Hormone therapy use and young-onset breast cancer: a pooled analysis of prospective cohorts included in the Premenopausal Breast Cancer Collaborative Group

Katie M O’Brien; Melissa G House; Mandy Goldberg; Michael E Jones; Clarice R Weinberg; Amy Berrington de Gonzalez; Kimberly A Bertrand; William J Blot; Jessica Clague DeHart; Fergus J Couch; Montserrat Garcia-Closas; Graham G Giles; Victoria A Kirsh; Cari M Kitahara; Woon-Puay Koh; Hannah Lui Park; Roger L Milne; Julie R Palmer; Alpa V Patel; Thomas E Rohan; Minouk J Schoemaker; Anthony J Swerdlow; Lauren R Teras; Celine Vachon; Kala Visvanathan; Jian-Min Yuan; Wei Zheng; Hazel B Nichols; Dale P Sandler

doi:10.1016/S1470-2045(25)00211-6

. Author manuscript; available in PMC: 2026 Jul 1.

Published in final edited form as: Lancet Oncol. 2025 Jul;26(7):911–923. doi: 10.1016/S1470-2045(25)00211-6

Hormone therapy use and young-onset breast cancer: a pooled analysis of prospective cohorts included in the Premenopausal Breast Cancer Collaborative Group

Katie M O’Brien ¹, Melissa G House ², Mandy Goldberg ¹, Michael E Jones ³, Clarice R Weinberg ⁴, Amy Berrington de Gonzalez ³, Kimberly A Bertrand ⁵, William J Blot ⁶, Jessica Clague DeHart ⁷, Fergus J Couch ⁸, Montserrat Garcia-Closas ³, Graham G Giles ^9,^10,¹¹, Victoria A Kirsh ^12,¹³, Cari M Kitahara ¹⁴, Woon-Puay Koh ^15,¹⁶, Hannah Lui Park ¹⁷, Roger L Milne ^9,^10,¹¹, Julie R Palmer ⁵, Alpa V Patel ¹⁸, Thomas E Rohan ¹⁹, Minouk J Schoemaker ^3,²⁰, Anthony J Swerdlow ^3,²¹, Lauren R Teras ¹⁸, Celine Vachon ²², Kala Visvanathan ²³, Jian-Min Yuan ^24,²⁵, Wei Zheng ⁶, Hazel B Nichols ²⁶, Dale P Sandler ¹

¹Epidemiology Branch, National Institute of Environmental Health Sciences, Research Triangle Park, NC, USA (K M O’Brien PhD, M Goldberg PhD, D P Sandler PhD);

²Westat, Durham, NC, USA (M G House MS);

³Division of Genetics and Epidemiology, The Institute of Cancer Research, London, UK (M E Jones PhD, A Berrington de Gonzalez DPhil, M Garcia-Closas MD DrPH, M J Schoemaker PhD);

⁴Biostatistics and Computational Biology Branch, National Institute of Environmental Health Sciences, Research Triangle Park, NC, USA (C R Weinberg PhD);

⁵Slone Epidemiology Center at Boston University, Boston, MA, USA (K A Bertrand ScD, J R Palmer ScD);

⁶Division of Epidemiology, Department of Medicine, Vanderbilt University Medical Center, Nashville, TN, USA (W J Blot PhD, W Zheng MD PhD);

⁷School of Community and Global Health, Claremont Graduate University, Claremont, CA, USA (J Clague DeHart PhD);

⁸Department of Laboratory Medicine and Pathology, Mayo Clinic, Rochester, MN, USA (F Couch PhD);

⁹Cancer Epidemiology Division, Cancer Council Victoria, Melbourne, Victoria, Australia (G G Giles PhD, R L Milne PhD);

¹⁰Centre for Epidemiology and Biostatistics, Melbourne School of Population and Global Health, The University of Melbourne, Victoria, Australia (G G Giles PhD, R L Milne PhD);

¹¹Precision Medicine, School of Clinical Sciences at Monash Health, Monash University, Clayton, Victoria, Australia (G G Giles PhD, R L Milne PhD);

¹²Ontario Institute for Cancer Research, Toronto, ON, Canada (V A Kirsh PhD);

¹³Dalla Lana School of Public Health, University of Toronto, Toronto, ON, Canada (V A Kirsh PhD);

¹⁴Radiation Epidemiology Branch, National Cancer Institute, Rockville MD, USA (C Kitahara PhD);

¹⁵Healthy Longevity Translational Research Programme, Yong Loo Lin School of Medicine, National University of Singapore, Singapore, Singapore (W-P Koh PhD);

¹⁶Singapore Institute for Clinical Sciences, Agency for Science Technology and Research, Singapore, Singapore (W-P Koh PhD);

¹⁷Department of Pathology and Laboratory Medicine, Department of Epidemiology, UC Irvine School of Medicine, Irvine, CA, USA (H Lui Park, PhD);

¹⁸Department of Population Science, American Cancer Society, Atlanta, GA, USA (A V Patel PhD, L R Teras PhD);

¹⁹Epidemiology and Population Health, Albert Einstein College of Medicine, Bronx, New York, USA (T E Rohan MBBS, DHSc);

²⁰Real World Solutions, IQVIA, Amsterdam, the Netherlands (M J Schoemaker PhD);

²¹Division of Breast Cancer Research, The Institute of Cancer Research, London, United Kingdom (A J Swerdlow PhD);

²²Department of Quantitative Health Sciences, Division of Epidemiology, Mayo Clinic, Rochester, MN, USA (C Vachon PhD);

²³Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA (K Visvanathan MD MHS);

²⁴Cancer Epidemiology and Prevention Program, University of Pittsburgh Medical Center Hillman Cancer Center, Pittsburgh, PA, USA (J-M Yuan MD PhD);

²⁵Department of Epidemiology, School of Public Health, University of Pittsburgh, Pittsburgh, PA, USA (J-M Yuan MD PhD);

²⁶Department of Epidemiology, Gillings School of Global Public Health at the University of North Carolina, Chapel Hill, NC, USA (H B Nichols PhD)

Contributions: Drs. O’Brien and Jones and Ms. House had full access to all the data in the study. Drs. O’Brien and Sandler take responsibility for the integrity of the data and accuracy of the data analysis. Dr. O’Brien and Ms. House were responsible for data verification.

Study concept and design: all authors

Data curation: O’Brien, House, Jones, Schoemaker, Swerdlow, Nichols, Sandler

Methodology, Formal analysis: O’Brien and Sandler

Funding acquisition, Project administration, Resources: O’Brien, Jones, Berrington de Gonzalez, Bertrand, Blot, Clague DeHart, Couch, Garcia-Closas, Giles, Kirsh, Kitahara, Koh, Lui Park, Milne, Palmer, Patel, Rohan, Schoemaker, Swerdlow, Teras, Vachon, Visvanathan, Yuan, Zheng, Nichols, Sandler

Supervision: Sandler and Nichols

Writing- original draft: O’Brien, House, Goldberg, Jones, Weinberg, Nichols, Sandler

Writing- reviewing and editing: all authors

^✉

Corresponding author: Katie M. O’Brien, Epidemiology Branch, National Institute of Environmental Health Sciences, 111 TW Alexander Drive, Research Triangle Park, NC, USA, 27709; katie.obrien@nih.gov

PMCID: PMC12233149 NIHMSID: NIHMS2089962 PMID: 40609572

Summary

Background

Estrogen plus progestin hormone therapy (EP-HT) is an established risk factor for breast cancer in postmenopausal women. We examined the less well-studied association between exogenous hormones and breast cancer in premenopausal women, who may use hormone therapy following gynecologic surgery or to relieve peri-menopausal symptoms.

Methods

We investigated this relationship using pooled data from 10–13 prospective cohorts from North America, Europe, Asia and Australia. This included 459,476 women aged 16–54 (mean=42), 8,455 of whom developed young-onset breast cancer (YOBC; diagnosed <55 years) (median follow-up=7.8 years). We used cohort-stratified, multivariable-adjusted Cox proportional hazards regression to estimate hazard ratios (HRs) and 95% confidence intervals (CI) for associations of hormone therapy with incident YOBC. We also estimated risk differences (RD) based on cumulative risk up to age 55.

Findings:

Overall, 15% of participants reported ever using hormone therapy, with EP-HT (6%) and unopposed estrogen (E-HT; 5%) the most common types. Cumulative risk of YOBC was 4.1% in non-users. Hormone therapy of any type was not associated with incident YOBC (HR=0.96, CI: [0.88–1.04]), but ever E-HT use was inversely associated (HR=0.86 [0.75–0.98]; RD= −0.5% [−1.0, −0.0]). The HR for ever EP-HT and YOBC was 1.10 (0.98–1.24), with positive associations observed for longer-term use (HR=1.18 [1.01–1.38] for >2 years) and use among women without hysterectomy or bilateral oophorectomy (HR=1.15 [1.02–1.31]). The E-HT and YOBC association was similar for all subtypes, but EP-HT was more strongly associated with estrogen receptor negative (HR=1.44 [1.11–1.88]) and triple-negative disease (HR=1.50 [1.02–2.20]), than other subtypes.

Interpretation:

E-HT use was inversely associated with YOBC, and EP-HT was associated with higher YOBC incidence among women with intact uterus and ovaries. These findings largely parallel results from studies of hormone use and later-onset breast cancer and provide novel evidence for establishing clinical recommendations among younger women.

Keywords: young-onset breast cancer, hormone therapy, premenopausal breast cancer, pooled analysis

Research from the Women’s Health Initiative (WHI) hormone therapy randomized clinical trial and the Million Women Study (MWS) established estrogen plus progestin combination hormone therapy (EP-HT) as a risk factor for breast cancer among postmenopausal women.^1,2 The association between unopposed estrogen hormone therapy (E-HT) and risk of postmenopausal breast cancer is less certain. The WHI trial found an inverse association among women without a uterus,³ but the MWS and a later collaborative group study⁴ reported positive associations that increased with duration of use.

Exposure to endogenous and exogenous estrogen or progesterone is generally thought to initiate or promote the growth of hormone-sensitive tumors, including most breast cancers.⁵ Women may seek hormone therapy to manage symptoms related to natural or surgical menopause or other hormone-related conditions.^6–9 Natural menopause occurs at age 52, on average,¹⁰ though women generally start experiencing vasomotor symptoms several years prior to their last menstrual period.¹¹ While women younger than 55 may therefore experience clinical benefits from hormone therapy, they represent a small proportion of those given a prescription.¹² Young women have been largely excluded from major studies of hormone therapy and breast cancer, so risks associated with pre- and perimenopausal hormone therapy use are not known.

We aimed to address this research gap by investigating the association between hormone therapy use and incident young-onset breast cancer (YOBC), defined here as breast cancer diagnosed before age 55 years. We pooled individual-level data from prospective cohort studies participating in the Premenopausal Breast Cancer Collaborative Group¹³ and investigated the association between type of hormone therapy (EP-HT, E-HT, others) and breast cancer incidence, overall and by breast cancer subtype, including considerations of how these associations differed by duration of use, age at first use, or gynecological surgery status. In secondary analyses, we examined associations within subgroups defined by age, menopausal status, body mass index (BMI), race, geographic region, birth cohort, and family history of breast cancer. The overall goal of this research was to provide empirical evidence that might be useful for informing clinical recommendations for younger women considering hormone therapy.

Methods

Study design

The Premenopausal Breast Cancer Collaborative Group has harmonized and pooled data from females participating in prospective cohort studies who were younger than 55 years of age at enrollment and had not been previously diagnosed with breast cancer. Contributing studies represent populations in North America, Europe, Asia, and Australia. Since the group’s initial publications,^13–15 additional cohorts have been added, as have time-dependent covariate data and new incident cases, when available. Participating cohorts followed women for incident breast cancer until at least age 55 years.

For this analysis of hormone therapy and breast cancer incidence, 13 cohorts met initial eligibility criteria. Of these, 10 provided data on type of hormone therapy (Black Women’s Health Study,¹⁶ the Generations Study,¹⁷ California Teachers’ Study¹⁸ [CTS], Cancer Prevention Study 3,¹⁹ Canadian Study of Diet, Lifestyle, and Health²⁰ [CSDLH], Campaign Against Cancer and Heart Disease²¹ [CLUEII], Mayo Mammography Health Study,²² Singapore Chinese Health Study,²³ Sister Study²⁴ [SIS], and US Radiologic Technologists Cohort²⁵ [USRTC]). Three others (Melbourne Collaborative Cohort Study,²⁶ Shanghai Women’s Health Study,²⁷ and Southern Community Cohort Study²⁸) provided data on ever versus never use, but not type. All participants provided study-specific written informed consent, and each cohort had approval and oversight by a relevant institutional review board.

Participants

From the 13 cohorts with data on hormone therapy use we excluded 8,946 participants (1.6%) whose data on hormone therapy use was missing, leaving 547,751 eligible women. Of these, 9,482 developed incident YOBC (median follow-up=7.8 years). In the subset of 10 cohorts with data on hormone therapy type, 459,476 women contributed data, including 8,455 with incident YOBC. One of the cohorts, CSDLH, provided data for a case-cohort sample (total n=1574, with 376 cases), which we weighted according to their inverse probability of selection into the sample (1.0 for cases, 23.1 for non-cases) to represent the original cohort.²⁹

Hormone Therapy

Each of the included cohorts was asked to provide detailed information on hormone therapy use as of study enrollment, including current use (never, former, current, current non-user with unknown history), age at first use, age at last use, and total cumulative years of use. Cohorts with data on hormone therapy type were asked to harmonize responses into the following mutually exclusive categories: never use, E-HT only, EP-HT only, both E-HT and EP-HT, other hormone therapy (O-HT), and unknown. No additional information was collected for women who responded that they used “other” hormone therapy and cohorts were not asked to specify the route of administration (e.g. oral, transdermal, vaginal). We defined long-term users as those reporting more than two years of cumulative use, which was approximately half of users.

In addition to providing information about hormone use prior to study enrollment, four of the cohorts (CTS, CLUEII, SIS, and USRTC) also provided data updating hormone use during follow-up. Pooling across these studies, we estimated the associations of time-varying hormone therapy use with YOBC.

Incident YOBC

All invasive or in situ breast cancers diagnosed after enrollment and before age 55 years were considered YOBC events. Many of the studies verified cases via cancer registries,^{16–23,26–28} while others relied on self-report augmented by medical record validation.^24,30 Data on disease stage, estrogen receptor (ER) status, progesterone receptor (PR) status, and human epidermal growth factor receptor-2 (HER2) status were included, when possible. Follow-up was defined on the age time scale, starting at enrollment and continuing until YOBC event, with censoring at the minimum of age 55, loss to follow-up, cohort-defined end of follow-up, or death.

Potential Confounders and Other Covariates

Covariates were considered confounders if they were potentially causally related to YOBC and associated with, but not affected by, hormone therapy use. All were based on status at enrollment and included: age, attained education level (≤high school, some college, college graduate), body mass index (BMI; kg/m², continuous), age at menarche (continuous), birth cohort (1928–1945, 1946–1954, 1955–1964, ≥1965), race (White, Black, Asian, or a category combining all other races), menopausal status (with postmenopausal defined as not having menstruated in at least 12 months), ever hysterectomy, ever bilateral oophorectomy, oral contraceptive use (never, former, current), number of births (0, 1, 2, ≥3), age at first birth (an interaction term with ever parous; age <20, 20–24, 25–29, 30–34, ≥35 years), and age at menopause (age at last menstrual period with a menopausal status interaction term; continuous).

Because some of the covariates had high proportions of missingness (Table 1, Supplementary Table 1), we implemented multiple imputation with chained equations (MICE; SAS v9.4). All potential confounders were included in the imputation model, in addition to cohort, geographic location (North America, Europe, Asia/Australia), regular alcohol use (ever, never, as defined by study), first-degree family history of breast cancer, case status, and crude cumulative hazard estimates.³¹ We also imputed breast cancer subtype (ER, PR and HER2 status) with these models, including for studies without tumor data.

Table 1.

Covariate distributions at enrollment for 10 cohorts with data on hormone therapy type, Premenopausal Breast Cancer Collaborative Group (n=459,476)

	Full pooled sample^a	Breast Cancer Cases (n=8,455) (diagnosed age <55)
Age at enrollment (years); Mean (IQR)	42.0 (35.5, 49.2)	40.7 (35.3, 46.5)
Age at end of follow-up (years); Mean (IQR)	51.0 (48.4, 55.0)	48.1 (45.0, 52.3)
Follow-up Time (years); Median (IQR)	7.8 (5.2, 11.2)	5.9 (2.8, 11.0)
Body mass index (BMI) (kg/m ² ); Mean (IQR)	26.5 (21.9, 29.3)	25.3 (21.3, 27.9)
Age at menarche (years); Mean (IQR)	12.6 (12.0, 13.0)	12.5 (12.0, 13.0)
Race; N (%)
White	354,108 (79)	6,473 (77)
Black	65,526 (14)	1,495 (18)
Asian	21,518 (5)	236 (3)
Mixed or other	10,952 (2)	149 (2)
Birth cohort
1928–1945	46,509 (11)	672 (8)
1946–1954	103,910 (24)	2,590 (31)
1955–1964	154,214 (33)	3,699 (44)
≥1965	154,843 (32)	1,494 (18)
Geographic region; N (%)
North America	371,127 (82)	7,280 (86)
Europe	72,296 (15)	1,041 (12)
Asia	16,053 (3)	134 (2)
Attained education; N (%)
High school or less	49,549 (11)	647 (9)
Some college	167,079 (38)	3,287 (43)
Completed college/university	197,359 (51)	3,656 (48)
Postmenopausal; N (%)	98,068 (21)	1,134 (14)
Age at menopause (years)^b; Mean (25^th, 75^th %ile)	45.0 (41.0, 51.0)	42.0 (37.0, 48.0)
Hormone therapy; N (%)
Never	388,911 (85)	7,513 (89)
Unopposed estrogen (E-HT)	24,932 (5)	290 (3)
Estrogen plus progestin (EP-HT)	23,863 (6)	334 (4)
Both E-HT and EP-HT	8,160 (2)	75 (1)
Other (O-HT)	6,828 (1)	134 (2)
User, unknown type	6,782 (1)	109 (1)
Bilateral oophorectomy; N (%)	23,444 (5)	266 (5)
Hysterectomy; N (%)	56,262 (12)	757 (9)
Oral contraceptive use; N (%)
Never	67,347 (16)	1,242 (15)
Former	289,664 (68)	5,589 (68)
Current	67,257 (15)	1,272 (15)
User, unknown timing	5,782 (1)	137 (2)
Ever pregnant; N (%)	345,330 (76)	6,388 (76)
Parity; N (%)
No births	127,712 (28)	2,407 (29)
1 birth	72,203 (16)	1,510 (18)
2 births	139,724 (31)	2,734 (33)
≥3 births	76,859 (17)	1,343 (16)
Parous, unknown number of births	37,611 (8)	409 (5)
Age at first birth (years), among parous; N (%)
<20	30,774 (9)	457 (8)
20–24	90,653 (27)	1,544 (26)
25–29	111,754 (35)	2,208 (37)
30–34	54,673 (17)	1,200 (20)
35+	16,522 (5)	394 (7)
Birth at unknown age	22,021 (7)	193 (3)
Ever Regular Drinker; N (%)	279,996 (73)	4,399 (69)
1^st degree family history of breast cancer; N (%)	66,206 (15)	1,909 (24)

Open in a new tab

IQR = interquartile range

Restricted to cohorts and participants with data on ever/never hormone therapy use and type. Other missing data: BMI 7,605; age at menarche 79,284; race 7,372; education 45,489; menopause status 3,805; age at menopause 4,052; bilateral oophorectomy 7,795; hysterectomy 8,571; oral contraceptive use 29,426; ever pregnant 5,411; parity 5,367; alcohol use 75,108; family history of breast cancer 31,782.

The Canadian Study of Diet, Lifestyle, and Health provided case-cohort data (1574 total women, with 376 cases; representative of 28,000 total women). When calculating means (IQRs) and percentages, we weighted non-cases in the sub-cohort based on their probability of selection into the sub-cohort (4.3%, weight=23.1).

Among postmenopausal participants

Statistical Analysis

We examined covariate distributions in the pooled sample, first for the combined cohorts (weighted by inverse probability of sampling) and then separately among cases. Within the 10 cohorts with hormone therapy type data, we compared covariate distributions by hormone therapy type.

Associations were initially assessed via random effects meta-analysis (implemented in R 4.4.0),³² using Cox regression models to generate cohort-specific multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs). We adjusted for the above listed covariates, as defined at enrollment, plus a BMI by menopause status interaction term.^15,33 This was done with complete case analysis, but if a cohort was completely missing specific confounders, those covariates were omitted from the adjusted model. Study-specific HRs were combined via meta-analyses to assess effects of ever versus never hormone therapy use (all 13 cohorts) and type of hormone therapy relative to never use (10 cohorts). We used I² values and Cochran’s Q statistics to quantify heterogeneity across cohorts.

We next estimated pooled hazard ratios (HRs) and 95% confidence intervals (CIs) for the 10 cohorts with data on hormone therapy type (implemented in SAS v9.4). The HRs were averaged over 10 multiply imputed copies of the pooled data set using Rubin’s rules,³⁴ with verification that estimates remained consistent across multiple seeds. We calculated both age- and multivariable-adjusted HR estimates using cohort-stratified Cox regression models. In addition to hormone therapy type, we also considered duration of use (never, ≤2 years, >2 years), age at first use (never, <45, 45–49, ≥50), and current use (never, former, current), overall and by type. To further explore duration and age at initiation, we also considered models with interaction terms for ever use by continuous duration or ever use by continuous age.

For each primary exposure comparison, we additionally estimated cumulative risk of YOBC and risk differences (RDs) at age 55. To implement this, we estimated the baseline hazard function using the Breslow method³⁵ and then compared cumulative risk estimates across exposure groups, assuming covariate distributions identical to those observed in our sample.

We also considered factors that could potentially modify the association between E-HT or EP-HT and incident YOBC. We were particularly interested in gynecological surgery status, as women who undergo hysterectomy and/or oophorectomy often take hormone therapy,^8,36,37 though E-HT is contraindicated in women with a uterus due to its known link to increased risk of endometrial cancer.³⁸ Other potential modifiers of interest included age, menopausal status (accounting for changes over follow-up), overweight status among premenopausal women, race, geographic region, birth cohort (as a proxy for changes in prescribing patterns and decreases in standard dosages over time^4,12,39,40), and first-degree family history of breast cancer. We also estimated HRs for subsets of breast cancer cases defined by invasiveness (in situ or invasive) and subtype (ER, PR, and/or HER2 status). Here, women with breast cancers that did not meet the case definition of interest were censored at diagnosis.

Time-varying analyses

Finally, we examined the association between time-varying hormone therapy use and YOBC across the 4 cohorts with data on use during follow-up. Because these cohorts had different follow-up intervals and lengths, we harmonized the data by generating a dataset with one observation per 2-year period per participant. For each participant’s initial period, their covariates were assigned based on enrollment data. If a new questionnaire was administered before the end of the 2-year period, the covariate values for the second cycle were updated, but if no new data were collected then prior values were carried forward. This was continued every 2 years until the end of follow-up. Because covariate values could not change until the follow-up cycle after a questionnaire, the prospective nature of the data was preserved.

We again used Cox proportional hazards models to estimate pooled HRs and 95% CIs, this time stratifying baseline hazard functions by both follow-up period (i.e., 2-year interval) and cohort. We first estimated the association between YOBC and time-varying hormone therapy type, relative to never use. Next, we looked at current hormone therapy use by type, where current use was defined by use during the last 2 years. Here, we additionally adjusted for ever prior hormone therapy use as a potential confounder. Lastly, we considered duration of hormone therapy use by type and age at start of use by type, updating as needed for new or ongoing users.

Role of the funding source

The funding sources had no role in the design, conduct, or interpretation of the study.

Results

Most of the participants included in the 10 cohorts with hormone therapy type data identified as White (79%) and lived in North America (82%) or Europe (15%) (Table 1). They ranged in age from 16–54 (mean=42.0, interquartile range [IQR]=35.5–55.0) years at enrollment. Women who developed YOBC during follow-up tended to be slightly younger at enrollment (mean age=40.7 years, IQR=35.3–46.5) and were less likely to be postmenopausal (14% versus 21% overall). Women born 1955–1964 made up the largest group (33%; 44% of cases), followed closely by those born after 1964 (32%; 18% of cases). Overall, 15% of women reported a first-degree family history of breast cancer, including 24% of cases. Fifteen percent of the sample reported ever using hormone therapy prior to enrollment (currently or formerly), with 6% reporting EP-HT only, 5% E-HT only, 2% both E-HT and EP-HT, 1% hormone therapy other than E-HT or EP-HT (O-HT), and 1% unknown type. Among cases, 11% reported having used hormone therapy (3% E-HT, 4% EP-HT, <1% both E-HT and EP-HT, 2% O-HT, and 1% unknown type).

E-HT users were more likely to have had a hysterectomy (72%) or bilateral oophorectomy with or without hysterectomy (50%) than non-hormone therapy users (7% and 1%, respectively; Supplementary Table 2). Mean duration of use was 4.7 years (IQR:1.0–7.0) and mean age at initiation was 42.2 years (IQR:39.0–47.0). EP-HT users were also more likely to have had hysterectomies (16%) or bilateral oophorectomies (12%) than non-users. Mean duration of EP-HT use was 3.3 years (IQR:1.0–4.0) and mean age at initiation was 44.2 years (IQR:40.0–50.0).

In the meta-analysis of all 13 cohorts (Supplementary Table 3, Supplementary Figure 1), ever use of any hormone therapy was not associated with incident YOBC (meta-HR=1.03, 95% CI: 0.94–1.11), with no evidence of heterogeneity across cohorts (heterogeneity p-value=0.23). In analyses that pooled data from the 10 cohorts with type data, ever use was inversely associated with YOBC in age-adjusted models (HR=0.87, 0.82–0.94), but the estimate attenuated with covariate adjustment (HR=0.96, 0.88–1.04; Table 2).

Table 2.

Association between hormone therapy use prior to enrollment and incident young-onset breast cancer (age<55) among 10 cohorts (n=459,476) with data on type of hormone therapy

	Non-cases; N (%) n=451,021	Cases; N (%) n=8,455	Age-adjusted HR^a (95% CI)	Multivariable-adjusted^a,b HR (95% CI)	Cumulative risk of breast cancer by age 55^b,c	Multivariable-adjusted risk difference (%), age 55^b,c (95% CI)
Hormone therapy
Never Use	381,407 (84)	7,513 (89)	1.00	1.00	4.1%	0.00
Ever Use	69,514 (16)	942 (11)	0.87 (0.82, 0.94)	0.96 (0.88, 1.04)	4.0%	−0.2 (−0.5, 0.2)
Hormone therapy type
Never Use	381,407 (84)	7,513 (89)	1.00	1.00	4.1%	0.00
Unopposed estrogen (E-HT)	27,297 (6)	326 (4)	0.75 (0.67, 0.83)	0.86 (0.75, 0.98)	3.6%	−0.5 (−1.0, −0.0)
Estrogen plus progestin (EP-HT)	25,362 (6)	366 (4)	1.05 (0.95, 1.17)	1.10 (0.98, 1.24)	4.5%	0.4 (−0.1, 0.9)
Both E-HT and EP-HT	9,646 (2)	101 (1)	0.86 (0.70, 1.06)	0.93 (0.75, 1.15)	3.9%	−0.2 (−1.0, 0.6)
Other (O-HT)	7,309 (2)	149 (2)	0.86 (0.72, 1.02)	0.87 (0.74, 1.03)	3.6%	−0.5 (−1.1, 0.1)
Duration of hormone therapy use
Never Use	381,407 (84)	7,513 (89)	1.00	1.00	4.1%	0.00
≤2 year	31,542 (7)	481 (6)	0.92 (0.83, 1.01)	0.98 (0.88, 1.08)	4.0%	−0.1 (−0.5, 0.3)
>2 year	38,072 (9)	461 (5)	0.83 (0.76, 0.92)	0.94 (0.83, 1.05)	3.9%	−0.2 (−0.7, 0.2)
Start of hormone therapy use
Never use	381,407 (84)	7513 (89)	1.00	1.00	4.1%	0.00
<45 years	35,961 (8)	535 (6)	0.80 (0.73, 0.87)	0.87 (0.78, 0.97)	3.6%	−0.5 (−0.9, −0.2)
45–49 years	19,911 (5)	262 (3)	0.99 (0.87, 1.13)	1.08 (0.94, 1.23)	4.5%	0.4 (−0.2, 0.9)
50+ years	13,742 (3)	145 (2)	1.05 (0.87, 1.25)	1.12 (0.92, 1.35)	4.6%	0.5 (−0.3, 1.3)
Current Hormone Therapy Use
Never Use	381,407 (84)	7,513 (89)	1.00	1.00	4.1%	0.00
Former Use	27,031 (6)	340 (4)	0.82 (0.74, 0.92)	0.88 (0.78, 0.98)	3.6%	−0.5 (−0.9, −0.1)
Current Use	42,583 (10)	601 (7)	0.91 (0.83, 0.99)	1.04 (0.94, 1.15)	4.3%	0.2 (−0.2, 0.6)
Hormone therapy, type and duration ^c,d
Never	381,407 (86)	7,513 (90)	1.00	1.00	4.1%	0.00
E-HT, ≤2 years	10,228 (2)	152 (2)	0.84 (0.72, 0.99)	0.94 (0.79, 1.12)	3.9%	−0.2 (−0.9, 0.5)
E-HT, >2 years	17,069 (4)	174 (2)	0.68 (0.59, 0.80)	0.80 (0.66, 0.96)	3.3%	−0.8 (−1.4, −0.2)
EP-HT, ≤2 years	12,455 (3)	180 (2)	1.01 (0.86, 1.17)	1.04 (0.89, 1.22)	4.2%	0.1 (−0.6, 0.7)
EP-HT, >2 years	12,907 (3)	186 (2)	1.12 (0.96, 1.30)	1.18 (1.01, 1.38)	4.8%	0.7 (−0.0, 1.4)
Both E-HT and EP-HT, ≤2 years	4,462 (1)	50 (1)	0.88 (0.65, 1.20)	0.92 (0.68, 1.26)	4.0%	−0.1 (−1.2, 1.0)
Both E-HT and EP-HT, >2 years	5,184 (1)	51 (1)	0.85 (0.63, 1.14)	0.93 (0.68, 1.27)	3.8%	−0.3 (−1.4, 0.7)
Hormone therapy, type and age at start ^c,e
Never	381,407 (86)	7,513 (90)	1.00	1.00	4.1%	0.00
E-HT, <45 years	14,966 (3)	192 (2)	0.68 (0.59, 0.78)	0.77 (0.65, 0.92)	3.2%	−1.0 (−1.5, −0.4)
E-HT, ≥45 years	12,331 (3)	135 (2)	0.88 (0.74, 1.06)	0.99 (0.82, 1.19)	4.2%	0.0 (−0.7, 0.8)
EP-HT, <45 years	8,736 (2)	164 (2)	1.02 (0.87, 1.21)	1.07 (0.90, 1.28)	4.4%	0.2 (−0.4, 0.9)
EP-HT, ≥45 years	16,626 (4)	202 (2)	1.09 (0.94, 1.26)	1.14 (0.97, 1.33)	4.6%	0.5 (−0.2, 1.2)
Both E-HT and EP-HT, <45 years	6,856 (1)	66 (1)	0.78 (0.59, 1.02)	0.84 (0.63, 1.11)	3.3%	−0.9 (−1.7, −0.0)
Both E-HT and EP-HT, ≥45 years	2,790 (1)	35 (0)	1.09 (0.77, 1.55)	1.14 (0.80, 1.62)	5.2%	1.1 (−0.4, 2.7)
Current hormone therapy by type ^c
Never	381,407 (86)	7,513 (90)	1.00	1.00	4.1%	0.00
E-HT, Former	7,892 (2)	80 (1)	0.69 (0.55, 0.87)	0.77 (0.62, 0.97)	3.1%	−1.0 (−1.7, −0.2)
E-HT, Current	19,405 (4)	246 (3)	0.77 (0.68, 0.87)	0.91 (0.77, 1.07)	3.8%	−0.3 (−0.9, 0.3)
EP-HT, Former	8,566 (2)	98 (1)	0.88 (0.72, 1.09)	0.92 (0.75, 1.14)	3.6%	−0.5 (−1.2, 0.3)
EP-HT, Current	16,796 (4)	268 (3)	1.14 (1.01, 1.30)	1.21 (1.06, 1.38)	5.0%	0.8 (0.2, 1.5)
Both E-HT and EP-HT, Former	5,254 (1)	54 (1)	0.96 (0.72, 1.28)	1.01 (0.75, 1.36)	4.3%	0.2 (−0.9, 1.3)
Both E-HT and EP-HT, Current	4,392 (1)	47 (1)	0.78 (0.58, 1.04)	0.86 (0.64, 1.16)	3.5%	−0.6 (−1.6, 0.5)

Open in a new tab

Includes Black Women’s Health Study; Generations Study; California Teachers Study; Cancer Prevention Study 3; Canadian Study of Diet, Lifestyle and Health (included as weighted case-cohort), Campaign Against Cancer and Heart Disease, Mayo Mammography Health Study, Singapore Chinese Health Study, Sister Study, and US Radiologic Technologists Cohort; samples sizes are based on rounded counts after imputation, frequency distributions are weighted by inverse probability of selection weights

All models use age as the time scale, starting at enrollment and stratified by cohort; participants are weighted by their inverse probability of selection.

Additionally adjusted for BMI (continuous), attained education (high school equivalent or less, some college, college graduate), age at menarche (continuous), birth cohort (1928–45, 1946–54, 1955–64, ≥1965), race (White, Black, Asian, Mixed/other), menopausal status (pre/postmenopausal), bilateral oophorectomy (yes/no), hysterectomy (yes/no), oral contraceptive use (never, former, current), parity (0, 1, 2, 3+ births), ever parous by age at first birth (<20, 20–24, 25–29, 30–34, 35+) interaction, menopausal status by age at menopause (continuous) interaction, and BMI by menopausal status interaction. We used multiple imputation with chained equations to impute missing data for hormone therapy type and other covariates. The imputation model included the above potential confounders in addition to case status, breast cancer subtype (estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2 positivity), geographic location (North America, Europe, Asia), regular alcohol use (ever/never, as defined by study), first-degree family history of breast cancer (yes/no), and a crude cumulative hazard estimate.

Women taking other types of hormone therapy excluded from this analysis

HRs for ever use by continuous duration interaction terms (per year of use): HT-E HR=0.98 (0.96–1.00), HT-EP HR=1.01 (0.99–1.03), both E-HT and EP-HT HR=1.00 (0.96–1.04)

HRs for ever use by continuous age at initiation interaction terms (per initiation age in years): HT-E HR=1.01 (0.99–1.03), HT-EP HR=1.00 (0.98–1.02), both E-HT and EP-HT HR=1.00 (0.98–1.03)

When we considered hormone therapy type, relative to never use of any type, E-HT was inversely associated with YOBC (pooled, covariate-adjusted HR=0.86, 0.75–0.98). This corresponded to a RD of −0.5% (−1.0, −0.0), relative to an estimated cumulative risk of 4.1% for non-users. For EP-HT the HR was 1.10 (0.98–1.24) and the RD was 0.4% (−0.1, 0.9). HRs for O-HT (HR=0.88, 0.74–1.04) or both E-HT and EP-HT (HR=0.88, 0.71–1.10) were below 1.0 but statistically imprecise. Relative to never E-HT users, inverse associations were most evident for longer-term users (HR=0.80, 0.66–0.96 for >2 years), earlier initiators (HR=0.77, 0.65–0.92 for <age 45) and former users (HR=0.77, 0.62–0.97). Longer duration of use also amplified the positive EP-HT and YOBC association (HR=1.18, 1.01–1.38 for >2 years), as did current use (HR=1.21, 1.06–1.38). In models with interaction terms for continuous duration or age at use, only the E-HT-by-duration term indicated a consistent trend (HR=0.98, 0.96–1.00 per year of use). Post-imputation, cohort-specific HRs for hormone therapy by type are included in Supplementary Table 4.

For E-HT, associations were similar for all categories of gynecological surgery (Table 3), but for EP-HT we noted stronger positive associations among women without surgery (HR=1.15, 1.02–1.31; RD=0.6%; −0.0, 1.2). These patterns remained consistent in analyses jointly considering surgical status and hormone therapy relative to a common referent group of no surgery, no hormone therapy (Supplementary Table 5). HR estimates for E-HT or EP-HT and YOBC associations were largely similar across categories of menopausal status, age, BMI, race, geographic region, and birth year (Table 3). We observed possible differences by first-degree family history of breast cancer, with E-HT showing a stronger inverse association among women with ≥1 affected relative (HR=0.77, 0.60–1.00) and EP-HT showing a stronger positive association among women without an affected relative (HR=1.17, 1.02–1.34).

Table 3.

Association between hormone therapy use at enrollment, by type^a, and incident young-onset breast cancer (age<55) among subgroups (n=459,476)

	Sample size: cases / cohort	Unopposed estrogen (E-HT)		Estrogen plus progestin (EP-HT)
	Sample size: cases / cohort	% users (non-cases / cases)	Hazard Ratio^c (95% CI)	% users (non-cases / cases)	Hazard Ratio^c (95% CI)
Overall	8,455 / 459,476	6% / 4%	0.86 (0.75, 0.98)	6% / 4%	1.10 (0.98, 1.24)
Premenopausal person-time	4,585 / 357,362	2% / 1%	0.71 (0.49, 1.03)	2% / 1%	1.03 (0.76, 1.40)
Premenopausal, BMI ≤25 kg/m²	2,936 / 199,308	1% / 1%	0.79 (0.48, 1.30)	2% / 1%	1.08 (0.73, 1.58)
Premenopausal, BMI >25 kg/m²	1,650 / 158,054	2% / 1%	0.62 (0.35, 1.10)	2% / 1%	0.94 (0.56, 1.58)
Postmenopausal person-time	3,870 / 265,623	10% / 8%	0.82 (0.70, 0.95)	9% / 8%	0.97 (0.85, 1.10)
Age ^b
Age <50	4,801 / 364,723	4% / 2%	0.81 (0.64, 1.02)	3% / 2%	0.89 (0.70, 1.13)
Age <45	2,111 / 266,349	2% / 1%	0.85 (0.52, 1.40)	1% / 1%	1.14 (0.74, 1.76)
Gynecological Surgery Status
No Surgery	7,655 / 399,822	2% / 1%	0.81 (0.64, 1.01)	6% / 4%	1.15 (1.02, 1.31)
Hysterectomy without bilateral oophorectomy	494 / 34,127	17% / 12%	0.78 (0.57, 1.04)	4% / 3%	0.85 (0.50, 1.46)
Bilateral oophorectomy with or without hysterectomy	306 / 25,528	61% / 61%	0.90 (0.65, 1.25)	12% / 13%	1.09 (0.70, 1.68)
Race
White	6,567 / 360,696	6% / 4%	0.88 (0.75, 1.03)	7% / 5%	1.12 (0.98, 1.26)
Black	1,498 / 65,866	7% / 4%	0.76 (0.55, 1.05)	3% / 2%	0.89 (0.61, 1.29)
Asian	236 / 21,518	5% / 3%	0.62 (0.14, 2.76)	3% / 3%	1.00 (0.42, 2.35)
Mixed or another race	153 / 11,395	6% / 3%	1.15 (0.38, 3.47)	5% / 6%	1.81 (0.89, 3.68)
Geographic Region
North America	7,280 / 371,127	6% / 4%	0.86 (0.74, 0.99)	6% / 5%	1.10 (0.97, 1.24)
Europe	1,041 / 72,296	4% / 2%	0.82 (0.52, 1.30)	4% / 3%	0.92 (0.63, 1.34)
Asia	134 / 16,053	5% / 4%	0.44 (0.05, 4.20)	2% / 3%	1.47 (0.40, 5.40)
Birth cohort
Born 1928–45	672 / 46,509	15% / 10%	1.06 (0.74, 1.54)	20% / 16%	1.28 (0.99, 1.64)
Born 1946–54	2,590 / 103,910	8% / 5%	0.80 (0.64, 1.01)	8% / 6%	1.19 (0.99, 1.43)
Born 1955–64	3,699 / 154,214	6% / 3%	0.83 (0.67, 1.04)	4% / 2%	0.82 (0.65, 1.03)
Born ≥1965	1,494 / 154,843	2% / 1%	0.64 (0.32, 1.28)	1% / 1%	1.03 (0.63, 1.71)
Family History of Breast Cancer
No first-degree family history	6,465 / 389,214	6% / 4%	0.91 (0.77, 1.07)	6% / 4%	1.17 (1.02, 1.34)
≥1 first-degree female relative with breast cancer	1,990 / 70,262	8% / 5%	0.77 (0.60, 1.00)	7% / 4%	0.94 (0.73, 1.19)

Open in a new tab

Includes Black Women’s Health Study; Generations Study; California Teachers Study; Cancer Prevention Study 3; Canadian Study of Diet, Lifestyle and Health (as weighted case-cohort), Campaign Against Cancer and Heart Disease, Mayo Mammography Health Study, Singapore Chinese Health Study, Sister Study, and US Radiologic Technologists Cohort; samples sizes are based on rounded counts after imputation, frequency distributions are weighted by inverse probability of selection

Effect estimates for users of both E-HT and EP-HT and for O-HT are not presented, though these were retained as exposure categories in the model

Includes person-time and events occurring up to the specified age, starting at age at enrollment (i.e., censoring at upper age cut-off).

All models use age as the time scale, starting at enrollment and stratified by cohort; participants are weighted by their inverse probability of selection. Adjusted models include BMI (continuous), attained education (high school equivalent or less, some college, college graduate), age at menarche (continuous), birth cohort (1928–45, 1946–54, 1955–64, ≥1965), race (White, Black, Asian, Mixed/other), menopausal status (pre/postmenopausal), bilateral oophorectomy (yes/no), hysterectomy (yes/no), oral contraceptive use (never, former, current), parity (0, 1, 2, 3+ births), ever parous by age at first birth (<20, 20–24, 25–29, 30–34, 35+) interaction, menopausal status by age at menopause (continuous) interaction, and BMI by menopausal status interaction. We used multiple imputation with chained equations to impute missing data for hormone therapy type and other covariates. The imputation model included the above potential confounders in addition to case status, breast cancer subtype (estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2 positivity), geographic location (North America, Europe, Asia), regular alcohol use (ever/never, as defined by study), first-degree family history of breast cancer (yes/no), and a crude cumulative hazard estimate.

E-HT was inversely associated with most subtypes of breast cancer (Table 4). However, for EP-HT, positive associations were observed for ER-negative (HR=1.44, 1.11–1.88) and triple-negative (HR=1.50, 1.02–2.20) YOBC.

Table 4.

Association of hormone therapy use, by type, and incident young-onset breast cancer (age<55) by breast cancer subtype (n=459,476)

	Cases	Unopposed Estrogen (E-HT)		Estrogen plus Progestin (EP-HT)
	Cases	% users cases	Hazard Ratio^a (95% CI)	% users cases	Hazard Ratio^a (95% CI)
Overall	8,455	4%	0.86 (0.75, 0.98)	4%	1.10 (0.98, 1.24)
Invasiveness
In situ	1,875 (22%)	3%	0.73 (0.53, 1.00)	4%	1.05 (0.82, 1.35)
Invasive	6,580 (78%)	4%	0.89 (0.76, 1.04)	4%	1.12 (0.98, 1.27)
Estrogen Receptor (ER) Status
ER-	1,922 (23%)	5%	0.91 (0.67, 1.23)	5%	1.44 (1.11, 1.88)
ER+	6,580 (78%)	4%	0.84 (0.72, 0.99)	4%	1.02 (0.89, 1.18)
Immunohistochemical Subtypes
ER or PR+, HER2-	5,104 (60%)	4%	0.86 (0.71, 1.04)	4%	1.04 (0.88, 1.23)
ER or PR+, HER2+	1,531 (18%)	4%	0.83 (0.57, 1.23)	4%	1.07 (0.70, 1.62)
HER2+/ER-	652 (8%)	4%	1.16 (0.49, 2.76)	3%	0.99 (0.64, 1.54)
ER-, PR-, HER2- (Triple-negative)	1,169 (14%)	4%	0.76 (0.48, 1.22)	5%	1.50 (1.02, 2.20)

Open in a new tab

All models use age as the time scale, starting at enrollment and stratified by cohort; participants are weighted by their inverse probability of selection. Adjusted models include BMI (continuous), attained education (high school equivalent or less, some college, college graduate), age at menarche (continuous), birth cohort (1928–45, 1946–54, 1955–64, ≥1965), race (White, Black, Asian, Mixed/other), menopausal status (pre/postmenopausal), bilateral oophorectomy (yes/no), hysterectomy (yes/no), oral contraceptive use (never, former, current), parity (0, 1, 2, 3+ births), ever parous by age at first birth (<20, 20–24, 25–29, 30–34, 35+) interaction, menopausal status by age at menopause (continuous) interaction, and BMI by menopause status interaction. We used multiple imputation with chained equations to impute missing data for hormone therapy type and other covariates. The imputation model included the above potential confounders in addition to case status, breast cancer subtype (estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2 positivity), geographic location (North America, Europe, Asia), attained education level (high school or less, some college, completed college/university), regular alcohol use (ever/never, as defined by study), first-degree family history of breast cancer (yes/no), and a crude cumulative hazard estimate.

In analyses where covariate information was updated throughout follow-up, estimates for ever E-HT were consistent with those assuming no covariate changes (HR=0.86, 0.73–1.01; Table 5, compared to HR=0.86, 0.75–0.98 initially), as were estimates for EP-HT (HR=1.06, 0.93–1.21 versus HR=1.10, 0.98–1.24). The other HRs from time-updated analyses were also generally consistent with the original estimates, including a strong inverse association for E-HT use <age 45 (HR=0.79, 0.65–0.97) and YOBC.

Table 5.

Time-varying^a associations of hormone therapy use and incident young-onset breast cancer (age<55)

	Pooled sample n=122,093 (3,673 cases)	CTS n=68,946 (1,642 cases)	CLUEII n=5,433 (125 cases)	SIS n=23,970 (816 cases)	USRTC n=23,744 (1,090 cases)
Hormone therapy status at end of follow-up Ever use / Current use / ≥5 years of use
E-HT	12% / 10% / 7%	11% / 11% / 7%	13% / 10% / 8%	13% / 4% / 5%	11% / 11% / 8%
EP-HT	13% / 5% / 6%	15% / 6% / 8%	9% / 2% / 4%	9% / 1% / 2%	12% / 6% / 6%
E-HT and EP-HT	5% / 1% / 3%	7% / 1% / 4%	3% / <1% / 2%	3% / <1% / 1%	2% / <1% / 1%
O-HT^a	4%	4%	2%	8%	<1%
HR (95% CI) for ever hormone therapy by type
Never use	1.00	1.00	1.00	1.00	1.00
E-HT	0.86 (0.73, 1.01)	0.83 (0.66, 1.04)	1.09 (0.46, 2.53)	0.93 (0.68, 1.27)	0.92 (0.62, 1.36)
EP-HT	1.06 (0.93, 1.21)	1.03 (0.86, 1.24)	1.32 (0.65, 2.71)	0.82 (0.59, 1.16)	1.31 (1.00, 1.72)
E-HT and EP-HT	0.79 (0.61, 1.02)	0.73 (0.54, 0.99)	0.75 (0.18, 3.22)	0.84 (0.45, 1.56)	1.11 (0.49, 2.55)
O-HT	0.91 (0.76, 1.10)	0.90 (0.72, 1.14)	0.70 (0.16, 3.13)	0.91 (0.64, 1.28)	0.88 (0.33, 2.34)
HR (95% CI)^b for current hormone therapy use by type
Never use	1.00	1.00	1.00	1.00	1.00
E-HT, former	0.99 (0.75, 1.30)	0.84 (0.50, 1.42)	2.00 (0.57, 7.03)	1.01 (0.65, 1.56)	1.16 (0.67, 2.01)
E-HT, current	1.06 (0.81, 1.38)	1.11 (0.68, 1.83)	1.52 (0.44, 5.27)	0.88 (0.53, 1.46)	1.12 (0.67, 1.87)
EP-HT, former	0.97 (0.67, 1.42)	0.96 (0.55, 1.68)	2.06 (0.41, 10.2)	1.67 (0.74, 3.76)	1.03 (0.31, 3.73)
EP-HT, current	1.11 (0.77, 1.60)	1.08 (0.61, 1.92)	0.56 (0.04, 7.22)	1.97 (0.72, 5.38)	1.37 (0.51, 3.72)
E-HT and EP-HT, former	1.21 (0.49, 2.97)	2.01 (0.55, 7.30)	NE	0.67 (0.06, 7.57)	NE
E-HT and EP-HT, current	1.12 (0.53, 2.37)	0.98 (0.36, 2.64)	NE	1.99 (0.32, 12.4)	NE
HR (95% CI) for duration of hormone therapy use by type
Never use	1.00	1.00	1.00	1.00	1.00
E-HT, <5 years	0.89 (0.73, 1.08)	0.83 (0.63, 1.10)	1.00 (0.35, 2.91)	0.95 (0.68, 1.33)	0.82 (0.47, 1.43)
E-HT, ≥5 years	0.84 (0.66, 1.06)	0.80 (0.58, 1.11)	1.31 (0.43, 3.96)	0.95 (0.58, 1.56)	0.99 (0.59, 1.65)
EP-HT, <5 years	1.02 (0.87, 1.21)	1.00 (0.80, 1.25)	1.63 (0.73, 3.66)	0.75 (0.50, 1.12)	1.29 (0.91, 1.83)
EP-HT, ≥5 years	1.14 (0.93, 1.40)	1.09 (0.85, 1.42)	0.83 (0.22, 3.26)	1.20 (0.64, 2.23)	1.32 (0.87, 2.01)
E-HT and EP-HT, <5 years	0.86 (0.62, 1.19)	0.92 (0.64, 1.32)	NE	0.67 (0.25, 1.79)	NE
E-HT and EP-HT, ≥5 years	0.74 (0.51, 1.08)	0.56 (0.35, 0.91)	1.49 (0.28, 7.93)	1.06 (0.46, 2.44)	NE
HR (95% CI) for age at start of hormone therapy use by type
Never use	1.00	1.00	1.00	1.00	1.00
E-HT, <45 years	0.79 (0.65, 0.97)	0.73 (0.54, 0.98)	1.12 (0.41, 3.01)	0.84 (0.57, 1.23)	1.04 (0.66, 1.62)
E-HT, ≥45 years	0.99 (0.79, 1.24)	0.98 (0.73, 1.33)	1.06 (0.31, 3.57)	1.08 (0.71, 1.63)	0.74 (0.37, 1.45)
EP-HT, <45 years	1.09 (0.90, 1.32)	1.06 (0.81, 1.39)	1.06 (0.38, 2.99)	0.89 (0.57, 1.37)	1.45 (0.99, 2.12)
EP-HT, ≥45 years	1.06 (0.74, 1.26)	1.05 (0.84, 1.31)	1.73 (0.67, 4.43)	0.75 (0.44, 1.28)	1.19 (0.83, 1.71)
E-HT and EP-HT, <45 years	0.74 (0.55, 1.00)	0.67 (0.46, 0.96)	0.45 (0.06, 3.32)	0.78 (0.37, 1.65)	1.18 (0.51, 2.75)
E-HT and EP-HT, ≥45 years	1.01 (0.65, 1.58)	0.98 (0.59, 1.62)	NE	0.98 (0.32, 3.06)	NE

Open in a new tab

NE = Not evaluated

Hormone therapy use status and other time-varying covariates (BMI, menopausal status, bilateral oophorectomy status, hysterectomy status, parity, and age at first birth) were updated, as needed, at the time of each follow-up questionnaire; all follow-up periods were broken down into 2 year time periods and analyzed using period-stratified Cox proportional hazards model. If follow-up information for covariates was missing, values were carried forward from the prior period. Age at menarche, birth cohort, attained education, race, and oral contraceptive use status were considered time-fixed. All models use age as the time scale, starting at enrollment and stratified by cohort; participants are weighted by their inverse probability of selection. Adjusted models include BMI (continuous), attained education (high school equivalent or less, some college, college graduate), age at menarche (continuous), birth cohort (1928–45, 1946–54, 1955–64, ≥1965), race (White, Black, Asian, Mixed/other), menopausal status (pre/postmenopausal), bilateral oophorectomy (yes/no), hysterectomy (yes/no), oral contraceptive use (never, former, current), parity (0, 1, 2, 3+ births), ever parous by age at first birth (<20, 20–24, 25–29, 30–34, 35+) interaction, menopausal status by age at menopause (continuous) interaction, and BMI by menopause status interaction. We used multiple imputation with chained equations to impute missing data for hormone therapy type and other covariates. The imputation model included the above potential confounders in addition to case status, breast cancer subtype (estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2 positivity), geographic location (North America, Europe, Asia), attained education level (high school or less, some college, completed college/university), regular alcohol use (ever/never, as defined by study), first-degree family history of breast cancer (yes/no), and a crude cumulative hazard estimate.

Percentages for ever use only. Duration of use and current/former use of other types of hormone therapy use not evaluated (included as non-users).

Current hormone therapy use defined by use during the most recent 2-year follow-up period. All models estimating the effect of current use were additionally adjusted for prior hormone therapy use (i.e., former use by type), as a potential confounder.

Discussion

In this large analysis that pooled data from 10 cohorts, we observed an inverse association between ever use of E-HT and incident breast cancer diagnosed before age 55, which was consistent across breast cancer subtypes and was potentially stronger with earlier or longer-term use. In contrast, EP-HT was positively associated with YOBC, particularly among longer-term users and women with no history of gynecologic surgery. The positive EP-HT and YOBC association may be specific to ER– (including triple-negative) breast cancers.

Our results parallel those previously reported for the WHI clinical trials conducted among postmenopausal women. Specifically, Anderson et al. (2004)³ reported an inverse association between E-HT and invasive breast cancer risk among hysterectomized women aged 50–79 (mean=64) years (HR=0.77, 0.59–1.01), which persisted in a re-analysis that incorporated additional follow-up (HR=0.77, 0.62–0.95).⁴¹ Among women aged 50–59 at randomization, the HR was 0.80 (0.54–1.20). Given that a large proportion of E-HT users in our pooled sample had also undergone hysterectomy (with or without bilateral oophorectomy), the combined evidence supports that there may be health benefits, at least in terms of breast cancer risk, of using exogenous estrogen to replace depleted endogenous estrogen. The inverse associations observed among women who started E-HT at younger ages or who used it for longer durations lends further support to the idea that women who undergo menopause earlier, either naturally or surgically, may derive the most benefit from estrogen replacement.

In the WHI trial of EP-HT versus placebo, EP-HT use was associated with an increase in postmenopausal invasive breast cancer in the initial report (HR=1.26, 1.00–1.59)¹ and after extended follow-up (HR=1.24, 1.01–1.53).⁴² While older than our pooled sample, these participants had a uterus at enrollment. Thus, the trial results are most comparable to our results for women with no prior gynecologic surgery, a group where we too observed an elevated HR. Altogether, this suggests that among women with stable or naturally declining hormone levels, exogenous estrogen may be carcinogenic,⁵ with exogenous progestin also potentially playing an antagonistic role.⁴³ The positive association we observed for EP-HT and ER-negative or triple-negative breast cancer, specifically, is a novel finding that deserves further exploration,⁴⁴ but if replicated could help us better understand the role of hormones in initiating or promoting the growth of even hormone receptor-negative tumors.

In the MWS,² past use of hormone therapy (any type) was not associated with breast cancer incidence (relative risk [RR]= 1.01, 0.95–1.08). In contrast, current use of either E-HT (RR=1.30, 1.22–1.38) or EP-HT (RR=2.00, 1.91–2.09) versus never use of any type, was associated with increased risk. However, critiques of the study have theorized that the positive associations were largely driven by over-selection of current hormone therapy users and women with known breast abnormalities seeking screening, as demonstrated by the short average time to diagnosis (1.2 years).^45,46 Similar patterns of increased screening may have contributed to the elevated HR for current HT-EP users in our sample as well, though the stronger HR for EP-HT and YOBC in women without a breast cancer family history goes against this assumption.

A multi-study collaborative group analysis of postmenopausal women that included the MWS also reported evidence for a positive association between duration of E-HT and EP-HT use and incident breast cancer among both current and past users.⁴ However, a supplementary analysis of the collaborative group data produced findings similar to those observed here for cancer diagnosed before age 55, albeit with lower statistical precision (5–14 years of current E-HT use versus never, relative risk= 0.95, 0.80–1.12; 5–14 years of current EP-HT use versus never, relative risk=1.71, 1.45–2.03).

Some additional discrepancies in findings across studies may be attributable to differences in risk factors for younger versus later-onset breast cancer. This pattern is well-documented for BMI, which is inversely associated with premenopausal breast cancer but positively associated with postmenopausal breast cancer.^15,47 Then again, our results were not entirely consistent with 3 small case-control studies of young-onset breast cancer.^48–50 While Shantakumar et al.⁴⁸ also reported positive associations with risk of premenopausal breast cancer for EP-HT or mixed regimens, they did not observe inverse associations for E-HT and premenopausal breast cancer. In a sister-matched case-control study,⁵⁰ a statistically non-significant inverse association was observed for EP-HT and breast cancer diagnosed before age 50, but, as with the current pooled analysis, the E-HT association was strongly inverse (Odds Ratio [OR]=0.58, 0.34–0.99). Palmer et al.⁴⁹ did not observe inverse associations for E-HT and breast cancer before age 50.

Strengths and Limitations

YOBC and hormone use before age 55 are both uncommon, and pooling individual level data from multiple cohorts allowed us to conduct well-powered, comprehensive analyses with adjustment for potential confounding factors beyond what could have been accomplished with a single study or meta-analysis. However, harmonizing data across cohorts has inherent limitations, including the need to align exposure metrics with the least detailed study(s). For example, although some studies provided hormone therapy dose, it was not consistently collected, and we could not capture type-specific data for women who reported multiple types. We employed multiple imputation to account for missing data, including missing hormone therapy type, but this approach was limited by the covariate information available and data on hysterectomy, oophorectomy, or oral contraceptive use was sparse in some studies. Although only a small proportion of participants were classified as O-HT users, we lacked detail on whether these women could have been classified to specific categories of hormone therapy (e.g., progestin alone or tibolone) that are potentially of interest. We also lacked data on indications for use, which may have allowed us to better disentangle the independent effects of hormone therapy from gynecologic surgery status or other co-exposures, and on progestin/estrogen type or route of administration, which varies by population and may impact hormonal absorption.^40,51 Lastly, we note that most cohorts only provided data on hormone therapy status up until enrollment, meaning that women who started using hormone therapy after baseline but before age 55 were miscategorized as unexposed throughout follow-up. However, results from the secondary analyses conducted among the 4 cohorts with time-updated data were largely consistent with the main findings. Because the majority of cohorts only provided information on hormone use at the time of enrollment, these data were ill-suited for target trial emulation⁵² or other approaches to modeling the effects of drug initiation.

Bias from unmeasured or poorly measured confounders may be present. We did not adjust for alcohol use or physical activity because the included cohorts had heterogeneous definitions and data quality for these factors. Further, while we adjusted for attained education, this has different meanings across countries and cohorts and is likely insufficient to fully control for confounding by socioeconomic position, a likely correlate of hormone therapy use. Lastly, we did not have data on BRCA1/2 or other known breast cancer predisposition genes and could not evaluate whether any of the observed associations were modified by genetic factors.

We initially included cohorts from 4 continents and many countries, but the final pooled sample was predominantly White women from North America. Therefore, these results may not be generalizable to all groups. Additional studies on women from other continents and ancestries are warranted, as are studies of women from more recent birth cohorts or with greater socioeconomic disadvantage. Studies of these groups would also add information on different hormone regimens and use patterns.

The major strengths of this analysis were the large sample size and detailed data enabling an examination of hormone therapy type and subtypes of YOBC, accounting for many known confounders and effect modifiers. The time-varying analyses, though limited to data from 4 cohorts, are also an important strength, as these help to bolster the results of the original analysis and offer some additional insights.

Young women may use hormone therapy to relieve symptoms from hormone-related conditions, menopause, or gynecologic surgery, but the risks and benefits of doing so have not been thoroughly investigated. In this large collaborative study of hormone therapy use and breast cancer incidence among younger women, we observed an inverse association for ever use of E-HT and breast cancer before age 55. EP-HT was modestly positively associated with YOBC rates, and the observed stronger association among women who had not had a hysterectomy or bilateral oophorectomy deserves further consideration given that E-HT is contraindicated in women with a uterus. Altogether, the results of this large-scale investigation, in conjunction with mostly consistent results from prior studies of postmenopausal women, provide key information that may be useful for establishing clinical recommendations regarding hormone therapy use in young women.

Supplementary Material

NIHMS2089962-supplement-Supplementary_Material.pdf^{(472.5KB, pdf)}

Research in Context.

Evidence before the study

The association between exogenous hormone therapy and breast cancer risk has been studied extensively, including in clinical trials. However, the vast majority of these studies have been limited to postmenopausal women, ignoring the potentially substantial number of women who initiate hormone therapy during peri-menopause. Though our literature review was primarily focused on a handful of well-cited large clinical trials, observational studies and pooled analyses conducted among post-menopausal women, we also did a targeted search current through September 2024 for studies in young or premenopausal women using the following PubMed search terms: “breast cancer and (unopposed estrogen or estrogen plus progestin) and (young onset or premenopausal)”, and then noting any additional references cited within relevant publications. This search led us to 3 prior studies- Palmer et al., 1991; Shantakumar et al., 2007 and O’Brien et al., 2015- all of which are discussed in the manuscript.

Added value of this study

As previous investigations of hormone therapy use and breast cancer risk in young women were conducted in small, stand-alone observational studies, this pooled analysis is by far the largest and most comprehensive study of the topic and, to our knowledge, the first to offer substantive evidence to help inform clinical recommendations for hormone therapy use in young women, who may greatly benefit from hormone therapy to alleviate post-surgical or peri-menopausal symptoms.

Implications of all of the available evidence

The results of our pooled analysis suggest that the associations between hormone therapy use and breast cancer risk are largely consistent across groups defined by menopausal status and age. Although the strength of these associations may vary by age at first use, duration of use, gynecological surgery status, and other factors, unopposed estrogen hormone therapy use appears to decrease breast cancer risk and estrogen plus progestin therapy appears to increase breast cancer risk. The findings reported in the pooled analysis can be used to augment clinical recommendations for hormone therapy use in young women, where guidance was previously lacking.

Acknowledgements

The authors would like to acknowledge the contribution to this study from central cancer registries supported through the Centers for Disease Control and Prevention’s National Program of Cancer Registries (NPCR) and/or the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) Program. Central registries may also be supported by state agencies, universities, and cancer centers. Participating central cancer registries include the following: AL, AR, AZ, CA, CO, CT, DE, DC, FL, GA, HI, IA, IL, IN, KY, LA, MD, MA, MI, MO, MS, NE, NJ, NM, NY, NC, OH, OK, OR, PA, SC, TN, TX, VA, WA, WI. The content is solely the responsibility of the authors and does not necessarily represent the official views of the U.S. Department of Health and Human Services, the National Institutes of Health, the National Cancer Institute, or the state cancer registries. We thank participants and staff of the BWHS for their contributions. The authors express sincere appreciation to all Cancer Prevention Study-3 participants, and to each member of the study and biospecimen management group. The authors would like to acknowledge the contribution to this study from central cancer registries supported through the Centers for Disease Control and Prevention’s National Program of Cancer Registries and cancer registries supported by the National Cancer Institute’s Surveillance Epidemiology and End Results Program

Funding:

The Black Women’s Health Study was funded by the National Institutes of Health (U01-CA164974). The California Teachers Study and the research reported in this publication were supported by the National Cancer Institute of the National Institutes of Health under award number U01-CA199277; P30-CA033572; P30-CA023100; UM1-CA164917; and R01-CA077398. The collection of cancer incidence data used in the California Teachers Study was supported by the California Department of Public Health pursuant to California Health and Safety Code Section 103885; Centers for Disease Control and Prevention’s National Program of Cancer Registries, under cooperative agreement 5NU58DP006344; the 3 National Cancer Institute’s Surveillance, Epidemiology and End Results Program under contract HHSN261201800032I awarded to the University of California, San Francisco, contract HHSN261201800015I awarded to the University of Southern California, and contract HHSN261201800009I awarded to the Public Health Institute. The American Cancer Society funds the creation, maintenance, and updating of the Cancer Prevention Study-II cohort (and/or Cancer Prevention Study-3). The Generations Study was funded by Breast Cancer Now, the UK National Health Service, and the Institute of Cancer Research. The Mayo Mammography Health Study was supported in part by the National Institutes of Health grant R35CA253187. The Singapore Chinese Health Study was supported by the National Institutes of Health (grants # R01CA080205, R01CA144034, and UM1CA182876). The Sister Study was funded by the National Institute of Environmental Health Sciences intramural research program (Z01-ES044005). Dr. Rohan is supported by the Breast Cancer Research Foundation (BCRF-24-140).

The Avon Foundation, the Institute of Cancer Research, and the Intramural Program of National Institute of Environmental Health Sciences provided funding for the creation of the Premenopausal Breast Cancer Collaborative Group.

Funding source:

NIH Intramural Research Program

Footnotes

Conflict of interest: The authors have no conflicts of interest to declare.

Data sharing:

The raw data for this manuscript cannot be shared publicly due to privacy reasons, data sharing policies in the countries of origin, and the terms of the data sharing agreement when the Premenopausal Breast Cancer Collaborative Group was formed. Data can be requested from individual cohorts following their specific data sharing protocols.

References

1.Rossouw JE, Anderson GL, Prentice RL, LaCroix AZ, Kooperberg C, Stefanick ML, et al. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results From the Women’s Health Initiative randomized controlled trial. JAMA. 2002. Jul;288(3):321–33. [DOI] [PubMed] [Google Scholar]
2.Beral V, Million Women Study Collaborators. Breast cancer and hormone-replacement therapy in the Million Women Study. Lancet. 2003. Aug;362(9382):419–27. [DOI] [PubMed] [Google Scholar]
3.Anderson GL, Limacher M, Assaf AR, Bassford T, Beresford SAA, Black H, et al. Effects of conjugated equine estrogen in postmenopausal women with hysterectomy: the Women’s Health Initiative randomized controlled trial. JAMA. 2004. Apr;291(14):1701–12. [DOI] [PubMed] [Google Scholar]
4.Type and timing of menopausal hormone therapy and breast cancer risk: individual participant meta-analysis of the worldwide epidemiological evidence. The Lancet. 2019. Sep;394(10204):1159–68. [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Bernstein L Epidemiology of Endocrine-Related Risk Factors for Breast Cancer. Jounral of Mammary Gland Biology and Neoplasia. 7(1):3–15. [DOI] [PubMed] [Google Scholar]
6.Stuenkel CA, Davis SR, Gompel A, Lumsden MA, Murad MH, Pinkerton JV, et al. Treatment of Symptoms of the Menopause: An Endocrine Society Clinical Practice Guideline. The Journal of Clinical Endocrinology & Metabolism. 2015. Nov;100(11):3975–4011. [DOI] [PubMed] [Google Scholar]
7.the American College of Obstetricians and Gynecologists. Practice Bulletin: Clinical Management Guidelines for Obstetrician-Gynecologists. Obstet Gynecol. 2014;123(1):202–16.24463691 [Google Scholar]
8.The 2017 hormone therapy position statement of The North American Menopause Society. Menopause. 2017. Jul;24(7):728–53. [DOI] [PubMed] [Google Scholar]
9.Warnock JK, Cohen LJ, Blumenthal H, Hammond JE. Hormone-Related Migraine Headaches and Mood Disorders: Treatment with Estrogen Stabilization. Pharmacotherapy. 2017. Jan;37(1):120–8. [DOI] [PubMed] [Google Scholar]
10.Shifren JL, Gass MLS. The North American Menopause Society Recommendations for Clinical Care of Midlife Women. Menopause. 2014. Oct;21(10):1038–62. [DOI] [PubMed] [Google Scholar]
11.Tepper PG, Brooks MM, Randolph JF, Crawford SL, El Khoudary SR, Gold EB, et al. Characterizing the trajectories of vasomotor symptoms across the menopausal transition. Menopause. 2016. Oct;23(10):1067–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Sprague BL, Trentham-Dietz A, Cronin KA. A sustained decline in postmenopausal hormone use. Obstet Gynecol. 2012. Sep;120(3):595–603. [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Nichols HB, Schoemaker MJ, Wright LB, McGowan C, Brook MN, McClain KM, et al. The Premenopausal Breast Cancer Collaboration: A pooling project of studies participating in the National Cancer Institute Cohort Consortium. Cancer epidemiology, biomarkers & prevention. 2017;26(September). [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Nichols HB, Schoemaker MJ, Cai J, Xu J, Wright LB, Brook MN, et al. Breast Cancer Risk After Recent Childbirth. Annals of Internal Medicine [Internet]. 2018; Available from: http://annals.org/article.aspx?doi=10.7326/M18-1323 [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Schoemaker MJ, Nichols HB, Wright LB, Brook MN, Jones ME, O’Brien KM, et al. Association of Body Mass Index and Age With Subsequent Breast Cancer Risk in Premenopausal Women. JAMA Oncology. 2018;1–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Rosenberg L, Palmer JR, Wise LA, Adams-Campbell LL. A Prospective Study of Female Hormone Use and Breast Cancer Among Black Women. Arch Intern Med. 2006;166. [DOI] [PubMed] [Google Scholar]
17.Jones ME, Schoemaker MJ, Wright L, McFadden E, Griffin J, Thomas D, et al. Menopausal hormone therapy and breast cancer: what is the true size of the increased risk? Br J Cancer. 2016. Aug;115(5):607–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Saxena T, Lee E, Henderson KD, Clarke CA, West D, Marshall SF, et al. Menopausal hormone therapy and subsequent risk of specific invasive breast cancer subtypes in the California Teachers Study. Cancer Epidemiol Biomarkers Prev. 2010. Sep;19(9):2366–78. [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Patel AV, Jacobs EJ, Dudas DM, Briggs PJ, Lichtman CJ, Bain EB, et al. The American Cancer Society’s Cancer Prevention Study 3 (CPS-3): Recruitment, study design, and baseline characteristics. Cancer. 2017. Jun;123(11):2014–24. [DOI] [PubMed] [Google Scholar]
20.Rohan TE, Soskolne CL, Carroll KK, Kreiger N. The Canadian Study of Diet, Lifestyle, and Health: Design and characteristics of a new cohort study of cancer risk. Cancer Detection and Prevention. 2007;31(1):12–7. [DOI] [PubMed] [Google Scholar]
21.Gross AL, Newschaffer CJ, Hoffman-Bolton J, Rifai N, Visvanathan K. Adipocytokines, Inflammation, and Breast Cancer Risk in Postmenopausal Women: A Prospective Study. Cancer Epidemiology, Biomarkers & Prevention. 2013. Jul 1;22(7):1319–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Olson JE, Sellers TA, Scott CG, Schueler BA, Brandt KR, Serie DJ, et al. The influence of mammogram acquisition on the mammographic density and breast cancer association in the mayo mammography health study cohort. Breast Cancer Res. 2012. Dec;14(6):R147. [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Wu AH, Koh WP, Wang R, Lee HP, Yu MC. Soy intake and breast cancer risk in Singapore Chinese Health Study. Br J Cancer. 2008. Jul;99(1):196–200. [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Sandler DP, Hodgson ME, Deming-Halverson SL, Juras PJ, D’Aloisio AD, Suarez L, et al. The Sister Study: Baseline methods and participant characteristics. Environ Health Perspect. 2017;125(12):127003. [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Preston DL, Kitahara CM, Freedman DM, Sigurdson AJ, Simon SL, Little MP, et al. Breast cancer risk and protracted low-to-moderate dose occupational radiation exposure in the US Radiologic Technologists Cohort, 1983–2008. Br J Cancer. 2016. Oct;115(9):1105–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Gertig DM, Fletcher AS, English DR, MacInnis RJ, Hopper JL, Giles GG. Hormone therapy and breast cancer: what factors modify the association?: Menopause. 2006. Mar;13(2):178–84. [DOI] [PubMed] [Google Scholar]
27.Zheng W, Chow WH, Yang G, Jin F, Rothman N, Blair A, et al. The Shanghai Women’s Health Study: Rationale, Study Design, and Baseline Characteristics. American Journal of Epidemiology. 2005. Dec 1;162(11):1123–31. [DOI] [PubMed] [Google Scholar]
28.Signorello LB, Hargreaves MK, Steinwandel MD, Zheng W, Cai Q, Schlundt DG, et al. Southern Community Cohort Study: Establishing a Cohort to Investigate Health Disparities. Journal of the National Medical Association. 2005;97(7). [PMC free article] [PubMed] [Google Scholar]
29.O’Brien KM, Lawrence KG, Keil AP. The case for case-cohort: An applied epidemiologist’s guide to re-framing case-cohort studies to improve usability and flexibility. Epidemiology. 33(3):354–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Schonfeld SJ, Pfeiffer RM, Lacey JV, Berrington De Gonzalez A, Doody MM, Greenlee RT, et al. Hormone-related Risk Factors and Postmenopausal Breast Cancer Among Nulliparous Versus Parous Women: An Aggregated Study. American Journal of Epidemiology. 2011. Mar 1;173(5):509–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
31.White IR, Royston P. Imputing missing covariate values for the Cox model: Imputing missing covariate values for the Cox model. Statist Med. 2009. Jul 10;28(15):1982–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
32.DerSimonian R, Laird N. Meta-analysis in clinical trials. Controlled Clinical Trials. 1986;7(3):177–88. [DOI] [PubMed] [Google Scholar]
33.Von Holle A, Adami HO, Baglietto L, Berrington De Gonzalez A, Bertrand KA, Blot W, et al. BMI and breast cancer risk around age at menopause. Cancer Epidemiology. 2024. Apr;89:102545. [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Rubin Donald. Multiple Imputation for Nonresponse in Surveys. John Wiley & Sons, Inc; 1987. [Google Scholar]
35.Breslow NE. Discussion of Professor Cox’s paper. Journal of the Royal Statistical Society, Series A (Statistics in Society). 1972;34:216. [Google Scholar]
36.Kingsberg SA, Larkin LC, Liu JH. Clinical Effects of Early or Surgical Menopause. Obstetrics & Gynecology. 2020. Apr;135(4):853–68. [DOI] [PubMed] [Google Scholar]
37.Lovett SM, Sandler DP, O’Brien KM. Hysterectomy, bilateral oophorectomy, and breast cancer risk in a racially diverse prospective cohort study. JNCI: Journal of the National Cancer Institute. 2023. Jun 8;115(6):662–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Sjögren LL, Mørch LS, Løkkegaard E. Hormone replacement therapy and the risk of endometrial cancer: A systematic review. Maturitas. 2016. Sep;91:25–35. [DOI] [PubMed] [Google Scholar]
39.Warren MP. Historical Perspectives in Postmenopausal Hormone Therapy: Defining the Right Dose and Duration. Mayo Clinic Proceedings. 2007. Feb;82(2):219–26. [DOI] [PubMed] [Google Scholar]
40.Hersh AL, Stefanick ML, Stafford RS. National use of postmenopausal hormone therapy: annual trends and response to recent evidence: Obstetrics & Gynecology. 2004. Apr;103(4):796. [DOI] [PubMed] [Google Scholar]
41.Anderson GL, Chlebowski RT, Aragaki AK, Kuller LH, Manson JE, Gass M, et al. Conjugated equine oestrogen and breast cancer incidence and mortality in postmenopausal women with hysterectomy: extended follow-up of the Women’s Health Initiative randomised placebo-controlled trial. Lancet Oncol. 2012. May;13(5):476–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Chlebowski RT, Rohan TE, Manson JE, Aragaki AK, Kaunitz A, Stefanick ML, et al. Breast Cancer After Use of Estrogen Plus Progestin and Estrogen Alone: Analyses of Data From 2 Women’s Health Initiative Randomized Clinical Trials. JAMA Oncol. 2015. Jun 1;1(3):296. [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Tuesley KM, Spilsbury K, Pearson SA, Donovan P, Obermair A, Coory MD, et al. Long-acting, progestin-based contraceptives and risk of breast, gynecological, and other cancers. JNCI: Journal of the National Cancer Institute. 2025. Jan 14;djae282. [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Phipps AI, Malone KE, Porter PL, Daling JR, Li CI. Reproductive and hormonal risk factors for postmenopausal luminal, HER-2-overexpressing, and triple-negative breast cancer. Cancer. 2008. Oct;113(7):1521–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Whitehead M, Farmer R. The million women study. 24(3). [DOI] [PubMed] [Google Scholar]
46.Shapiro S The Million Women Study: potential biases do not allow uncritical acceptance of the data. Climacteric. 2004. Jan;7(1):3–7. [DOI] [PubMed] [Google Scholar]
47.van den Brandt PA, Spiegelman D, Yaun SSSS, Adami HOHOO, Beeson L, Folsom AR, et al. Pooled analysis of prospective cohort studies on height, weight, and breast cancer risk. Am J Epidemiol. 2000;152(6):514–27. [DOI] [PubMed] [Google Scholar]
48.Shantakumar S, Terry MB, Paykin A, Teitelbaum SL, Britton JA, Moorman PG, et al. Age and menopausal effects of hormonal birth control and hormone replacement therapy in relation to breast cancer risk. Am J Epidemiol. 2007. May;165(10):1187–98. [DOI] [PubMed] [Google Scholar]
49.Palmer JR, Rosenberg L, Clarke EA, Miller DR, Shapiro S. Breast cancer risk after estrogen replacement therapy: results from the Toronto Breast Cancer Study. Am J Epidemiol. 1991;134(12):1386–95. [DOI] [PubMed] [Google Scholar]
50.O’Brien KM, Fei C, Sandler DP, Nichols HB, DeRoo LA, Weinberg CR. Hormone therapy and young-onset breast cancer. American Journal of Epidemiology. 2015. Feb;181(10):799–807. [DOI] [PMC free article] [PubMed] [Google Scholar]
51.Racine A, Bijon A, Fournier A, Mesrine S, Clavel-Chapelon F, Carbonnel F, et al. Menopausal hormone therapy and risk of cholecystectomy: a prospective study based on the French E3N cohort. CMAJ. 2013. Apr 16;185(7):555–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
52.Hernán MA, Robins JM. Using Big Data to Emulate a Target Trial When a Randomized Trial Is Not Available. American Journal of Epidemiology. 2016. Mar;183(8):758–64. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material

NIHMS2089962-supplement-Supplementary_Material.pdf^{(472.5KB, pdf)}

Data Availability Statement

[R1] 1.Rossouw JE, Anderson GL, Prentice RL, LaCroix AZ, Kooperberg C, Stefanick ML, et al. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results From the Women’s Health Initiative randomized controlled trial. JAMA. 2002. Jul;288(3):321–33. [DOI] [PubMed] [Google Scholar]

[R2] 2.Beral V, Million Women Study Collaborators. Breast cancer and hormone-replacement therapy in the Million Women Study. Lancet. 2003. Aug;362(9382):419–27. [DOI] [PubMed] [Google Scholar]

[R3] 3.Anderson GL, Limacher M, Assaf AR, Bassford T, Beresford SAA, Black H, et al. Effects of conjugated equine estrogen in postmenopausal women with hysterectomy: the Women’s Health Initiative randomized controlled trial. JAMA. 2004. Apr;291(14):1701–12. [DOI] [PubMed] [Google Scholar]

[R4] 4.Type and timing of menopausal hormone therapy and breast cancer risk: individual participant meta-analysis of the worldwide epidemiological evidence. The Lancet. 2019. Sep;394(10204):1159–68. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R5] 5.Bernstein L Epidemiology of Endocrine-Related Risk Factors for Breast Cancer. Jounral of Mammary Gland Biology and Neoplasia. 7(1):3–15. [DOI] [PubMed] [Google Scholar]

[R6] 6.Stuenkel CA, Davis SR, Gompel A, Lumsden MA, Murad MH, Pinkerton JV, et al. Treatment of Symptoms of the Menopause: An Endocrine Society Clinical Practice Guideline. The Journal of Clinical Endocrinology & Metabolism. 2015. Nov;100(11):3975–4011. [DOI] [PubMed] [Google Scholar]

[R7] 7.the American College of Obstetricians and Gynecologists. Practice Bulletin: Clinical Management Guidelines for Obstetrician-Gynecologists. Obstet Gynecol. 2014;123(1):202–16.24463691 [Google Scholar]

[R8] 8.The 2017 hormone therapy position statement of The North American Menopause Society. Menopause. 2017. Jul;24(7):728–53. [DOI] [PubMed] [Google Scholar]

[R9] 9.Warnock JK, Cohen LJ, Blumenthal H, Hammond JE. Hormone-Related Migraine Headaches and Mood Disorders: Treatment with Estrogen Stabilization. Pharmacotherapy. 2017. Jan;37(1):120–8. [DOI] [PubMed] [Google Scholar]

[R10] 10.Shifren JL, Gass MLS. The North American Menopause Society Recommendations for Clinical Care of Midlife Women. Menopause. 2014. Oct;21(10):1038–62. [DOI] [PubMed] [Google Scholar]

[R11] 11.Tepper PG, Brooks MM, Randolph JF, Crawford SL, El Khoudary SR, Gold EB, et al. Characterizing the trajectories of vasomotor symptoms across the menopausal transition. Menopause. 2016. Oct;23(10):1067–74. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R12] 12.Sprague BL, Trentham-Dietz A, Cronin KA. A sustained decline in postmenopausal hormone use. Obstet Gynecol. 2012. Sep;120(3):595–603. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R13] 13.Nichols HB, Schoemaker MJ, Wright LB, McGowan C, Brook MN, McClain KM, et al. The Premenopausal Breast Cancer Collaboration: A pooling project of studies participating in the National Cancer Institute Cohort Consortium. Cancer epidemiology, biomarkers & prevention. 2017;26(September). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R14] 14.Nichols HB, Schoemaker MJ, Cai J, Xu J, Wright LB, Brook MN, et al. Breast Cancer Risk After Recent Childbirth. Annals of Internal Medicine [Internet]. 2018; Available from: http://annals.org/article.aspx?doi=10.7326/M18-1323 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R15] 15.Schoemaker MJ, Nichols HB, Wright LB, Brook MN, Jones ME, O’Brien KM, et al. Association of Body Mass Index and Age With Subsequent Breast Cancer Risk in Premenopausal Women. JAMA Oncology. 2018;1–10. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R16] 16.Rosenberg L, Palmer JR, Wise LA, Adams-Campbell LL. A Prospective Study of Female Hormone Use and Breast Cancer Among Black Women. Arch Intern Med. 2006;166. [DOI] [PubMed] [Google Scholar]

[R17] 17.Jones ME, Schoemaker MJ, Wright L, McFadden E, Griffin J, Thomas D, et al. Menopausal hormone therapy and breast cancer: what is the true size of the increased risk? Br J Cancer. 2016. Aug;115(5):607–15. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R18] 18.Saxena T, Lee E, Henderson KD, Clarke CA, West D, Marshall SF, et al. Menopausal hormone therapy and subsequent risk of specific invasive breast cancer subtypes in the California Teachers Study. Cancer Epidemiol Biomarkers Prev. 2010. Sep;19(9):2366–78. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R19] 19.Patel AV, Jacobs EJ, Dudas DM, Briggs PJ, Lichtman CJ, Bain EB, et al. The American Cancer Society’s Cancer Prevention Study 3 (CPS-3): Recruitment, study design, and baseline characteristics. Cancer. 2017. Jun;123(11):2014–24. [DOI] [PubMed] [Google Scholar]

[R20] 20.Rohan TE, Soskolne CL, Carroll KK, Kreiger N. The Canadian Study of Diet, Lifestyle, and Health: Design and characteristics of a new cohort study of cancer risk. Cancer Detection and Prevention. 2007;31(1):12–7. [DOI] [PubMed] [Google Scholar]

[R21] 21.Gross AL, Newschaffer CJ, Hoffman-Bolton J, Rifai N, Visvanathan K. Adipocytokines, Inflammation, and Breast Cancer Risk in Postmenopausal Women: A Prospective Study. Cancer Epidemiology, Biomarkers & Prevention. 2013. Jul 1;22(7):1319–24. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R22] 22.Olson JE, Sellers TA, Scott CG, Schueler BA, Brandt KR, Serie DJ, et al. The influence of mammogram acquisition on the mammographic density and breast cancer association in the mayo mammography health study cohort. Breast Cancer Res. 2012. Dec;14(6):R147. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R23] 23.Wu AH, Koh WP, Wang R, Lee HP, Yu MC. Soy intake and breast cancer risk in Singapore Chinese Health Study. Br J Cancer. 2008. Jul;99(1):196–200. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R24] 24.Sandler DP, Hodgson ME, Deming-Halverson SL, Juras PJ, D’Aloisio AD, Suarez L, et al. The Sister Study: Baseline methods and participant characteristics. Environ Health Perspect. 2017;125(12):127003. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R25] 25.Preston DL, Kitahara CM, Freedman DM, Sigurdson AJ, Simon SL, Little MP, et al. Breast cancer risk and protracted low-to-moderate dose occupational radiation exposure in the US Radiologic Technologists Cohort, 1983–2008. Br J Cancer. 2016. Oct;115(9):1105–12. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R26] 26.Gertig DM, Fletcher AS, English DR, MacInnis RJ, Hopper JL, Giles GG. Hormone therapy and breast cancer: what factors modify the association?: Menopause. 2006. Mar;13(2):178–84. [DOI] [PubMed] [Google Scholar]

[R27] 27.Zheng W, Chow WH, Yang G, Jin F, Rothman N, Blair A, et al. The Shanghai Women’s Health Study: Rationale, Study Design, and Baseline Characteristics. American Journal of Epidemiology. 2005. Dec 1;162(11):1123–31. [DOI] [PubMed] [Google Scholar]

[R28] 28.Signorello LB, Hargreaves MK, Steinwandel MD, Zheng W, Cai Q, Schlundt DG, et al. Southern Community Cohort Study: Establishing a Cohort to Investigate Health Disparities. Journal of the National Medical Association. 2005;97(7). [PMC free article] [PubMed] [Google Scholar]

[R29] 29.O’Brien KM, Lawrence KG, Keil AP. The case for case-cohort: An applied epidemiologist’s guide to re-framing case-cohort studies to improve usability and flexibility. Epidemiology. 33(3):354–61. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R30] 30.Schonfeld SJ, Pfeiffer RM, Lacey JV, Berrington De Gonzalez A, Doody MM, Greenlee RT, et al. Hormone-related Risk Factors and Postmenopausal Breast Cancer Among Nulliparous Versus Parous Women: An Aggregated Study. American Journal of Epidemiology. 2011. Mar 1;173(5):509–17. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R31] 31.White IR, Royston P. Imputing missing covariate values for the Cox model: Imputing missing covariate values for the Cox model. Statist Med. 2009. Jul 10;28(15):1982–98. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R32] 32.DerSimonian R, Laird N. Meta-analysis in clinical trials. Controlled Clinical Trials. 1986;7(3):177–88. [DOI] [PubMed] [Google Scholar]

[R33] 33.Von Holle A, Adami HO, Baglietto L, Berrington De Gonzalez A, Bertrand KA, Blot W, et al. BMI and breast cancer risk around age at menopause. Cancer Epidemiology. 2024. Apr;89:102545. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R34] 34.Rubin Donald. Multiple Imputation for Nonresponse in Surveys. John Wiley & Sons, Inc; 1987. [Google Scholar]

[R35] 35.Breslow NE. Discussion of Professor Cox’s paper. Journal of the Royal Statistical Society, Series A (Statistics in Society). 1972;34:216. [Google Scholar]

[R36] 36.Kingsberg SA, Larkin LC, Liu JH. Clinical Effects of Early or Surgical Menopause. Obstetrics & Gynecology. 2020. Apr;135(4):853–68. [DOI] [PubMed] [Google Scholar]

[R37] 37.Lovett SM, Sandler DP, O’Brien KM. Hysterectomy, bilateral oophorectomy, and breast cancer risk in a racially diverse prospective cohort study. JNCI: Journal of the National Cancer Institute. 2023. Jun 8;115(6):662–70. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R38] 38.Sjögren LL, Mørch LS, Løkkegaard E. Hormone replacement therapy and the risk of endometrial cancer: A systematic review. Maturitas. 2016. Sep;91:25–35. [DOI] [PubMed] [Google Scholar]

[R39] 39.Warren MP. Historical Perspectives in Postmenopausal Hormone Therapy: Defining the Right Dose and Duration. Mayo Clinic Proceedings. 2007. Feb;82(2):219–26. [DOI] [PubMed] [Google Scholar]

[R40] 40.Hersh AL, Stefanick ML, Stafford RS. National use of postmenopausal hormone therapy: annual trends and response to recent evidence: Obstetrics & Gynecology. 2004. Apr;103(4):796. [DOI] [PubMed] [Google Scholar]

[R41] 41.Anderson GL, Chlebowski RT, Aragaki AK, Kuller LH, Manson JE, Gass M, et al. Conjugated equine oestrogen and breast cancer incidence and mortality in postmenopausal women with hysterectomy: extended follow-up of the Women’s Health Initiative randomised placebo-controlled trial. Lancet Oncol. 2012. May;13(5):476–86. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R42] 42.Chlebowski RT, Rohan TE, Manson JE, Aragaki AK, Kaunitz A, Stefanick ML, et al. Breast Cancer After Use of Estrogen Plus Progestin and Estrogen Alone: Analyses of Data From 2 Women’s Health Initiative Randomized Clinical Trials. JAMA Oncol. 2015. Jun 1;1(3):296. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R43] 43.Tuesley KM, Spilsbury K, Pearson SA, Donovan P, Obermair A, Coory MD, et al. Long-acting, progestin-based contraceptives and risk of breast, gynecological, and other cancers. JNCI: Journal of the National Cancer Institute. 2025. Jan 14;djae282. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R44] 44.Phipps AI, Malone KE, Porter PL, Daling JR, Li CI. Reproductive and hormonal risk factors for postmenopausal luminal, HER-2-overexpressing, and triple-negative breast cancer. Cancer. 2008. Oct;113(7):1521–6. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R45] 45.Whitehead M, Farmer R. The million women study. 24(3). [DOI] [PubMed] [Google Scholar]

[R46] 46.Shapiro S The Million Women Study: potential biases do not allow uncritical acceptance of the data. Climacteric. 2004. Jan;7(1):3–7. [DOI] [PubMed] [Google Scholar]

[R47] 47.van den Brandt PA, Spiegelman D, Yaun SSSS, Adami HOHOO, Beeson L, Folsom AR, et al. Pooled analysis of prospective cohort studies on height, weight, and breast cancer risk. Am J Epidemiol. 2000;152(6):514–27. [DOI] [PubMed] [Google Scholar]

[R48] 48.Shantakumar S, Terry MB, Paykin A, Teitelbaum SL, Britton JA, Moorman PG, et al. Age and menopausal effects of hormonal birth control and hormone replacement therapy in relation to breast cancer risk. Am J Epidemiol. 2007. May;165(10):1187–98. [DOI] [PubMed] [Google Scholar]

[R49] 49.Palmer JR, Rosenberg L, Clarke EA, Miller DR, Shapiro S. Breast cancer risk after estrogen replacement therapy: results from the Toronto Breast Cancer Study. Am J Epidemiol. 1991;134(12):1386–95. [DOI] [PubMed] [Google Scholar]

[R50] 50.O’Brien KM, Fei C, Sandler DP, Nichols HB, DeRoo LA, Weinberg CR. Hormone therapy and young-onset breast cancer. American Journal of Epidemiology. 2015. Feb;181(10):799–807. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R51] 51.Racine A, Bijon A, Fournier A, Mesrine S, Clavel-Chapelon F, Carbonnel F, et al. Menopausal hormone therapy and risk of cholecystectomy: a prospective study based on the French E3N cohort. CMAJ. 2013. Apr 16;185(7):555–61. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R52] 52.Hernán MA, Robins JM. Using Big Data to Emulate a Target Trial When a Randomized Trial Is Not Available. American Journal of Epidemiology. 2016. Mar;183(8):758–64. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Hormone therapy use and young-onset breast cancer: a pooled analysis of prospective cohorts included in the Premenopausal Breast Cancer Collaborative Group

Katie M O’Brien, PhD

Melissa G House, MS

Mandy Goldberg, PhD

Michael E Jones, PhD

Clarice R Weinberg, PhD

Amy Berrington de Gonzalez, DPhil

Kimberly A Bertrand, ScD

William J Blot, PhD

Jessica Clague DeHart, PhD

Fergus J Couch, PhD

Montserrat Garcia-Closas, MD DrPH

Graham G Giles, PhD

Victoria A Kirsh, PhD

Cari M Kitahara, PhD

Woon-Puay Koh, PhD

Hannah Lui Park, PhD

Roger L Milne, PhD

Julie R Palmer, ScD

Alpa V Patel, PhD

Thomas E Rohan, MBBS, DHSc

Minouk J Schoemaker, PhD

Anthony J Swerdlow, PhD

Lauren R Teras, PhD

Celine Vachon, PhD

Kala Visvanathan, MD, MHS

Jian-Min Yuan, MD PhD

Wei Zheng, MD PhD

Hazel B Nichols, PhD

Dale P Sandler, PhD

Summary

Background

Methods

Findings:

Interpretation:

Methods

Study design

Participants

Hormone Therapy

Incident YOBC

Potential Confounders and Other Covariates

Table 1.

Statistical Analysis

Time-varying analyses

Role of the funding source

Results

Table 2.

Table 3.

Table 4.

Table 5.

Discussion

Strengths and Limitations

Supplementary Material

Research in Context.

Evidence before the study

Added value of this study

Implications of all of the available evidence

Acknowledgements

Funding:

Funding source:

Footnotes

Data sharing:

References

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases