Introduction
Breast cancers are biologically heterogeneous. Gene expression profiling of breast tumor tissues has identified reliable patterns indicative of clinically distinct subtypes [
1‐
4]. At this time, the subtype classifications most often used in clinical settings are based on the commonly measured tumor markers estrogen receptor (ER), progesterone receptor (PR), and HER2/neu (HER2), which offer imperfect but practical surrogates for genomic profiling [
5]. It is increasingly recognized that breast cancer subtypes vary in occurrence (especially by race/ethnicity) [
5‐
8], in their detection by screening mammography [
9,
10], and in their risk associations with other factors [
11‐
15]. Treatment options and prognosis also depend on breast cancer subtype [
9,
16‐
18].
Despite accumulating evidence that breast cancer subtypes should be considered separately, it is still routine to present statistics that consider the disease as a single entity. Perhaps most commonly cited is the 12% lifetime probability statistic [
19], prompting the widespread perception that 'one in eight' US women will develop the disease. This single estimate does not convey race-specific variation in breast cancer risks. Moreover, although some groups are reported to have greater relative risk of specific breast cancer subtypes, there are no data with which to counsel patients about the absolute magnitude of these risks in comparison with other threats to their health; one clinically important example is the ER-, PR-, and HER2-negative (triple-negative) breast cancer subtype among black women [
5]. To provide estimates relevant to patient care and health policy, we took advantage of recent data on subtype-specific incidence patterns (collected in the large and diverse population of California) to calculate absolute lifetime risks of developing a first primary breast cancer according to breast cancer subtype and presented those calculations separately for women of four racial/ethnic groups.
Conclusions
We present lifetime and age-specific probabilities of developing luminal (ER- or PR-positive or both and HER2-negative), HER2-positive (ER- and PR-positive or -negative and HER2-positive), and triple-negative (ER-, PR-, and HER2-negative) subtypes of breast cancer for women from four racial/ethnic groups and use the most recently available data from the large and diverse population of California. These estimates refine the frequently cited 'one in eight' statistic [
12], which fails to capture the substantial differences in epidemiology and prognosis among breast cancer subtypes [
6‐
9,
16,
17]. Most importantly, these estimates facilitate clinically relevant discussion between patients and physicians. For women considering prevention strategies such as prophylactic tamoxifen or raloxifene, which reduce the incidence of only certain subtypes of breast cancer [
30,
31], or screening methods such as magnetic resonance imaging (MRI), which may contribute more to the detection of triple-negative than luminal cancers [
9,
10], our estimates may inform decisions about managing breast cancer risk. A woman at low risk for a specific subtype might choose to forego particular interventions and their side effects (for example, stroke and uterine cancer from tamoxifen or false-positive biopsies from screening mammogram or MRI) [
30,
32], depending on the relative importance of such side effects and her competing health risks.
We present statistics separately for women in four major racial/ethnic groupings because lifetime risks for breast cancer as a whole vary substantially by these groups. Most notable were the significantly increased risks of luminal breast cancer among whites and of triple-negative breast cancer among black women. Our findings are consistent with studies reporting greater relative risks of triple-negative breast cancer among black women [
5,
7,
8,
33] and strengthen the rationale for investigating genetic, reproductive, and lifestyle factors that may mediate this racial difference, such as age at menarche, family cancer history, breastfeeding, and abdominal adiposity [
34,
35]. In all groups, the luminal subtype was the most common one. This universal predominance of luminal (ER- or PR-positive or both and HER2-negative) breast cancer, regardless of race, may be reassuring since this subtype has the best survival [
5,
17], can be targeted by existing chemoprevention agents [
30,
31], and may be most readily detectable by screening mammography [
9,
10,
36,
37]. Although the disproportionately increased risk among black women of poor-prognosis triple-negative breast cancer warrants further study and targeted interventions, black women may be reassured to learn that they also have a high probability of avoiding this disease over their lifetimes.
It is essential to differentiate risks according to a woman's current age. A prior analysis using SEER data characterized qualitative patterns of breast cancer incidence according to ER status and reported an age-related crossover between black and white women [
38]. We found that, for all races and for all subtypes, absolute breast cancer risks were low between ages 40 and 49 years: less than 1% per subtype and less than 2% for all subtypes combined. For women between ages 50 and 59, risks of each subtype increased substantially, and the greatest increase was for luminal breast cancers in white women. Nearly half the lifetime risk of luminal breast cancer, the dominant subtype for all racial/ethnic groups, occurred at or after age 70. These findings are important to the ongoing critical examination of mammographic screening guidelines [
32,
36,
39] and may warrant extending recommendations for mammographic screening beyond the current upper limit of 69 years of age [
32].
Our analyses have certain limitations, which should be considered in interpreting our results. Given the DevCan program's competing-risks methodology, cause-specific mortality is required to calculate incidence of the cancer in question [
23,
25]; we used overall US breast cancer mortality rates to calculate subtype-specific incidence [
28] because subtype-specific mortality rates are not available. However, since overall breast cancer mortality is low at the population level, this is unlikely to affect our risk estimates substantially. Racial misclassification might present another potential source of bias, but given that prior studies of the CCR found that race data derive from patient self-report in more than 80% of cases [
21], it seems improbable that a large proportion were incorrectly classified. We excluded 7,737 cases (18.9%) from analysis because of missing ER, PR, or HER2 information; since there were no major differences in race or age distribution between the excluded and included cases, the lack of information on these cases seems unlikely to have biased our findings. Although defining subtypes by ER, PR, and HER2 expression does not entirely approximate results of genomic profiling, this classification offers a practical substitute that is increasingly well characterized in published literature [
5,
7‐
9,
40‐
42] and that guides breast cancer treatment [
43].
This study reports average lifetime risks at the population level; it does not address the urgent need for more accurate risk stratification of individual patients or the limitations of current breast cancer risk prediction models [
44]. Genetic mutations such as
BRCA1 convey dramatically increased risks of triple-negative breast cancer [
45,
46], and the results of genome-wide association studies may eventually guide even more personalized risk prediction [
47,
48]. Our estimates may inform health policy and resource planning across diverse populations and may help patients and clinicians to weigh the average probabilities of developing specific breast cancer subtypes against other competing health risks.
Acknowledgements
This study was supported by the National Cancer Institute (NCI) Surveillance, Epidemiology and End Results (SEER) program under contract N01-PC-35136. The collection of cancer incidence data used in this study was supported by the California Department of Public Health as part of the statewide cancer reporting program mandated by California Health and Safety Code Section 103885; the NCI SEER program under contract N01-PC-35136 awarded to the Cancer Prevention Institute of California (formerly the Northern California Cancer Center), contract N01-PC-35139 awarded to the University of Southern California, and contract N01-PC-54404 awarded to the Public Health Institute; and the Centers for Disease Control and Prevention's National Program of Cancer Registries, under agreement 1U58DP00807-01 awarded to the Public Health Institute. The ideas and opinions expressed herein are those of the authors, and endorsement by the State of California, Department of Public Health, the NCI, or the Centers for Disease Control and Prevention or their contractors and subcontractors is not intended nor should it be inferred.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
AWK conceived of the study, participated in study design and data interpretation, and drafted the manuscript. KF and SJS participated in study design and performed statistical analysis. CAC conceived of the study; participated in study design, analysis, and interpretation; and helped to draft the manuscript. All authors read and approved the final manuscript.