Increased mammographic breast density has been widely viewed as a strong predictor for the development of breast cancer. A 2006 meta-analysis by McCormack et al. showed that women with the highest breast densities are at 4-6 times higher risk for developing breast cancer [1]. Further, it showed that breast density was a stronger risk predictor than either Wolfe grade or BIRADS classification for future development of malignancy [1]. Given that mammography is less sensitive in women with dense breasts, the study suggested that this estimate actually underestimates the positive predictive value of breast density on development of malignancy [1,2]. Interestingly, breast cancer risk increases with age and weight, even as breast density decreases [3,4]. The Pike model (1983) suggests an explanation for this apparent paradoxical relationship, attributing cancer incidence to cumulative breast tissue hormonal exposure [3,4]. Hormonal exposure was proposed as one of the contributing factors and remains the focus of most predictive models and questionnaires for female patients [3,4].

      The effect of HRT on breast density and breast cancer has been the focus of a recent study. The Women’s Health Initiative (WHI) Postmenopausal Hormone Therapy Trials examined the effect of hormone replacement on women’s health [5-7]. In 2003, the WHI reported that combined HRT (estrogen plus progestin) increases risk of breast cancer and those women were diagnosed with larger tumors and at later stages [7-9]. An ancillary study by McTiernan et al. looked at breast density in postmenopausal patients assigned to combined HRT or placebo after 1 year [10]. The study showed that the mean mammographic density increased by 6.0% in the HRT group, whereas the placebo group had a .9% decrease [8]. Subsequently, the WHI Trials found that women with previous hysterectomies given estrogen alone therapy were not at increased risk of cancer at 7 years. They did, however, require more frequent mammography (secondary to increased number of abnormalities on mammogram) [10].

      Though the use of HRT has been linked to increased mammographic abnormalities and breast cancer, it can provide relief from moderate to severe hot flashes and symptoms of vulvar and vaginal atrophy. These symptoms can be particularly severe in women who have undergone surgical menopause. A 2011 study by Finch et al showed that vasomotor symptoms and decline in sexual function were significantly more dramatic in women who underwent Bilateral-Salpingo-Oophorectomy (BSO) prior to menopause than in women who underwent natural menopause, and that these symptoms were at least partially mitigated by the use of HRT [10,11]. After BSO, women have an abrupt rather than gradual reduction in hormone production [10,11]. Although 4-6% of women in the US undergo surgical menopause, the effect of HRT on breast density this cohort has not been thoroughly examined. The purpose of our study was to investigate the relationship of mammographic breast density, clinical and tumor characteristics in women who had BSO and used HRT.

Study participants

      The Breast Cancer Database (BCD) at our medical center is a longitudinal registry that was established in January 2010. All patients undergoing definitive breast cancer surgery at our institution are eligible to enroll in the BCD. The variables collected include information on personal and family history, screening history, methods of diagnosis, stage at diagnosis, details of treatment and outcomes. All clinical data were obtained from detailed questionnaires filled out at the time of diagnosis, and review of the electronic medical records. This study was approved by the Institutional Review Board and was compliant with the standards of the Health Insurance Portability and Accountability Act.

      Patients included in this study were enrolled in the Breast Cancer Database between January 2010 and February 2016. The BCD was queried for all patients who had reported menopause following BSO (which may have been performed for reasons including malignancy, benign diagnosis or preventative with BRCA diagnosis or family history of malignancy), prior to diagnosis of malignancy and then were subsequently diagnosed with invasive breast cancer. BSO may have occurred with or without concurrent hysterectomy. Patients were considered as being on HRT if by questionnaire or on chart review they reported taking any type of HRT: combined, or estrogen or progestin only. Women who had a history of HRT usage without history of BSO were excluded from the analyses. Demographic information, indication for the examination and information including mammographic density and BI-RADS assessment was obtained from the electronic medical record.

Statistical Analyses

      Patients who had BSO were categorized into three groups: 1) BSO with no HRT 2) BSO with prior HRT usage or 3) BSO and current use of HRT at the time of diagnosis. The patients who had BSO were then compared to patients who had never taken HRT and reported natural (non-surgical) menopause. The variables of interest included clinical and tumor characteristics and mammographic breast density at the time of malignancy diagnosis. Pearson’s chi-square and Fisher’s exact tests were used to assess differences in stage, histology, ER/PR/HER2 positivity and breast density between those who had never taken HRT and those who were on HRT at the time of cancer diagnosis with a significance level of 0.05. The same analysis was completed for those who had never taken HRT and those who had previously taken HRT but were not on HRT at the time of diagnosis. All analyses were performed using SAS version 9.3 (SAS Institute Inc., Cary, NC).

      During the study period, a total of 1106 women with breast cancer were eligible for analysis. The median age was 59 years (39-95 years). The majority of women were Caucasian (74%). Of these, 976 (88%) women had natural menopause with no HRT, 63 (6%) underwent a BSO with no HRT, 51 (5%) had a BSO with previous history of HRT, and 16 (1%) had a BSO and were using HRT at time of cancer diagnosis. Patients who used HRT were using combined estrogen and progesterone (63%) or estrogen only (25%) (Table 1). There was no statistically significant difference in age of diagnosis between the groups. However, we found that women in the BSO groups underwent menopause at earlier ages (47,46,47 years) compared to the natural menopause group (51 years) (Table 1). The majority of patients (73% and 94% in previous and current users respectively), reported use of HRT for >1 year.

Table 1

Table 1: Clinical Characteristics.

      Though not statistically significant, women with BSO and current HRT use had more dense breasts (69% heterogeneously or extremely dense) compared to women who had natural menopause with no HRT (44% heterogeneously or extremely dense). However, women with BSO and past HRT use had significantly less dense breasts (67%) compared to women who underwent natural menopause without HRT usage (56%) (p=0.007).

      Women with BSO and no HRT use had a significantly higher proportion of mammographically occult malignancies compared to patients in other groups (p=.0006).In this group, only 49% of patients had malignancies detected on mammogram and 29% were detected with self or clinical breast exam, which is significantly different from the other groups (p=.0001).

      We found no statistical difference between all patients for tumor characteristics including stage of diagnosis, histology and hormone receptor status among the four groups (Table 2). Patients with BSO and no HRT usage were significantly more likely to have BRCA mutation than patients without BSO or HRT usage and were more likely to have used chemoprevention prior to malignancy diagnosis. The majority of patients were diagnosed with early stage 0,1 disease (65-82%).

Table 2

Table 2: Tumor Characteristics.

      In our study population, women with current HRT usage at time of cancer diagnosis had a higher breast density compared to women who had never reported use of HRT. HRT use in women with natural menopause has been shown to result in an increase in mammographic breast density [9-12]. Even in patients with BSO, the trend was persistent towards increased breast density with use of exogenous hormonal therapy. This is a surprising finding, as the doses of exogenous hormone given to women with BSO are much lower than basal levels of ovarian hormone production would have been had surgical menopause not occurred [11].

      However, women with BSO and previous HRT use had remarkably less dense breast tissue than patients who had natural menopause and never used HRT. This surprising finding could indicate that the effects of HRT in this patient population did not persist past the discontinuation of this therapy. Indeed, past studies on patients who took HRT for natural menopause have shown the effects on mammographic density did not persist at 5 years from time of discontinuation [1].

      Surprisingly, despite increased breast density, patients who used HRT did not have more mammographically occult malignancies than patients who did not use HRT. Additionally, patients had malignancies which did not differ in histology or stage irrespective of past HRT usage. This may indicate that despite the use of HRT, patients are not at higher risk for more advanced stage malignancies. This is a surprising finding as the 2003 WHI trial indicated that HRT users are more likely to be diagnosed at later stages [9]. This may reflect an increased commitment to screening in our HRT patient population. Hormone receptor positivity was also similar for patients who had ever taken HRT and those who had never taken HRT.

      This study has several limitations since it is a retrospective review. We did not have available information for many patients on time between HRT discontinuation and malignancy diagnosis, which may play a role in mammographic density. This study was no sufficiently powered to examine the effects of race and BMI on mammographic breast density.

      In our study cohort, women with history of surgical menopause who were on HRT at the time of breast cancer diagnosis had a trend towards increased breast density. Surprisingly, women who were prior users of HRT who were not on HRT at the time of malignancy diagnosis had less dense breasts than women who never used HRT. Despite the change observed in breast density, the tumor characteristics do not differ with HRT usage, suggesting that variability in breast density did not correlate with stage of disease at presentation. Further research is warranted in looking at the effects of HRT usage in the population of women with BSO and the impact on long term outcomes, including breast cancer recurrences.