Purpose

Over the past half century, breast cancer incidence in the USA has undergone considerable fluctuation, potentially influenced by implementation of mammography screening [1], increasing obesity [2], and changes in hormone therapy use [3, 4]. The relative contributions of these factors to breast cancer incidence have not been resolved. As breast cancer is the most common cancer in US women [5], clarification of these relationships has implications for future clinical and health policy decisions.

Methods

Annual breast cancer incidence trends (1975–2015) for US women 50–64 years of age were assembled from nine of the national cancer institute’s surveillance, epidemiology, and end results (SEER) registries. Annual incidence rates were age adjusted to the US standard population in 2000 (19 age groups; Census P25-1130 series) and were adjusted for reporting delays. Trends in mammography among women aged 50–64 years were from Table 33 [1] mammography use among women aged ≥ 50, from the center for disease control national health interview survey, and conducted 15 times between 1987 and 2018 combined with estimates of 0% through 1980; before mammographic screening was generally available [1]. Trends in hormone therapy use were based on findings from the collaborative group on hormonal factors in breast cancer report [6]. Annual obesity prevalence estimates (BMI ≥ 30 kg/m2) were from the non-communicable diseases (NCD) Risk Factor Collaboration, a global network providing data on risk factors for diseases including obesity trends from 1975 to 2016 from 3,300 population-based studies with nearly 200 million participants [7].

Results

Findings regarding the annual breast cancer incidence by trends in hormone therapy use, mammography screening, and obesity as backdrop are depicted in Fig. 1. Visual inspection of the year-to-year trend in age-adjusted annual breast cancer incidence indicates a modest reduction in breast cancer incidence beginning in 1975, coincident with reduced hormone therapy use (Fig. 1) following the estrogen-alone association with higher endometrial cancer incidence [8]. Around 1982, a sustained 20-year increase in breast cancer incidence began, coincident with the transitional phase of mammogram use, shortly followed by wide-scale implementation of mammogram screening beginning around 1992 [9].

Fig. 1
figure 1

Annual incidence of breast cancer using trends in hormone therapy use, mammography screening, and obesity as a backdrop. Incident breast cancer rates (95% CI; dot whiskers) among women 50 − 64 years old and number of menopausal hormone therapy users in the U.S.A. (black line) are shown over time during distinct phases of mammographic screening (pre-screening, transitional, screening) as indicated by the dotted vertical lines and quantified by mammography rates (red-connected dots), during which time obesity rates (blue line) steadily increased

Full size image

The year-to-year, age-adjusted increase in breast cancer incidence peaked around 1999 as mammography screening continued at a high level which has largely persisted, with modest decrease, to current time [1]. After presentation of findings in 2002 [10] and 2003 [11] from the Women’s Health Initiative (WHI) randomized trial, where estrogen plus progestin increased breast cancer incidence and impaired breast cancer detection, a substantial and sustained decrease in menopausal hormone therapy use was seen [3, 6].

Subsequently, the first decrease in breast cancer incidence in two decades emerged for women ≥ 50 years old and was attributed to reduced hormone therapy use [4], a controversial hypothesis, corroborated with both longer follow-up of the WHI trial (n = 16,608) comparing estrogen plus progestin to placebo influence on breast cancer incidence and a comparable observational cohort of WHI women (n = 41,449), consisting of participants that reported no use of menopausal hormone therapy and those that reported use of estrogen plus progestin at baseline. All trial participants were instructed to immediately stop taking study pills (active or placebo) in a letter coincident with the publication of the trial results on July 8, 2002 [10]. Subsequently, a sudden decrease in breast cancer incidence was seen only in the combined hormone therapy group and not in the placebo group [12]. Similar patterns were also observed for the observational cohort [12]. Lower incidence in the U.S.A., beginning in 2003, has been largely maintained through 2020 at levels last seen in 1993 [5, 6], despite continued high mammography participation. While obesity prevalence had year-to-year increases over this nearly five-decade period, increasing obesity did not track changes in breast cancer incidence (Fig. 1).

Discussion

There have been substantial changes in year-to-year age-adjusted breast cancer incidence rates for women 50 to 65 years of age over the past 50 years. The introduction and increasing use of mammography track the incident breast cancer rate from the transitional phase of mammography introduction in the mid-1980s through 2002. However, the subsequent continued high mammogram adherence rate does not track the substantial decline in breast cancer seen beginning in 2003. Menopausal hormone therapy use does track the modest decline in breast cancer incident rates seen from about 1975 to 1980 and tracts the sudden, substantial decline, and the maintained lower breast cancer incidence seen beginning in 2003 and maintained subsequently. The year-to-year increase in obesity prevalence seen from 1975 does not appear related to major changes in breast cancer incidence rates.

The decades long increase in breast cancer incidence seen with estrogen plus progestin use [13]  likely were mediated by a progestin-induced breast epithelium stem cell pool increase, resulting in a persistent cell population at higher breast cancer risk [14]. Thus, the sustained lower breast cancer incidence in the US population seen since 2003 could be related to fewer women who would otherwise initiate estrogen plus progestin and subsequently be at long-term, higher breast cancer risk. By one estimate, in comparison to 2002, there have been 126,000 fewer breast cancers through 2012 in the USA than would have occurred without the WHI findings [15]. Because the lower age-adjusted breast cancer incidence has persisted for another decade, there were likely 10,000 fewer breast cancers yearly from 2003 through 2022 related to the WHI reports.

This report did not consider the influence of breast cancer prevention programs. However, despite the success of endocrine-targeting agents to reduce breast cancer incidence in randomized trials [16, 17], the uptake of incorporating these agents in breast cancer prevention programs in clinical practice has been limited [18]. Also not addressed were trends in breast cancer mortality. While age-adjusted breast cancer incidence was increasing from the mid-1980s to 2002, then suddenly decreased in 2003 to a lower incidence level, breast cancer mortality has been steadily decreasing In US women since about 1990 [5, 19] largely related to screening mammography early detection and adjuvant therapy advances including neoadjuvant therapy to define additional therapy needs and biomarkers to guide adjuvant endocrine and chemotherapy. Thus, trends in breast cancer incidence and breast cancer mortality are driven by different factors.

Study strengths include the use of reliable sources for changes in breast cancer incidence, mammography, hormone therapy, and obesity. Limitations relate to visual rather than statistical comparisons, across population-level summaries, for associations of trends in mammography, obesity, and hormone therapy to trends in breast cancer incidence.

Conclusion

Based on consideration of trends in mammography, obesity, and menopausal hormone therapy, the sustained reduction in hormone therapy use beginning in 2003 provides a plausible explanation for most of the lower age-adjusted breast cancer incidence seen in US postmenopausal women during the last two decades. The strong observational study association of obesity with higher breast cancer risk is not reflected in population trends.