Introduction: Mammographic density (MD), after adjustment for a women’s age and body mass index, is a strong and independent risk factor for breast cancer (BC). Although the BC risk attributable to increased MD is significant in healthy women, the biological basis of high mammographic density (HMD) causation and how it raises BC risk remain elusive. We assessed the histological and immunohistochemical differences between matched HMD and low mammographic density (LMD) breast tissues from healthy women to define which cell features may mediate the increased MD and MD-associated BC risk. Methods: Tissues were obtained between 2008 and 2013 from 41 women undergoing prophylactic mastectomy because of their high BC risk profile. Tissue slices resected from the mastectomy specimens were X-rayed, then HMD and LMD regions were dissected based on radiological appearance. The histological composition, aromatase immunoreactivity, hormone receptor status and proliferation status were assessed, as were collagen amount and orientation, epithelial subsets and immune cell status. Results: HMD tissue had a significantly greater proportion of stroma, collagen and epithelium, as well as less fat, than LMD tissue did. Second harmonic generation imaging demonstrated more organised stromal collagen in HMD tissues than in LMD tissues. There was significantly more aromatase immunoreactivity in both the stromal and glandular regions of HMD tissues than in those regions of LMD tissues, although no significant differences in levels of oestrogen receptor, progesterone receptor or Ki-67 expression were detected. The number of macrophages within the epithelium or stroma did not change; however, HMD stroma exhibited less CD206⁺ alternatively activated macrophages. Epithelial cell maturation was not altered in HMD samples, and no evidence of epithelial–mesenchymal transition was seen; however, there was a significant increase in vimentin⁺/CD45⁺ immune cells within the epithelial layer in HMD tissues. Conclusions: We confirmed increased proportions of stroma and epithelium, increased aromatase activity and no changes in hormone receptor or Ki-67 marker status in HMD tissue. The HMD region showed increased collagen deposition and organisation as well as decreased alternatively activated macrophages in the stroma. The HMD epithelium may be a site for local inflammation, as we observed a significant increase in CD45⁺/vimentin⁺ immune cells in this area.

Macrophages play diverse roles in mammary gland development and breast cancer. CC-chemokine ligand 2 (CCL2) is an inflammatory cytokine that recruits macrophages to sites of injury. Although CCL2 has been detected in human and mouse mammary epithelium, its role in regulating mammary gland development and cancer risk has not been explored. Transgenic mice were generated wherein CCL2 is driven by the mammary epithelial cell-specific mouse mammary tumour virus 206 (MMTV) promoter. Estrous cycles were tracked in adult transgenic and non-transgenic FVB mice, and mammary glands collected at the four different stages of the cycle. Dissected mammary glands were assessed for cyclical morphological changes, proliferation and apoptosis of epithelium, macrophage abundance and collagen deposition, and mRNA encoding matrix remodelling enzymes. Another cohort of control and transgenic mice received carcinogen 7,12-Dimethylbenz(a)anthracene (DMBA) and tumour development was monitored weekly. CCL2 protein was also quantified in paired samples of human breast tissue with high and low mammographic density. Overexpression of CCL2 in the mammary epithelium resulted in an increased number of macrophages, increased density of stroma and collagen and elevated mRNA encoding matrix remodelling enzymes lysyl oxidase (LOX) and tissue inhibitor of matrix metalloproteinases (TIMP)3 compared to non-transgenic controls. Transgenic mice also exhibited increased susceptibility to development of DMBA-induced mammary tumours. In a paired sample cohort of human breast tissue, abundance of epithelial-cell-associated CCL2 was higher in breast tissue of high mammographic density compared to tissue of low mammographic density. Constitutive expression of CCL2 by the mouse mammary epithelium induces a state of low level chronic inflammation that increases stromal density and elevates cancer risk. We propose that CCL2-driven inflammation contributes to the increased risk of breast cancer observed in women with high mammographic density.

CCL2 is an inflammatory cytokine that regulates macrophage activity and is implicated in increased mammographic density and early breast tumorigenesis. The role of CCL2 in mediating stromal interactions that contribute to breast tumorigenesis has yet to be fully elucidated. THP-1-derived macrophages and mammary fibroblasts were co-cultured for 72 h. Fibroblasts and macrophages were analysed for phenotype, expression of inflammatory and ECM-regulatory genes and collagen production. Mice overexpressing CCL2 in the mammary glands were analysed for global gene expression by RNAseq at 12 weeks of age. These mice were cross-bred with PyMT mammary tumour mice to examine the role of CCL2 in tumorigenesis. The co-culture of macrophages with fibroblasts resulted in macrophage polarization towards an M2 phenotype, and upregulated expression of CCL2 and other genes associated with inflammation and ECM remodelling. CCL2 increased the production of insoluble collagen by fibroblasts. A global gene expression analysis of CCL2 overexpressing mice revealed that CCL2 upregulates cancer-associated gene pathways and downregulates fatty acid metabolism gene pathways. In the PyMT mammary tumour model, CCL2 overexpressing mice exhibited increased macrophage infiltration and early tumorigenesis. Interactions between macrophages and fibroblasts regulated by CCL2 can promote an environment that may increase breast cancer risk, leading to enhanced early tumorigenesis.

Background: There is growing awareness of breast density in women attending breast cancer screening; however, it is unclear whether this awareness is associated with increased knowledge. This study aims to evaluate breast density knowledge among Australian women attending breast cancer screening. Method: This cross-sectional study was conducted on women undergoing breast cancer screening at The Queen Elizabeth Hospital Breast/Endocrine outpatient department. Participants were provided with a questionnaire to assess knowledge, awareness, and desire to know their own breast density. Result: Of the 350 women who participated, 61% were familiar with ‘breast density’ and 57% had ‘some knowledge’. Prior breast density notification (OR = 4.99, 95% CI = 2.76, 9.03; p = 0.004), awareness (OR = 4.05, 95% CI = 2.57, 6.39; p = 0.004), younger age (OR = 0.97, 95% CI = 0.96, 0.99; p = 0.02), and English as the language spoken at home (OR = 3.29, 95% CI = 1.23, 8.77; p = 0.02) were independent predictors of ‘some knowledge’ of breast density. A significant proportion of participants (82%) expressed desire to ascertain their individual breast density. Conclusions: While knowledge of breast density in this Australian cohort is generally quite low, we have identified factors associated with increased knowledge. Further research is required to determine optimal interventions to increase breast density knowledge.

Background: To progress research into the biological mechanisms that link mammographic breast density to breast cancer risk, fibroglandular breast density can be used as a surrogate measure. This study aimed to develop a computational tool to classify fibroglandular breast density in hematoxylin and eosin (H&E)-stained breast tissue sections using deep learning approaches that would assist future mammographic density research. Methods: Four different architectural configurations of transferred MobileNet-v2 convolutional neural networks (CNNs) and four different models of vision transformers were developed and trained on a database of H&E-stained normal human breast tissue sections (965 tissue blocks from 93 patients) that had been manually classified into one of five fibroglandular density classes, with class 1 being very low fibroglandular density and class 5 being very high fibroglandular density. Results: The MobileNet-Arc 1 and ViT model 1 achieved the highest overall F1 scores of 0.93 and 0.94, respectively. Both models exhibited the lowest false positive rate and highest true positive rate in class 5, while the most challenging classification was class 3, where images from classes 2 and 4 were mistakenly classified as class 3. The area under the curves (AUCs) for all classes were higher than 0.98. Conclusions: Both the ViT and MobileNet models showed promising performance in the accurate classification of H&E-stained tissue sections across all five fibroglandular density classes, providing a rapid and easy-to-use computational tool for breast density analysis.

Background: To study the biological underpinning of mammographic breast density, fibroglandular breast density can be used as a surrogate marker of breast cancer risk. This study aimed to develop a computational tool to classify fibroglandular breast density in hematoxylin and eosin (H&E)-stained breast tissue sections using deep learning approaches that would assist future mammographic density research. Methods: Four different architectural configurations of transferred MobileNet-v2 convolutional neural networks (CNNs) and four different models of vision transformers (ViTs) were developed and trained on a database of H&E-stained normal human breast tissue sections (965 tissue blocks from 93 patients) that had been manually classified into one of five fibroglandular density classes, with class 1 being very low fibroglandular density and class 5 being very high fibroglandular density. Results: The MobileNet-Arc 1 and ViT model 1 achieved the highest overall F1-scores of 0.93 and 0.94, respectively. Both models exhibited the lowest false positive rate and highest true positive rate in class 5, while the most challenging classification was class 3, where images from classes 2 and 4 were mistakenly classified as class 3. The area under the curves (AUCs) for all classes were higher than 0.98. Conclusions: Both the ViT and MobileNet models showed promising performance in the accurate classification of H&E-stained tissue sections across all five fibroglandular density classes, providing a rapid and easy-to-use computational tool for breast density analysis.