Effect of Preventive Hormonal Therapy on Breast Density: A Systematic Qualitative Review

Table 1

Tamoxifen (TAM): characteristics of the reviewed studies (n: 10).

Authors/year	Endpoint	Design	High risk definition	Population/number of cases	Age	Method of BD measurement/time	Mammograms readers	Reproducibility	BD baseline
Cuzick et al. 2004 [32]	Effect of tamoxifen on breast density	Subgroup retrospective analysis within IBIS 1 randomized double-blind study	Family history of BC or proliferative benign breast disease	7152 healthy women at high risk of BC/818 BC—free women (388 in the TAM group and 430 in the placebo group)	35–70 y	Visually according to the Boyd classification scale/at baseline and during 54 months of treatment, at 12- to 18-month intervals	A single consultant radiologist blinded for treatment group + specially trained research nurse	In a subset of mammograms from 70 women/correlation coefficients: 0.91 (baseline readings), 0.88 (18 months), and 0.74 (54 months)	Placebo: 42.6% (95% CI: 39.6%–45.6%); TAM: 41.9% (95% CI: 38.8%–45.0%)
Decensi et al. 2009 [33]	Effect of tamoxifen^1 and fenretinide (alone and in combination) on IGF-1 and breast density	Randomized, double-blind, placebo-controlled trial with a 2×2 factorial design	Intraepithelial neoplasia or small invasive BC in the 3 yrs before random assignment or a Gail 5-year risk for BC ≥ 1.3%	880 premenopausal women at high risk of BC/235 (TAM alone: 55; fenretinide alone: 56; combination: 59; placebo: 57)	TAM alone: 46.2 y ± 5.0 (range 32–57), fenretinide alone: 46.2 y ± 5.2 (range 30–56), combination: 46.9 y ± 4.5 (range 38–54), placebo: 46.5 y ± 4.3 (range 36–54)	Visually according to the Boyd classification scale and computer-assisted calculation/at baseline and at 12 and 24 months on analog screen films or digital scans	Single trained radiologist, blinded as to time sequence and treatment arm	NA	IEN/T1 ^2: 46.6% (SD ± 16.4%); Gail: 49.5% (SD ± 15.2%)^3
Chow et al. 2000 [34]	Effect of tamoxifen on breast density	Secondary analysis of a prospective safety and feasibility study	Diagnosis of DCIS, LCIS, or atypical hyperplasia or a Gail 5-year risk for BC ≥ 1.7%	32 women at high risk of BC/TAM group: 28 patients and control group: 20 patients from another protocol	TAM: mean 49.5 y (range 36–74 y) and controls: mean 51 y (range 42–57 y)	4 methods: (a) Wolfe and BIRADs; (b) Boyd classification scale; (c) computer aided calculation from digitized mammograms/4 days to 9 months (mean 2.3 months) before the start of TAM therapy and annual mammograms during 2 y	Two independent readers for methods (a) and (b); single reader for method (c)	High interobserver variability for (a) and (b). For (c): in a subset: differences between second set and first set: mean −2.2±6.9% (range −15.5% to 11.0%).	Initial digital scores: TAM: 31.9% (SD ± 19.0%) and controls: 29.7% (SD ± 12.0%)
Brisson et al. 2000 [35]	Effect of tamoxifen on breast density	Subgroup analysis within NSABP P1 randomized double-blind study	Diagnosis of LCIS or a Gail 5-year risk for BC ≥ 1.7%	184 healthy women at high risk of BC/69 (TAM group: 36; placebo: 33)	TAM: mean 50.9±8.0 y and placebo: mean 50.3 ± 6.4 y	Visually according to Wolfe’s parenchymal pattern and a 22-score percentage of the breast showing densities/pre- and posttreatment mammograms	Two independent blinded readers (third reading if necessary)	Weighted kappa comparing assessments (pre/posttreatment): 0.82/0.83. Intraobserver correlation coefficients: 0.92/0.89	Mean BD at randomization: TAM: 60.3%±18.5% and placebo:  60.5%±24.2%
Atkinson et al. 1999 [36]	Effect of tamoxifen on breast density	Retrospective case control study, 1 : 2 matching by age and menopausal status	BC	282 (cases: 94 BC patients who had received TAM and control group: 188 age matched cancer free women)	Mean 59 ± 3.3 y (range 50–64 y)	Visually according to Wolfe’s parenchymal pattern/at baseline and at 19.6 months (cases) and 14.3 months (controls)	A single reader not blinded	For a subset (13%) of mammograms intraobserver agreement. Correlation coefficient: 0.88	Number of patients in %: TAM: N1: 7.4, P1: 20.2, P2: 62.8, DY: 9.6; controls: N1: 19.7, P1: 38.3, P2: 37.2, DY: 4.8
Hong and Ki 1999 [37]	Effect of tamoxifen on breast density in women who had undergone surgery for BC	Retrospective case control study	BC	152 BC patients/172 (group I: 102 BC patients (TAM only: 40; +chemotherapy: 36; +radiotherapy: 13; +radiotherapy and chemotherapy: 13), group II: 50 BC patients without TAM, group III: 20 healthy women)	Mean age: group I: 49 y (range 28–67 y); group II: 47 y (range 32–58 y); and group III: 47 y (range 34–62 y)	Visually according to a 3-score classification; decrease of BD: 0 = <10%, 1 = 10–20%, 2 = 20%/at baseline: BC patients: before and after surgery and for control group: once a year for at least 2 years, and on follow-up	Two readers	NA	Number of patients^4: dense pattern: group I: 19 (18.63%), II: 13 (26%), III: 7 (35%). Heterogeneously dense: group I: 44 (43.13%), II: 21 (42%), III: 9 (45%). Fatty pattern: group I: 39 (38.24%), II: 16 (32%), III: 4 (20%)
Konez et al. 2001 [38]	Effect of tamoxifen on breast density	Retrospective cohort study	BC	27 BC patients without chemotherapy/24 BC patients	Mean 67 y (32–81 y)	Visually and using densitometric values (ratio dense/fat)/at baseline, within 2 or 3 yrs, at 5 yrs and 1 yr after cessation of treatment	Two radiologists blinded to the time sequence of the mammograms	Agreement in 24/27 cases^5	Mean values of the relative glandular densities (glandular tissue/fat density) at baseline: 0.652
Meggiorini et al. 2008 [39]	Effect of tamoxifen on breast density	Retrospective case control study	BC	148 BC patients/cases: 68 BC patients who receive tamoxifen alone or ± radiotherapy and chemotherapy; controls: 80 BC patients ± radiotherapy and chemotherapy	TAM group: mean 58.5 ± 9.3 y, median 56.5 y, range 41–78 y; controls group: mean 63.9±9.2 y, median 63.5 y, range 49–78 y	Visually according to BIRADs and by a computer-assisted method after digitizing images/at baseline and annually for six years	Two radiologists blinded to the treatment	Agreement between methods for baseline mammograms: cases = 0.994 𝑃<0.00001, controls = 0.985 𝑃<0.00001 and mammograms at 1 yr: cases = 1, controls = 1	Basal mammography with classification BIRADS (number of patients): in TAM group: A: 22, B: 14, C: 22, D: 10 and control group: A: 30, B: 12, C: 26, D: 12
Chen et al. 2011 [40]	Effect of tamoxifen on breast density	Retrospective cohort study	BC	17 BC patients without chemotherapy/16 BC patients treated by TAM after surgery and without chemotherapy	Mean 43 y, range 33–51 y	Based on 3D MRI T1-weighted images. The changes in breast volume (ΔBV), fibroglandular tissue volume (ΔFV), and percent density (Δ%BD) between 2 MRI studies were analyzed/pretreatment and follow-up studies ranged from 8 to 26 months (17.5±5.7 months)	NA	NA	Baseline %BD ranged from 5.1% to 39.5% (22.1±2.6%)
Cuzick et al. 2011 [41]	Effect of tamoxifen-induced reductions in breast density on BC risk	Nested case control study within a randomized prevention trial (IBIS 1)	At least twice the average BC risk of a 50-year-old woman: either benign proliferative breast disease or a strong family history of BC^6	1288 women (potential control subjects without BC: 𝑛=1064; potential case subjects with BC: 𝑛=224)/1065 women with mammograms at baseline and follow-up (12–18 months after trial entry). Cases: 123 women diagnosed with BC; control subjects: 942 women without BC	Mean 51 ± 6 y, range 35–70 y	Visual assessment of the proportion of the total breast area that was composed of dense tissue (to the nearest 5%)/at baseline (at or up to 12 months before randomization), and at first follow-up (18–23 months)	One radiologist blinded to treatment arm and not case control status	Reproducibility assessed for 48 women by 5 readers at baseline and follow-up (Pearson correlation coefficient r ranged from 0.48 to 0.98)	Case subjects: mean MD at entry: 51%; control subjects: 44% (P = 0.02)

BC: breast cancer. ¹Low-dose tamoxifen of 5 mg. ²IEN: intraepithelial neoplasia including ductal and lobular carcinoma in situ; T1: pT1 a or pT1mic N0. ³For 40 women BD was not measured at baseline. DB with digital measurement: 16% lower compared to those with analog film screen. DCIS: ductal carcinoma in situ. LCIS: lobular carcinoma in situ. NA: nonavailable or not applicable; SD: standard deviation. Wolfe classification: N1: nondense, no ducts visible; P1: prominent ductal pattern occupying less than one-fourth of the breast; P2: prominent ductal pattern occupying more than one-fourth of the breast; DY: homogenous, plaque-like areas of density. ⁴Personal calculation for percentages. ⁵For 24 cases: results revealed parallelism in the findings of the two radiologists. MD: mammographic density. BI-RADS breast categories: (1) almost entirely fat, (2) fatty with scattered fibroglandular densities, (3) heterogeneously dense breast tissue, and (4) extremely dense breast tissue. ⁶i.e., a mother or sister who developed BC before the age of 50.