Multi-ethnic differences in breast cancer: current concepts and future directions.

Abstract

Breast cancer poses a serious public health concern throughout the world in both developed and developing nations. Recent data show a small decline in breast cancer mortality in the United States and Northern Europe where the disease has been a leading cause of death.1 This reduction has been attributed in part to the early detection of breast cancer in addition to advances in clinical management. The decline in mortality has also been postulated to be due to a decreased risk in women developing breast cancer in the last 2 decades associated with increased fertility as part of the post-Second World War baby boom.1 Despite these encouraging results, other European nations such as Spain, Portugal, Greece, Hungary, Poland and Italy have not reported a reduction in breast cancer mortality. The true reasons for these breast cancer trends, particularly across continents, remain perplexing to both epidemiologists and clinicians. Any observed variation of incidence and oncologic outcome between different populations and ethnic backgrounds may relate to the underlying biological behavior of breast cancer at the cellular and molecular level. Application of biomarker studies could therefore enhance the information obtained from classical study designs and further expand the areas of scientific inquiry to which epidemiology can contribute. This approach may yield important clues in breast cancer pathogenesis, develop potential preventive strategies, improve early detection and treatment of disease with distinctive protocols, tailored to the needs of individual target populations. Conclusions from the workshop on Multicultural Aspects of Breast Cancer Etiology in Washington DC by the Etiology working group of the National Action Plan on Breast Cancer (NAPBC) in March 1999 highlighted the importance of “ethno-oncology” that addresses the comparison of population groups with extreme differences in the rates of incidence, mortality and survival.2 This article reviews the evidence for global multiethnic differences in breast cancer outcome and discusses the future clinical implications of ethno-oncology. Breast cancer statistics may be collated from randomized clinical trials, hospital-based studies and from population based cancer survival registries. Although randomized trials are considered the gold standard, national survival information that may be derived is limited. Population-based studies provide broad data more applicable to screening and organization of treatment services. These studies, however, may be restricted by a shortage of information specific to host and tumor factors, overall accuracy and data quality. Reliable sources that compare global differences in breast cancer demography and patterns of molecular biology are lacking. Many studies have been done in the United States, investigating potential differences between immigrant African Americans and Caucasians. In 2000, the International Union Against Cancer reported the lowest incidence of breast cancer in less industrialized countries epitomized in parts of Africa, South Central Asia and Melanesia in contrast to the 8–10-fold higher risk in industrialized countries such as North America, Western and Northern Europe (Fig. 1). These data illustrate the wide variation of breast cancer statistics worldwide, including the distribution of breast cancer incidence within a continent or a region (Fig. 2). In the United States of America, breast carcinoma mortality appears to be highest for African Americans, next among Caucasians, followed by native Hawaiians. The reasons for these disparities in survival following breast cancer treatment are unclear. Worldwide Age Standardized Rates (ASR) on breast cancer incidence and death. Reference: http://www-dep.iarc.fr/globocan/globocan.html Age Standardized Rates (ASR) of breast cancer in Northern Europe. Reference: http://www-dep.iarc.fr/globocan/globocan.html All women are at risk of developing breast cancer during their lifetime, irrespective of racial or ethnic origin. Many socioeconomic and host factors appear to influence the development of breast carcinoma. Inherent genetic makeup, an individual's reproductive experience, the effect of endogenous and exogenous hormones and environmental factors that influence host vulnerability each have a role to play in the etiology of breast cancer. Epidemiologic studies focusing on variations in genetic and environmental factors will probably provide insight into distinct breast cancer phenotypic expression in different circumstances. The proportion of diversity that exists between human populations is relatively low compared to other species, although Jorde et al. have shown a relatively higher level of genetic variation among Africans than in nonAfrican populations.3 Only about 5–10% of all breast cancers in Western populations may be attributed to a high-risk cancer predisposition gene. When an ethnic group is made up of a small number of people of common origin, differences in genetic predisposition compared to more diverse populations could occur by the founder effect based upon a high penetrance gene mutation.4 Although these high-risk predisposition genes are rare, when the population expands, the mutation in these founders can become prevalent in larger proportions. Over 300 mutations have been identified in BRCA1 and BRCA2 carriers worldwide. Within these populations, founder mutations have been described amongst the Ashkenazi Jews, Swedes, Norwegians, Dutch, Russians, Japanese, African Americans, Finns and French Canadians.4 In African American women, who have a higher incidence of early onset breast cancer with poor overall survival, 3 novel BRCA1 mutations have been identified, which are distinct from those identified amongst Caucasians.4 Mutations in high-risk susceptibility genes for breast cancer such as BRCA-1 and BRCA-2 have high penetrance with a 60–80% risk of carriers developing breast cancer. Mutations in low-risk susceptibility genes on the other hand are found with a higher frequency in the population and multiple low-risk susceptibility gene mutations may exist in individuals. Low-risk susceptibility genes have the potential to interact with adverse environmental risk factors. The penetrance of such genes is low and may account for breast cancer that occurs in the absence of a strong family history at a relatively late age at diagnosis. The origin of low penetrance genes is largely unexplained and is not attributed to a founder effect in an ethnic group.5 Early migration studies done by Regina et al. have shown that Chinese, Japanese and Filipino ethnic groups who emigrated to America had a higher death rate from breast carcinoma compared to their native compatriots in the country of origin. Further analysis identified a gradual mounting risk within the émigrés, which increased over the subsequent generations.6 Kliewer and Smith analyzed breast carcinoma mortality among immigrants in Australia and Canada, concluding that environmental and life style factors associated with the new place of residence influenced the subsequent development of breast cancer.7 This important concept has led the National Institute of Environmental Health Sciences (NIEHS) to develop new initiatives on the Environmental Genome Project, which aim to define gene polymorphism that can enhance or reduce an individual's susceptibility to the adverse effects of environmental agents. Hormone dependence is the common denominator in recognized pathogenic pathways leading to breast cancer. Premenopausal women thus possess an ideal host environment for breast cancer development that is not diminished by cessation of ovulation. Early age at menarche and late age of menopause are both established as modest risk factors for breast cancer. A recent study on the age at menarche of 1,166 British teenagers in 10 different regions by Whincup et al. reported the median menarcheal age as 12 years and 11 months,8 with no significant difference conferred by social class or ethnic group. In Asian third world countries, the median age of menarche is often much later, typically after the age of 14 years.9 Age at menarche may contribute to the geographical variation of breast cancer incidence worldwide, as early menarche appears to increase the risk of breast cancer by approximately 4–5% each year of earlier onset.10 The importance of age at first full-term pregnancy (FFTP) and overall parity as breast cancer risk factors were first observed in 1926 by Lane-Claypon. Although this was confirmed in many subsequent studies,11, 12 the protective effect may paradoxically only apply to the lifetime risk amongst parous women rather than age specific risk. There is strong epidemiological evidence that indicates that childbearing age and parity have a reverse effect on subsequent breast carcinoma risk (the so-called “crossover effect”), with parous women at an initially higher risk than the nulliparous but with later reversal of risk at older age13 (Fig. 3). Logan first reported this “crossover” in 1953 on breast cancer mortality.14 The effect was subsequently reported by Janerich and Hoff, and Lubin and Pathak, in similar studies.15, 16, 17 Based on the Nurses' Health Study data, Pathak et al.17 observed the crossover effect, especially for women of parity-2 relative to nulliparous women, with a relative risk of 1.85, 1.32 and 0.93 in the age groups of 30–39, 40–49 and 50–55 years, respectively. This crossover observation was further analyzed by Pike et al.18 who concluded that the risk of malignancy was assumed to be proportional to the power of the accumulated “breast tissue age.” In that model, the breast tissue hypothetically ages at a constant rate between age at menarche and age at FFTP, which continues until the perimenopausal years and decreases linearly up to the menopause. The rate of aging remains low after menopause, accounting for a lesser relative risk. Simultaneously, at each full-term pregnancy, the hormonal milieu of pregnancy causes a one-time increase in “breast tissue age.” It was further observed that each one-time increase raises the risk of malignancy compared to nulliparous women. Although at each pregnancy the “breast tissue age” could increase, as the rate of aging reduces in the perimenopausal period, breast cancer risk could become far less compared to nulliparous women.13, 18 This would partly explain the multiethnic differences of breast cancer in the United States of America where breast cancer incidence in women less than 45 years at diagnosis is higher among African-Americans than Caucasians, despite a higher observed parity.13, 19, 20 “Cross-over” Effect. “One-time” increase in the cumulative breast tissue age at pregnancy (P) increases the risk of breast cancer in the parous (dark curve and shaded area) more than the nulliparous women (light curve and shaded area) at an early age and subsequent reversal at older age. The surface area under the curve represents cumulative breast tissue age. The role of dietary fat in the etiology and outcome of breast cancer remains controversial. Wynder et al.21 observed that the typical Japanese diet, a nation with low breast cancer rates, is 10–25% of calories from fat sources, compared to the USA, where 40–45% of total calories come from fat. Although meta-analyses of case-control studies have found an association between dietary fat and breast carcinoma, these results may be affected by recall bias.22, 23 Several studies have been done to investigate the relationship between dietary fat and breast carcinoma in the USA, including the National Health And National Examination Survey — NHANES II (1976–1980), the Department of Agriculture — USDA (1985–1986) and the National Health Interview Survey — NHIS (1992). These trials have unfortunately not included sufficient numbers of ethnic groups, such as Asian Americans, Native Hawaiians or Native Americans to allow separate comparison. Potential differences in the incidence of breast cancer worldwide may relate to variation in the intake of specific fatty acid constituents. There is a paucity of epidemiological studies in multicultural populations that compare the relative levels and composition of fatty acids that contribute to the total fat consumption. Simonsen et al.24 have suggested that an increase in the ratio of omega-3 to total omega-6 fatty acids in adipose tissue is associated with an inverse risk of developing breast carcinoma. A large number of epidemiological studies have shown an association between alcohol intake and breast cancer risk. In an overview of 38 such studies and a similar meta-analysis of 6 prospective studies, strong evidence of a dose-response relationship between alcohol consumption and the risk of developing breast carcinoma was found.25. Although a higher alcohol intake among white women compared to African American and Latino women were found,26 the absolute extent to which higher alcohol consumption could contribute to increased breast cancer risk remains unknown. There is some evidence that dietary fibber and phytoestrogens (soy foods) may play a role in reducing the risk of breast cancer. A high fiber diet may eliminate estrogen by the fecal route through reduced intestinal reabsorption.27 The higher fibber intake in Latinos and Asian populations may thus confer a lower breast cancer risk in these ethnic groups. Customary dietary differences amongst different ethnic groups with respect to specific vegetables, fruits, legumes and soy therefore provide an excellent opportunity to investigate the relationship between specific micronutrients and breast cancer risk. Soy (isoflavone, 1 class of phytoestrogen), which is staple in the Asian diet, is considered to be a selective estrogen receptor modulator with nonhormonal properties. A significant protective effect of soy intake among Asian immigrants but not amongst the Asians born in the United States of America was reported by Wu AH et al.28 Studies done in Singapore29 and Japan30 have identified a significant reduction in the risk of breast cancer among women with high soy dietary intake. In a similar Australian study, the risk of breast cancer was found to be 50% less in women with high soy intake.31 A single larger study done in China, however, failed to identify any significant effect of soy on breast cancer risk.32 Ethnic differences in survival from breast cancer in the United States was reported in the National Cancer Institute Black/White cancer survival study by Eley et al.33 in 1994. In this study, multivariate analysis of prognostic factors including stage, tumor size, treatment, co-morbid, social and demographic factors revealed a 2.2-fold higher relative risk of dying from breast cancer in black compared to white women. The most important factor determining survival in the black populations was advanced stage at presentation. Several authors therefore concluded that this could be resolved by improving access to medical care and screening facilities to the black population. Dayal,34 Gordon,35 Berg36 and Freeman37 have similarly shown that after adjusting for socioeconomic status (SES), survival between these 2 racial groups remain identical. Other studies support the opposite view that SES alone does not account for the observed differences in survival between different ethnic groups.38, 39, 40 Wojcik et al.41 retrospectively analyzed 5,879 whites and 698 African Americans diagnosed with breast cancer that had equal access to the US military health care facilities in the Department of Defense. Although the African American women with access to military care fared better than African American women who did not, an unexplained 1.45 (95% confidence interval, 1.20–1.76) higher risk of age-adjusted death in blacks was observed compared to white women. The African American women in this study however had significantly larger tumors at diagnosis and were younger than the white women. However, Aziz et al.42 showed that African Americans presented with breast cancer 10 years earlier than Caucasian women after adjusting for stage and grade. A recently reported meta-analysis of 14 studies done in the USA concluded that African-American ethnicity is a significant independent predictor of outcome in breast cancer, even after adjusting for SES. In the aggregate analysis of this study, the mortality odds ratio for African-American patients compared to the Caucasians was 1.22 (95% CI, 1.13–1.30). A mortality odds ratio of 1.35 (95% CI, 1.00–1.82) was also noted in 3 studies done in patients who had equal-access systems, further supporting the view that SES and presentation delay did not account for poorer survival in African-American women.43 As breast cancer survival has been consistently reported to be 12–18% lower among African Americans than white wo-men,44, 45, 46, 47, 49 it is reasonable to hypothesize that poor survival in certain ethnic groups may be related to the presence of indigenous poor prognostic factors. Such prognostic variables might include younger age at presentation, a greater proportion of tumors with higher tumor stage and grade and lymph node involvement. Several studies have found differences in tumor estrogen receptor (ER) and progesterone receptor (PgR) content in breast cancers between ethnic groups. A consistent finding in African American populations with breast cancer is low levels of ER and PgR, with higher S-phase fraction, but no observed difference in HER-2/neu and p53 expression between the different ethnic groups.46, 48. The study of ethno-oncology is a challenging field that requires close collaboration between clinicians, protagonists of traditional epidemiology and laboratory science. With advancements in molecular biology and cancer genetics, efforts should be directed to include multiethnic populations in all aspects of independent or comparative research. This could provide new mechanisms to identify etiologic pathways that may be utilized for better breast cancer screening protocols, disease prevention, early detection and subsequent outcome. In ethno-oncological research, many questions may arise if a difference in molecular determinants were to be found between different ethnic groups. An important issue that needs to be addressed is whether different ethnic groups with molecular diversity should be treated with similar protocols. There have been no studies to evaluate cancer outcomes after similar treatment regimens in distinct ethnic groups. Individualized treatment strategies may be developed to meet the needs of ethnic subgroups. The reasons for the observed differences in breast cancer incidence, mortality and survival among different ethnic populations are not well understood. Future research should be directed to identify potential genetic or environmental factors. Several investigators have emphasized the relative importance of SES in influencing access to better health care and screening services. These data do not explain the observation that black population still have a worse prognosis when comparable stage for stage of breast cancer is considered. Coates et al.49 in the M. D. Anderson Cancer Center and by Wojcik et al.41 at the US Ministry of Defense independently concluded that delays in seeking treatment alone did not account for ethnic differences in outcome from breast cancer treatment. If the black women were experiencing a delay in diagnosis, their age at diagnosis should have been older and not younger.41, 42 Pierce50 and Kimmick51 reported worse outcomes for black patients, despite similar stage and treatment, respectively. It has therefore been postulated that different biologic determinants of breast cancer might account for at least a proportion of the disparity in breast cancer survival in multiethnic populations. Molecular markers of breast cancer therefore should be further investigated in multiethnic populations to determine their role as predictors or prognostic indicators of disease in different subgroups. On a global perspective, breast cancer risk varies by several fold between developed and less developed nations. Although this variation has been attributed to screening, early detection and modern therapeutic strategies, other demographic factors and differences appear to be involved. Distinct etiologic factors may affect the biological expression, disease phenotype and treatment outcome amongst multicultural populations. In the future, well-matched descriptive studies, comparing groups with significant differences in incidence, mortality or survival rates, may provide answers to these perplexing problems.