The "Epidemic" of Breast Cancer in the U.S.--Determining the Factors

Introduction

Between 1973 and 1991, breast cancer incidence rates in the United States increased by 24%. In contrast, mortality from breast cancer during the same period remained constant. Changes in the 5-year relative survival rate and the incidence rates of different stages of breast cancer have also been documented. In this article, we will outline these recent trends in breast cancer incidence and mortality and offer possible explanations for them. In particular, we will explore the impact of increased mammography screening on breast cancer incidence. We will also discuss breast cancer risk factors that may vary by birth cohort, including age at menarche, age at first birth, physical activity, obesity, diet, alcohol(Drug information on alcohol) intake, estrogen therapy, and exposure to organochlorines.

Recent Trends

Incidence

Breast cancer incidence in the United States has risen continually during the past several decades. According to statistics from the National Cancer Institute's Surveillance, Epidemiology and End Results (SEER) Program, age-adjusted, invasive breast cancer incidence rates rose by 24% between 1973 and 1991, with an estimated average annual increase of 1.7% [1]. Data from the Connecticut Tumor Registry, extending back to 1940, indicate that breast cancer incidence rates have been increasing annually by at least 1.0% since 1940 [2]. Evidence from the Connecticut Tumor Registry also indicates that breast cancer incidence has been increasing with successive birth cohorts [3]. Campbell et al [4] have modeled the probability of breast cancer by birth cohort and have determined that the risk of getting breast cancer by age 85 increased from 1 in 29 for women born between 1888 and 1892 to 1 in 10 for women born between 1948 and 1952.

Breast cancer incidence has increased in both US white and black populations. Although black women have experienced a higher percentage increase in breast cancer incidence than white women during the last 2 decades, particularly among women under 50 years of age (26% vs 8.5%), white women have continually demonstrated the highest rates [1]. In 1991, the breast cancer incidence rate for white females, age-adjusted to the 1970 US population, was 113.6 per 100,000 women. The comparable rate for black women was 95.1 per 100,000 women (Figures 1 and 2).

Breast cancer incidence rates increase dramatically with age. The 1987 to 1991 age-specific breast cancer incidence rates for 30- to 34-year-old women of all races was 26 per 100,000. This rate increased to 229 per 100,000 for women 50 to 54 years old and increased further to 450 per 100,00 for those 70 to 74 years old [1]. A graph of the age-specific breast cancer incidence rates plotted on a semilogarithmic scale illustrates that the rate of change in breast cancer incidence among premenopausal women is greater than the rate of change among postmenopausal women (Figure 3). The difference in breast cancer rates between 30 and 50 year olds is almost 10-fold, while the difference between 50 and 70 year olds is only 2-fold.

According to cross-sectional SEER data, most of the increase in breast cancer incidence rates during the past 2 decades has been experienced by postmenopausal women. Between 1973 and 1991, women under 50 years of age experienced only a 9% increase in breast cancer incidence rates [1]. During this same period, the increase in breast cancer incidence for women over age 50 was 29%. However, Holford et al [3], using data from the Connecticut Tumor Registry beginning in 1950, reported that breast cancer incidence models that account for the effect of birth cohort, in addition to age at diagnosis and calendar period, showed similar patterns of increase in breast cancer incidence in premenopausal and postmenopausal women.

Mortality

Cross-sectional breast cancer mortality has remained fairly stable during the last 2 decades (Figure 1). Data from the Connecticut Tumor Registry indicate that mortality for breast cancer has been declining in more recent birth cohorts [3]. Although white women have higher breast cancer incidence rates, black women have disproportionately higher breast cancer mortality. Mortality for black women of all ages increased from 26.3 per 100,000 in 1973 to 31.9 per 100,000 in 1991-an increase of 20% [1]. Breast cancer mortality for white women rose by only 0.6% during this same period. Among white women under 65 years of age, breast cancer mortality declined by 8.6% during the past 2 decades, whereas in black women under age 65, breast cancer mortality increased by 12.2%.

Racial differences are also evident in the estimated lifetime risks of developing and dying from breast cancer. Based on data from SEER population-based cancer registries in nine US geographic areas from 1989 to 1991, white women have a 12.9% lifetime risk of being diagnosed with breast cancer and a 3.7% lifetime risk of dying from breast cancer [1]. Black women, on the other hand, have a 8.8% lifetime risk of developing breast cancer, while their lifetime risk of dying from breast cancer is 3.3%. The ratio of the age-adjusted death rate to the age-adjusted incidence rate during a specific time interval suggests the likelihood of a case fatality for each incident or newly diagnosed case of breast cancer. In 1973, this ratio was 0.32 in white women and 0.38 in black women; in 1991, the ratio decreased in white women (0.24) and, to a lesser extent, in black women (0.34).

Survival

Survival rates for breast cancer have been increasing during the past 2 decades. In 1973, the 5-year relative survival rate for women of all races with breast cancer, adjusted for other causes of mortality, was 72%.1 This rose significantly to 80.4% by 1986. Survival rates are substantially higher among whites than blacks. Between 1983 and 1990, the average 5-year relative survival rate for breast cancer of all stages was 81.6% in white women, as compared with 65.8% in black women.

There are also considerable differences in age-specific survival rates. Younger women have worse breast cancer survival rates than older women. The 5-year relative survival rate for women under 45 years old diagnosed in 1983 to 1990 was 76.3%, while women 65 to 74 years old had a survival rate of 82.7% [1]. Between 1973 and 1986, the 5-year relative survival rate for women diagnosed with cancer when under 50 years of age rose from 74% to 78.1%; however, for women over age 50, the survival rate increased from 71.3% to 81.3%.

Breast Cancer Incidence by Stage

Between 1983 and 1991, there have also been striking trends in breast cancer incidence by stage. Among US white women, in situ breast cancer incidence increased by 105% in those under age 50, and by 180% in those age 50 and over; localized (stage I) breast cancer incidence increased by 62% in women under age 50 and by 88% in older women [1]. During the same period, US black women under age 50 experienced a 175% increase in the incidence of in situ breast cancer and women age 50 and over experienced a 138% increase.

Stages II, III, and IV breast cancer incidence remained fairly constant during this period and even fell among certain age groups. Miller et al [2] reported that the incidence of tumors < 2 cm in diameter increased steeply from 1982 to 1988 and has since leveled off. In 1991, there were significant inequalities in localized stage distribution between blacks and whites; 40% of incident breast cancers in white women were stage I, as compared with only 26% in blacks. These differences in staging help explain some of the discrepancy in breast cancer mortality between black and white women.

International Trends in Incidence

The United States has the highest cross-sectional breast cancer incidence rates in the world, but other countries are experiencing similar increases in breast cancer incidence (Figures 4 and 5) [5,6]. Japanese women have manifested significant increases in incidence between the periods of 1973 to 1977 and 1983 to 1987. While breast cancer incidence in US white women rose by 28% between these two time periods, incidence in Japanese women increased by 59%. However, the age-adjusted breast cancer incidence rates in Japan are currently less than one-third those in the United States. Other populations, such as the Chinese in Singapore and Eastern Europeans in several countries, have also experienced increases in breast cancer incidence, although none has been as dramatic as the rise in incidence observed in Japan.

Kelsey et al have noted international differences in the shape of the cross-sectional age-specific breast cancer incidence curves [7]. Countries with the highest incidence rates of breast cancer have a steeper rise in age-specific incidence during the premenopausal years and a continuing but lesser rate of increase in incidence during the postmenopausal years. Countries with intermediate breast cancer incidence rates demonstrate a similar pattern of increase during the premenopausal years and a relatively constant rate during the postmenopausal years. Areas of lowest breast cancer incidence also have a steep rise in premenopausal rates, but postmenopausal rates appear to decline with increasing age (Figure 3) [5,6]. Moolgavkar argues, however, that these differences in age-specific breast cancer incidence curves are due to birth cohort effects. Once such cohort effects are adjusted for in the analysis, the curves appear to be similar, and only the magnitude of the rates differs among countries [8].

Within the United States, several minorities, especially Asian immigrants, have also experienced large increases in breast cancer incidence. Breast cancer incidence rates in Filipino-Americans doubled during the 1970s and '80s, while rates in Japanese-Americans rose by almost 50% [8]. Shimizu et al [9] demonstrated that breast cancer incidence rates are further determined by the time since immigration. Japanese immigrants who migrated late in life have breast cancer incidence rates 3.6 times that of Japanese living in Japan, and Japanese-Americans who were born in the United States have incidence rates 4.1 times those of indigenous Japanese.

Mammography Screening and Breast Cancer Incidence

The long-term secular trend in age-adjusted breast cancer incidence in the United States during 1940 to 1982 has reflected an annual rate of increase of about 1%. This pattern changed dramatically from 1982 to 1987, during which time the rate of increase was about 4% per year. Studies by White et al, Feuer and Wun, Lantz et al, and Liff et al noted that the increase was due primarily to a higher frequency of in situ (intraductal) carcinomas and early-stage invasive cancers, ie, those < 2 cm in diameter [10-13]. The increase in the incidence of breast cancer during 1982 to 1987 was more pronounced in women age 60 years and over (about 5% per year) than in women under age 60 (about 2.5% per year). Since 1987, overall incidence rates of breast cancer have declined slightly (eg, 293 per 100,000 population in 1987, 287 per 100,000 in 1988, and 284 per 100,000 in 1990). During this period, annual cross-sectional breast cancer mortality in the United States was relatively stable.

In the population-based studies of Reeves et al in Wisconsin and Farwell et al in Vermont, 35% to 36% of breast cancers diagnosed in the late 1980s were first detected by screening mammography [14,15]. The percentage of women having mammograms and the reported incidence rates for breast cancer in the United States both began to rise dramatically after 1982. Thus, the studies of Miller et al, Lantz et al, and Feuer and Wun concluded that the incidence trends between 1982 and 1987 were generally consistent with population-based increases in mammographic screening and were not due to the emergence of a new risk factor [2,11,12]. The assumption of a direct correlation between increasing screening-detected breast cancer cases and the rising incidence of breast cancer is probably valid during the initial years of dynamically changing screening practices in a geographic area, and would persist in relation to the estimated lead times for various age groups, which increase on average with increasing age.