Abstract
Background
Postmenopausal hormone therapy (HT) cessation is associated with a decrease in mammographic density (MD), but it is unknown whether this effect is modified by woman-level characteristics. We investigated whether we could identify characteristics of women who were most likely to experience a decrease in MD due to HT cessation.
Methods
Postmenopausal HT users with a prior screening mammogram (n=1,168) were randomized to continue HT, or to suspend HT for 1 month or 2 months before their next screening mammogram. We estimated relative risks (RR) and attributable risks with 95% confidence intervals (CI) of a ≥7.5% decrease in percentage MD (%MD) versus no change associated with HT cessation, stratified by age, BMI, parity, and other factors.
Results
HT cessation increased a woman’s likelihood of experiencing a ≥7.5% decrease in %MD by 30% (95% CI=1.03–1.7), but we found little evidence of effect modification by age, race, BMI, change in BMI, baseline %MD, parity, family history of breast cancer, HT type, or duration of HT use.
Conclusions
Woman-level factors do not appear to explain why some women experience a decrease in %MD following HT cessation and others do not.
Impact
We were unable to identify subgroups of women who are more likely to experience a decrease in MD due to HT cessation; other factors, such as genetic factors, may be important determinants of HT-related changes in MD.
Keywords: breast density, mammographic density, hormone therapy, cessation, breast cancer
Introduction
Mammographic breast density (MD) is a strong marker of breast cancer risk (1). Postmenopausal hormone therapy (HT) increases MD (2,3), particularly combined HT (2), which also increases breast cancer risk (4). Postmenopausal HT cessation decreases MD in some women (5); reasons for this are not well understood. We examined whether the association between HT cessation and decrease in percentage MD (%MD) is modified by woman-level factors related to MD, including age, race, BMI, change in BMI between mammograms, baseline %MD, parity, and family history of breast cancer, as well as HT characteristics (type, duration of use) (2,3,5). The purpose of this study was to gain insight into the relationship between HT and MD change and to identify subgroups of women who might be more likely to decrease due to HT cessation.
Methods
Methods used in the Radiological Evaluation and Breast Density (READ) randomized trial have been described previously (5). Briefly, we identified women ages 45–80 who reported HT use at a screening (index) mammogram conducted at Group Health Cooperative within the past 24 months. Women due for another screening (study) mammogram who had filled ≥2 HT prescriptions in the 6 months before recruitment were eligible. Of those who completed the study mammogram (n=1,471), we excluded women with a baseline %MD of <10% (n=290), as well as women missing information on BMI (n=13). Study methods were approved by the Institutional Review Boards of Group Health and the US Department of Defense.
Women randomized to 1 month or 2 months of HT cessation were considered “exposed” (n=738); those randomized to continue HT were “unexposed” (n=430). We gathered data on potential effect modifiers (age, race, BMI, change in BMI between mammograms, baseline %MD, parity, and family history of breast cancer, HT type, duration of HT use) from automated databases and study questionnaires (5). We defined change in BMI between mammograms as a change of ≥1.5 units (kg/m2), corresponding to a weight change of approximately 7–11 pounds for height 4′11″-6′0″. We used Cumulus software to measure dense area and total breast area in digitized left breast craniocaudal projections; %MD was calculated as dense area divided by total area (5).
We report median change in %MD between mammograms with 95% confidence intervals (CI) by woman-level characteristics, separately by exposure status. We used generalized linear models (GLM) with a log link and a robust variance estimator to estimate relative risks (RRs) of decreasing %MD vs. no change associated with HT cessation (see definitions of change below). We used GLM with an identity link to estimate attributable risks (ARs). We report RRs and ARs with 95% CIs by subgroup of each potential effect modifier to assess whether certain women were more likely to experience a decrease in %MD due to cessation. To estimate RR and AR, we compared women who decreased ≥7.5% MD versus women with no change (defined as an absolute change of ≤3% MD). A 7.5% decrease in %MD was approximately the median level of decrease among women whose %MD decreased (>3%), and may represent a clinically relevant level of decrease. Studies suggest that women with a given %MD have a 15–20% higher risk of breast cancer compared to women whose %MD is lower by 10% (6,7). All analyses were conducted using STATA, v.11.1 (StataCorp. College Station, TX).
Results
Women randomized to suspend HT had a greater median decrease in %MD between mammograms than continued users (−2.4% versus −0.7%), and a greater proportion who decreased by ≥7.5% MD (21% vs 15%) (Table 1). Overall, HT cessation was associated with a 30% increased likelihood of a ≥7.5% MD decrease vs. no change (RR=1.3, 95% CI=1.03–1.7), and an attributable risk of 9% (95% CI=2–17%). We found little evidence of effect modification by the factors studied, on either a multiplicative or an additive scale (Table 2).
Table 1.
Change in percent mammographic density (%MD) for continued-use and HT cessation groups by demographic and reproductive factors.
| Continued-use group | HT cessation group | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Median change in %MD | N | Outcome: %MD change: Row percent
|
Median change in %MD | N | Outcome: %MD change: Row percent
|
|||||||
| Decrease | No change | Increase | Decrease | No change | Increase | |||||||
| % (95% CI) | ≥ 7.5% | >3 to < 7.5% | ±3% | >3% | % (95% CI) | ≥7.5% | >3 to <7.5% | ±3% | >3% | |||
| Overall | −0.7 (−1.3, −0.2) | 430 | 15 | 20 | 36 | 29 | −2.4 (−2.7,−1.4) | 738 | 21 | 24 | 32 | 23 |
|
| ||||||||||||
| Age, years | ||||||||||||
| 45–55 | −0.6 (−1.8, 0.5) | 116 | 17 | 17 | 35 | 30 | −3.4 (−4.4, −2.4) | 199 | 26 | 25 | 28 | 20 |
| 56–65 | −0.8 (−1.8, 0.6) | 214 | 15 | 21 | 35 | 29 | −1.7 (−2.6, −0.7) | 381 | 18 | 24 | 32 | 25 |
| 66+ | −0.6 (−1.9, 0.9) | 100 | 13 | 21 | 39 | 27 | −1.4 (−3.1, −0.3) | 158 | 19 | 24 | 36 | 21 |
| Caucasian | ||||||||||||
| No | −0.2 (−3.0, 3.1) | 31 | 16 | 16 | 32 | 35 | 0.7 (−2.2, 3.0) | 70 | 17 | 20 | 24 | 39 |
| Yes | −0.8 (−1.4, −0.2) | 398 | 15 | 21 | 36 | 28 | −2.4 (−3.0, −1.7) | 664 | 21 | 25 | 33 | 22 |
| Baseline BMI (kg/m2) | ||||||||||||
| <25 | −0.2 (−1.4, 1.0) | 191 | 17 | 15 | 34 | 34 | −1.6 (−2.6, −0.3) | 311 | 20 | 22 | 31 | 27 |
| 25–30 | −0.7 (−1.7, −0.1) | 147 | 12 | 23 | 39 | 26 | −2.0 (−3.0, −0.7) | 244 | 20 | 24 | 32 | 24 |
| >30 | −1.4 (−3.5, 0.05) | 92 | 17 | 25 | 35 | 23 | −3.4 (−4.4, −2.0) | 183 | 22 | 29 | 33 | 16 |
| Baseline %MD | ||||||||||||
| 10–<20% | −0.3 (−1.4, 0.6) | 137 | 4 | 21 | 50 | 26 | −0.8 (−1.8, 0.3) | 241 | 9 | 27 | 40 | 24 |
| 20–<30% | 0.2 (−1.0, 2.0) | 119 | 12 | 19 | 34 | 35 | −1.7 (−2.5, −0.6) | 201 | 17 | 24 | 36 | 23 |
| 30–<40% | −2.6 (−4.4, −0.1) | 87 | 25 | 23 | 26 | 25 | −3.4 (−4.8, −2.0) | 152 | 28 | 25 | 26 | 20 |
| 40+ % | −2.6 (−4.9, −0.3) | 87 | 29 | 16 | 26 | 29 | −5.1 (−7.0, −2.7) | 144 | 38 | 20 | 18 | 24 |
| Parity | ||||||||||||
| Parous | −0.9 (−1.5, −0.2) | 347 | 16 | 20 | 36 | 27 | −2.3 (−2.9, −1.5) | 583 | 21 | 25 | 32 | 23 |
| Nulliparous | −0.2 (−1.1, 1.8) | 82 | 12 | 18 | 35 | 34 | −1.5 (−3.2, −0.3) | 153 | 20 | 24 | 33 | 24 |
| Family history of Breast Cancer (1st degree) | ||||||||||||
| No | −0.4 (−0.9, 0.2) | 330 | 16 | 18 | 36 | 30 | −2.3 (−3.0, −1.5) | 589 | 21 | 24 | 31 | 23 |
| Yes | −2.2 (−3.4, −0.4) | 89 | 12 | 30 | 31 | 26 | −1.7 (−3.2, −0.5) | 130 | 19 | 24 | 35 | 22 |
| Change in BMI (kg/m2) | ||||||||||||
| Decrease>1.5 | 1.1 (−3.4, 4.3) | 37 | 11 | 24 | 30 | 35 | −0.9 (−3.6, 0.2) | 64 | 16 | 23 | 38 | 23 |
| No change (≤1.5) | −0.4 (−1.4, 0.4) | 286 | 16 | 18 | 35 | 31 | −2.0 (−2.6, −1.1) | 486 | 21 | 24 | 30 | 26 |
| Increase >1.5 | −1.8 (−3.5, −0.7) | 84 | 17 | 24 | 39 | 20 | −2.9 (−4.1, −1.7) | 159 | 22 | 27 | 33 | 18 |
| HT Type | ||||||||||||
| ET | −0.9 (−1.7, −0.2) | 264 | 16 | 22 | 35 | 27 | −2.1 (−2.7, −1.1) | 450 | 19 | 25 | 33 | 23 |
| E+P | −0.2 (−1.4, 0.8) | 166 | 14 | 16 | 37 | 32 | −2.4 (−3.4, −1.2) | 288 | 24 | 24 | 30 | 23 |
| Duration of prior HT use at baseline | ||||||||||||
| <5 years | 0.5 (−1.1, −2.5) | 57 | 16 | 14 | 37 | 33 | −1.7 (−5.1, 1.7) | 86 | 31 | 13 | 24 | 31 |
| 5–9 years | −1.2 (−2.8, 0.5) | 93 | 18 | 18 | 35 | 28 | −2.7 (−3.9, −1.6) | 159 | 23 | 25 | 35 | 17 |
| 10–14 yrs | −0.7 (−1.8, 2.0) | 83 | 14 | 20 | 29 | 36 | −2.2 (−3.5, −1.0) | 167 | 17 | 28 | 32 | 23 |
| 15+ years | −0.9 (−1.6, 0.2) | 173 | 14 | 22 | 38 | 25 | −2.0 (−3.1, −0.6) | 277 | 20 | 25 | 33 | 22 |
Row percents may not add to 100 because of rounding to the nearest 1%; N’s may not sum to totals due to missing values
The study had 88% power to detect a 7% change in density among exposed women, if there were no change among unexposed
Table 2.
Effect of HT cessation on likelihood of a decrease in %MD between mammograms: Relative risk (RR) and attributable risk (AR) of a decrease in %MD of ≥7.5% vs. no change (± 3%), overall and by reproductive and demographic subgroups.
| Continued HT use (referent group) | 1- to 2-month HT cessation | 1- to 2-month vs. no HT cessation | ||||
|---|---|---|---|---|---|---|
| No change n (%)a | Decrease ≥7.5% n (%) | No change n (%) | Decrease ≥7.5% n (%) | RRb (95%CI) | ARb,c (95% CI) | |
| Overall Effect | 154 (100) | 66 (100) | 235 (100) | 152 (100) | 1.3 (1.03, 1.7) | 9% (2, 17) |
| Age (years) | ||||||
| 66+ | 39 (25) | 20 (30) | 57 (24) | 30 (20) | 1.5 (1.0, 2.2) | 9% (−6, 25) |
| 56–65 | 74 (48) | 33 (50) | 122 (52) | 70 (46) | 1.2 (0.8, 1.7) | 6% (−5, 17) |
| 45–55 | 41 (27) | 20 (30) | 56 (24) | 52 (34) | 1.4 (0.8, 2.4) | 15% (0, 30) |
| BMI (kg/m2)d | ||||||
| <25 | 65 (42) | 32 (48) | 96 (41) | 54 (40) | 1.2 (0.9, 1.7) | 7% (−5, 19) |
| 25–30 | 57 (37) | 18 (27) | 79 (34) | 48 (32) | 1.6 (1.0, 2.5) | 14% (1, 27) |
| >30 | 32 (21) | 16 (24) | 60 (26) | 41 (27) | 1.2 (0.8, 1.9) | 7% (−9, 24) |
| Baseline %MDd | ||||||
| 10–<20% | 68 (44) | 5 (8) | 97 (41) | 21 (14) | 2.6 (1.02, 6.6) | 11% (2, 20) |
| 20–<30% | 40 (26) | 14 (21) | 72 (31) | 34 (22) | 1.2 (0.7, 2.1) | 6% (−9, 21) |
| 30–<40% | 23 (15) | 22 (33) | 40 (17) | 43 (28) | 1.1 (0.7, 1.5) | 3% (−15, 21) |
| 40+ % | 23 (15) | 25 (38) | 26 (11) | 54 (36) | 1.3 (0.9, 1.8) | 15% (−2, 33) |
| Parity | ||||||
| Parous | 125 (81) | 56 (85) | 185 (79) | 121 (80) | 1.3 (1.0, 1.7) | 9% (0, 17) |
| Nulliparous | 29 (19) | 10 (15) | 50 (21) | 31 (20) | 1.5 (0.8, 2.7) | 13% (−5, 30) |
| Family history of breast cancer (1st degree) | ||||||
| Yes | 28 (18) | 11 (17) | 45 (19) | 25 (16) | 1.3 (0.7, 2.3) | 10% (1, 19) |
| No | 120 (78) | 52 (79) | 185 (79) | 125 (82) | 1.3 (1.0–1.7) | 8% (−11, 26) |
| Change in BMI | ||||||
| Decrease > 1.5 | 11 (7) | 4 (6) | 24 (10) | 10 (7) | 1.1 (0.4–3.0) | 3% (−24, 30) |
| No change (≤1.5) | 99 (64) | 46 (70) | 145 (62) | 100 (66) | 1.3 (1.0–1.7) | 9% (0, 19) |
| Increase > 1.5 | 33 (21) | 14 (21) | 53 (23) | 35 (23) | 1.3 (0.8–2.2) | 10% (−7, 27) |
| HT type | ||||||
| ET | 92 (60) | 42 (64) | 149 (63) | 84 (55) | 1.2 (0.8, 1.6) | 5% (−5, 15) |
| E+ P | 62 (40) | 24 (36) | 86 (37) | 68 (45) | 1.6 (1.1, 2.3) | 16% (4, 29) |
| HT duratione | ||||||
| <5 years | 21 (14) | 9 (14) | 21 (9) | 27 (18) | 1.9 (1.0, 3.4) | 26% (5, 48) |
| 5–9 years | 33 (21) | 17 (26) | 55 (23) | 37 (24) | 1.2 (0.7, 1.9) | 6% (−10, 23) |
| 10–14 yrs | 24 (16) | 12 (18) | 54 (23) | 28 (18) | 1.0 (0.6, 1.8) | 1% (−18, 19) |
| 15+ years | 66 (43) | 25 (38) | 91 (39) | 55 (36) | 1.4 (0.9, 2.0) | 10% (2, 22) |
Percents may not add to 100 because of rounding to the nearest 1%
Unadjusted (Assignment to the exposure was randomized and is not associated with other factors (5))
AR: Cumulative incidence (proportion) of women in the cessation group who experienced a decrease in MD of ≥7.5% due to HT cessation
At index mammogram
Duration of prior HT use at baseline
Discussion
We did not observe strong evidence that the effect of HT cessation on %MD varies by woman-level characteristics. Our findings suggest that HT cessation may have a similar biological effect on %MD regardless of age, BMI, baseline %MD, and the other factors we examined. Although we used continuous measures of %MD, which are more precise than commonly used categorical measures, and a single trained reader interpreted all mammograms, measurement error of %MD remains a concern. Quality control sampling indicated that <10% of mammography re-reads differed by ≥7.5% MD. We were unable to measure %MD immediately before HT cessation. However, our comparison of two large, randomized groups allowed us to infer this change with reasonable confidence. Additional strengths include a large sample size, randomization of exposure, high compliance with the intervention (8), and high-quality administrative data. Our findings suggest that HT-cessation-associated decrease in %MD is not predicted by woman-level factors. Other explanations for why subsets of women experience %MD decreases due to HT cessation, such as genetic factors, should be considered.
Acknowledgments
Financial support: The READ study was funded by the Department of Defense (PI, D Buist; DAMD17-03-1-0447). Registered clinical trial number: NCT00117663. Study participants were recruited from the Group Health Breast Cancer Screening Program funded by the National Cancer Institute (PI: D Buist, U01CA63731).
This study could not have been completed without the assistance of Tammy Dodd, Linda Palmer, Dawn Fitzgibbons, Deborah Seger, Susan Reed, Katherine Newton or Melissa Rabelhofer. We would further like to thank members of our advisory board: Hermien Watkins, Paula Hoffman, Deb Schiro, and Margrit Schubiger; members of the Data Safety and Monitoring Board: Susan Heckbert, MD, PhD, Chair, University of Washington Department of Epidemiology; Ben Anderson, MD, University of Washington; Mary Anne Rossing, DVM, PhD, Fred Hutchinson Cancer Research Center; Robert D. Rosenberg, MD, University of New Mexico; Thomas Lumley, PhD, University of Washington Department of Biostatistics; and Elizabeth Lin, MD, Group Health Permanente medical monitor. We would further like to thank Robert Karl, MD, Donna White, MD and Jo Ellen Callahan for their support for implementing this trial at Group Health. We would also like to thank Noel Weiss, MD, MPH, and Jessica Chubak, PhD, MBHL, for providing feedback on methods and analyses. We further acknowledge Stephen Taplin, MD, MPH for his collaboration in getting this study funded when he was an investigator at Group Health Cooperative.
Footnotes
Conflicts of Interest: There are no conflicts of interest to disclose.
References
- 1.Boyd NF, Martin LJ, Bronskill M, Yaffe MJ, Duric N, Minkin S. Breast tissue composition and susceptibility to breast cancer. J Natl Cancer Inst. 2010;102(16):1224–37. doi: 10.1093/jnci/djq239. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.McTiernan A, Chlebowski RT, Martin C, Peck JD, Aragaki A, Pisano ED, et al. Conjugated equine estrogen influence on mammographic density in postmenopausal women in a substudy of the women’s health initiative randomized trial. J Clin Oncol. 2009;27(36):6135–43. doi: 10.1200/JCO.2008.21.7166. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.McTiernan A, Martin CF, Peck JD, Aragaki AK, Chlebowski RT, Pisano ED, et al. Estrogen-plus-progestin use and mammographic density in postmenopausal women: Women’s Health Initiative randomized trial. J Natl Cancer Inst. 2005;97(18):1366–76. doi: 10.1093/jnci/dji279. [DOI] [PubMed] [Google Scholar]
- 4.Chlebowski RT, Hendrix SL, Langer RD, Stefanick ML, Gass M, Lane D, et al. Influence of estrogen plus progestin on breast cancer and mammography in healthy postmenopausal women: the Women’s Health Initiative Randomized Trial. JAMA. 2003;289(24):3243–53. doi: 10.1001/jama.289.24.3243. [DOI] [PubMed] [Google Scholar]
- 5.Buist DS, Anderson ML, Reed SD, Aiello Bowles EJ, Fitzgibbons ED, Gandara JC, et al. Short-term hormone therapy suspension and mammography recall: a randomized trial. Ann Intern Med. 2009;150(11):752–65. doi: 10.7326/0003-4819-150-11-200906020-00003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Byrne C, Schairer C, Wolfe J, Parekh N, Salane M, Brinton LA, et al. Mammographic features and breast cancer risk: effects with time, age, and menopause status. J Natl Cancer Inst. 1995;87(21):1622–9. doi: 10.1093/jnci/87.21.1622. [DOI] [PubMed] [Google Scholar]
- 7.Boyd NF, Lockwood GA, Byng JW, Tritchler DL, Yaffe MJ. Mammographic densities and breast cancer risk. Cancer Epidemiol Biomarkers Prev. 1998;7(12):1133–44. [PubMed] [Google Scholar]
- 8.Newton KM, Anderson ML, Reed SD, Bowles EJ, Buist DS. Factors associated with non-compliance with hormone therapy cessation before screening mammography. Climacteric. 2011;14(2):268–74. doi: 10.3109/13697137.2010.520172. [DOI] [PubMed] [Google Scholar]