Hormone Replacement Therapy, Breast Cancer Risk Factors, and Breast Cancer Risk: A Nationwide Population-Based Cohort

Hormone replacement therapy (HRT) alleviates the climacteric symptoms of menopause. The use of HRT increased phenomenally after its approval in 1988 by the FDA for prevention of osteoporosis, but decreased after publication of the Women's Health Initiative (WHI) study in 2002 (1, 2) The risk of breast cancer associated with HRT was widely known, and women stopped or refused HRT due to fear of breast cancer (1, 3, 4). The association between the risk of breast cancer and HRT has been confirmed by multiple observational studies, and was indirectly verified by the concurrent decrease in the incidence of breast cancer from 2002 to 2003 (5–8). However, whether HRT is related to the risk of breast cancer in all postmenopausal women is still unclear.

Most randomized controlled and observational cohort studies of the link between HRT and breast cancer were performed in Western countries (2, 6, 9). However, the epidemiology of breast cancer differs considerably between Western and Asian populations. The age-specific incidence of breast cancer differs between Western and Asian countries; in the Republic of Korea its incidence peaks at age 50 years and decreases thereafter (10). The distribution of breast cancer subtypes also differs among regions and countries; the Korean population has a high proportion of estrogen receptor–negative breast cancer (11).

The HRT-related risk of breast cancer may also differ according to other risk factors for breast cancer. Obesity is a risk factor for breast cancer in postmenopausal women. Breast density is also a strong and independent risk factor for breast cancer and is affected by HRT use (12). Reproductive factors such as oral contraceptive use, parity, and breastfeeding history influence both the incidence of breast cancer and use of HRT (13, 14). The clinical decision to use HRT involves balancing its risks and benefits, and the heterogeneity of the risk of breast cancer among HRT users must be considered.

In this study, we investigated the association between HRT and breast cancer in a nationwide cohort with risk stratification according to risk factors for breast cancer.

The National Health Insurance Service (NHIS) is a nonprofit, single-payer organization run by the Korean government. Almost all Korean citizens (97.2%, ∼50 million) are covered by the NHIS (15). The NHIS database includes information on demographics, medical examination results, treatment information, and International Classification of Diseases (ICD)-10 codes registered by clinicians. The NHIS also provides regular health checkups every 1 or 2 years to insured adults >40 years of age and employees >20 years of age. The Korean National Health Screening database includes data on anthropometric parameters, responses to health questionnaires, and laboratory findings. Data from the regular health checkups can be merged with the NHIS insurance claim data and are provided for research purposes, being handled with very high confidentiality.

We analyzed all women who participated in the breast cancer screening program from 2009 to 2014 (n = 6,612,839; Fig. 1). The NHIS provides breast cancer screening via mammography every 2 years, starting from the age of 40 years, to all insured women. Women who underwent breast cancer screening along with regular annual/biennial health checkups within 30 days were identified (n = 6,414,645). Among them, only women who self-reported as menopausal were included (n = 4,775,420). Patients with a history of breast cancer were excluded. Also, patients who were diagnosed with breast cancer within 6 months of screening were assumed to have undiagnosed breast cancer at the point of screening and were excluded. Finally, a total of 4,558,376 women were included in this study. The cohort was divided into four groups according to the self-reported history of HRT: none, <2 years, 2 to <5 years, and ≥5 years.

Newly diagnosed breast cancer was identified using the rare, incurable disease registry of the NHI (15). The registration program provides copayment reduction for patients with severe and/or rare, intractable diseases, including malignant neoplasms. Breast cancer is registered by physicians based on strict diagnostic criteria. Women who were registered for breast cancer 6 months after participating in the breast cancer screening program were defined as newly diagnosed patients with breast cancer. The diagnosis of breast cancer was followed-up until 2016.

The breast cancer screening program includes a standardized questionnaire about the following risk factors for breast cancer: age at menarche and menopause, parity, history and duration of breastfeeding, history of oral contraceptives (OC) and HRT, previous diagnosis of a benign breast mass, and family history of breast cancer (Supplementary Materials and Methods). Data on history of HRT were obtained from this questionnaire but it has limitations, as the question asking about any hormone medications to relieve menopausal symptoms is quite broad. Data on the family history of breast cancer were not used in this study due to the high rate of missing data. Breast density by mammography was divided into four categories according to the Breast Imaging-Reporting and Data System (BI-RADS) classification (16).

The following health checkup data were obtained: body weight (kg) and height (cm) measured using an electronic scale, and waist circumference measured at the umbilicus level by trained examiners. Body mass index (BMI) was calculated as body weight in kilograms divided by the square of height in meters (kg/m²). Lifestyle factors including smoking, alcohol consumption, and regular exercise were assessed using standardized questionnaires. Smoking status was categorized into three groups: never, former, and current smoker. Alcohol consumption was categorized into three groups: none, moderate, and heavy. Heavy drinking was defined as consumption of >30 g/day. Regular exercise was defined as ≥30 minutes of moderate physical activity ≥5 times per week or ≥20 minutes of strenuous physical activity ≥3 times per week.

Comorbidities were defined by the ICD-10 codes, use of related medications, or an abnormal measurement in the health examination. Hypertension was defined as an ICD-10 code of I10-13/15, a history of related medications, or systolic blood pressure ≥140 mm Hg or diastolic blood pressure ≥90 mm Hg. Type 2 diabetes mellitus (DM) was defined as an ICD-10 code of E11-14, a history of related medications, or a fasting glucose level of ≥126 mg/dL. Dyslipidemia was defined as an ICD-10 code of E78, a history of related medications, or a fasting total cholesterol level of ≥240 mg/dL.

This study was approved by the Institutional Review Board of the Catholic University of Korea (IRB no. KC18ZESI0157). Anonymous and deidentified information were used in the analyses.

The characteristics of the four groups defined by HRT use were compared using the χ² test and one-way ANOVA. The cumulative incidence of breast cancer according to HRT duration was analyzed by the Kaplan–Meier method and the log-rank test. A Cox proportional hazards regression analysis was performed to calculate the HR for the incidence of breast cancer according to the duration of HRT. Adjustments for age, parity, history of breastfeeding, breast density, OC history, smoking, alcohol drinking, exercise, BMI, height, DM, and menarche age were conducted. Covariates were categorized as shown in Table 1 (baseline characteristics).

The interactions between HRT and risk factors for breast cancer were analyzed by Cox proportional hazards regression analysis, adjusting for the above factors. All analyses were performed with SAS (version 9.4; SAS Institute). Statistical significance was assumed at P < 0.05.

A total of 4,558,376 women underwent breast cancer screening from 2009 to 2014. Among them, 696,084 (15.3%) women reported current or previous HRT use. The number of women exposed to HRT for <2 years was 410,304 (9.0%), the number exposed for 2 to <5 years was 158,514 (3.5%), and the number exposed for ≥5 years was 127,266 (2.8%). The patients' characteristics according to the duration of HRT are listed in Table 1. Women who had no exposure to HRT were more likely to have a history of breastfeeding, have a fatty breast, and no personal history of benign breast diseases. Women who were nonsmokers and did not drink alcohol were less likely to use HRT. Nonuse of HRT was also associated with a higher BMI, DM, hypertension, and dyslipidemia. In contrast, HRT duration was linearly related to women with no live births, longer duration of OC, and regular exercise.

During a median follow-up of 5.35 years (Q1–Q3, 3.26–6.41), 26,797 (0.6%) women were newly diagnosed with breast cancer. Among them, 92.9% had invasive breast cancer and 7.1% had ductal carcinoma in situ (DCIS). The HR of the risk of breast cancer in HRT users was 1.25 [95% confidence interval (CI), 1.22–1.29] compared with HRT nonusers. The results of a multivariable analysis of the incidence of breast cancer according to HRT duration are shown in Table 2. The incidence of breast cancer significantly increased with increasing HRT duration. The adjusted HRs for breast cancer incidence in women who had used HRT for <2 years, 2 to <5 years, and ≥5 years were 1.08 (95% CI, 1.04–1.12), 1.33 (95% CI, 1.25–1.40), and 1.72 (95% CI, 1.63–1.82), respectively. This proportional increase in the incidence of breast cancer according to HRT duration was evident in women with invasive breast cancer and those with DCIS (Table 2). The cumulative incidence of breast cancer was linearly related to the duration of HRT (Fig. 2).

Associations with HRT according to risk factors for breast cancer are shown in Table 3. Women >65 years of age had a higher risk of breast cancer related to HRT compared with younger women (P_interaction < 0.0001). The association between HRT and the risk of breast cancer was stronger in women of normal weight (HR = 1.29; 95% CI, 1.24–1.34) compared with overweight women (HR = 1.19; 95% CI, 1.13–1.25; P_interaction = 0.0035). The association also varied by mammographic breast density, as women with dense breasts (density = 51%–75%; HR = 1.28; 95% CI, 1.22–1.35; density = 76%–100%; HR = 1.29; 95% CI, 1.18–1.41) had a higher risk of breast cancer compared with women with scattered density breasts (density = 25%–50%; HR = 1.18; 95% CI, 1.12–1.24; P_interaction = 0.0243). The statistical significance was not derived from the high sample size, as when breast density was dichotomized by 50% the P value for interaction was not significant (P_interaction = 0.6348). The influence of HRT on the incidence of breast cancer did not differ according to personal history of benign breast disease, parity, breastfeeding history, OC use, or age at menarche.

In this nationwide population-based cohort study, we confirmed that HRT users were at greater risk of breast cancer than nonusers. Use of HRT increased risk of both invasive and in situ cancer, and also increased proportionally with longer use. HRT-related increased breast cancer risk was greater for lean women and women with high-density breasts. This is the largest cohort study to address this issue, especially in the Asian population.

In our study, the adjusted HR for breast cancer related to HRT use was 1.25, with an incidence of 1.62 per 1,000 persons. The breast cancer risk increment is similar to a recently published individual participant meta-analysis of 58 studies, which reports a modest but significant increase in risk irrespective of type of HRT (relative risk = 1.26; 95% CI, 1.24–1.28; ref. 9). In this meta-analysis, the excess risk was clear even in current or past HRT users of only 1 to 4 years (relative risk = 1.17; 95% CI, 1.10–1.26), and it also increased steadily with duration of use (9). That finding is consistent with our study; the increment in the risk of breast cancer was 7.9% among women who had used HRT for under 2 years, but 72.2% in women who had used HRT for over 5 years.

In this study, HRT use increased the breast cancer risk not only in invasive cancer but also in DCIS. Despite the small number of events, DCIS risk increased proportionally with HRT duration. Although HRT use is a relatively well-established risk factor for invasive breast cancer, reports on the association between HRT use and DCIS risk show conflicting results (17–21). The first randomized trial to investigate this issue was the report from the WHI trial, which reported a nonsignificant trend to an increased DCIS risk related to the combined use of estrogen and progesterone (18). Several cohort studies showed comparable results, also demonstrating a risk increase proportional to HRT duration, similar to this study, whereas other cohort and case–control studies found no association between HRT use and DCIS risk (17, 19–22). The inconsistent results across studies are due to various factors, such as HRT regimen, duration of use, pattern of use, and outcomes (combining DCIS with lobular carcinoma in situ or reporting separately) investigated (19). Also, studies about risk factors of DCIS are prone to screening bias, as most DCIS cases are diagnosed through screening programs. However, all women included in this cohort study underwent mammography screening, eliminating the possibility of this. Although the relationship between HRT use and DCIS risk is still unclear, this large-scale population-based cohort study adds data suggesting a positive relationship.

There are relatively few reports on Asian populations regarding the effect of HRT use on breast cancer risk and these have tended to be small studies with conflicting results (23–25). A recent meta-analysis pooling these data suggested that Asians have a higher HRT-related risk of breast cancer than Caucasians (pooled HR for Asians = 1.84 vs. Caucasians = 1.34; ref. 26). However, this analysis included only 541 cancer events, which constitute insufficient data with which to draw a conclusion. Our findings provide real-word insight into the influence of HRT use on breast cancer risk in an Asian population, indicating a modest risk increase, of a degree similar to that in Western populations.

Regarding breast cancer risk factors, normal-weight women and women with dense breasts had a higher risk of breast cancer related to HRT use. Obesity is reported to attenuate the association between HRT and the risk of breast cancer (9, 27, 28). In postmenopausal women, body weight is positively associated with the risk of breast cancer due to the increased conversion of adrenal androgens to estrogens within the larger adipose stores of obese women. This leads to relatively higher circulating estrogen levels in obese women and, for HRT users, the proportional increase in the estrogen level may be smaller in obese compared with normal-weight women (27, 28). Furthermore, breast density may be linked to the relationship between body weight and the risk of breast cancer, as leaner women are more likely to have denser breasts, which increases the risk of breast cancer.

Women with extremely dense breasts have a four-fold higher risk of breast cancer compared with women with a breast density of <5%, indicating a strong positive relationship between breast density and the risk of breast cancer (29, 30). In addition, HRT use increases breast density (31, 32). Similar to our results, the risk of breast cancer in HRT users reportedly differs according to breast density, being generally higher in women with high-density breasts (27, 33). The biological mechanism underlying the relationship between high breast density and the effect of HRT on the risk of breast cancer is unknown. The use of HRT in postmenopausal women may slow the normal process of breast involution during aging, resulting in a sustained high breast density and an increased risk of breast cancer (34). Alternatively, the presence of exogenous estrogen due to HRT use may stimulate hyperplasia of epithelial cells, the number of which is greater in high-density breasts, promoting tumorigenesis and increasing the risk of breast cancer (34, 35). Although the causality is unclear, the higher risk of HRT users with denser breasts could be useful for risk stratification when considering HRT for postmenopausal women.

The main strengths of this study are the large sample size and its longitudinal nationwide population-based cohort design. In addition, use of health checkup and breast cancer screening data enabled adjustment not only for risk factors for breast cancer but also comorbidities and lifestyle factors. Furthermore, the possibility of screening bias can be eliminated as all of the women underwent screening mammograms.

The limitations of this study include the lack of information on the formulation and timing of HRT. Therefore, the effects of estrogen–progesterone and estrogen-only agents were not separated and analysis according to current/past user or the timing of HRT initiation was not feasible. Previous reports describe different risk ratios according to these factors (2, 9, 36). Data for history of hysterectomy or bilateral oophorectomy were also not available. Another limitation is the lack of data on the subtype and histologic type of breast cancer. In addition, history of HRT use was assessed by self-reporting and, thus, may have been affected by recall bias. However, a report from Korea shows that the compliance of HRT is low (60.9% for 1 year, 49.6% for 3 years), which implies that self-reported data may be more accurate than prescription data (37). Yet, the broadness of the question asking of any hormone therapy (Supplementary Materials and Methods) still limits the accuracy of data for HRT use.

In this nationwide population-based cohort study, HRT use was associated with an increased risk of breast cancer. The risk of breast cancer increased proportionally with the duration of HRT and was greater in normal-weight women and women with dense breasts. Risk stratification should be performed when deciding whether to apply HRT for relief of menopausal symptoms.

No potential conflicts of interest were disclosed.

Conception and design: T.-K. Yoo, K.D. Han, W-C. Park, B.J. Chae

Development of methodology: T.-K. Yoo, K.D. Han, W-C. Park

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K.D. Han, J. Ahn

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): T.-K. Yoo, K.D. Han, D. Kim, B.J. Chae

Writing, review, and/or revision of the manuscript: T.-K. Yoo, B.J. Chae

Study supervision: W-C. Park, B.J. Chae

This work was supported by YuYu Pharma, Inc., and the Basic Science Research Program of the National Research Foundation funded by the Ministry of Education (2017R1D1A1B03033486 to B.J. Chae).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.