Med Oncol (2014) 31:96 DOI 10.1007/s12032-014-0096-3
ORIGINAL PAPER
Prognostic value of mammographic breast density in patients with metastatic breast cancer Shereef Elsamany • Abdullah Alzahrani • Seham Abd Elkhalik Omaima Elemam • Elham Rawah • Mian U. Farooq • Musab H. Almatrafi • Feras K. Olayan
•
Received: 30 May 2014 / Accepted: 23 June 2014 / Published online: 11 July 2014 Ó Springer Science+Business Media New York 2014
Abstract Breast density is a modifiable trait linked with breast cancer predisposition. However, the relation between mammographic breast density and survival outcome is not yet clarified. The present study aims to study the prognostic value of mammographic density in patients diagnosed with metastatic breast cancer. In this observational study, breast cancer patients with metastatic disease at diagnosis were enrolled. Two-view mammograms were performed at diagnosis, and breast density was quantitatively assessed. Progression-free survival (PFS) was correlated with breast density and other prognostic variables in univariate and multivariate analyses. PFS, stratified by different prognostic factors, was assessed in low compared
S. Elsamany A. Alzahrani S. A. Elkhalik O. Elemam Oncology, King Abdullah Medical City, Makkah, Saudi Arabia S. Elsamany O. Elemam Medical Oncology, Oncology Center, Mansoura University, Mansoura, Egypt S. Elsamany (&) Medical Oncology Department, Oncology Center, King Abdullah Medical City, 2677 Al-Mashaeer District, Makkah 57657, Saudi Arabia e-mail: [email protected] S. A. Elkhalik Clinical Oncology and Nuclear Medicine, Mansoura University Hospital, Mansoura, Egypt E. Rawah Radiology, King Abdullah Medical City, Makkah, Saudi Arabia M. U. Farooq Research, King Abdullah Medical City, Makkah, Saudi Arabia M. H. Almatrafi F. K. Olayan Medicine, Um Alqora University, Makkah, Saudi Arabia
to high density patients to check for possible differential survival outcome in patients’ subgroups. Among the sixty enrolled patients, median PFS in low density patients was significantly better than those with high density (18.4 months, 95 % CI 14.88–22.15 vs. 9.3 months, 95 % CI 8.51–13.60, respectively, p = 0.002). Significant correlation of breast density with PFS persisted after adjustment by body mass index (p = 0.003) and after multivariate analysis incorporating other prognostic variables (HR 6.16, 95 % CI (2.17–17.48), p = 0.001). PFS was better in low density patients older than 40 years at diagnosis (p = 0.001), with HER2-negative disease (p = 0.015), hormonal receptor-positive phenotype (p = 0.020), patients with single site of metastasis (p = 0.006), and patients with bone-only metastases (p = 0.042). Breast density assessed at the time of diagnosis was significantly correlated with PFS of metastatic breast cancer patients. Survival outcome is improved in certain patients’ subgroups with low breast density. Keywords Metastatic
Prognosis Breast cancer Density
Introduction Mammographic density is a modifiable trait that has been linked with breast cancer predisposition [1]. High mammographic density was found to increase the risk of breast cancer by four to six folds [2]. Similarly, an increase in mammographic density has been correlated with increased breast cancer risk in postmenopausal women who use estrogen/progestin therapies [3]. The mammographic appearance is determined by the amount of radio-dense tissue relative to the breast volume. The radio-dense tissue
123
96
Page 2 of 7
consists of stromal and epithelial components, and thereby, it is assumed to reflect the target tissue for breast cancer [4]. However, the amount of connective tissue is far greater than the glandular tissue and contributes more to the percentage of the dense area. It appears that connective tissue stroma contributes to neoplastic progression, and its disturbance may precede epithelial changes [5]. Noteworthy, the stroma associated with mammographic density may enhance tumor formation through epithelial–stromal interactions which increase the likelihood of growth and progression [6]. Many studies have tried to identify factors that cause and/or influence the fibroglandular breast pattern. Factors including age, body mass index (BMI), parity, passing through the menopause, and smoking have been shown to decrease the relative density. Late age at first birth, use of hormone therapy, and alcohol consumption have been found to increase breast density [4]. Furthermore, several reports have examined the influence of mammographic density on the outcome of breast cancer patients. Two studies found higher local recurrence rates with higher density [7, 8]. In addition, several studies have investigated breast density as a predictor of breast cancer-specific survival with inconsistent results [9–11]. Furthermore, the change in mammographic density seems to reflect the efficacy of tamoxifen used in primary prevention of breast cancer. Women who had at least a 10 % reduction in mammographic density over the first 1.5 years of tamoxifen prophylaxis had 63 % reduction in breast cancer risk [12]. In addition, in breast cancer patients receiving adjuvant tamoxifen, mammographic density reduction appears to serve as a marker for improved longterm survival [13]. Several prognostic factors have been described in metastatic breast cancer; however, the link between mammographic density and survival outcome in those patients is not yet clarified. The present study aims to study the utility of mammographic density as a prognostic factor in patients presented with metastatic breast cancer at diagnosis in addition to evaluating possible interaction with other variables, known to affect the outcome of those patients.
Patients and methods Study population The study group consisted of female patients with histologically confirmed breast cancer who presented to king Abdullah Medical City, Saudi Arabia from May 2011 to November 2012. Only patients with an evidence of metastatic disease at the time of diagnosis were enrolled. Patients must have a baseline mammogram at the time of
123
Med Oncol (2014) 31:96
diagnosis. Those with only non-measurable disease other than bony metastases were excluded. Seventy-three patients were screened; 13 patients were excluded; 7 had no baseline mammogram; and 6 had only non-measurable metastases (pleural effusion). Study design and procedures This is an observational prospective study. Data regarding age at diagnosis, menopausal status, BMI, site and number of metastases, tumor phenotype (ER, PR, HER2 status), type of therapy (chemotherapy and/or hormonal therapy), mammographic breast density, date of disease progression, and death if any were recorded. ER/PR were considered positive if [1 % of cells showed positive staining. Hormonal receptor positivity was defined as ER and/or PR positive, while patients with negative staining for both ER and PR receptors were considered hormonal receptornegative. HER2 status was assessed by immune-histochemistry (IHC) in addition to FISH confirmation in cases with (??) by IHC. Breast density was assessed in twoview mammograms and performed at the time of diagnosis before starting therapy. The mammogram was interpreted through visual assessment by one radiologist at our institution without the assistance of computer-aided system. Breast density was assessed in the diseased breast, and it was quantitatively classified based on the categories originally proposed by Wolfe as follows [14]: low density indicating radio-dense fibroglandular tissue B25 %; moderate density, radio-dense fibroglandular tissue [25–50 %; and high density, radio-dense fibroglandular tissue [50 %. Progression-free survival (PFS) was correlated with breast density at baseline mammogram in addition to other variables. Statistical analysis The data were analyzed by using SPSS version 17 (SPSS Inc., Chicago, IL, USA) and have been subjected to descriptive analysis. Age and BMI have been considered both as continuous as well as categorical variables, i.e., age (B40 and [40 years) and BMI (normal, 18.5–\25; overweight, 25–\30; and obese, C30). We examined the distribution of patients and treatment characteristics in low compared to moderate/high breast density to describe the study population and to identify possible associations with different breast density categories. These categorical variables were compared using Chi-square test or Fisher’s exact test as appropriate. PFS was defined as the time from starting treatment to the first documented tumor progression or death from any cause. PFS was assessed using survival analysis (Kaplan– Meier curve), while the differences in survival distributions
Med Oncol (2014) 31:96
according to age categories, breast density levels, menopausal status, tumor phenotype, receiving chemotherapy, hormonal therapy, and the site of metastasis, i.e., bone-only versus others and single versus multiple, were evaluated via log-rank (Mantel-Cox) test. Moreover, we stratified the study group by characteristics potentially related to PFS, such as age (B40 vs. [40), premenopausal versus postmenopausal, and hormonal receptors (positive vs. negative). PFS in each stratum was assessed in low compared to moderate/high density patients to check for possible differential survival outcome in patients’ subgroups. The breast density was initially adjusted for BMI, age, and for both (simple adjustment) to find out any possible difference in the impact on PFS after adjustment from that found with unadjusted breast density. Multivariate analysis by Cox proportional hazards model has been used to check for independent prognostic factors associated with PFS. Multivariate analysis including unadjusted breast density was first performed and then repeated with BMI-adjusted breast density (fully adjusted multivariate analysis). Two approaches have been used for breast density adjustment; age and BMI were first used as continuous and then as categorical variables. Furthermore, logistic regression analysis was performed to find out independent predictors for breast cancer progression. An alpha level of \0.05 has been considered significant for each analysis.
Results Sixty patients diagnosed with metastatic breast cancer were included. Breast density in baseline mammogram at the time of diagnosis was low in 30, moderate in 18, and high in 12 patients. To have enough number of patients for comparison, moderate and high density patients were grouped together. Invasive ductal carcinoma was the pathological type in 58 patients, while the other two had invasive lobular carcinoma. Forty-six patients (76.7 %) received chemotherapy (10, anathracycline/cyclophosphamide; 19, single-agent taxan; 17, both anthracycline and taxan). None of our patients received hormonal replacement therapy. Distribution of patient characteristics by breast density categories Compared to moderate/high density cases, patients with low breast density were more likely to be older than 40 years at diagnosis (p = 0.010), postmenopausal (p = 0.001), and with bone-only metastasis (p = 0.039). In addition, low density patients tended to have positive hormonal receptors (p = 0.060) and single site of metastasis (p = 0.067). However, no significant difference in
Page 3 of 7 96
HER2 status (p = 0.180) was found between low and moderate/high density patients. Noteworthy, disease progression was more frequent in moderate/high compared to low density cases (76.6 vs. 43.3 %, respectively, p = 0.017) after median follow-up of 18 months (Table 1).
Survival outcome Median PFS in patients with low density was significantly better than those with moderate density/high density (18.4 months, 95 % CI 14.88–22.15 vs. 9.3 months, 95 % CI 8.51–13.60, respectively, p = 0.002) (Table 2; Fig. 1). Breast density retained significant correlation with PFS after adjustment by BMI (p = 0.003, HR 2.90, 95 % CI 1.43–5.89), adjustment by age (p = 0.000, HR 5.29, 95 % CI 2.14–13.04), and adjustment by both age and BMI (p = 0.000, HR 5.42, 95 % CI 2.16–13.58). In addition, we explored the prognostic value of different variables in univariate and multivariate analyses. In univariate analysis, PFS was significantly better in hormonal receptor-positive patients (18.4 months, 95 % CI 13.49–19.26 vs. 9.4 months, 95 % CI 7.93–12.75, p = 0.006), those who received hormonal therapy (18.4 months, 95 % CI 13.79–19.61 vs. 9.4 months, 95 % CI 8.06–12.64, p = 0.003), and patients with bone-only metastases (27.0 months, 95 % CI 16.78–36.72 vs. 9.6 months, 95 % CI 9.98–14.05, p = 0.003) (Table 2; Fig. 2). No significant difference was found in PFS according to other variables (Table 2). In multivariate Cox regression analysis including unadjusted breast density, age [40 years (HR 3.09, 95 % CI 1.18–8.13, p = 0.021), postmenopausal status (HR 3.77, 95 % CI 1.25–11.36, p = 0.018), hormonal receptor negativity (HR 6.08, 95 % CI 2.31–16.03, p = 0.000), no chemotherapy (HR 2.97, 95 % CI 1.03–8.61, p = 0.040), and moderate density/high density (HR 8.62, 95 % CI 2.67–27.03, p = 0.000) were independent predictors of worse PFS. Furthermore, breast density adjusted by BMI retained its significant relation with PFS in fully adjusted multivariate analysis when age and BMI were included as categorical as well as continuous variables (Table 3). In the adjusted analyses, bone-only metastasis had independent prognostic value which was not found in the BMI-unadjusted multivariate analysis. Noteworthy, menopausal status lost its prognostic value when age and BMI were included as continuous variables (Table 3). To assess the impact of breast density on the risk of disease progression, we performed multivariate logistic regression analysis including different prognostic variables. Moderate density/ high density (adjusted with BMI) increased the hazard of disease progression when BMI and age were included as continuous variables (OR 3.27, 95 % CI 1.02–10.51,
123
96
Page 4 of 7
Table 1 Baseline characteristics of the study group
Med Oncol (2014) 31:96
Variables
All patients No. (%)
Low mammographic density N = 30
High mammographic density N = 30
Age
0.010
B40
13 (21.7)
2 (6.7)
11 (36.7)
[40
47 (78.3)
28 (93.3)
19 (63.3)
Premenopausal
30 (50.0)
8 (26.7)
22 (73.3)
Postmenopausal
30 (50.0)
22 (73.3)
8 (26.7)
Hormone receptors Negative
22 (36.7)
7 (23.3)
15 (50.0)
38 (63.3)
23 (76.7)
15 (50.0)
Negative
38 (63.3)
22 (73.3)
16 (53.3)
Positive
22 (36.7)
8 (26.7)
14 (46.7)
26 (43.3)
17 (56.7)
9 (30.0)
Menopausal status
Positive
0.001
0.060
HER2 status
0.180
Site of metastasis Single Multiple
0.067 34 (56.7)
13 (43.3)
21 (70.0)
Bone-only metastasis
16 (26.7)
12 (40.0)
4 (13.3)
Other than bone-only metastasis
44 (73.3)
18 (60.0)
26 (86.7)
14 (23.3)
9 (30.0)
5 (16.7)
46 (76.7)
21 (70.0)
25 (83.3)
No
27 (45)
10 (33.3)
17 (56.7)
Yes
33 (55.0)
20 (66.7)
13 (43.3)
Chemotherapy No Yes
0.039
0.360
Hormonal treatment
0.119
Disease progression
0.017
No
24 (40.0)
17 (56.7)
7 (23.3)
Yes
36 (60.0)
13 (43.3)
23 (76.7)
p = 0.047) and also as categorical variables (OR 3.44, 95 % CI 0.99–12.04, p = 0.05). Prognostic value of breast density in different patients’ subgroups We evaluated the prognostic value of breast density stratified by different prognostic variables to assess for subgroups of patients who may have better PFS with low density. Compared to patients with moderate density/high density, PFS was better in low density patients older than 40 years at diagnosis (p = 0.001), with HER2-negative disease (p = 0.015), hormonal receptor-positive phenotype (p = 0.020) in addition to those who received hormonal therapy (p = 0.034), patients with single site of metastasis (p = 0.006), and patients with bone-only metastasis (p = 0.042) (Table 4). Low density retained its favorable prognostic outcome irrespective of the receipt of chemotherapy (chemotherapy, p = 0.021; no chemotherapy, p = 0.002) and menopausal status (premenopausal, p = 0.030; postmenopausal, p = 0.021) (Table 4).
123
p value
Discussion In this observational study, we assessed the prognostic value of mammographic breast density in patients diagnosed with metastatic breast cancer. Moderate/high breast density patients had worse PFS in unadjusted as well as in simple and fully adjusted models. Moreover, breast density was found to be independently correlated with the risk of disease progression. Noteworthy, breast density in our study was assessed in diagnostic mammogram in contrast to most of the other studies where density was mostly assessed in mammograms taken few years before diagnosis. Prediagnostic mammograms were previously proven to predict the risk of developing breast cancer; however, the relation of this prediagnostic assessment with survival outcome is less clear with somewhat contradictory results between different studies, and none of them was designed exclusively for stage IV patients. An inverse association of borderline significance was detected between breast density at screening mammogram and survival outcome in a Swedish cohort [9], but no
Med Oncol (2014) 31:96
Page 5 of 7 96
Table 2 Univariate analyses of relations of different prognostic variables with PFS Variables
Median PFS
95 % CI Lower
p value Upper
Age B40
14.8
10.39
18.02
Age [40
11.0
11.53
17.07
Low density High density
18.4 9.3
14.88 8.51
22.15 13.60
0.002
Premenopausal
9.6
10.46
15.98
0.443
Postmenopausal
12.5
12.05
19.21
HER2-negative
12.5
12.65
19.41
HER2-positive
9.4
8.84
14.58
No chemotherapy
11.0
8.66
15.13
Chemotherapy
12.5
11.96
16.99
Hormone receptor-negative
9.4
7.93
12.75
Hormone receptor-positive
18.4
13.49
19.26
9.4
8.06
12.64
Hormonal therapy
18.4
13.79
19.61
Single site of metastasis
12.5
11.67
20.08
9.6
10.62
15.39
27.0
16.78
36.72
9.6
9.98
14.05
No hormonal therapy
Multiple sites of metastases Bone-only metastasis Metastases other than bone
0.948
0.055 0.630 0.006 0.003 0.257 0.003
PFS progression-free survival
Fig. 1 PFS in low compared to moderate/high breast density patients in univariate analysis
adverse effect of mammographic density on survival was found in the British or US studies [10, 11]. Noteworthy, in the US study, there was an increased risk of breast cancer death in low density patients which was not statistically significant. In 74.5 % of patients included in that study, mammogram was done [1 year before diagnosis [11]. In
the British study, density was assessed in the contralateral breast [10] with the caveat of discrepancy of density between the diseased and the contralateral breast. This may spoil the correlation between breast density and survival outcome as density in the diseased breast may be more representative of disease biology. In another study from Hawaii including all disease stages, high density in the prediagnostic mammogram was associated with worse survival outcome in the whole group; however, there was no special analysis for stage IV patients [2]. Meanwhile, Cil et al. [7] assessed breast density at diagnostic mammogram, and they found an increased risk of disease recurrence in high density patients following conservative breast surgery. They used similar method of breast density assessment as that utilized in our study. In our opinion, breast density, to be utilized as a prognostic tool, should be assessed at the time of diagnosis given that breast density is a dynamic process that can change over time. Breast density is affected by several genetic, hormonal, biological, and environmental factors that may change the amount of fibroglandular tissue; thus, the final phenotype measured by mammography may differ accordingly [15]. Actually, there is evidence from animal models that density can change with variations of hormonal exposure [16]. So, prediagnostic images taken several years before actual diagnosis of cancer can predict the risk of developing breast cancer, but may not reflect the tumor microenvironment, biology, and aggressiveness at the time of diagnosis and, in our opinion, should not be used to predict survival outcome. In the previous studies, no multivariate analysis specified for stage IV patients was performed which may explain the inconsistency of data. For example, in the US study [11], several prognostic factors that affect the outcome of metastatic patients were not included in the multivariate analysis, such as tumor phenotype, site, and number of metastases. This is mostly related to the design of the study which was not specified for metastatic patients. In the British study [10], breast density was not adjusted with BMI, and no multivariate analysis was performed. The association of high density with poor prognostic factors such as large tumors or lymph node positivity was reported in the Swedish study [9] and the study of Aiello et al. [17] which evaluated the effect of breast density at screening mammogram on patients’ outcome. In our study, moderate/high density cases were associated with non-bone metastases and negative hormonal receptors which were found to be predictors of poor outcome in univariate analyses. However, in fully adjusted model, breast density retained its prognostic value which suggests that the association of density with poor PFS is not related to the association with these characteristics.
123
96
Page 6 of 7
Med Oncol (2014) 31:96
Fig. 2 Comparisons of PFS according to different patients’ characteristics in univariate analyses. a HER2 positive versus HER2 negative, b ER/ PR positive versus ER/PR negative, c hormonal therapy versus no hormonal therapy, d bone-only metastasis versus others
Table 3 Multivariate Cox regression analysis of different prognostic variables in relation to PFS
BMI body mass index, HR hazard ratio, CI confidence interval
Variables
Multivariate analysis (BMI, age; continuous variables)
Multivariate analysis (BMI, age; categorical variables)
HR (95 % CI)
p value
HR (95 % CI)
p value
Age [40
2.77 (1.27–6.11)
0.001
3.22 (1.22–8.49)
0.018
High density
6.16 (2.17–17.48)
0.001
7.51 (2.42–23.28)
0.000
No chemotherapy
2.98 (1.14–7.84)
0.026
3.43 (1.33–8.87)
0.011
Negative hormone receptors Other than bone-only metastasis
3.53 (1.52–8.16) 4.20 (1.22–14.39)
0.003 0.022
4.91 (1.88–12.81) 2.73 (0.83–8.98)
0.001 0.099
Postmenopausal
1.02 (0.27–3.82)
0.974
3.39 (1.23–9.32)
0.018
We identified subsets of breast cancer patients where low density was associated with good outcome such as those with positive hormonal receptors, HER2-negative disease, single site of metastasis, and patients with boneonly metastases. These findings highlight the possibility of integrating breast density with other prognostic factors to
123
allow for better assessment of the overall prognosis of metastatic breast cancer patients. Several methods of breast density evaluation have been utilized; however, the quantitative visual assessment of mammographic density utilized in our study carries some points of attraction. It is a simple method that can be
Med Oncol (2014) 31:96
Page 7 of 7 96
Table 4 Comparisons of PFS in low versus high density patients in different subgroups Variables
Low density median PFS (95 % CI)
High density median PFS (95 % CI)
p value
Premenopausal
NA
9.4 (8.96–9.83)
0.030
Postmenopausal
18.4 (9.38–27.42)
6.6 (4.28–8.92)
0.021
Age B40
NA
9.6 (8.90–19.30)
0.16
Age [40
18.4 (11.0–27.0)
7.8 (5.20–9.60)
0.001
HER2-negative
27.0 (11.10–27.0)
9.1 (6.62–11.58)
0.015
HER2-positive
18.4 (9.03–27.77)
9.3 (8.74–9.86)
0.299
No chemotherapy
12.4 (10.90–NA)
6.6 (5.11–8.09)
0.002
Chemotherapy
18.4 (10.97–25.83)
9.4 (9.06–9.74)
0.021
Hormone receptor-negative
11.0 (7.92–14.08)
8.9 (5.23–12.57)
0.221
Hormone receptorpositive
27.0 (12.0–27.0)
9.6 (3.68–15.52)
0.020
No hormonal therapy
10.9 (8.17–13.63)
8.9 (5.36–12.44)
0.165
Hormonal therapy
27.0 (12.0–27.0)
9.6 (3.63–15.57)
0.034
Single site of metastasis
27.0 (12.0–27.0)
5.2 (4.67–5.73)
0.006
Multiple sites of metastases
10.9 (3.02–18.78)
9.4 (9.10–9.70)
0.282
Bone-only metastasis
27.0 (NA–29.0)
7.8 (0.000–16.48)
0.042
Other than boneonly metastases
11.0 (8.89–13.11)
9.4 (8.85–9.95)
0.240
PFS progression-free survival, CI confidence interval
assessed easily with routine mammograms without requiring special software, and it categorizes patients into easily interpretable classes. However, it lacks the accurate estimation of breast density provided by computer-aided assessment in addition to the hazard of inter-observer variation in breast density estimation. At the time point, overall survival (OS) of the study group was not assessed as only nine cases died till the time of this analysis. It will be reported later after longer followup to check for possible impact of breast density on OS. In conclusion, breast density assessed at the time of diagnosis was significantly correlated with PFS in metastatic breast cancer patients both in unadjusted and adjusted models. Certain groups of patients seem to have better outcome if they have low density such as in those older than 40 years at diagnosis, patients with HER2-negative disease, and those with hormonal receptor-positive phenotype. Conflict of interest of interest.
References 1. Boyd NF, Rommens JM, Vogt K, et al. Mammographic breast density as an intermediate phenotype for breast cancer. Lancet Oncol. 2005;6:798–808. 2. Maskarinec G, Pagano I, Little M. Mammographic density as a predictor of breast cancer survival: the multiethnic cohort. Breast Can Res. 2013;15:R7. 3. McTiernan A, Martin CF, Peck JD, et al. Estrogen-plus-progestin use and mammographic density in postmenopausal women: women’s health initiative randomized trial. J Natl Cancer Inst. 2005;97:1366–76. 4. Haars G, Van Noord P, Gils C, et al. Measurements of breast density: no ratio for a ratio. Cancer Epidemol Biomark Prev. 2005;1411(11):2634–40. 5. Alowami S, Troup S, Al-Haddad S, et al. Mammographic density is related to stroma and stromal proteoglycan expression. Breast Cancer Res. 2003;5:R129–35. 6. Provenzano PP, Inman DR, Eliceiri KW, et al. Collagen density promotes mammary tumor initiation and progression. BMC Med. 2008;6:11. 7. Cil T, Fishell E, Hanna W, et al. Mammographic density and the risk of breast cancer recurrence after breast-conserving surgery. Cancer. 2009;115:5780–7. 8. Park CC, Rembert J, Chew K, et al. High mammographic breast density is independent predictor of local but not distant recurrence after lumpectomy and radiotherapy for invasive breast cancer. Int J Radiat Oncol Biol Phys. 2009;73:75–9. 9. Chiu SY, Duffy S, Yen AM, et al. Effect of baseline breast density on breast cancer incidence, stage, mortality, and screening parameters: 25-year follow-up of a Swedish mammographic screening. Cancer Epidem Biom Prev. 2010;19:1219–28. 10. Porter GJ, Evans AJ, Cornford EJ, et al. Influence of mammographic parenchymal pattern in screening-detected and interval invasive breast cancers on pathologic features, mammographic features, and patient survival. AJR. 2007;188:676–83. 11. Gierach GL, Ichikawa L, Kerlikowske K, et al. Relationship between mammographic density and breast cancer death in the breast cancer surveillance consortium. J Natl Cancer Inst. 2012;104:1218–27. 12. Cuzick J, Warwick J, Pinney E, et al. Tamoxifen-induced reduction in mammographic density and breast cancer risk reduction: a nested case-control study. J Natl Cancer Inst. 2011;103:744–52. 13. Li J, Humphreys K, Eriksson L, et al. Mammographic density reduction is a prognostic marker of response to adjuvant tamoxifen therapy in postmenopausal patients with breast cancer. JCO. 2013;31(18):2249–56. 14. Wolfe J. Breast patterns as an index of risk for developing breast cancer. AJR. 1976;126:1130–9. 15. Boyd NF, Martin LJ, Yaffe MJ, et al. Mammographic density: a hormonally responsive risk factor for breast cancer. J Br Menopause Soc. 2006;12(4):186–93. 16. Chew GL, Huang D, Huo CW, et al. Dynamic changes in high and low mammographic density human breast tissues maintained in murine tissue engineering chambers during various murine peripartum states and over time. Breast Cancer Res Treat. 2013;140(2):285–97. 17. Aiello EJ, Buist DS, White E, et al. Association between mammographic breast density and breast cancer tumor characteristics. Cancer Epidem Bioma Prev. 2005;14:662–8.
The authors declare that they have no conflict
123
ORIGINAL PAPER
Prognostic value of mammographic breast density in patients with metastatic breast cancer Shereef Elsamany • Abdullah Alzahrani • Seham Abd Elkhalik Omaima Elemam • Elham Rawah • Mian U. Farooq • Musab H. Almatrafi • Feras K. Olayan
•
Received: 30 May 2014 / Accepted: 23 June 2014 / Published online: 11 July 2014 Ó Springer Science+Business Media New York 2014
Abstract Breast density is a modifiable trait linked with breast cancer predisposition. However, the relation between mammographic breast density and survival outcome is not yet clarified. The present study aims to study the prognostic value of mammographic density in patients diagnosed with metastatic breast cancer. In this observational study, breast cancer patients with metastatic disease at diagnosis were enrolled. Two-view mammograms were performed at diagnosis, and breast density was quantitatively assessed. Progression-free survival (PFS) was correlated with breast density and other prognostic variables in univariate and multivariate analyses. PFS, stratified by different prognostic factors, was assessed in low compared
S. Elsamany A. Alzahrani S. A. Elkhalik O. Elemam Oncology, King Abdullah Medical City, Makkah, Saudi Arabia S. Elsamany O. Elemam Medical Oncology, Oncology Center, Mansoura University, Mansoura, Egypt S. Elsamany (&) Medical Oncology Department, Oncology Center, King Abdullah Medical City, 2677 Al-Mashaeer District, Makkah 57657, Saudi Arabia e-mail: [email protected] S. A. Elkhalik Clinical Oncology and Nuclear Medicine, Mansoura University Hospital, Mansoura, Egypt E. Rawah Radiology, King Abdullah Medical City, Makkah, Saudi Arabia M. U. Farooq Research, King Abdullah Medical City, Makkah, Saudi Arabia M. H. Almatrafi F. K. Olayan Medicine, Um Alqora University, Makkah, Saudi Arabia
to high density patients to check for possible differential survival outcome in patients’ subgroups. Among the sixty enrolled patients, median PFS in low density patients was significantly better than those with high density (18.4 months, 95 % CI 14.88–22.15 vs. 9.3 months, 95 % CI 8.51–13.60, respectively, p = 0.002). Significant correlation of breast density with PFS persisted after adjustment by body mass index (p = 0.003) and after multivariate analysis incorporating other prognostic variables (HR 6.16, 95 % CI (2.17–17.48), p = 0.001). PFS was better in low density patients older than 40 years at diagnosis (p = 0.001), with HER2-negative disease (p = 0.015), hormonal receptor-positive phenotype (p = 0.020), patients with single site of metastasis (p = 0.006), and patients with bone-only metastases (p = 0.042). Breast density assessed at the time of diagnosis was significantly correlated with PFS of metastatic breast cancer patients. Survival outcome is improved in certain patients’ subgroups with low breast density. Keywords Metastatic
Prognosis Breast cancer Density
Introduction Mammographic density is a modifiable trait that has been linked with breast cancer predisposition [1]. High mammographic density was found to increase the risk of breast cancer by four to six folds [2]. Similarly, an increase in mammographic density has been correlated with increased breast cancer risk in postmenopausal women who use estrogen/progestin therapies [3]. The mammographic appearance is determined by the amount of radio-dense tissue relative to the breast volume. The radio-dense tissue
123
96
Page 2 of 7
consists of stromal and epithelial components, and thereby, it is assumed to reflect the target tissue for breast cancer [4]. However, the amount of connective tissue is far greater than the glandular tissue and contributes more to the percentage of the dense area. It appears that connective tissue stroma contributes to neoplastic progression, and its disturbance may precede epithelial changes [5]. Noteworthy, the stroma associated with mammographic density may enhance tumor formation through epithelial–stromal interactions which increase the likelihood of growth and progression [6]. Many studies have tried to identify factors that cause and/or influence the fibroglandular breast pattern. Factors including age, body mass index (BMI), parity, passing through the menopause, and smoking have been shown to decrease the relative density. Late age at first birth, use of hormone therapy, and alcohol consumption have been found to increase breast density [4]. Furthermore, several reports have examined the influence of mammographic density on the outcome of breast cancer patients. Two studies found higher local recurrence rates with higher density [7, 8]. In addition, several studies have investigated breast density as a predictor of breast cancer-specific survival with inconsistent results [9–11]. Furthermore, the change in mammographic density seems to reflect the efficacy of tamoxifen used in primary prevention of breast cancer. Women who had at least a 10 % reduction in mammographic density over the first 1.5 years of tamoxifen prophylaxis had 63 % reduction in breast cancer risk [12]. In addition, in breast cancer patients receiving adjuvant tamoxifen, mammographic density reduction appears to serve as a marker for improved longterm survival [13]. Several prognostic factors have been described in metastatic breast cancer; however, the link between mammographic density and survival outcome in those patients is not yet clarified. The present study aims to study the utility of mammographic density as a prognostic factor in patients presented with metastatic breast cancer at diagnosis in addition to evaluating possible interaction with other variables, known to affect the outcome of those patients.
Patients and methods Study population The study group consisted of female patients with histologically confirmed breast cancer who presented to king Abdullah Medical City, Saudi Arabia from May 2011 to November 2012. Only patients with an evidence of metastatic disease at the time of diagnosis were enrolled. Patients must have a baseline mammogram at the time of
123
Med Oncol (2014) 31:96
diagnosis. Those with only non-measurable disease other than bony metastases were excluded. Seventy-three patients were screened; 13 patients were excluded; 7 had no baseline mammogram; and 6 had only non-measurable metastases (pleural effusion). Study design and procedures This is an observational prospective study. Data regarding age at diagnosis, menopausal status, BMI, site and number of metastases, tumor phenotype (ER, PR, HER2 status), type of therapy (chemotherapy and/or hormonal therapy), mammographic breast density, date of disease progression, and death if any were recorded. ER/PR were considered positive if [1 % of cells showed positive staining. Hormonal receptor positivity was defined as ER and/or PR positive, while patients with negative staining for both ER and PR receptors were considered hormonal receptornegative. HER2 status was assessed by immune-histochemistry (IHC) in addition to FISH confirmation in cases with (??) by IHC. Breast density was assessed in twoview mammograms and performed at the time of diagnosis before starting therapy. The mammogram was interpreted through visual assessment by one radiologist at our institution without the assistance of computer-aided system. Breast density was assessed in the diseased breast, and it was quantitatively classified based on the categories originally proposed by Wolfe as follows [14]: low density indicating radio-dense fibroglandular tissue B25 %; moderate density, radio-dense fibroglandular tissue [25–50 %; and high density, radio-dense fibroglandular tissue [50 %. Progression-free survival (PFS) was correlated with breast density at baseline mammogram in addition to other variables. Statistical analysis The data were analyzed by using SPSS version 17 (SPSS Inc., Chicago, IL, USA) and have been subjected to descriptive analysis. Age and BMI have been considered both as continuous as well as categorical variables, i.e., age (B40 and [40 years) and BMI (normal, 18.5–\25; overweight, 25–\30; and obese, C30). We examined the distribution of patients and treatment characteristics in low compared to moderate/high breast density to describe the study population and to identify possible associations with different breast density categories. These categorical variables were compared using Chi-square test or Fisher’s exact test as appropriate. PFS was defined as the time from starting treatment to the first documented tumor progression or death from any cause. PFS was assessed using survival analysis (Kaplan– Meier curve), while the differences in survival distributions
Med Oncol (2014) 31:96
according to age categories, breast density levels, menopausal status, tumor phenotype, receiving chemotherapy, hormonal therapy, and the site of metastasis, i.e., bone-only versus others and single versus multiple, were evaluated via log-rank (Mantel-Cox) test. Moreover, we stratified the study group by characteristics potentially related to PFS, such as age (B40 vs. [40), premenopausal versus postmenopausal, and hormonal receptors (positive vs. negative). PFS in each stratum was assessed in low compared to moderate/high density patients to check for possible differential survival outcome in patients’ subgroups. The breast density was initially adjusted for BMI, age, and for both (simple adjustment) to find out any possible difference in the impact on PFS after adjustment from that found with unadjusted breast density. Multivariate analysis by Cox proportional hazards model has been used to check for independent prognostic factors associated with PFS. Multivariate analysis including unadjusted breast density was first performed and then repeated with BMI-adjusted breast density (fully adjusted multivariate analysis). Two approaches have been used for breast density adjustment; age and BMI were first used as continuous and then as categorical variables. Furthermore, logistic regression analysis was performed to find out independent predictors for breast cancer progression. An alpha level of \0.05 has been considered significant for each analysis.
Results Sixty patients diagnosed with metastatic breast cancer were included. Breast density in baseline mammogram at the time of diagnosis was low in 30, moderate in 18, and high in 12 patients. To have enough number of patients for comparison, moderate and high density patients were grouped together. Invasive ductal carcinoma was the pathological type in 58 patients, while the other two had invasive lobular carcinoma. Forty-six patients (76.7 %) received chemotherapy (10, anathracycline/cyclophosphamide; 19, single-agent taxan; 17, both anthracycline and taxan). None of our patients received hormonal replacement therapy. Distribution of patient characteristics by breast density categories Compared to moderate/high density cases, patients with low breast density were more likely to be older than 40 years at diagnosis (p = 0.010), postmenopausal (p = 0.001), and with bone-only metastasis (p = 0.039). In addition, low density patients tended to have positive hormonal receptors (p = 0.060) and single site of metastasis (p = 0.067). However, no significant difference in
Page 3 of 7 96
HER2 status (p = 0.180) was found between low and moderate/high density patients. Noteworthy, disease progression was more frequent in moderate/high compared to low density cases (76.6 vs. 43.3 %, respectively, p = 0.017) after median follow-up of 18 months (Table 1).
Survival outcome Median PFS in patients with low density was significantly better than those with moderate density/high density (18.4 months, 95 % CI 14.88–22.15 vs. 9.3 months, 95 % CI 8.51–13.60, respectively, p = 0.002) (Table 2; Fig. 1). Breast density retained significant correlation with PFS after adjustment by BMI (p = 0.003, HR 2.90, 95 % CI 1.43–5.89), adjustment by age (p = 0.000, HR 5.29, 95 % CI 2.14–13.04), and adjustment by both age and BMI (p = 0.000, HR 5.42, 95 % CI 2.16–13.58). In addition, we explored the prognostic value of different variables in univariate and multivariate analyses. In univariate analysis, PFS was significantly better in hormonal receptor-positive patients (18.4 months, 95 % CI 13.49–19.26 vs. 9.4 months, 95 % CI 7.93–12.75, p = 0.006), those who received hormonal therapy (18.4 months, 95 % CI 13.79–19.61 vs. 9.4 months, 95 % CI 8.06–12.64, p = 0.003), and patients with bone-only metastases (27.0 months, 95 % CI 16.78–36.72 vs. 9.6 months, 95 % CI 9.98–14.05, p = 0.003) (Table 2; Fig. 2). No significant difference was found in PFS according to other variables (Table 2). In multivariate Cox regression analysis including unadjusted breast density, age [40 years (HR 3.09, 95 % CI 1.18–8.13, p = 0.021), postmenopausal status (HR 3.77, 95 % CI 1.25–11.36, p = 0.018), hormonal receptor negativity (HR 6.08, 95 % CI 2.31–16.03, p = 0.000), no chemotherapy (HR 2.97, 95 % CI 1.03–8.61, p = 0.040), and moderate density/high density (HR 8.62, 95 % CI 2.67–27.03, p = 0.000) were independent predictors of worse PFS. Furthermore, breast density adjusted by BMI retained its significant relation with PFS in fully adjusted multivariate analysis when age and BMI were included as categorical as well as continuous variables (Table 3). In the adjusted analyses, bone-only metastasis had independent prognostic value which was not found in the BMI-unadjusted multivariate analysis. Noteworthy, menopausal status lost its prognostic value when age and BMI were included as continuous variables (Table 3). To assess the impact of breast density on the risk of disease progression, we performed multivariate logistic regression analysis including different prognostic variables. Moderate density/ high density (adjusted with BMI) increased the hazard of disease progression when BMI and age were included as continuous variables (OR 3.27, 95 % CI 1.02–10.51,
123
96
Page 4 of 7
Table 1 Baseline characteristics of the study group
Med Oncol (2014) 31:96
Variables
All patients No. (%)
Low mammographic density N = 30
High mammographic density N = 30
Age
0.010
B40
13 (21.7)
2 (6.7)
11 (36.7)
[40
47 (78.3)
28 (93.3)
19 (63.3)
Premenopausal
30 (50.0)
8 (26.7)
22 (73.3)
Postmenopausal
30 (50.0)
22 (73.3)
8 (26.7)
Hormone receptors Negative
22 (36.7)
7 (23.3)
15 (50.0)
38 (63.3)
23 (76.7)
15 (50.0)
Negative
38 (63.3)
22 (73.3)
16 (53.3)
Positive
22 (36.7)
8 (26.7)
14 (46.7)
26 (43.3)
17 (56.7)
9 (30.0)
Menopausal status
Positive
0.001
0.060
HER2 status
0.180
Site of metastasis Single Multiple
0.067 34 (56.7)
13 (43.3)
21 (70.0)
Bone-only metastasis
16 (26.7)
12 (40.0)
4 (13.3)
Other than bone-only metastasis
44 (73.3)
18 (60.0)
26 (86.7)
14 (23.3)
9 (30.0)
5 (16.7)
46 (76.7)
21 (70.0)
25 (83.3)
No
27 (45)
10 (33.3)
17 (56.7)
Yes
33 (55.0)
20 (66.7)
13 (43.3)
Chemotherapy No Yes
0.039
0.360
Hormonal treatment
0.119
Disease progression
0.017
No
24 (40.0)
17 (56.7)
7 (23.3)
Yes
36 (60.0)
13 (43.3)
23 (76.7)
p = 0.047) and also as categorical variables (OR 3.44, 95 % CI 0.99–12.04, p = 0.05). Prognostic value of breast density in different patients’ subgroups We evaluated the prognostic value of breast density stratified by different prognostic variables to assess for subgroups of patients who may have better PFS with low density. Compared to patients with moderate density/high density, PFS was better in low density patients older than 40 years at diagnosis (p = 0.001), with HER2-negative disease (p = 0.015), hormonal receptor-positive phenotype (p = 0.020) in addition to those who received hormonal therapy (p = 0.034), patients with single site of metastasis (p = 0.006), and patients with bone-only metastasis (p = 0.042) (Table 4). Low density retained its favorable prognostic outcome irrespective of the receipt of chemotherapy (chemotherapy, p = 0.021; no chemotherapy, p = 0.002) and menopausal status (premenopausal, p = 0.030; postmenopausal, p = 0.021) (Table 4).
123
p value
Discussion In this observational study, we assessed the prognostic value of mammographic breast density in patients diagnosed with metastatic breast cancer. Moderate/high breast density patients had worse PFS in unadjusted as well as in simple and fully adjusted models. Moreover, breast density was found to be independently correlated with the risk of disease progression. Noteworthy, breast density in our study was assessed in diagnostic mammogram in contrast to most of the other studies where density was mostly assessed in mammograms taken few years before diagnosis. Prediagnostic mammograms were previously proven to predict the risk of developing breast cancer; however, the relation of this prediagnostic assessment with survival outcome is less clear with somewhat contradictory results between different studies, and none of them was designed exclusively for stage IV patients. An inverse association of borderline significance was detected between breast density at screening mammogram and survival outcome in a Swedish cohort [9], but no
Med Oncol (2014) 31:96
Page 5 of 7 96
Table 2 Univariate analyses of relations of different prognostic variables with PFS Variables
Median PFS
95 % CI Lower
p value Upper
Age B40
14.8
10.39
18.02
Age [40
11.0
11.53
17.07
Low density High density
18.4 9.3
14.88 8.51
22.15 13.60
0.002
Premenopausal
9.6
10.46
15.98
0.443
Postmenopausal
12.5
12.05
19.21
HER2-negative
12.5
12.65
19.41
HER2-positive
9.4
8.84
14.58
No chemotherapy
11.0
8.66
15.13
Chemotherapy
12.5
11.96
16.99
Hormone receptor-negative
9.4
7.93
12.75
Hormone receptor-positive
18.4
13.49
19.26
9.4
8.06
12.64
Hormonal therapy
18.4
13.79
19.61
Single site of metastasis
12.5
11.67
20.08
9.6
10.62
15.39
27.0
16.78
36.72
9.6
9.98
14.05
No hormonal therapy
Multiple sites of metastases Bone-only metastasis Metastases other than bone
0.948
0.055 0.630 0.006 0.003 0.257 0.003
PFS progression-free survival
Fig. 1 PFS in low compared to moderate/high breast density patients in univariate analysis
adverse effect of mammographic density on survival was found in the British or US studies [10, 11]. Noteworthy, in the US study, there was an increased risk of breast cancer death in low density patients which was not statistically significant. In 74.5 % of patients included in that study, mammogram was done [1 year before diagnosis [11]. In
the British study, density was assessed in the contralateral breast [10] with the caveat of discrepancy of density between the diseased and the contralateral breast. This may spoil the correlation between breast density and survival outcome as density in the diseased breast may be more representative of disease biology. In another study from Hawaii including all disease stages, high density in the prediagnostic mammogram was associated with worse survival outcome in the whole group; however, there was no special analysis for stage IV patients [2]. Meanwhile, Cil et al. [7] assessed breast density at diagnostic mammogram, and they found an increased risk of disease recurrence in high density patients following conservative breast surgery. They used similar method of breast density assessment as that utilized in our study. In our opinion, breast density, to be utilized as a prognostic tool, should be assessed at the time of diagnosis given that breast density is a dynamic process that can change over time. Breast density is affected by several genetic, hormonal, biological, and environmental factors that may change the amount of fibroglandular tissue; thus, the final phenotype measured by mammography may differ accordingly [15]. Actually, there is evidence from animal models that density can change with variations of hormonal exposure [16]. So, prediagnostic images taken several years before actual diagnosis of cancer can predict the risk of developing breast cancer, but may not reflect the tumor microenvironment, biology, and aggressiveness at the time of diagnosis and, in our opinion, should not be used to predict survival outcome. In the previous studies, no multivariate analysis specified for stage IV patients was performed which may explain the inconsistency of data. For example, in the US study [11], several prognostic factors that affect the outcome of metastatic patients were not included in the multivariate analysis, such as tumor phenotype, site, and number of metastases. This is mostly related to the design of the study which was not specified for metastatic patients. In the British study [10], breast density was not adjusted with BMI, and no multivariate analysis was performed. The association of high density with poor prognostic factors such as large tumors or lymph node positivity was reported in the Swedish study [9] and the study of Aiello et al. [17] which evaluated the effect of breast density at screening mammogram on patients’ outcome. In our study, moderate/high density cases were associated with non-bone metastases and negative hormonal receptors which were found to be predictors of poor outcome in univariate analyses. However, in fully adjusted model, breast density retained its prognostic value which suggests that the association of density with poor PFS is not related to the association with these characteristics.
123
96
Page 6 of 7
Med Oncol (2014) 31:96
Fig. 2 Comparisons of PFS according to different patients’ characteristics in univariate analyses. a HER2 positive versus HER2 negative, b ER/ PR positive versus ER/PR negative, c hormonal therapy versus no hormonal therapy, d bone-only metastasis versus others
Table 3 Multivariate Cox regression analysis of different prognostic variables in relation to PFS
BMI body mass index, HR hazard ratio, CI confidence interval
Variables
Multivariate analysis (BMI, age; continuous variables)
Multivariate analysis (BMI, age; categorical variables)
HR (95 % CI)
p value
HR (95 % CI)
p value
Age [40
2.77 (1.27–6.11)
0.001
3.22 (1.22–8.49)
0.018
High density
6.16 (2.17–17.48)
0.001
7.51 (2.42–23.28)
0.000
No chemotherapy
2.98 (1.14–7.84)
0.026
3.43 (1.33–8.87)
0.011
Negative hormone receptors Other than bone-only metastasis
3.53 (1.52–8.16) 4.20 (1.22–14.39)
0.003 0.022
4.91 (1.88–12.81) 2.73 (0.83–8.98)
0.001 0.099
Postmenopausal
1.02 (0.27–3.82)
0.974
3.39 (1.23–9.32)
0.018
We identified subsets of breast cancer patients where low density was associated with good outcome such as those with positive hormonal receptors, HER2-negative disease, single site of metastasis, and patients with boneonly metastases. These findings highlight the possibility of integrating breast density with other prognostic factors to
123
allow for better assessment of the overall prognosis of metastatic breast cancer patients. Several methods of breast density evaluation have been utilized; however, the quantitative visual assessment of mammographic density utilized in our study carries some points of attraction. It is a simple method that can be
Med Oncol (2014) 31:96
Page 7 of 7 96
Table 4 Comparisons of PFS in low versus high density patients in different subgroups Variables
Low density median PFS (95 % CI)
High density median PFS (95 % CI)
p value
Premenopausal
NA
9.4 (8.96–9.83)
0.030
Postmenopausal
18.4 (9.38–27.42)
6.6 (4.28–8.92)
0.021
Age B40
NA
9.6 (8.90–19.30)
0.16
Age [40
18.4 (11.0–27.0)
7.8 (5.20–9.60)
0.001
HER2-negative
27.0 (11.10–27.0)
9.1 (6.62–11.58)
0.015
HER2-positive
18.4 (9.03–27.77)
9.3 (8.74–9.86)
0.299
No chemotherapy
12.4 (10.90–NA)
6.6 (5.11–8.09)
0.002
Chemotherapy
18.4 (10.97–25.83)
9.4 (9.06–9.74)
0.021
Hormone receptor-negative
11.0 (7.92–14.08)
8.9 (5.23–12.57)
0.221
Hormone receptorpositive
27.0 (12.0–27.0)
9.6 (3.68–15.52)
0.020
No hormonal therapy
10.9 (8.17–13.63)
8.9 (5.36–12.44)
0.165
Hormonal therapy
27.0 (12.0–27.0)
9.6 (3.63–15.57)
0.034
Single site of metastasis
27.0 (12.0–27.0)
5.2 (4.67–5.73)
0.006
Multiple sites of metastases
10.9 (3.02–18.78)
9.4 (9.10–9.70)
0.282
Bone-only metastasis
27.0 (NA–29.0)
7.8 (0.000–16.48)
0.042
Other than boneonly metastases
11.0 (8.89–13.11)
9.4 (8.85–9.95)
0.240
PFS progression-free survival, CI confidence interval
assessed easily with routine mammograms without requiring special software, and it categorizes patients into easily interpretable classes. However, it lacks the accurate estimation of breast density provided by computer-aided assessment in addition to the hazard of inter-observer variation in breast density estimation. At the time point, overall survival (OS) of the study group was not assessed as only nine cases died till the time of this analysis. It will be reported later after longer followup to check for possible impact of breast density on OS. In conclusion, breast density assessed at the time of diagnosis was significantly correlated with PFS in metastatic breast cancer patients both in unadjusted and adjusted models. Certain groups of patients seem to have better outcome if they have low density such as in those older than 40 years at diagnosis, patients with HER2-negative disease, and those with hormonal receptor-positive phenotype. Conflict of interest of interest.
References 1. Boyd NF, Rommens JM, Vogt K, et al. Mammographic breast density as an intermediate phenotype for breast cancer. Lancet Oncol. 2005;6:798–808. 2. Maskarinec G, Pagano I, Little M. Mammographic density as a predictor of breast cancer survival: the multiethnic cohort. Breast Can Res. 2013;15:R7. 3. McTiernan A, Martin CF, Peck JD, et al. Estrogen-plus-progestin use and mammographic density in postmenopausal women: women’s health initiative randomized trial. J Natl Cancer Inst. 2005;97:1366–76. 4. Haars G, Van Noord P, Gils C, et al. Measurements of breast density: no ratio for a ratio. Cancer Epidemol Biomark Prev. 2005;1411(11):2634–40. 5. Alowami S, Troup S, Al-Haddad S, et al. Mammographic density is related to stroma and stromal proteoglycan expression. Breast Cancer Res. 2003;5:R129–35. 6. Provenzano PP, Inman DR, Eliceiri KW, et al. Collagen density promotes mammary tumor initiation and progression. BMC Med. 2008;6:11. 7. Cil T, Fishell E, Hanna W, et al. Mammographic density and the risk of breast cancer recurrence after breast-conserving surgery. Cancer. 2009;115:5780–7. 8. Park CC, Rembert J, Chew K, et al. High mammographic breast density is independent predictor of local but not distant recurrence after lumpectomy and radiotherapy for invasive breast cancer. Int J Radiat Oncol Biol Phys. 2009;73:75–9. 9. Chiu SY, Duffy S, Yen AM, et al. Effect of baseline breast density on breast cancer incidence, stage, mortality, and screening parameters: 25-year follow-up of a Swedish mammographic screening. Cancer Epidem Biom Prev. 2010;19:1219–28. 10. Porter GJ, Evans AJ, Cornford EJ, et al. Influence of mammographic parenchymal pattern in screening-detected and interval invasive breast cancers on pathologic features, mammographic features, and patient survival. AJR. 2007;188:676–83. 11. Gierach GL, Ichikawa L, Kerlikowske K, et al. Relationship between mammographic density and breast cancer death in the breast cancer surveillance consortium. J Natl Cancer Inst. 2012;104:1218–27. 12. Cuzick J, Warwick J, Pinney E, et al. Tamoxifen-induced reduction in mammographic density and breast cancer risk reduction: a nested case-control study. J Natl Cancer Inst. 2011;103:744–52. 13. Li J, Humphreys K, Eriksson L, et al. Mammographic density reduction is a prognostic marker of response to adjuvant tamoxifen therapy in postmenopausal patients with breast cancer. JCO. 2013;31(18):2249–56. 14. Wolfe J. Breast patterns as an index of risk for developing breast cancer. AJR. 1976;126:1130–9. 15. Boyd NF, Martin LJ, Yaffe MJ, et al. Mammographic density: a hormonally responsive risk factor for breast cancer. J Br Menopause Soc. 2006;12(4):186–93. 16. Chew GL, Huang D, Huo CW, et al. Dynamic changes in high and low mammographic density human breast tissues maintained in murine tissue engineering chambers during various murine peripartum states and over time. Breast Cancer Res Treat. 2013;140(2):285–97. 17. Aiello EJ, Buist DS, White E, et al. Association between mammographic breast density and breast cancer tumor characteristics. Cancer Epidem Bioma Prev. 2005;14:662–8.
The authors declare that they have no conflict
123
