Mammographically dense human breast tissue stimulates MCF10DCIS.com progression to invasive lesions and metastasis
In this within-individual matched-sample study, we found that HMD breast tissue led to significantly increased tumour weight, greater proportions of high-grade DCIS and grade 3 IDCs, and metastasis of DCIS.com cells compared with LMD tissue from the same woman. The tumour-promoting effect of HMD was observed across women, despite the heterogeneous demographic characteristics of our cohort (mixed menopausal statuses and risk profiles). The finding of high-grade DCIS with grade 3 IDC was consistent with the characteristics of DCIS.com cells to form high-grade DCIS that progresses to correspondingly high-grade IDC in vivo [22]. To our knowledge, we are the first to demonstrate a causal relationship between HMD tissue and BC progression and metastasis in an in vivo setting. We also found a trend of increased CTCs in mice carrying biochambers implanted with HMD tissue compared with LMD and DCIS.com only; however, this was not statistically significant, which may be due to the variation in CTC numbers per mouse for each woman and between women, as well as to our small sample size.
There is little evidence on whether HMD directly affects the progression and metastasis of already established tumours; however, the pathobiology of HMD does support stimulated cancer progression [16, 31]. Boyd and colleagues found that HMD is associated with increased breast tissue stiffness [31], and HMD stroma has increased collagen organisation compared with LMD from the same woman [16, 32, 33]. Specifically, McConnell and colleagues found that increased collagen stiffness and organisation, not abundance, correlated with HMD in a cohort of 22 post-menopausal women (4 cancer-free and 18 with BC). In their study, the tissues from 18 women were sampled at least 4 cm away from tumours, albeit within the same breasts where cancer had initially arisen, and a total of 6 HMD samples were compared with 6 LMD tissue specimens of different women matched for age and menopausal status, but not for BMI or other confounding factors [32]. Their findings contrast with our earlier association of HMD with increased stroma and collagen content [16], where paired HMD breast tissue showed increased collagen organisation as well as abundance compared with LMD tissue of the same woman in a group of 15 cancer-free women. As McConnell et al. used Picrosirius red staining and atomic force microscopy, whereas we used second harmonic generation imaging coupled with grey-level co-occurrence matrix analysis; variable methodologies may also contribute to the differences in the results. Collagen of altered alignment, through mechanical and other unknown properties, have been shown to facilitate tumour growth [14, 34, 35]. The breast stroma is a rich source of numerous cell types, including fibroblasts, adipocytes and extracellular matrix (ECM) proteins [36–38]. ECM comprises not only collagen but also fibronectin, proteoglycans and matrix metalloproteinase (MMP) inhibitors, which have also been shown to enhance collagen stiffness and regulate growth factors and susceptibility to BC [3, 39–43]. Although BC is of epithelial cell origin and HMD is associated with increased benign epithelial lesions [44, 45], an increasing body of data supports the hypothesis that perturbations in stromal architecture are key to establishing a pro-neoplastic environment that enhances cancer growth [46, 47]. In addition, fibroblasts are a major stromal component and have been implicated in pro-malignant activity through the production and/or modification of cytokines, growth factors, ECM components and MMPs [48, 49].
Our results show that the incorporation of LMD breast tissue into DCIS.com cell inoculations reduced tumour weight, lowered the proportion of high-grade DCIS with grade 3 IDCs and led to less metastasis compared with incorporating HMD tissue, suggesting a protective role of the adipose-rich, dense, connective tissue-poor LMD tissue. Consistent with the trends and effects seen in our studies, the association of absolute dense area with BC risk was found to be decreased for larger breasts [50]. Investigators in two large case-control studies (634 cases:1880 controls [51] and 1424 cases:2660 controls [52]) and a prospective study of 111 cases of BC [53] all found statistically significant inverse associations of non-dense breast area with BC risk. There is limited evidence on how adipose tissue modifies BC risk and cancer progression. Fatty breast tissue secretes leptin, which was found to enhance BC cell proliferation, as well as adiponectin, which limits cell proliferation and promotes apoptosis of aberrant cells [54]. Whilst it is unknown whether adipose tissue produces more adiponectin than leptin, a balance between the two was proposed to alter BC risk [55]. Adipose tissue also stores vitamin D, known for its protective effect against cancer development through a wide range of roles, including cell-cycle arrest, apoptosis, repair and promotion of differentiation [56–58]. Little is currently known about cancer-associated adipose tissue, although it is known to secrete a range of cytokines (interleukin [IL]-6, IL-8, chemokine [C-C motif] ligand 5) and collagen VI, promoting BC progression and metastasis [59–63]. Further work on cytokines and ECM from HMD and LMD tissue is required to understand the relationship of the decreased adipose content in HMD and BC.
Strengths and limitations
Previous work conducted by Chew and colleagues showed that human HMD and LMD breast tissues that were sampled from prophylactic mastectomy specimens, mechanically minced, mixed with Matrigel™ and then incubated in murine biochambers for 6 weeks remained viable, maintained their original histological characteristics and their MD status [15]. HE, Masson’s trichrome blue and vimentin staining showed increased collagen and stromal content and a lower fat percentage in HMD chamber tissue compared with that in LMD, correlating with the histological composition of the original mastectomy specimens. Thus, we believe that mechanical mincing in preparation for chamber implantation and supplementation with Matrigel™ did not have a significant impact on the histological composition or collagen content of implanted tissues, and that any effect would be equally distributed to both HMD and LMD groups. In addition, we have unpublished work showing that, when the ECM structure is disrupted through collagenase, hyaluronidase and trypsin digestion of the samples, the original histology and MD status of the input material was not maintained in murine chambers [64]. Prior to chamber implantation, second harmonic generation imaging found that HMD breast tissue had a higher degree of stromal collagen organisation than LMD tissue [16]. However, future studies are needed to further assess whether mincing and addition of Matrigel™ would affect the mechanical stiffness, composition or actions of MMPs in the ECM of implanted breast tissue. Another potential limitation of our study is that our group of ten women was relatively young (mean age 45 years), and five of them were pre-menopausal. The results therefore cannot be directly extrapolated to the average population, in particular for post-menopausal women. However, studying normal breast tissue of different MDs enabled us to examine features of MD that may elevate BC risk prior to established tumour burdens in the breast. The within-individual study design also allowed us to compare HMD and LMD breast tissues of the same patient, eliminating all important confounding factors, such as age, BMI and menopausal status, which can be difficult to adjust for in across-patient studies.