Figures
Abstract
Background
Mammographic density (MD) has been found to be an independent risk factor for breast cancer. Although data from twin studies suggest that MD has a strong genetic component, the exact genes involved remain to be identified. Alterations in stromal composition and the number of epithelial cells are the most predominant histopathological determinants of mammographic density. Interactions between the breast stroma and epithelium are critically important in the maturation and development of the mammary gland and the cross-talk between these cells are mediated by paracrine growth factors and cytokines. The potential impact of genetic variation in growth factors and cytokines on MD is largely unknown.
Methods
We investigated the association between 89 single nucleotide polymorphisms (SNPs) in 7 cytokine/growth-factor genes (FGFR2, IGFBP1, IGFBP3, TGFB1, TNF, VEGF, IL6) and percent MD in 301 premenopausal women (aged 50 to 55 years) participating in the Norwegian Breast Cancer Screening Program. We evaluated the suggestive associations in 216 premenopausal Singapore Chinese Women of the same age.
Citation: Ozhand A, Lee E, Wu AH, Ellingjord-Dale M, Akslen LA, McKean-Cowdin R, et al. (2013) Variation in Inflammatory Cytokine/Growth-Factor Genes and Mammographic Density in Premenopausal Women Aged 50–55. PLoS ONE 8(6): e65313. https://doi.org/10.1371/journal.pone.0065313
Editor: Xiao-Ping Miao, MOE Key Laboratory of Environment and Health, School of Public Health, Tongji Medical College, Huazhong University of Science and Technology, China
Received: September 4, 2012; Accepted: April 28, 2013; Published: June 7, 2013
Copyright: © 2013 Ozhand et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: No current external funding sources for this study.
Competing interests: The authors have declared that no competing interests exist.
Introduction
High mammographic density (MD) is an established risk factor for breast cancer. Women with extensive MD have been found to have four to six times the risk of breast cancer compared to women with little or no density [1]–[3]. MD is influenced by several breast cancer risk factors including age, body mass index (BMI), parity, age at first birth, hormone therapy use and physical activity; these variables jointly explain approximately 30% of the variability in MD [4]. It is likely that genetic variation is another key factor influencing variability in MD. Twin studies suggest that genetic factors account for 30–60% of the variance in MD [5]–[7]. However, the genetic determinants of MD have not yet been identified. In a recent combined meta-analysis of data from five genome wide association studies (GWAS) among women of European descent, one locus (ZNF365- rs10995190) was reported as highly associated with MD after correction for age and BMI. Although highly statistically significant (combined P = 9×6·10−10), this SNP explains only 0.5% of the variance in MD [8]. It seems likely that there will be multiple other loci involved not detected in the GWAS given the low statistical power of GWAS [9], [10].
The histopathological composition of dense breast tissue consists of both stroma and concentrated epithelial tissue [11]. Mammographically dense breasts have been shown to have higher amounts of collagen, more extensive stromal fibrosis, and higher numbers of epithelial cells when compared with breasts with little density [11]–[14]. Breast stroma and epithelium interact by means of paracrine cytokines and growth factors, which is a necessary process in the normal maturation and development of the mammary gland [15]–[17].
The stroma includes fibrous connective tissue, extracellular matrix (ECM) proteins, fibroblasts, adipocytes, endothelial cells, and innate immune cells. Stroma provides physical structure for the gland and stromal cells secrete signals that are important in the development and function of the epithelium [18]. The extracellular matrix (ECM) together with growth factors/cytokines and cell-cell interactions, modulate the shape, polarity and behavior (survival, proliferation, differentiation, or migration) of cells in mammary tissue [19]. The interactions between cells and ECM are also crucial in determining the organization of the ECM itself [20], [21]. Both cell behavior and tissue structure is therefore affected by cell-ECM interactions. Thus, studying the growth factor/cytokines, as the important signals in the mammary tissue microenvironment, and their role in determining mammographic density, as a marker of the tissue structure and breast cancer, is crucial for understanding mechanisms of breast cancer development.
A number of studies have suggested an association between growth factors and cytokines and MD. Specifically, serum levels of IGF-I and IGF binding proteins have been associated with MD [22]–[24]; findings have been more consistent in premenopausal than in postmenopausal women [24]–[26]. Further, quantitative microscopy using immunoreactive staining has shown higher amounts of IGF-I in dense breasts compared with lower density breasts, especially in women younger than 50 years of age [13]. Genetic variations in IGF and IGF binding proteins have been associated with MD in several studies [7], [27]–[30]. The role of other growth factors and cytokines such as transforming growth factor-beta (TGF-β), interleukins and tumor-necrosis-factor-alpha (TNF-α) with MD has not been well described. A gene expression analysis found decreased levels of TGF-β signaling in women with increased MD [31]. One study observed a positive association between serum levels of interleukin-6, TNF-α, and C-reactive protein (CRP) with MD. Although that association did not remain statistically significant after adjusting for BMI [32], the sum of the findings to date was supportive, and we decided to further study the association between growth factor genetic variants and MD.
Given the biological constituents of MD, the known role of hormone therapy on MD [33], [34], and the individual variability in such hormonal effects, we recently investigated the association between genetic variants in 23 hormone metabolism genes and 7 growth factor genes and MD in postmenopausal participants of the Norwegian Breast Cancer Screening Program (NBCSP). That analysis suggested that there was an association with genetic variants in PRL and CYP1B1 in hormone users (most of whom had used norethisterone acetate preparations). In women who had never used hormone therapy, it was not a hormone gene, but a growth factor gene that was most important (genetic variants in TNF-α.) This suggests that genetic determinants of MD may vary depending on women’s hormonal milieu, and indicated that in never users of hormone therapy growth factor genes may play a role [35].
We therefore explored the role of variation in growth factor genes in premenopausal women participating in NBCSP and compared the results with our previous findings in postmenopausal women. We also decided to test any association in an independent sample of similarly aged premenopausal Singapore Chinese women.
Materials and Methods
Study Population: Norwegian Breast Cancer Screening Program (NBCSP) Participants
The NBCSP is a governmentally funded program which provides biennial screening mammograms to all Norwegian women 50–69 years of age. The screening program began as a four-year pilot project in 1995–96 in four counties of Norway. The project was expanded to all 19 counties and became a nationwide program in 2004. As part of the NBCSP, all women of the appropriate age are sent an invitation letter to receive a bilateral two-view mammogram biennially. Each woman is given an appointment time and location for receiving the mammogram. During the first 10 years (1996–2005), 76.2% of invited women participated in the screening program [36].
In 2004, 17,050 female residents of the three largest counties in Norway (Oslo, Akershus, and Hordaland) were invited to participate in the current study at the same time as they were mailed the official NBCSP invitation letter. This study has previously been described [37]. In brief, participants were asked to complete a risk factor questionnaire which included questions on menstrual and reproductive history, oral contraceptive and menopausal hormone use, family history of breast cancer, current weight and height, alcohol and smoking. Subjects were asked to bring the completed questionnaire and informed consent to the clinic on the day of their scheduled mammogram. Approximately 71% (N = 12,056) of the invited women attended the scheduled mammographic examination and 66% of the attendees aged 50 to 69 (N = 7,941) completed the risk factor questionnaire.
Buccal kits were mailed to 7,174 of the 7,941 women who completed the mammogram and questionnaire to collect DNA for genetic testing. A total of 3,728 women (51% of the 7,174 women) provided a buccal sample. We requested mammograms from the radiological facilities on all 3,728 women with a completed questionnaire and a buccal sample. After excluding women with only a digital mammogram (n = 300), we were able to obtain analog mammograms from the year 2004 on 2,876 women. Of these, 121 women were excluded for the following reasons; history of breast or any cancers (N = 17), undetermined breast area (N = 3), missing age (N = 28), missing BMI (N = 73) (height = 46/weight = 67). After the exclusions, a total of 2,755 women aged 50 to 69 had usable analog mammogram and complete risk factor data. All the participants signed an informed consent and the study was approved by the USC institutional review board, the Norwegian regional ethics committee and the Norwegian Data Inspectorate.
Mammographic Density Assessment
Left craniocaudal mammograms were scanned using a Kodak Lumisys 85 scanner. MD was assessed by a trained reader (GU) using a previously validated computer-assisted method (the University of Southern California Madena software) [38]. The reader assessed the absolute MD by outlining all dense areas within the breast except white artifacts, prominent fibrous strands, vasculature or the pectoralis muscle. The total area of the breast was assessed by a research assistant who was trained by GU. MD was calculated as the absolute density divided by the total area of the breast.
Tagging SNP Selection and Genotyping
We selected genes encoding growth factors (VEGF), growth factor receptors (FGFR2, GHRHR), growth factor binding proteins (IGFBP1; IGFBP3), and cytokines (TGFB1, TNF, IL6). For VEGF, IGFBP1;IGFBP3, TGFB1, TNF, and IL6, we selected tagging SNPs to capture the genetic variation in each gene with an R2>0.80. Tagging SNPs were selected from 20 kb upstream of 5′ untranslated region (UTR) to 10 kb downstream of 3′ UTR that tagged all common SNPs (minor allele frequency ≥5%) among the non-Hispanic white or Chinese population. This selection was done using the Snagger [39] software and a custom database of the Hapmap CEU data (http://hapmap.ncbi.nlm.nih.gov); release 24) merged with the Affymetrix 500 K panel as well as the Hapmap CHB data release 24. For FGFR2 and GHRHR, we selected one SNP of interest for each gene.
Due to restricted funding, DNA extraction and genotyping were performed on 3,317 of the 3,728 participants who donated buccal samples. DNA was extracted from buccal swabs using the standard protocol for the QIAamp blood DNA kit (Qiagen, Valencia, CA). We genotyped the selected SNPs using an Illumina BeadLab System (San Diego, CA) with GoldenGate®. Genotyping was completed in the USC Genomics Center under the direction of Dr. David Van Den Berg. Briefly, samples were run in a 96-well format using the Illumina Sentrix Array technology, scanned on a BeadArray Reader, and analyzed using BeadStudio Software (v.3.0.9) with Genotyping Module (v.3.0.27) (Illumina). The SNPs with <85% call rates were excluded: this resulted in the exclusion of 4% of SNPs. The genotyping concordance rate based on 57 duplicate samples was 98%. Out of 97 SNPs in this pathway, 8 SNPs were excluded due to departure from Hardy-Weinberg equilibrium (HWE) (P<0.001), leaving 89 SNPs for further analysis.
Of the genotyped 3,317 samples, 241 samples were excluded from the analysis due to low overall genotype call-rates (less than 80%). In total, 2,397 women (2,055 postmenopausal, 342 peri- or premenopausal at the time of mammography) had complete information on genotype, MD and breast cancer risk factors. Of the 342 peri- or premenopausal women, 301 were premenopausal and aged 55 or younger at the time of mammogram (they were still menstruating and were not taking any type of hormones).
Statistical Analysis
We explored the association between MD and potential risk factors (age, BMI, age at full-term pregnancy, number of children, age at menarche, family history of breast cancer, and level of education) using categorical variables. We used analysis of covariance (ANCOVA) to calculate age adjusted least-square mean of MD in each category. A test of trend across these categories was generated using linear regression models after adjusting for age; BMI was further included in the models [40].
We investigated the association between each genetic variant and MD based on additive models, which estimate the difference in the continuous dependent variable (MD) per copy of the minor allele of each polymorphism after adjustments for age and BMI. In order to explore the potential modifying effect of BMI on the findings, we repeated this analysis separately in women with BMI below as well as above 25 kg/m2. We considered a two-sided P value of <0.05 as statistically significant.
Replication Study and Combined Analysis
We evaluated the statistically significant associations observed in the NBCSP participants using data from 163 premenopausal Singapore Chinese women of similar age, who were participants of the genetic study component of the Mammography Subcohort of the Singapore Chinese Health Study (SCHS). Participants of the Mammography Subcohort were enrolled in both the SCHS and the Singapore Breast Screening Project (SBSP); details have been described previously [41], [42]. Briefly, 35,298 Chinese women and 27,959 men, ages 45–74 years, enrolled in SCHS during 1993–1998. Subjects were residents of government housing estates; during the enrollment period 86% of the Singapore population resided in such housing facilities. During 1994 to 1997, Singaporean women ages 50–64 years were invited for a screening mammography as part of the SBSP [43]. Through a computer linkage, a total of 3,777 women common to the SBSP and SCHS databases were identified. Of these, mammograms were successfully retrieved from 3,702 women. We excluded 6 women due to missing information on key variables; 1 woman who was later found not to be a Singapore resident. Mammograms of the Mammography Subcohort of the SCHS were scanned using a Cobrascan 812T scanner (Radiographic Digital Imaging Inc., Compton, California). Images were read using the same procedures and software by GU. The total breast area was assessed by two assistants and the average of the two readings was used. Of the 3,695 women in the Mammography Subcohort [41], [44], [45], DNA samples were available on 2,164 women (1,848 blood, 316 buccal). Twenty tagging SNPs in the IL6 locus were selected and genotyped using the same methods used for the NBCSP participants; 1 SNP with a genotyping call rate <85% and 7 SNPs with a MAF<0.01 in Chinese population were excluded, leaving 12 IL6 SNPs for statistical analyses. 2,038 samples of the 2,164 genotyped samples had a genotyping success rate (call rate ≥85%). The mean age of the 2,038 participants were 57.2 (SD 4.3). Two hundred and sixteen women self-reported as premenopausal at time of mammography; of these, 163 women who were aged 55 or younger at mammography (range 46–55) and had never used hormone therapy were included in the current analysis. Genotyping concordance based on the 42 random duplicate samples was >99.9%. None of the 12 IL6 SNPs departed significantly from HWE (P≥0.01).
We combined the Norwegian and Singapore samples and assessed the association between 12 IL6 tagging SNPs and MD. In the combined analysis, we defined the risk allele as the minor allele in the Norwegian sample. We adjusted the models for age at mammogram (continuous), BMI at mammogram (continuous), and ethnic and dialect group (Norwegian, Cantonese, Hokkien).
Results
Baseline Characteristics of the Participants
The baseline characteristics of the postmenopausal sample have previously been described [35]. In brief, mean age at screening was 58.4 years, mean BMI 25.1, mean age at menarche 13.2 years, mean age at first pregnancy 22.0 years, mean number of children 2.0, and mean years of education 12.8. In premenopausal women (Table 1), mean MD decreased with increasing BMI after adjustment for age (P<0.0001). Older age at full term pregnancy was associated with higher MD after adjustment for age and BMI (P = 0.02). Higher level of education was associated with higher percent MD after adjustment for age (P = 0.011) but the association was no longer statistically significant after we further adjusted the model for BMI.
Associations between SNPs and Mammographic Density in NBCSP Participants
The effect of growth factor gene variants on MD was significantly modified by menopausal status (Table S1; see [35] for detailed results on postmenopausal women). The majority of statistically significant associations were observed among the premenopausal women only. In the remaining part of the results, we limit the analysis to this group of women.
Associations between SNPs and Mammographic Density in Premenopausal NBCSP Participants
In the additive genetic model, IL6 tagging SNPs rs6952003, rs10242595, rs11766273, rs1880241, rs1880242, rs2069833, rs2069840, rs4552807 and rs7776857 were associated with MD with P values less than 0.05 (Table 2 and Table 3). The estimated difference in MD per minor allele of each IL6 SNP ranged from 3–5%, with p-values ranging from 0.04 to 0.0002. One TNF tagging SNP (rs2857605) was also significantly associated with MD (beta = 2.99), however the level of significance was relatively low (P = 0.046). We did not find any statistically significant associations between the polymorphisms in VEGF, GHRHR, IGFBP1, IGFBP3, FGFR2, and TGFB1 and MD (Table 2 and Table S2).
In addition, we examined the associations separately in women with low and high BMI (using 25 kg/m2 as the cut-off value). The association between IL6 SNPs and MD appeared to be restricted to women with a BMI less than 25 kg/m2; 8 of 9 tagging IL6 SNPs that showed significant results in the overall analysis remained significant only in the low BMI group. For 5 of these 8 SNPs, the effect modification by BMI was statistically significant (Table 4).
Association between IL6 SNPs and MD in Singapore Chinese Women
Of the 12 evaluated IL6 SNPs, only rs10242595 was associated MD in the replication sample, with an estimated 10.6% increase in MD per A-allele (Table 5). In the pooled analysis with data from the Singapore Chinese women and the NBCSP, rs10242595 A-allele was associated with a 6.2% increase in MD (P = 0.0001).
Discussion
We studied the association between MD and the SNPs in 7 growth factor or cytokine genes including IGFBP1, IGFBP3, TNF, FGFR2, VEGF, GHRHR, and IL6. We observed statistically significant effect modification by menopausal status. While there were no significant associations for SNPs in 6 of the genes, 9 SNPs in the IL6 region (rs6952003, rs10242595, rs11766273, rs1880241, rs1880242, rs2069833, rs2069840, rs4552807, rs7776857) were each significantly associated with MD in premenopausal women. MD varied between 3.4% to 5.8% per allele for these SNPs. Several of the associations were statistically significantly modified by BMI; the associations were limited to women with low BMI. The association with rs10242595 was replicated in an independent study of Singapore Chinese women. Given that each 1% increment in MD has been shown to be associated with a 2% higher relative risk of breast cancer [46], the magnitude of these associations suggest that these variants could be clinically significant.
The lack of association we found between SNPs in most of these growth factor and cytokine genes (IGFBP1, IGFBP3, FGFR2, VEGF and GHRHR) and MD is consistent with results from the few studies that have been conducted on these genes and MD. Consistent with our findings, the majority of previous studies investigating IGFBP1 and IGFBP3 SNPs reported a lack of significant association between IGFBP1/IGFBP3 SNPs including rs2854746, rs1553009, rs1065780, rs2132570, rs3110697, rs35539615, rs4619, and rs6670 and MD. These studies include a cross-sectional study among 1,121 of premenopausal and postmenopausal women from the Nurses’ Health Study cohort investigating 13 tagging SNPs [27], a study of 819 pre- and postmenopausal women of Hawaiian, European, and Japanese ancestry from the Multiethnic Cohort study investigating 22 tagging SNPs [30], and a study of 1,916 premenopausal women within the Prospect-EPIC cohort investigating 11 tagging SNPs [47]. In the study by Tamimi et al., rs4619 in IGFBP1/IGFBP3 region was positively associated with increased MD in a mixed population of premenopausal and postmenopausal women [27], however, this association was not observed in another study [30] nor in our study. Results from the Multiethnic Cohort study showed no association between IGFBP1/IGFBP3 rs10228265, rs1496497 and rs3110697 and MD in the overall analysis, but a significant association was found when the analysis was limited to women with Hawaiian and Japanese descent [30]. In that study, the results were based on data from premenopausal and postmenopausal women pooled together. Our finding of no significant association between FGFR2 rs2981582 and MD is consistent with a study of 516 white (429 non-Hispanic, 87 Hispanic) women in the age range of 20 to 49 years [48] and in a study of 825 pre-and postmenopausal women within the Multiethnic Cohort study [49]. We looked at only one SNP in GHRHR gene (rs4988496) and found no significant association. Similarly, in a study of 177 premenopausal women [50] a different polymorphism, GHRHR A57T, was reported as not significantly associated with MD.
While there have been no previous studies on IL6 SNPs and MD, there are experimental data suggesting that our significant findings are biologically plausible. Cultures of normal mammary epithelial cells obtained from healthy women were shown to release interleukin-6 and express interleukin-6 receptor [51]. Data coming from in vitro studies supports the pleiotropic (having both tumor promoting and tumor-counteracting effects) nature of interleukin-6 in breast tissue [52]. It seems plausible that variations in the IL6 gene could have effects on cell growth and alter MD and eventually breast cancer risk.
Another plausible way to explain the effects of interleukin-6 on MD is indirectly through estrogen. Interleukin-6 has an important role in regulating estrogen synthesis in normal and malignant breast tissues. The activities of aromatase, estradiol 17β-hydroxysteroid dehydrogenase and estrone sulfatase have been shown to be influenced by interleukin-6 in these tissues [53].
Mammographic density is inversely associated with the amount of fat tissue in the breast. It is possible that genetic factors could influence MD by influencing the amount of fat in the breast. IL6 rs10242595-A allele was associated with decreased total body fat mass in one study where fat mass was measured with dual energy X-ray absorptiometry [54]. Our finding is consistent with this result; we found a significant positive association between this polymorphism and MD. IL6 rs1880242 has been significantly associated with decreased risk of obstructive sleep apnea syndrome; obesity is a strong risk factor for this syndrome [55]. Consistent with results from this study, we found a significant positive association between this polymorphism and MD. However, this may suggest that the observed association between this SNP and MD is driven by its association with non-dense breast area rather than the absolute density. When we tested the association with absolute density for this IL-6 SNP, the observed association was similar to percent MD.
In this study we found both positive and inverse associations with different IL6 SNPs. To what extent the IL-6 tagging SNPs modify IL-6 protein levels, and the direction of effect on protein levels is not yet clear. The negative association observed for some of these SNPs and MD does not necessarily represent a negative association between serum or tissue levels or function of IL-6 and MD. Future studies with available breast tissue samples, blood samples and mammographic density measurements would allow us to explore whether any of these associations represent tissue specific effects. However, it would be a challenge to assemble a large enough group of healthy women with breast tissue samples for such investigations.
In this study the association between 5 IL6 SNPs and MD was significantly modified by BMI (Table 4). Higher magnitude and more significant associations in women with the BMI of 25 or less, suggests that the role of IL6 variants in predicting MD is less important in obese women.
There were several strengths of our study. The study sample was selected from a population based study conducted within a national screening program, and the population studied is ethnically homogeneous. Further, we used previously validated MD assessment techniques, and collected detailed information regarding key MD risk factors. We also replicated our findings in a different ethnic group. BMI was considered a potentially confounding factor in this study; we controlled for this variable in all the analyses presented. Many previous studies of MD combined pre- and postmenopausal women, which could mask any findings in premenopausal women. Our analysis conducted separately in premenopausal and postmenopausal women, may have helped to clarify results in premenopausal women. A limitation of our study was using buccal samples for genotyping of the NBCSP samples, which resulted in a relatively lower call-rate compared to the results from studies using blood samples. Further, given the nature of these screening programs, relatively few women were eligible for our study of premenopausal women. Finally, although it could have been more informative to examine the association between genetic variants, serum or tissue levels of growth factors/cytokines as intermediates, and MD, we did not have serum or tissue available to perform such analyses.
Conclusions
Our study suggests that SNPs in the IL6 region may be associated with MD in premenopausal women. Future studies should be conducted to relate these SNPs and interleukin-6 concentrations as well as IL6 gene expression in the mammographically dense tissue to elucidate the mechanisms underlying this association.
Supporting Information
Table S1.
Association between polymorphisms in growth-factor/cytokine pathway genes and MD by menopausal status in Norwegian women.
https://doi.org/10.1371/journal.pone.0065313.s001
(XLSX)
Table S2.
Association between polymorphisms in growth-factor/cytokine pathway genes and MD in Norwegian women (n = 301).
https://doi.org/10.1371/journal.pone.0065313.s002
(XLS)
Author Contributions
Conceived and designed the experiments: AO GU AW. Performed the experiments: AO GU. Analyzed the data: AO EL GU. Contributed reagents/materials/analysis tools: AO EL GU ME. Wrote the paper: AO EL GU RM AW LA.
References
- 1. Boyd N, Lockwood G, Martin L, Knight J, Byng J, et al. (1998) Mammographic densities and breast cancer risk. Breast Dis 10: 113–126.
- 2. Boyd N, Guo H, Martin L, Sun L, Stone J, et al. (2007) Mammographic density and the risk and detection of breast cancer. N Engl J Med 356: 227–236.
- 3. McCormack V, dos Santos Silva I (2006) Breast density and parenchymal patterns as markers of breast cancer risk: a meta-analysis. Cancer Epidemiol Biomarkers Prev 15: 1159–1169.
- 4. Vachon C, Kuni C, Anderson K, Anderson V, Sellers T (2000) Association of mammographically defined percent breast density with epidemiologic risk factors for breast cancer (United States). Cancer Causes Control 11: 653–662.
- 5. Boyd N, Dite G, Stone J, Gunasekara A, English D, et al. (2002) Heritability of mammographic density, a risk factor for breast cancer. N Engl J Med 347: 886–894.
- 6. Ursin G, Lillie E, Lee E, Cockburn M, Schork N, et al. (2009) The relative importance of genetics and environment on mammographic density. Cancer Epidemiol Biomarkers Prev 18: 102–112.
- 7. Stone J, Gurrin L, Byrnes G, Schroen C, Treloar S, et al. (2007) Mammographic density and candidate gene variants: a twins and sisters study. Cancer Epidemiol Biomarkers Prev 16: 1479–1484.
- 8. Lindstrom S, Vachon CM, Li J, Varghese J, Thompson D, et al. (2011) Common variants in ZNF365 are associated with both mammographic density and breast cancer risk. Nat Genet 43: 185–187.
- 9. Fletcher O, Johnson N, Orr N, Hosking FJ, Gibson LJ, et al. (2011) Novel breast cancer susceptibility locus at 9q31.2: results of a genome-wide association study. J Natl Cancer Inst 103: 425–435.
- 10. Ahmed S, Thomas G, Ghoussaini M, Healey CS, Humphreys MK, et al. (2009) Newly discovered breast cancer susceptibility loci on 3p24 and 17q23.2. Nat Genet 41: 585–590.
- 11. Hawes D, Downey S, Pearce CL, Bartow S, Wan P, et al. (2006) Dense breast stromal tissue shows greatly increased concentration of breast epithelium but no increase in its proliferative activity. Breast Cancer Res 8: R24.
- 12. Boyd NF, Lockwood GA, Byng JW, Tritchler DL, Yaffe MJ (1998) Mammographic densities and breast cancer risk. Cancer Epidemiology, Biomarkers & Prevention 7: 1133–1144.
- 13. Guo Y, Martin L, Hanna W, Banerjee D, Miller N, et al. (2001) Growth factors and stromal matrix proteins associated with mammographic densities. Cancer Epidemiol Biomarkers Prev 10: 243–248.
- 14. Alowami S, Troup S, Al-Haddad S, Kirkpatrick I, Watson PH (2003) Mammographic density is related to stroma and stromal proteoglycan expression. Breast Cancer Res 5: R129–135.
- 15. Cullen KJ, Lippman ME (1992) Stromal-epithelial interactions in breast cancer. Cancer Treat Res 61: 413–431.
- 16. Sakakura T (1991) New aspects of stroma-parenchyma relations in mammary gland differentiation. Int Rev Cytol 125: 165–202.
- 17. Dickson RB, Lippman ME (1995) Growth factors in breast cancer. Endocr Rev 16: 559–589.
- 18. Sternlicht MD (2006) Key stages in mammary gland development: the cues that regulate ductal branching morphogenesis. Breast Cancer Res 8: 201.
- 19. Streuli CH, Akhtar N (2009) Signal co-operation between integrins and other receptor systems. Biochem J 418: 491–506.
- 20. Kadler KE, Hill A, Canty-Laird EG (2008) Collagen fibrillogenesis: fibronectin, integrins, and minor collagens as organizers and nucleators. Curr Opin Cell Biol 20: 495–501.
- 21. Kass L, Erler JT, Dembo M, Weaver VM (2007) Mammary epithelial cell: influence of extracellular matrix composition and organization during development and tumorigenesis. Int J Biochem Cell Biol 39: 1987–1994.
- 22. Bremnes Y, Ursin G, Bjurstam N, Rinaldi S, Kaaks R, et al. (2007) Insulin-like growth factor and mammographic density in postmenopausal Norwegian women. Cancer Epidemiol Biomarkers Prev 16: 57–62.
- 23. dos Santos Silva I, Johnson N, De Stavola B, Torres-Mejía G, Fletcher O, et al. (2006) The insulin-like growth factor system and mammographic features in premenopausal and postmenopausal women. Cancer Epidemiol Biomarkers Prev 15: 449–455.
- 24. Byrne C, Colditz GA, Willett WC, Speizer FE, Pollak M, et al. (2000) Plasma insulin-like growth factor (IGF) I, IGF-binding protein 3, and mammographic density. Cancer Res 60: 3744–3748.
- 25. Boyd NF, Stone J, Martin LJ, Jong R, Fishell E, et al. (2002) The association of breast mitogens with mammographic densities. Br J Cancer 87: 876–882.
- 26. Diorio C, Pollak M, Byrne C, Mâsse B, Hébert-Croteau N, et al. (2005) Insulin-like growth factor-I, IGF-binding protein-3, and mammographic breast density. Cancer Epidemiol Biomarkers Prev 14: 1065–1073.
- 27. Tamimi R, Cox D, Kraft P, Pollak M, Haiman C, et al. (2007) Common genetic variation in IGF1, IGFBP-1, and IGFBP-3 in relation to mammographic density: a cross-sectional study. Breast Cancer Res 9: R18.
- 28. Biong M, Gram I, Brill I, Johansen F, Solvang H, et al. (2010) Genotypes and haplotypes in the insulin-like growth factors, their receptors and binding proteins in relation to plasma metabolic levels and mammographic density. BMC Med Genomics 3: 9.
- 29. Verheus M, McKay JD, Kaaks R, Canzian F, Biessy C, et al. (2008) Common genetic variation in the IGF-1 gene, serum IGF-I levels and breast density. Breast Cancer Res Treat 112: 109–122.
- 30. Verheus M, Maskarinec G, Woolcott C, Haiman C, Le Marchand L, et al. (2010) IGF1, IGFBP1, and IGFBP3 genes and mammographic density: the Multiethnic Cohort. Int J Cancer 127: 1115–1123.
- 31. Yang WT, Lewis MT, Hess K, Wong H, Tsimelzon A, et al. (2010) Decreased TGFbeta signaling and increased COX2 expression in high risk women with increased mammographic breast density. Breast Cancer Res Treat 119: 305–314.
- 32.
Reeves KW, Weissfeld JL, Modugno F, Diergaarde B (2010) Circulating levels of inflammatory markers and mammographic density among postmenopausal women. Breast Cancer Res Treat.
- 33. Greendale GA, Reboussin BA, Slone S, Wasilauskas C, Pike MC, et al. (2003) Postmenopausal hormone therapy and change in mammographic density. J Natl Cancer Inst 95: 30–37.
- 34. McTiernan A, Martin CF, Peck JD, Aragaki AK, Chlebowski RT, et al. (2005) Estrogen-plus-progestin use and mammographic density in postmenopausal women: Women’s Health Initiative randomized trial. J Natl Cancer Inst 97: 1366–1376.
- 35. Ellingjord-Dale M, Lee E, Couto E, Ozhand A, Qureshi SA, et al. (2012) Polymorphisms in hormone metabolism and growth factor genes and mammographic density in Norwegian postmenopausal hormone therapy users and non-users. Breast Cancer Res 14: R135.
- 36. Hofvind S, Geller B, Vacek PM, Thoresen S, Skaane P (2007) Using the European guidelines to evaluate the Norwegian Breast Cancer Screening Program. Eur J Epidemiol 22: 447–455.
- 37. Qureshi SA, Couto E, Hilsen M, Hofvind S, Wu AH, et al. (2011) Mammographic density and intake of selected nutrients and vitamins in Norwegian women. Nutr Cancer 63: 1011–1020.
- 38. Ursin G, Ma H, Wu A, Bernstein L, Salane M, et al. (2003) Mammographic density and breast cancer in three ethnic groups. Cancer Epidemiol Biomarkers Prev 12: 332–338.
- 39. Edlund C, Lee W, Li D, Van Den Berg D, Conti D (2008) Snagger: a user-friendly program for incorporating additional information for tagSNP selection. BMC Bioinformatics 9: 174.
- 40.
Kleinbaum DG, Nizam A, Muller KE (2007) Applied Regression Analysis and Other Multivariable Methods: Duxbury Press.
- 41. Wu AH, Ursin G, Koh WP, Wang R, Yuan JM, et al. (2008) Green tea, soy, and mammographic density in Singapore Chinese women. Cancer Epidemiol Biomarkers Prev 17: 3358–3365.
- 42. Lee E, Hsu C, Van den Berg D, Ursin G, Koh WP, et al. (2012) Genetic variation in peroxisome proliferator-activated receptor gamma, soy, and mammographic density in singapore chinese women. Cancer Epidemiol Biomarkers Prev 21: 635–644.
- 43. Ng EH, Ng FC, Tan PH, Low SC, Chiang G, et al. (1998) Results of intermediate measures from a population-based, randomized trial of mammographic screening prevalence and detection of breast carcinoma among Asian women: the Singapore Breast Screening Project. Cancer 82: 1521–1528.
- 44. Jakes RW, Duffy SW, Ng FC, Gao F, Ng EH, et al. (2002) Mammographic parenchymal patterns and self-reported soy intake in Singapore Chinese women. Cancer Epidemiol Biomarkers Prev 11: 608–613.
- 45. Ursin G, Sun CL, Koh WP, Khoo KS, Gao F, et al. (2006) Associations between soy, diet, reproductive factors, and mammographic density in Singapore Chinese women. Nutr Cancer 56: 128–135.
- 46. Boyd N, Byng J, Jong R, Fishell E, Little L, et al. (1995) Quantitative classification of mammographic densities and breast cancer risk: results from the Canadian National Breast Screening Study. J Natl Cancer Inst 87: 670–675.
- 47. Taverne CW, Verheus M, McKay JD, Kaaks R, Canzian F, et al. (2010) Common genetic variation of insulin-like growth factor-binding protein 1 (IGFBP-1), IGFBP-3, and acid labile subunit in relation to serum IGF-I levels and mammographic density. Breast Cancer Res Treat 123: 843–855.
- 48. Lee E, Haiman CA, Ma H, Van Den Berg D, Bernstein L, et al. (2008) The role of established breast cancer susceptibility loci in mammographic density in young women. Cancer Epidemiol Biomarkers Prev 17: 258–260.
- 49. Woolcott CG, Maskarinec G, Haiman CA, Verheus M, Pagano IS, et al. (2009) Association between breast cancer susceptibility loci and mammographic density: the Multiethnic Cohort. Breast Cancer Res 11: R10.
- 50. Mulhall C, Hegele RA, Cao H, Tritchler D, Yaffe M, et al. (2005) Pituitary growth hormone and growth hormone-releasing hormone receptor genes and associations with mammographic measures and serum growth hormone. Cancer Epidemiol Biomarkers Prev 14: 2648–2654.
- 51. Basolo F, Conaldi P, Fiore L, Calvo S, Toniolo A (1993) Normal breast epithelial cells produce interleukins 6 and 8 together with tumor-necrosis factor: defective IL6 expression in mammary carcinoma. Int J Cancer 55: 926–930.
- 52. Knüpfer H, Preiss R (2007) Significance of interleukin-6 (IL-6) in breast cancer (review). Breast Cancer Res Treat 102: 129–135.
- 53. Purohit A, Newman SP, Reed MJ (2002) The role of cytokines in regulating estrogen synthesis: implications for the etiology of breast cancer. Breast Cancer Res 4: 65–69.
- 54. Andersson N, Strandberg L, Nilsson S, Adamovic S, Karlsson M, et al. (2010) A variant near the interleukin-6 gene is associated with fat mass in Caucasian men. Int J Obes (Lond) 34: 1011–1019.
- 55. Zhang X, Liu RY, Lei Z, Zhu Y, Huang JA, et al. (2009) Genetic variants in interleukin-6 modified risk of obstructive sleep apnea syndrome. Int J Mol Med 23: 485–493.