Increased peri-ductal collagen micro-organization may contribute to raised mammographic density

Twenty-two women (54–66 years) were recruited from the Nightingale Centre breast-screening clinic. Although this Centre is one of the largest breast clinics in the UK, only about 30 suitable patients within this age range undergo surgery each year. Of these patients we were able to achieve a consent rate of 50–60 % during the duration of the study period. The local Research Ethics Committee approved the study, and all patients gave written informed consent. Women on hormone replacement therapy or who had neoadjuvant breast cancer treatment were not included in the study cohort.

These studies fall under the ethics approvals for the MCRC Biobank, which is licensed by the Human Tissue Authority (licence number 30004) and has been ethically approved as a research tissue bank by the South Manchester Research Ethics Committee (Reference 07/H1003/161 + 5). Written informed consent was obtained at the time of tissue collection.

All of the study participants were characterised for breast cancer risk by oestrogen receptor (ER), human epidermal growth factor receptor 2 (HER-2) and breast cancer gene (BRCA) status – none of the individuals recruited were BRCA-positive. A clinical radiologist assessed digital mammograms from each patient, and identified the area of highest radiographic density. All studies were performed on tissue situated at least 4 cm distant from any neoplasia. There is clear evidence from assessments of adipokine expression, genetic changes, mechanical properties, and radiofrequency characteristics, that tissue distant from the tumour margin will not be subject to a field change and hence can be considered to be benign [15‐18]. Following surgery, highlighted tissue areas (5–10 mm³) were sampled from the area of high radiographic density in the resected breast, by a clinical histopathologist.

The samples were prepared for histological analysis of tissue architecture, and amorphous and aligned fibrillar collagen, by H&E, Masson’s trichrome and Picrosirius red (PSR) staining, respectively. MD was quantified by image analysis (Volpara® breast density [19]), validated against visual estimation recorded on a visual analogue scale (VAS) by a single experienced assessor. In six individuals with low and six with high MD (Table 1) micromechanical analysis was performed on sections cut from frozen tissue samples. Subsequently, ultrastructural and proteomic analyses were carried out on six tissue samples (n =3/group). Tissue samples were assessed by a histopathologist to confirm the absence of neoplasias.

For each patient, mediolateral oblique (MLO) and craniocaudal (CC) digital mammograms were processed by Volpara® to compute the percentage of the breast volume occupied by dense tissue [19, 20]. They were also assessed by an experienced radiologist who recorded the percentage of the breast area occupied by dense tissue on a VAS.

H&E and Masson’s trichrome staining were used to assess cellularity, stromal and fat content and the abundance of amorphous collagen [21, 22]. Tissue samples were bisected, one half was fixed overnight in 4 % paraformaldehyde in PBS before processing to wax with an ASP6025 automated tissue processor (Leica, Newcastle, UK). Sections (3 μm thick) were adhered to positively charged slides, before dewaxing and methanol processing prior to staining. Following batch staining with the histology stains H&E and Masson’s trichrome, mounted sections were imaged sequentially using an SCN400 digital slide scanner (Leica, Newcastle, UK), with a Leica NA Plan Apo × 20/0.75 objective. Slides were only scanned in areas where specimen tissue was present. Resulting images (0.46 μm/pixel resolution) were saved in the MIRAX format. Image analysis was performed with custom-written rule sets for the Tissue Studio 3 software package (Definiens, Munich, Germany) [23]. More specifically, images were thresholded using the hue, saturation and brightness colour space with epithelial regions identified as the areas of lowest brightness and hue, adipose regions as areas of highest brightness and hue and stromal regions as intermediate values.

As ECM organisation, and not just abundance, plays a key role in determining the mechanical properties of tissues [24], we additionally stained breast tissue samples with PSR (1 h incubation with 0.1 % sirius red F3BA in saturated aqueous picric acid at pH 2 followed by clearing in 0.1 % HAc, dehydration and mounting in DPX) [25]. When visualised under cross-polarised light, the resultant collagen-associated birefringence can be semiquantitatively assessed against total tissue area [26, 27]. To clarify whether this PSR enhanced birefringence was induced by collagen fibrils alone [28], or only by assembled collagen fibrils aligned into fibres [25], we assessed the ability of PSR/polarised light microscopy to visualise aligned collagen I fibrils in young Wistar rat tail tendon, and in non-aligned fibrils reconstituted in vitro from bovine collagen I (Thermo Fisher Scientific Ltd., Loughborough, UK). Collagen fibril gels were formed at 37 °C in a hydrated atmosphere, with a gelation time of approximately 30 minutes [29]. Additional quantitative analyses of collagen orientation (coherency) were conducted on PSR stained images using a well-established methodology [105].

We have previously shown that AFM can readily image nano-scale structural detail in unstained tissue cryosections [30]. We applied this technique to characterise the effects of MD on fibrillar collagen organisation in peri-ductal tissues, and to compare the ultrastructural architecture of collagen fibrils in rat tail tendon and reconstituted gels. The other halves of the bisected breast tissue samples, along with the rat tail tendon and reconstituted collagen fibrils were embedded in optimal cutting temperature cryo-sectioning media and snap frozen into a 2.5-cm mould with liquid nitrogen cooled isopentane [30]. Frozen sections were then cut to a nominal thickness of 5 μm in a cryostat microtome at −20 °C, air dried, washed in distilled H2O and stored at 4 °C. Subsequently these cryosections were imaged by a Bioscope Catalyst utilising Bruker’s ScanAsyst PeakForce mode (Bruker, Coventry, UK), a Bruker Scanasyst Air triangular silicon nitride cantilever (nominal spring constant of 0.2 N/m and a pyramidal tip of nominal radius of 2 nm). Height and amplitude images were captured at two scan sizes. For smaller scan areas, 10 × 10 μm was assessed at a scan frequency of 0.25 Hz and a sampling frequency of 1024 × 1024 (for a lateral spacing of 9.76 nm), whereas for large areas 150 x 150 μm was assessed at a scan frequency of 0.01 Hz and a sampling frequency 4992 x 4992 (for a lateral spacing of  30.05 nm).

Rat tail tendon was extracted from intact tail following sacrifice of 12-month-old Wistar rats. Whole collagen fibre bundles were removed from the tail, fixed overnight in 4 % paraformaldehyde in PBS before processing to wax (as before). Sections (3 μm thick) were adhered to positively charged slides and dewaxed before staining or AFM analysis. Re-constituted collagen gels were made from Bovine Collagen I (Gibco, Warrington, UK) by treating collagen at 5 mg/mL with 1 N NaOH on ice to produce a total volume of 2 ul. This liquid suspension was placed onto a charged slide where it was allowed to dry down overnight and rinsed with distilled water before being assessed for collagen fibril abundance and organisation by both polarised light microscopy of PSR stained sections and AFM..

Micro-indentation of rat tail tendon reconstituted collagen fibrils and peri-ductal breast tissue (from six patients with low MD and six with high MD) was carried out using 5 μm thick cryo-sections and a Bioscope Catalyst AFM (Bruker, Coventry, UK) mounted onto an Eclipse T1 inverted optical microscope (Nikon, Kingston, UK) fitted with a spherically tipped cantilever (nominal radius and spring constant of 1 μm and 3 Nm⁻¹ respectively: Windsor Scientific Ltd., Slough, UK,) running Nanoscope Software v 8.15 (Bruker, Coventry, UK). The local reduced modulus was determined for each of 400 points in a 25 × 25 μm region, indented at a frequency of 1 Hz with lateral spacing of 1.25 μm. The extend curve was used in conjunction with a contact-point-based model to calculate the reduced modulus for each indentation [31]. For each biological sample, three 25-μm² regions, and hence 1,200 force curves, were collected. Post hoc analyses of force curves were performed using Nanoscope Analysis v 1.40 (Bruker, Coventry, UK), whereby a baseline correction was applied to each curve before a force fit was applied using the Herzian (spherical) model and a maximum force fit of 70 %. Once all 400 force curves had been generated, quality control was applied, whereby any force values falling more than two standard deviations away from the mean value were discarded in order to account for failed indents. In general fewer than 10 % of force curves were excluded. (data not shown). Note that in the same breast samples, there was no significant difference between the reduced modulus of 5-uM-thick sections (440 kPa ± 12) and 20-uM-thick sections (450 kPa ± 17, n = 1,200, p >0.0001).

Frozen tissue (20 mg) from three individuals with low and three with high overall MD was used to assess protein content by mass spectrometry conducted in the Faculty of Life Sciences Biological Mass Spectrometry Facility (Bio-MS). Tissues were disrupted using a Fisher 120 sonic dismembrator (Thermo Fisher, Cramlington, UK) and resuspended in 8 M urea 0.1 M Tris HCl pH8.5 [32]. The solubilised protein level was quantified using a Direct Detect system (Millipore, Billerica, MA, USA) and 25 ug were taken for subsequent digestion. Proteins were then digested using a variant of the filter-aided sample preparation (FASP) method [33] whereby proteins were solubilised in urea instead of SDS. In brief, proteins were reduced with dithiothreitol and alkylated with iodoacetamide in the presence of 8 M urea in a Micron 30 kDa centrifugal unit followed by pre-digestion with LysC in 6 M urea before digestion with trypsin in 1.5 M urea. The resultant peptides were desalted into 0.1 % formic acid in 5 % acetonitrile using Poros R3 reversed phase chromatographic media (Life technologies, Carlsbad, CA, USA) housed in 0.20-um polyvinylidene fluoride (PVDF) filter 96-well plates (Corning, New York, NY, USA). Liquid chromatography–mass spectrometry (LC-MS/MS) was performed using an Orbitrap Elite™ Hybrid Ion Trap-Orbitrap Mass Spectrometer coupled with a nano U3000 chromatography system (both Thermo Fisher, as before). The data produced was quantified using Progenesis LC-MS (Non-Linear Dynamics, Newcastle, UK) and identified using Mascot (Matrix Science, London, UK). Proteins identified by mass spectrometry (MS) were allocated to categories according to their Gene Ontology (GO) cellular compartment annotation, and enrichment of GO terms was assessed using enrichment analysis in Cytoscape (NRNB, Bethesda, Maryland, USA) [34].

The non-parametric Mann–Whitney U test was applied to determine significance (Graphpad Instat La Jolla, CA, USA). Values are reported as mean ± SD. Analysis of AFM data was conducted using SPSS 20 (SPSS, Chicago, IL, USA). Normal distribution was confirmed by q-q plot. One-way analysis of variance (ANOVA) was used to compare means. For all analyses, a value p <0.05 was considered statistically significant.