Abstract
In post-menopausal breast cancer, obesity is related to invasive tumours that are resistant to chemotherapy and have a poor outcome. Recent research has revealed that adipocytes co-cultured with breast cancer cells (cancer-associated adipocytes or CAA) exhibit a decrease in late adipocyte differentiation markers, and increased expression of adipokines and extracellular matrix proteins such as collagen. In the breast tumour microenvironment collagen has been shown to exert a pro-tumorigenic influence by altering the tumour microenvironment to facilitate metastasis, and by acting as a ligand to stimulate breast cancer cells to undergo epithelial-mesenchymal transition (EMT).
Therefore, the overall aim of this study was to investigate the role of the CAA-collagen axis in human breast carcinoma specimens. Specific objectives included (1) determining the number and size of CAA at the invasive margin and within the tumour, (2) identifying the patterns of collagen deposition and alignment near CAA and within the tumour, (3) determining the relative levels and localisation of markers of breast cancer cell EMT, and (4) analysing this experimental data together with tumour pathology and patient clinical data.
The study cohort comprised 46 post-menopausal (aged 51yr+) women with invasive breast carcinoma, with formalin fixed paraffin embedded (FFPE) tumour samples and at least five years follow-up data available via the Cancer Society Tissue Bank Christchurch (CSTBC). Four patient groups were included: (1) Oestrogen receptor positive (ER+), Body Mass Index (BMI)=20-24.9 (n=11); (2) ER+, BMI>30 (n=12); (3) hormone receptor negative/triple negative breast cancer (TNBC), BMI=20-24.9 (n=10); and (4) TNBC, BMI >30 (n=13). The tumour samples were characterised by their clinicopathological parameters including histological subtype, Nottingham Prognostic Index (NPI), tumour size, tumour grade, lymph node status, stage and ER/PR/HER2 status. H&E stained tumour samples were assessed for the number and size of CAA along the interface and within the tumour, as well as the Tumour-Associated Collagen Signature (TACS) or pattern of collagen deposition and alignment, and these parameters were analysed together with patient clinicopathological data (Chapter 2). To investigate whether collagen was acting as a ligand to promote breast cancer cell metastasis, additional immunohistochemical analysis was used to determine the pattern of expression of Discoidin Domain Receptor 2 (DDR2) and markers of EMT (vimentin, Snail1) (Chapter 3). Experimental variables from both Chapters were analysed together with patient clinicopathological data including disease free survival and overall survival.
Adipocyte diameter (p<0.001), area (p<0.001), and number (p=0.008) were all significantly decreased within the tumour compared to at the tumour interface. As expected, there was strong positive correlation between adipocyte diameter and area at both the interface and within the tumour (p<0.01), and both parameters were negatively correlated with adipocyte count (p<0.01). Adipocyte count within the tumour was higher in grade 3 versus grade 1 and 2 tumours (p=0.22). Adipocyte diameter, area and number were not significantly different in tumours with TACS-1/2 versus TACS-3. However, Body Mass Index (BMI) was significantly higher in patients with TACS-3 compared to TACS-1/2 tumours (p=0.044), and this was due to patient weight (p=0.05) not height (p=0.94). In tumour sections stained with a pan-collagen stain (Masson’s Trichrome), cancer cells appeared to preferentially invade along straightened, aligned collagen fibres, which may provide ‘highways’ for intravasation. Immunohistochemical analysis showed DDR2 protein was present in mesenchymal stromal and immune cells at the tumour interface, but not in breast cancer cells. Vimentin was strongly expressed in TNBC samples and showed the expected positive correlations with tumour grade (p=0.02) and stage (p=0.07). Vimentin immunostaining was negatively correlated with adipocyte diameter (p=0.029) and area (p=0.013) at the tumour interface, adipocyte diameter (p=0.032) and area within the tumour (p=0.032), and a positive correlation with adipocyte count within the tumour (p=0.05). Snail1 immunostaining was strongest near the tumour interface and weaker in the centre of the tumour.
Overall, this study confirmed CAA were smaller at the tumour interface and further reduced in size within the tumour stroma. High patient BMI and weight were associated with increased TACS-3, and both TACS-3 and CAA number within the tumour were positively correlated with high grade (grade 3) breast tumours. CAA were often surrounded by collagen and were abundant in areas with strong vimentin immunostaining at the interface and within the tumour. This study did not provide direct evidence for collagen production by CAA, or for CAA-secreted collagen acting as a ligand to activate DDR2 and promote EMT. However, results from this study indicate the value of TACS-3 as a marker of localised invasion and metastasis, and indicate that further investigation of the role of CAA in the tumour microenvironment is warranted.