Hormone replacement therapy and mammographic density: a systematic literature review

Purpose Hormone replacement therapy (HRT) is used to reduce climacteric symptoms of menopause and prevent osteoporosis; however, it increases risk of breast cancer. Mammographic density (MD) is also a strong risk factor for breast cancer. We conducted this review to investigate the association between HRT use and MD and to assess the effect of different HRT regimens on MD. Methods Two of authors examined articles published between 2002 and 2019 from PubMed, Embase, and OVID using Covidence systematic review platform. Any disagreements were discussed until consensus was reached. The protocol used in this review was created in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). Quality of each eligible study was assessed using the Oxford Center for Evidence-Based Medicine (OCEBM) hierarchy. Results Twenty-two studies met the inclusion criteria. Six studies showed that using estrogen plus progestin (E + P) HRT was associated with higher MD than estrogen alone. Four studies reported that continuous estrogen plus progestin (CEP) users had higher MD than sequential estrogen plus progestin (SEP) and estrogen alone users. However, two studies showed that SEP users had slightly higher MD than CEP users and estrogen alone users. Conclusions Epidemiological evidence is rather consistent suggesting that there is a positive association between HRT use and MD with the highest increase in MD among current users, and CEP users. Our results suggest that due to increase in MD and masking effect, current E + P users may require additional screening procedures, shorter screening intervals, or using advanced imaging techniques. Electronic supplementary material The online version of this article (10.1007/s10549-020-05744-w) contains supplementary material, which is available to authorized users.


Introduction
Hormone replacement therapy (HRT) is recognized as an effective treatment for alleviating the climacteric symptoms of menopause such as hot flushes, sleeping disturbance, depressive mood, muscle and joint pain [1]. Large clinical trials showed that use of HRT prevents bone loss and decreases risk of osteoporosis and bone fractures in menopausal women [2,3]. The most common HRT regimens are: estrogen alone, combined estrogen plus progestin (E + P) either as continuous estrogen plus progestin (CEP) or sequential estrogen plus progestin (SEP) [4]. For many years, HRT was used widely to improve the quality of life of menopausal women. However, after the results from two large population-based studies, the Women's Health Initiative (WHI) study in the United States in 2003 [5] and Million Women Study (MWS) in the Unites Kingdom in 2003 [6] showing that HRT use increases risk of breast cancer and cardiovascular disease, the long-term benefits and potential adverse effects of HRT on menopausal women were reconsidered. After the results from these two studies, new guidelines concerning amount, types, and duration of HRT for menopausal women were released, leading to a decline in use of HRT along with a reduction in the rates of breast cancer incidence by 6.7% among American women. After 2003, dramatic decline in HRT consumption and breast cancer incidence were recognized as the consequence of the findings from WHI and MWS studies [7].
Mammographic density (MD) is a very strong predictor for breast cancer risk [8]. MD refers to the amount of radiologically dense breast consisting of epithelial or stromal tissue that appears light on a mammogram, whereas fat tissue appears dark on a mammogram [9]. There are different methods for measuring MD including percent mammographic density (PMD), Breast Imaging Reporting and Data System (BI-RADS), and Wolfe [10]. Women with very dense breasts (> 75% density in the breast) have a four to six times greater risk of breast cancer than women with little density (< 5% density) or fatty breasts [8,9]. Previous studies suggested that MD is influenced by several exogenous hormones which are known to influence breast cancer risk thus, MD is known an important surrogate marker for the effects of exogenous hormones on the risk of breast cancer [11,12]. This review used a systematic approach to explore the association between HRT use and MD. Furthermore, we investigated the effect of different HRT exposure states (never, former, current use) as well as different HRT regimens on MD.

Inclusion and exclusion criteria
Studies included in this review met the following criteria: original research in peer-reviewed journals, full-text available online, a randomized controlled trial (RCT) as the study design, a cohort (prospective-cohort or retrospective cohort), a case-control, or a cross-sectional with a clear description of the samples and methodology, and articles available in English language. We focused on studies that examined the associations between HRT use and MD. Additionally, the MD assessment criteria in the studies had to be based on either Wolfe, PMD, or BI-RADS category. We excluded any descriptive manuscripts which did not have the focus on the association between HRT and MD.

Search strategy
Epidemiological studies from July 2002 to 2019 were retrieved from the following databases: PubMed Central (US National Institutes of Health [NIH]), OVID, and Embase using the following combinations of MeSH terms: "hormone replacement therapy", "postmenopausal hormone replacement therapy", "estrogen-progestin hormone replacement therapy", "combined hormone replacement therapy", "HRT", and "estrogen alone hormone replacement therapy" in conjugation with "mammographic density" and "breast density". An example of the search strategy used in one of the search databases (PubMed) is found in "Appendix".

Study screening
We imported and managed all study citations identified from the search strategy using the Covidence systematic review platform [13]. Two reviewers (S.A, K.J) independently screened the titles, abstract and reviewed the bibliography of articles found through electronic search engines for eligibility. Two pairs of co-authors reviewed all the abstracts. Disagreements occurred in less than 5% of all articles; any disagreements were discussed until consensus was reached.

Review protocol
The protocol used in this systematic review was created in accordance with the Preferred Reporting Items for Systematic reviews and Meta-Analyses 2009 (PRISMA) statement and flowchart was used [14]. Supplementary Table 1 illustrates the PRISMA checklist of this systematic review. Furthermore, in this literature review, quality of each eligible study was assessed using the Oxford Centre for Evidence-based Medicine -Levels of Evidence (OCEBM) hierarchy [15]. The OCEBM levels is a widely used system, which categories studies into different levels ranging from 1 to 5 based on their study designs; it helps the researcher to evaluate the reported results. In The OCEBM system, levels 1, 2, 3, 4 and 5 represent well-designed and high quality RCTs, prospective and retrospective cohort studies, case-control studies, cross-sectional and case-series studies, and expert opinion or unpublished clinical observations, respectively [15]. Level 1 is the highest quality and level 5 is the lowest [15] Data extraction Data extraction included the information about author and year of publication, country, study design, quality of study design, sample size and characteristics of the participants, HRT regimens, MD assessment, confounders included in the final analysis, final results and findings are included in Table 1 from each study. The term "progestin" has been used to replace all progesterone synthetic names, such as progestogen and progesterone.

Ever and never HRT users
From all 22 studies, there were n = 12 (54.5%) which investigated the association between ever and never use of HRT and MD [4, 20-22, 26, 27, 31-33, 36]. All these studies reported a significant increase in MD among ever HRT users -29 citations were not full-text -10 citations were review papers -25 studies were not RCT, cohort, case control, or cross-sectional -9 studies used different methods than BI-RADS, PMD, and Wolfe for assessment of MD 22 articles selected by eligibility criteria and included in the systematic review compare to never users; however, the results from two studies did not reach the statistical significance (Table 1) [22,25]. In a RCT of 695 postmenopausal women, at 1-year follow-up the PMD change from the baseline was 1.4% for estrogen users and -0.8% for placebo group with a significant increase in MD among women assigned to HRT than women assigned to placebo [20]. In a case-control study, Harvey et al. showed that there is a statistically significant difference in MD among HRT ever and never users groups (p-value < 0.001) with a median difference in MD 54% for ever HER users and 31% for never users. They reported that percent fibrous stroma was 7% higher for ever users compared to never users; however, the result was not statistically significant [24]. In another case-control study, Jeon et al. found positive association between HRT and MD. The odds of having dense breasts increased by 2.13; 95% CI (1.09-4.16) for women who used HRT compared to never users [32].

Former, current and never HRT users
Five studies examined the association between HRT use and MD among former, current and never HRT users [16,17,19,23,28]. Three studies showed that MD increased in current HRT users compared to former and never users. Aiello et al. in a cross-sectional study of 39,296 women reported that odds of having high MD increased significantly in women who were the current HRT users (OR 1.91; 1.81-2.00) and former (OR 1.14; 1.08-1.21) as compared to never users [28]. Couto et al. have in a cross-sectional study of 2,424 postmenopausal women found that PMD was highest in current HRT users followed by former and never users with breast densities of 22.6%, 17.7%, and 16.3%, respectively [19]. Additionally, Gapstur et al. in a small cross-sectional study of 296 Hispanic women found that mean PMD was significantly higher in current HRT users than in former and never HRT users: 18.2%, 14%, and 14% (p-value = 0.02), respectively [23].
In a case-control study Boyed et al. showed that mean PMD among cases was greater in current HRT users than in never users (difference, 6%; p-value < 0.001) and greater in former users than never users (difference, 3.4% p-value 0.03). Among controls, current use of HRT was associated with a slightly greater mean PMD; however, the results were not statistically significant [17]. In another case control study, Boyd et al. found that among cases, mean PMD was greater in current (25.1%) and former (28.1%) HRT users than in never users (30.7%; p-value < 0.001). However, among controls the results were not statistically significant [16].
Finally, a study which examined the effect of HRT on MD among postmenopausal women who previously used HRT and women who never used HRT and that postmenopausal women with history of HRT use had increased odds of having higher MD (OR 2.1; 1.4-3.3) compared to never HRT users [35].

Estrogen alone versus E + P
The results from all five studies showed that E + P users had higher MD compared to estrogen alone users [19,20,27,28,30]. In a RCT of 695 postmenopausal women, Crandall et al. found that at 1-year follow-up changes in PMD were statistically greater among women assigned to E + P than women assigned to estrogen alone (p-value = 0.001) [20]. A cohort study of 1240 postmenopausal women showed that PMD at the first and second mammograms were higher in E + P users than in estrogen alone and never HRT users [27]. Additionally, in a large cross-sectional study of 39,296 postmenopausal women, Aiello et al. reported that current E + P users had significantly increased odds of having dense breasts (OR 1.98; 1.87-2.09) followed by estrogen alone users (OR 1.71, 1.95-1.87) compared to never HRT users [28]. Couto et al. have in a cross-sectional study showed that current E + P users had significantly higher PMD (25.4%) than current estrogen (18.9%) and never users (16.3%) [19]. Finally a small retrospective study of 467 postmenopausal study showed that women using E + P more than 4 years had significantly higher mean MD compared to women who used estrogen only (p-value = 0.01); however, after adjustment for other variables the association between MD and HRT regimens diminished [30].

Estrogen alone versus CEP and SEP
From six studies which compared the association between estrogen alone, CEP, and SEP regimens on MD [21,22,29,31,34,36], four studies showed that CEP users had significantly higher MD compared to SEP and estrogen alone users [29,31,34,36]. In a RCT study of 571 postmenopausal women randomly assigned to receive placebo, daily estrogen alone, SEP, or CEP, after 12 months of follow-up the absolute mean increase in PMD was observed in CEP and SEP users; 4.76% and 4.48%, respectively. No changes in MD were observed in estrogen alone users compared to never HRT users [31]. Additionally, in a retrospective study of 113 of healthy postmenopausal women, Topal et al. showed that 38.3% of CEP users, 12.5% of SEP users and 2.7% of estrogen alone users had increased in MD. Increase in MD was more pronounced among CEP users than other regimens of HRT (p-value = 0.009) [34]. Finally, Carmona-Sanchez et al. reported that after 5 years of HRT use, MD increased significantly in 21.8% women receiving SEP versus 38.8% of women receiving CEP (p-value = 0.039) [29].
Two studies showed that SEP users had slightly higher MD compared to CEP and estrogen alone users. In a RCT study by Crandall et al. at 12 months, PMD was significantly increased in SEP users by 4.6% and in CEP users by 4.4% [22]. In another RCT by Crandall et al. increase in PMD was 1.2% among estrogen users, 4.7% among CEP users, and 4.9% among SEP users [21].

CEP users versus never HRT users
Three studies addressed the increase in MD among CEP users compared to never HRT users. In a RCT study of 103 postmenopausal women, after 12 months 45.1% of CEP users had increased in MD, whereas no changes was observed among never HRT users. The difference between CEP users and never users were statistically significant (p-value < 0.001). In another RCT study of 413 postmenopausal women, McTiernan et al. reported an increase in mean PMD by 6% and decrease in MD in placebo group (p-value < 0.001) after 1-year follow-up. Finally a case-control study by Byrne et al. showed that women assigned to CEP group had a larger and boarder distribution of MD change (mean change = 9.49%) compared to women in never HRT/placebo group. Women assigned to placebo group experienced decrease in MD over 1 year (mean change = − 0.65%) after adjusting for possible confounders [18].

Discussion
Our review is the first to systematically investigate the association between HRT and MD, comparing different status of HRT exposure and MD, and finally reporting the effect of different HRT regimens on MD. A consistent finding in the literature is that MD was increased in ever HRT users compared to never users. Furthermore, the highest increase in MD was observed in current HRT users compared to never users [16,17,19,23,28]; in line with previous studies published before 2003 [37,38]. Marugg et al. found that, compared to never HRT users, 14.3% of women using HRT showed an increase in MD [37]. Another study concluded that current HRT users were more than twice as likely to have higher MD as never users (OR 2.48; 1.32-4.16) [38].
All selected studies which compared the effect of estrogen alone and E + P on MD showed that E + P users had a higher MD compared to estrogen alone users [19,20,27,28,30]. Similar to this result, previous studies showed a strong positive association between E + P use and increase in MD [37,39]. According to Marugg et al. 31% of women treated with E + P showed an increase in MD compared with only 8.7% in the group treated with estrogen alone [37]. Vachon et al. found that odds of having higher MD increased in women using E + P by 1.9 compared to women using estrogen alone [39].
Among the studies which provided data on the sub-type of HRT regimens (CEP, SEP and estrogen alone), four studies reported that women who used CEP, where both estrogen and progestin are taken daily, had higher MD than women who used SEP, where estrogen is used daily but progestin is taken only during a certain time of the month and estrogen alone users [29,31,34,36]. These results are in agreement with the results from other studies that examined the association between HRT and MD with respect to sub-types HRT regimens [40,41]. In a Swedish study of 31,498 women, Persson et al. has found that MD was significantly increased in 28% of CEP users, 10% of SEP and 5% of estrogen alone users [41]. Another Swedish study by Lundstrom et al. reported that MD was greater among CEP users (52%) than SEP users (13%), estrogen alone users (18%) over 2 years of follow-up [40]. However, two studies showed that SEP users had slightly higher MD compared to CEP and estrogen alone users [21,22]. It is important to mention that, different HRT regimens influence MD change differently but the mechanism for MD change among CEP and SEP users remain to be unclear. Lundstrom et al. claimed that the inconsistent result on the association between CEP and SEP users with MD maybe due to variation in progestin components, dosage, and duration of administration [40].
The biological explanation in increasing MD with respect to HRT use is not yet fully understood. However, one established hypothesis to explain the increase of MD in relation to use of HRT is based on the breast cell proliferation theory. In menstruating women, it has been observed that breast epithelial cell proliferation is increasing due to high levels of estrogen and progestin [42]. In a cross-sectional study of 56 pre-and 86 postmenopausal women, Hofseth et al. found that, use of HRT especially E + P is associated with higher level of breast epithelial cell proliferation in post-and premenopausal women compared to never HRT users [43]. Breast epithelial cell proliferation is also known as epithelial hyperplasia which defines as abnormal growth and accumulation of cells that line the ducts or the lobules in the breasts [44]. Hofseth et al. also found that breast epithelial cell density was significantly higher in women using HRT, especially in E + P users than estrogen alone and never HRT users [43]. Another hypothesis to explain the association between increasing MD with respect to use 1 3 of E + P is the stromal oedema theory. Stroma is the major tissue in the breasts therefor any changes in MD primarily reflect alteration of the stroma architecture and composition [44]. Longacre et al. found that stromal oedema is greater in the luteal phase of menstrual cycle which progestin is the main hormone associated with this stage [45]. Therefore receiving E + P should lead to greater MD than estrogen alone.
To our knowledge, this is the first systematic review, which studied not only the effect of different states of HRT exposure on MD but also the effect of different HRT regimens on MD. Strengths of our study are all studies included in this review appropriately reported their study design (RCT, cohort, case-control, and cross-sectional), inclusion criteria, and sampling methods. Sample size was quite large in observational studies and adequate samples were also reported in several RCT studies. Ethical considerations were reported according to the international standards in 18 papers [4, 16-24, 26-28, 31-33, 35, 36]. Most of studies addressed potential confounders, biases and a discussion of limitations. To minimize bias, in this review we used PRISMA checklist and follow chart to ensure transparency and completeness of the reporting (Supplementary material 1). In addition, we assess the quality of each eligible study based on their study designs using the OCEBM hierarchy. Another important strength of this review is that the results can be generalized since the selected studies were from different geographical regions (North America, Europe, and Asia) and these studies presented diverse ethnic groups (Asian, Hispanic, and White). Finally, this review included only peer-reviewed studies, since including unpublished and gray literature increase the risk of publication bias due to the absence of peer-review and low methodological quality. Moreover, the authors did not include their personal opinion or prior knowledge during the review process in order to avoid publication bias.
There are some limitations regarding the studies included in this review. By far the most common limitations mentioned in included studies were; misclassification in the assessment of MD, small sample size, unknown HRT composition, and lack of racial heterogeneity. Other limitations regarding this study are the search for the qualified articles was conducted only in English language, only articles that were accessible electronically were included, and this review found studies with different methodological designs, sample size and demographic factors, therefore due to methodological heterogeneity between studies included in this review it was not possible to conducted a meta-analysis.

Conclusions
In conclusion, this review showed that MD significantly increased in ever HRT users compared to never users with highest increase in MD among current HRT users. Furthermore, this review found that E + P users had a higher MD compared to estrogen alone users. Results with regards to HRT regimens and MD showed that CEP users had the highest increase in MD followed by SEP users and estrogen alone.
The findings from this systematic review on the association between HRT use and MD can be used in primary prevention of breast cancer incidence as well as secondary prevention of false-negative diagnosis of small tumours. From the primary prevention perspective, our results suggest that the use of HRT should be minimized at the lowest does needed for as short time as possible. In addition, from the secondary prevention perspective, women who are current HRT users and increase in MD is detected by mammography screening may require additional screening procedures, shorter screening intervals, and using advanced imaging techniques such as MRI/ultrasound for detecting small tumours.