There is a 4- to 5-fold variation in breast cancer incidence rates worldwide, with the highest in North America (99.4/100,000) and the lowest in Asia (22.1/100,000) and Africa (23.4/100,000).3 However, mortality rates are relatively less variable, with Africa (16.2/100,000) being similar to North America (19.2/100,000). These international variations are partly due to differences in environmental and lifestyle factors, screening practices, and treatment strategies.
In the past 50 years, breast cancer incidence has increased worldwide, including in the United States, where the highest rate is found.4 Data from the Surveillance, Epidemiology, and End Results (SEER) program show that incidence increased in the 1980s and 1990s in the United States but decreased after 2002 mainly in white women4 (Fig. 4-1). During the same period, 5-year survival also increased, reaching 90% at year 1997 for women 50 years or older and 87% for women under 50 at diagnosis (Fig. 4-2).4 The continued improvement in prognosis may be attributable to both screening, which increases the diagnosis of small, localized breast tumors, and treatment effectiveness.
Trends in female breast cancer incidence and death rates by race and ethnicity, United States. Rates are age-adjusted to the 2000 U.S. standard population. (Data from Ries L, Melbert D, Krapcho M, et al. SEER Cancer Statistics Review, 1975-2005. Bethesda: National Cancer Institute; 2008.)
Breast cancer survival trends in the United States. Values shown are 5-year relative survival (survival adjusted for life expectancy—an approximation to breast cancer-specific survival). (Data from Ries L, Melbert D, Krapcho M, et al. SEER Cancer Statistics Review, 1975-2005. Bethesda: National Cancer Institute; 2008.)
Race, Ethnicity, and Socioeconomic Status
There are racial and ethnic differences in breast cancer incidence and mortality in the United States (Fig. 4-1). In the past 30 years, incidence rates were higher in white women than in African American women. However, African American women had higher mortality rates than white women, and this racial disparity has been widening in recent years. The incidence and mortality rates in Asian, Hispanics, and Native American women are lower than those of non-Hispanic white and African American women.
African Americans are more likely to have regional and distant disease, contributing to the survival disparity (Fig. 4-4A). However, African Americans also have lower 5-year survival rates than whites within stage strata (Fig. 4-4B). In addition to later stage at diagnosis, socioeconomic factors such as limited access to quality health care and comorbidities contribute substantially to outcome disparities.5,6 However, some degree of difference in outcomes persists between blacks and whites after accounting for these factors, suggesting that cancer biological differences might also be relevant.7,8 Breast cancers in African Americans are more likely to be early-onset, higher-grade, and ER-negative compared with those in whites.5,6
A. Distribution of breast cancer stage at diagnosis in African-American and white women, United States, 2000-2005. B. Five-year relative survival for African-American and white women by breast cancer stage, United States, 1975-1979 and 1996-2004.4(Data from Ries L, Melbert D, Krapcho M, et al. SEER Cancer Statistics Review, 1975-2005. Bethesda: National Cancer Institute; 2008.)
Early age at menarche is a well-established risk factor for breast cancer in both premenopausal and postmenopausal women, with a reduction in risk of 5% to 10% for each year delay in age at menarche.9 Inaccurate recall of menarcheal age, especially in older women, may underestimate the strength of this association. Several mechanisms may explain the protective effects of late menarche. Early menarche may be associated with more rapid onset of regular, ovulatory menstrual cycles and hence longer duration of lifetime exposure to endogenous hormones.10 Estrogen levels are higher several years after menarche and remain so throughout the reproductive years in women with early menarche.11,12
Parous women have lower risk of breast cancer compared with nulliparous women, but the relationship between pregnancy and breast cancer is complex.10 Risk initially increases after the first pregnancy, then decreases after 10 years, and this protective effect is durable.13 In the long run, the protective effect outweighs the transient adverse effect. The short-term increased risk is thought to be due to significant elevated hormonal levels and rapid proliferation of breast epithelial cells, whereas the mechanisms for the long-term protection involves epithelial cell differentiation, which takes place largely after the first full-term pregnancy.14 Differentiated cells may be less sensitive to carcinogens due to longer cell cycles, allowing more time for DNA repair in the G1 phase.14 Additional births further reduce breast cancer risk though the effect is moderate. According to the collaborative reanalysis of 47 studies, each birth reduces the relative risk of breast cancer by 7%.15
Age at First Full-Term Pregnancy
Women with first full-term pregnancy after 35 years of age had 40% to 60% higher risk than those who had a first child before 20 years of age.10,16,17 However, more recent studies showed the effect was weaker,18 or did not exist for African Americans and indigenous Africans.18,19 For women who have later pregnancy, the transient postpregnancy risk increase mentioned earlier is more likely to manifest in epidemiologic studies, as age at delivery is closer to age at cancer diagnosis for these women. In addition, age at first birth is a less important factor among women with multiple children.19,20
Whether spontaneous and induced abortion are breast cancer risk factors is controversial.10 A meta-analysis documented an overall relative risk of 1.3 for induced abortion and no increased risk for spontaneous abortion, although there was a significant heterogeneity across studies.21 However, two large cohort studies in Denmark and the United States demonstrated that neither spontaneous nor induced abortion was associated with breast cancer (relative risk ≈ 1.0).22,23 A recent pooled analysis also concluded that abortions do not affect women's risk of developing breast cancer.24 Furthermore, recent studies conducted in China, where the prevalence of induced abortion is high, found no link between induced abortion and breast cancer risk.25,26 Taken together, available evidences does not support induced abortion increasing breast cancer risk.
Growing epidemiologic evidence supports risk reduction with prolonged breastfeeding, although studies reported varied magnitude of this protective effect.10,15,27 In the pooled analysis of data from 47 epidemiologic studies, the relative risk of breast cancer decreased by 4.3% for every 12 months of breastfeeding.15 There is apparently a dose–response relationship, with increasing total duration of breastfeeding decreasing the risk.10,15,27 Inconsistent findings with respect to the comparison of ever versus never breastfeeding may be due to international variation in duration. In modern Western countries, where few women breastfed for more than 1 year, the protective effect was moderate, for example, a 22% risk reduction in premenopausal women was documented.28 In contrast, a risk reduction of more than 50% was observed for women with at least 5 years breastfeeding in populations of China,29 Mexico,30 and Nigeria.19 Several mechanisms have been postulated to explain the protective effect of breastfeeding.10,27 Breastfeeding delays the reestablishment of ovulation, thereby reducing lifetime endogenous hormones exposure. Breastfeeding may result in further terminal differentiation of the breast epithelium, thus making it less susceptible to carcinogenic stimuli.
As shown in Figure 4-3, the slope of increase in breast cancer incidence slows around menopause. It is well established that late age at menopause is associated with higher risk of breast cancer. For each year delay in age of menopause, the risk for breast cancer increases by 3%.10,31 Artificial menopause through bilateral oophorectomy also decreases breast cancer risk, while simple hysterectomy does not.10 In women with BRCA1 or BRCA2 mutations, bilateral oophorectomy reduces the risk of breast cancer by more than 50%.32
Many epidemiologic studies have evaluated the relationship between risk of breast cancer and oral contraceptives that contain estrogen and progestin. Generally, a positive but rather weak association has been seen. In a pooled analysis of 54 studies, current use of oral contraceptives was associated with 24% greater risk compared with never-users.33 This elevated risk disappeared after discontinuing: the relative risks were 1.16, 1.07, and 1.01, respectively, for women 1 to 4 years, 5 to 9 years, and 10 or more years after stopping use.33 Duration of use has no independent effect. As women usually use oral contraceptives in their second and third decade of life when the absolute risk of breast cancer is low, and the excess risk decreases after cessation of use, there are likely few breast cancer cases due to oral contraceptive use. A recent large case-control study also confirmed that former oral contraceptive use did not increase the risk later in life, when the breast cancer incidence of is higher.34
Health benefits and risk of postmenopausal hormones have been evaluated by numerous studies. In a pooled analysis of 51 epidemiologic studies, the current or recent use of postmenopausal hormones was associated with increased risk of breast cancer, with a dose–response relationship based on duration of use.35 The risk increased by 2.3% for each year of use and the relative risk was 1.35 for women who had used hormones for 5 or more years. This effect disappeared 5 years after discontinuing, regardless of duration of use. The effect of postmenopausal hormones was stronger in lean women than in obese women, a finding confirmed by later studies.36,37 Mounting evidence suggests that estrogen plus progestin increases breast cancer risk more than estrogen alone.31,35,37,38 Estrogen alone is associated with a relative risk of 1.3, while for estrogen plus progestin, relative risk ranges from 1.5 to 2.0 among several cohort studies.31,35,38 In the Women's Health Initiative randomized trial, estrogen-only therapy did not increase risk, while estrogen plus progestin increased the risk by 26%.39,40
Attained height has been found to be positively associated with breast cancer in a large number of case-control and cohort studies.41-44 Numerous hypotheses related to nutritional influence on development have been proposed, including energy intake and growth during childhood (a hypothesis supported by variations in risk based on periods of nutritional deprivation as that which occurred in World War II), the interaction of nutrition with hormones during puberty, and the relationship of insulin-like growth factor with both height and breast cancer risk.
Weight and Body Mass Index
Excess body weight has been implicated as a risk factor, and has most often been examined via anthropometric measures such as body mass index (BMI—weight corrected for some function of total size). BMI has been convincingly correlated with breast cancer risk in numerous studies, although the relationship is often complex and involves additional modifying factors.36,44-51 In the Nurse's Health Study involving more than 90,000 women, both high BMI and weight gain increased postmenopausal breast cancer risk.36 A pooled analysis of 7 cohort studies estimated excess breast cancer risk of 26% for postmenopausal obese versus lean women.44 Numerous other studies have found 1.25- to 2-fold or greater excess risk among postmenopausal obese women.45-50 Interestingly, obesity is associated with decreased breast cancer risk in premenopausal women in most studies.44,46,48,51
The predominant hypothesized mechanism in postmenopausal women implicates increased endogenous estrogen resulting from conversion of androgens by the aromatase enzyme in adipose fat.52 The association between increased circulating estrogen and breast cancer risk supports this concept,53,54 as does the apparent protective effect of obesity on breast cancer risk in premenopausal women, in whom high BMI is associated with decreased serum estradiol.55,56
Lifestyle and Dietary Factors
Alcohol intake influences breast cancer risk, with a recent large meta-analysis of studies estimating relative risk increases of 32% for consumption of 35 to 44 g/day and 46% for 45 g/day or more, compared to women with no alcohol consumption.57
Tobacco use via smoking has not been consistently demonstrated to alter breast cancer risk, in part because this risk factor must be carefully evaluated in relation to other factors with which it may be correlated. In the above-referenced meta-analysis, the researchers identified a strong confounding influence of alcohol use on the effect of smoking, that when accounted for, resulted in no association remaining between smoking and breast cancer.57
Evidence from observational studies supports a modest protective effect of physical activity on breast cancer risk, with risk reductions in the range of 10% to 50%.58 Physical activity is nearly inextricably linked with a variety of nutritional, physiologic, and social factors, and may be difficult to characterize retrospectively throughout life, leading to limited reliable inference for this observation. As a modifiable risk factor with multiple potential benefits in addition to cancer, there is great interest in prospective interventional studies. Such studies are ongoing in both breast cancer patients and women at potential risk, but results have not been reported thus far.
Soy consumption has been suspected of providing a protective effect, based on global observations of differences in incidence that correlate with dietary patterns, coupled with the hypothesis regarding competition of phytoestrogens (known generally as isoflavones) with endogenous estrogen for binding to estrogen receptors, as well as other possible anticarcinogenic effects of these compounds.59 Observational studies support an association between soy intake and breast cancer risk, but it is dependent on a threshold level of consumption. Specifically, results of a meta-analysis showed that in Asian countries where soy consumption is high, women with over 20 mg/day of isoflavones had a 29% lower relative risk of breast cancer compared to those who consumed 5 or fewer mg/day.59 In the same investigation, in a synthesis of studies conducted in low soy consumption regions, where daily isoflavone intake ranged from 0.15 to 0.8 mg/day, no association could be found.
Dietary fat intake has long been implicated as a risk factor for breast cancer, in part as a result of the apparent strong correlation between breast cancer incidence and fat content in diet worldwide.60 As for most putative risk factors, observational studies of fat intake and breast cancer risk show mixed results. Meta-analyses have found relative risk increases ranging from 5% to nearly 50% among individuals with the highest fat intake.61 Different approaches and difficulties with accurate recording of fat intake, particularly in retrospective studies, likely account for some of the inconsistency in findings. For fruit, vegetable, and whole grain intake, observational study results have been similarly varied.62
Despite this uncertainty, diet content is one modifiable risk factor that has been evaluated in a prospective randomized study.63 In the Dietary Modification Trial of the Women's Health Initiative, women randomized to a low-fat, higher fruit and vegetable intake diet had a 9% lower risk of breast cancer over an 8-year period (risk ratio = 0.91, 95% confidence interval 0.83-1.01).63 Among women compliant with their assigned regimen, the effect was somewhat larger.
Exposure to moderate or high levels of ionizing radiation from various sources including medical treatment procedures and nuclear explosion increases the risk of breast cancer.64 The effect of radiation on breast cancer depends on age at exposure: the risk is higher in women exposed before age 20 years and is small for exposure after age 40 years.65 Before age 40, there is a positive correlation between radiation dose and breast cancer risk.64-66 These radiation-induced breast cancers typically do not occur until age 30 to 35 years, but the elevated risk persists through the woman's lifetime.64,66 The magnitude of effect can be large: an 8-fold increased risk was documented in women with Hodgkin disease who received radiotherapy dose of more than 40 Gy.67 It is unclear whether a low level of ionizing radiation (< 10 mGy), such as chest x-ray or mammography, increases breast cancer risk, but the cumulative dose is likely important.68
Other Environmental Factors
The impact of organochlorines including polychlorinated biphenyls (PCBs) and dichlorodiphenyl- trichloroethane (DDT) on breast cancer have been studied extensively. A pooled analysis of 5 studies found that PCBs and DDE levels in blood were similar between breast cancer cases and healthy controls.69 Studies that measured organochlorines using job history or residence location yielded inconsistent results.70 Other environmental pollutants have received less attention in epidemiologic studies.
Family history is a well-established risk factor for breast cancer. Women whose mother or sisters had breast cancer are at about 2-fold increased risk compared with general population.71 Multiple affected family members, early onset of disease, bilateral breast cancer, and affected male relatives further increase the risk. Familial risk could be attributable to shared environmental or genetic factors, or both. A twin study showed that inheritable genetic factors account for 27% of all breast cancer.72 A conventional genetic model of breast cancer is that disease risks are affected by mutations of several high-penetrance genes and common variants of many low-penetrance genes.71
In the 1990s, 2 major breast cancer suppressor genes, BRCA1 and BRCA2, were discovered using genetic linkage mapping and positional cloning.71 The BRCA1 gene is located on chromosome 17q21 and the BRCA2 gene is located on chromosome 13q12-13. A deleterious mutation in the 2 genes confers an over 10-fold relative risk. A recent meta-analysis of 10 studies estimated that the cumulative risk of breast cancer by age 70 was 57% and 49% for BRCA1 and BRCA2 mutation carriers, respectively.73 The prevalence of BRCA1 and BRCA2 mutation is generally low and varies across populations from different geographic regions and ethnicities (0.4%-7.0% for BRCA1 and 1%-3% for BRCA2).74
Germline mutations in P53 are linked to Li–Fraumeni syndrome, a condition characterized by increased risk of leukemias and cancers of the lung, brain, and breast.71,75 Mutations in PTEN causes Cowden syndrome.71,75 Both genes follow an autosomal dominant inheritance pattern and mutations are very rare. Linkage analyses using large numbers of families without BRCA1 or BRCA2 mutation failed to find additional genes, suggesting that other high-penetrance susceptibility genes, if these exist, may account for only a small fraction of the familial aggregation of breast cancer.76
As BRCA1 and BRCA2 are involved in the pathways of genomic integrity, DNA repair, and cell-cycle checkpoint controls, direct interrogation of these pathways have identified 6 candidates for breast cancer risk: CHEK2, ATM, BRIP1, PALB2, NBS1, and RAD50.75 Mutations in these genes were associated with 2- to 4-fold increased risk of breast cancer.71,75 However, mutation frequencies are very low in the general population: 1.1% have CHECK2*1100delC polymorphism, ~ 0.4% are heterozygous carriers of ATM mutations, and less than 0.1% are heterozygous carriers of BRIP1 or PALB2 mutations.71
Using a candidate gene approach, many studies have examined the association between breast cancer and common polymorphisms (>5%) of genes in the pathways of hormone synthesis and metabolism, carcinogen metabolism, and DNA repair. Most of the initial findings have not been replicated as the relative risks are presumably small. For example, the Breast Cancer Association Consortium has confirmed only two polymorphism in CASP8 and TGFB1, with allelic odds ratios of 0.88 and 1.08, respectively.77
Recently, genome-wide single-nucleotide polymorphism (SNP) association studies have found several breast cancer susceptibility loci, including FGFR2, TOX3, MAP3K1, LSP1, and 8q24.78-80 These susceptibility loci confer only a modest risk, with allelic odds ratio of approximately 1.3 and 1.2 for FGFR2 and TOX3, although the population-attributable risk might be high.78-80 Furthermore, both genes were found to have stronger association with ER+ than with ER- breast tumors.81 As most of the significant SNPs fall into introns or intergenic regions, further studies are warranted to identify causative alleles.
History of benign breast disease is associated with breast cancer diagnosis.82 Benign breast lesions are classified into three categories: nonproliferative breast disease (eg., fibroadenoma, cysts), proliferative breast disease without atypia (eg, adenosis, intraductal papilloma), and atypical hyperplasia.83 Nonproliferative breast disease, the most common lesion, is associated with small increased risk or no effect (relative risk 0.9-1.6).83-85 Women with proliferative breast disease without atypia had about 1.5- to 1.9-fold increased risk, whereas atypical hyperplasia produced about a 3- to 5-fold higher risk of breast cancer.83-85 In the high-risk atypical hyperplasia group, atypical lobular hyperplasia conferred an even higher risk than ductal hyperplasia.86 Of note, the subsequent breast cancer is slightly more likely to occur in ipsilateral (60%) than in the contralateral (40%) breast,86 and the ratio of ipsilateral to contralateral breast cancer was highest in the first 5 years after benign breast disease diagnosis,84 suggesting that benign breast disease may be both a cancer precursor and a biomarker for long-term risk.
The mammographic image is determined by the relative amounts of fat, connective tissue, and epithelial tissue.87 Connective and epithelial tissue has a high radiologic density, whereas fat appears translucent; thus mammographic density has been consistently observed to be inversely associated with both age and obesity.47,87-89 Women with the most extensive mammographic density (as a percentage of total area imaged) have a 2- to 6-fold greater risk, compared to those with low to absent density.88-90 Consistent with its risk association, mammographic density has been reported to increase with hormone replacement91,92 and to decrease with tamoxifen.93,94 In a recent study aimed towards developing a prospective risk prediction model using the Breast Cancer Surveillance Consortium (BCSC) cohort, breast density was found to be a strong contributor to risk, with 2- to 4-fold relative risks for extremely dense versus nondense breasts according to BI-RADS scoring.95 Breast density also figured prominently in a subsequent extension of the original Gail Risk Model.96 Among participants in a randomized trial for DCIS, highly dense breasts (according the area of the breast occupied with dense tissue) were associated with risk of subsequent invasive cancer.97 However, in another study using a subset of the BCSC cohort, BI-RADS density was not significantly associated with invasive cancer after DCIS.98