females during the same age interval. For most people, the strongest determinant of cancer risk is their age. That represents a good news/bad news situation for Minnesotans. The good news is that life expectancy in Minnesota is near the highest in the U.S. Current (1996) Minnesota mortality rates project that 50 percent of the males born today will be alive at age 79 and that 50 percent of the females born today will live to be age 84; while 25 percent of the males will survive to be age 86 and 25 percent of the females will live to age 92. The bad news is that since more people are living well into the ages with the highest cancer rates, more Minnesotans are also developing cancer. MCSS estimates of lifetime cancer risk take into account the greater life expectancy in Minnesota. Second, life expectancy will almost certainly change in the future, and to that extent, lifetime cancer risks would also be expected to change. As life expectancy increases, lifetime cancer risks will also increase if cancer rates remain the same. Third, our estimates of lifetime risks depend on current cancer incidence rates. Although overall cancer rates have changed (increased) only modestly over the last nine years, there have been significant changes (some increases, some decreases) in several specific types of cancer. For example, prostate cancer incidence increased 68 percent between 1988 and 1992 and then decreased by 25 percent over the next four years. It is difficult to predict what other changes may occur in the future, especially for specific cancers. Rather than attempt to predict future rates, the lifetime risks are based on current cancer incidence rates in Minnesota. Changes in rates in the future will change the lifetime risks proportionately. Fourth, lifetime risks cannot be meaningfully estimated for individuals. The risk to an individual depends on many factors such as smoking history, diet, reproductive history, other life-style factors, genetics, and other unknown factors. Consequently, lifetime risks do not apply necessarily to any one individual, just as "life expectancy" estimates do not apply to any one individual. Estimates reflect the average experience of the population. Furthermore, there is an additional problem in interpreting lifetime cancer risks that does not occur in examining life expectancy or other death statistics. Individuals can only die once, but individuals can and do develop more than one cancer during their lifetimes (either the same type or different types). Indeed, individuals with certain cancers are at much higher risk for several other cancers. Cancer incidence rates actually reflect the number of newly-diagnosed cancers in the population; this is not quite the same as the number of individuals who have been newly diagnosed with a cancer. For example, a person unfortunate enough to be diagnosed with two distinct (primary) cancers during the same year will add not one, but two cancers to the annual cancer rate. Because of the relative infrequency of multiple cancers in a given year, public health officials can (and do) ignore the distinction between "cancers" and "individuals with cancer" when presenting annual cancer rates. However, that distinction becomes more important when considering cancer risks over the lifetimes of individuals. Most estimates of lifetime risk, including those presented in this section, do not take this factor into account and, therefore, slightly overestimate the average risk to individuals. However, when evaluating the total burden of cancer, it is the number of cancers and not the number of individuals with cancer that is of interest. In this regard, the lifetime risks presented here are appropriate.

In light of the difficulties described above, the MCSS developed the following approach to estimating lifetime cancer risks. It was assumed that: (1) People born today will have the same mortality rates (life expectancy) that currently (1996) exist in Minnesota; and (2) People born today will experience the same cancer incidence rates that currently (1996) exist in Minnesota. The lifetime risks are then expressed as the number of cancers that are expected to occur throughout the lifetimes of 1,000 people. In other words, a hypothetical sample of 1,000 individuals is "aged" from birth to death based on current Minnesota death rates. Current cancer incidence rates are applied to those who are still "alive" at each age. The number of cancers in each age category is then summed, and the total represents the number of cancers expected to occur during the lives of 1,000 people.

Estimates of the number of cancers expected in the lifetimes of 1,000 Minnesotans born today are given in Table 2.5. For all cancers combined, 518 malignancies are expected among males, while 447 malignancies are expected among females. The expected number of prostate cancers in males (172) exceeds the total number expected for lung, colon, bladder, and rectum cancers combined (164). The difference between previous estimates and the current estimates of lifetime risk is dictated by the changing incidence of prostate cancer. Currently, prostate cancer alone accounts for a third of the total expected cancers among men, but has been declining in recent years (see Section 4). In females, the expected number of breast cancers (141) also exceeds the total number of cancers expected for the colon, lung, bladder, and rectum (124). Breast cancer accounts for 32 percent of the total expected cancers in women.

Lifetime cancer risks are also commonly expressed as "one out of X people will eventually develop cancer," where "X" is some number. This expression of risk can be obtained by dividing the number 1,000 by the lifetime cancer risk in Table 2.5. Using breast cancer as an example, 1,000 divided by 141 gives 7.09. Thus, 141 per 1,000 is numerically the same as 1 in 7.09. Rounding off to the nearest whole number, it might be said that the lifetime risk of breast cancer is almost one in seven. Presented this way, the breast cancer risk in Minnesota would appear higher than national estimates, which have been reported as one in eight. However, Minnesota and national rates for breast cancer are virtually identical. The greater life expectancy of Minnesota women is the major reason for this difference in lifetime risk. Differences between Minnesota and national lifetime risks may be due to any of the following: rounding, differences in incidence rates, life expectancy, and the methods and assumptions used in calculating the risks. Although it is much easier for most people to understand a risk that is presented as one in seven (or one in x), it may also be subject to greater misinterpretation. Many people will interpret this statistic to mean that a woman living today has a one in seven (or eight or nine) chance of developing breast cancer. This is not exactly what this statistic means. First, the lifetime risks in Table 2.5 are for women born today, throughout their lifetimes, not for women living today born up to 90 or more years ago. Second, the above interpretation conveys the sense that all women have a uniform risk. Some women will have much higher risks and some women will have lower risks based upon their personal set of risk factors. For example, a woman who has had a mother and a sister with breast cancer at an early age may have a risk approaching 50 percent.

The lifetime cancer risk is useful in communicating the absolute or total risk of developing cancer. Age-specific or age-standardized rates fail to communicate the total risk. For example, the 1996 Minnesota age-standardized rate of all cancers in males and females for all races is 389 per 100,000 per year. This rate translates into an average individual risk per year of 0.00389, which can also be expressed as a risk of 1 in 256 or 3.9 cancers per 1,000 persons per year. This risk can be compared to a lifetime cancer risk of 479 cancers per 1,000 Minnesotans. An age-specific rate conveys even a greater sense of the rarity of cancer for people under age 50. For example, the annual cancer incidence rate for the age group 30-34 (which includes the median age for Minnesotans) is 1 in 1,399 for males and 1 in 924 for females. The annual statistics convey the sense that cancer is rare; the lifetime cancer risk statistic indicates that over a lifetime cancer is extremely common. Leukemia is an example of a malignancy that is perceived by the public as very rare. The lifetime cancer risk for leukemias for both sexes combined is 14 per 1,000. That approximately 1 in 100 people is expected to develop leukemia in his/her lifetime is in sharp contrast to the public perception that these diseases are extremely rare and indicates the importance of the lifetime cancer risk for communicating the true lifetime risk of cancer. Additional commentary on the application of the lifetime risk statistic to cancer clusters, environmental concerns, and cancer care are given in Section 4 of the MCSS 1993 report.

Childhood Cancers in Minnesota 1992-1996

Although childhood cancers account for fewer than one percent of all cancers diagnosed each year in the United States, great effort has been made to improve upon their detection and treatment. Althought survival rates have noticeably risen, the overall national incidence rate among children under the age of 15 has also risen over the last few decades. Available data suggest that the rates have increased in Minnesota, as well. See "Childhood Cancer Incidence and Trends in Minnesota, 1988-1994" in Minnesota Medicine, Volume 81, pages 27-32 (December 1998) for more information.

During the years 1992-1996, 803 cancers (including benign tumors of the central nervous system) were diagnosed in Minnesota children younger than 15 years of age (Table 2.6). Cancer in children differs significantly from cancer in adults in the pattern of cancer cell types. The vast majority of cancers in adults arise from epithelial tissue (outer body surfaces, glands, lining of intestinal tract). In contrast, less than five percent of the 803 childhood cancers were epithelial in origin. The difference between the cell types in childhood and adult cancers is suggestive of different causes of these diseases. The epithelial origin of most adult cancers is probably a reflection of life-style experiences such as smoking and diet. Although a number of genetic, birth, and environmental factors have been clearly associated with the risk of certain childhood cancers, the major determinants of risk for childhood cancers remain largely unknown.

As with adult cancers, relatively few of the many types of cancer account for most of the cancers that occur in children. Twenty-eight percent of all childhood cancers in Minnesota were leukemias, with acute lymphocytic leukemia (ALL), the single most common type of cancer comprising 23 percent of the total cancers. Tumors of the brain and central nervous system (including benign tumors) were the next most common, accounting for nearly 26 percent of the total. ALL, tumors of the brain and central nervous system, together with Hodgkin's lymphoma, non-Hodgkin's lymphoma, and cancers of the bone, soft tissue, and kidney accounted for nearly 73 percent of Minnesota's childhood cancers.

Table 2.7 contains the incidence rates for the most common childhood cancers for Minnesota and the U.S. SEER estimates. Male children are at an increased risk of developing cancer compared to females. During 1992-1996 in Minnesota, the male incidence rate of 16.5 per 100,000 was 28 percent greater than the female incidence 12.9 per 100,000. The higher risk in males is seen nationally and in other countries. ALL, non-Hodgkin's lymphoma, and brain cancer are major contributors to the greater male cancer risk.

For all cancers combined, the average yearly childhood cancer incidence in Minnesota is approximately 14.7 per 100,000. Nationally (1991-1995), the overall rate was 14.1 for all races and 14.4 for whites. Thus, the Minnesota rate was within 10 percent of the national rates. Table 2.8 contains estimates of the number of cancers expected per 1,000 children by the age of 15 in Minnesota. For males, 2.5 cancers and for females, 2.1 cancers are expected. Overall for both sexes, the estimate is 2.3 cancers per 1,000 children prior to their 15th birthday. This figure translates into an estimate of one cancer per 435 children by the age of 15.