Collagen architecture in triple negative breast cancer

Abstract

This study evaluated collagen properties in TNBC samples collected from different racial groups to determine the presence of variance in matrix architecture. African American (AA) breast cancer patients have a significantly higher mortality rate and nearly a three-fold increased prevalence of triple negative breast cancer (TNBC) when compared to Caucasian (C) patients. The extracellular matrix region surrounding tumors contains abundant collagen fibers, and these fibers undergo remodeling throughout cancer progression, promote metastasis, and impede treatment response. High mammographic density, commonly known as dense breast tissue, is hypothesized to be more prevalent in AA women and characterized by increased collagen deposition and associated with more aggressive cancers. The aim of this research was to investigate fibrillar collagen architecture in TNBC samples from AA and C patients using two-photon microscopy with second harmonic generation (SHG), an intrinsic optical signal produced by fibrillar collagen. Twenty tissue regions per tumor sample were randomly selected for SHG microscopy, and image processing was conducted using the Fiji macro TWOMBLI to quantify mesoscopic fibrillar morphological properties and nanoscopic fibrillar properties with the Forward-Backward SHG ratio. Compared to the images from C tumor samples, those from AA samples exhibited a significant increase in parameters including fiber area, total length, and number of endpoints and branchpoints, but had decreased lacunarity. Collagen microstructure, including fibril arrangement and packing density, did not significantly differ between the groups. These results illustrate that the TNBC samples analyzed from AA patients may have macrostructural collagen characteristics associated with aggressive phenotype tumor formation.

Fig 1. a) ECM components and collagen structure change with tumor progression, which influences cancer cell proliferation and metastasis. b) SHG visual representation of collagen patterns from ten unique patient samples where the upper row is from five C patient samples and the lower row is from five AA patient samples.

Fig 2. a) FB image processing featuring (1) FB map, (2) TWOMBLI mask, (3) FB-mask multiplied image, and (4) Ridge detection selection of contours. b) TWOMBLI Output Descriptions. (1) Area: cross-sectional area of all fibers, (2) Lacunarity: number and size of gaps in the ECM, (3) Total Length: sum length of all drawn fibers, (4) Endpoints: sum of fiber terminus points, (5) Hyphal Growth Unit: number of end points per unit length, (6) Branchpoints: number of fiber intersections, (7) Box-Counting Fractal Dimension: indicates self-similarity and space-filling properties, (8) Curvature: change in angle moving along individual fibers, (9) Percent High-Density Matrix: proportion of white pixels with respect to dark pixels, (10) Alignment: degree to which fibers are oriented in a similar direction.

Fig 3. Boxplots of morphological parameters between racial groups.

(a) Area increased in AA samples, (b) Lacunarity decreased in AA samples, (c) Total length increased in AA samples, (d) Endpoints increased in AA samples, (e) Branchpoints increased in AA samples, (f) Box-counting fractal dimension increased in AA samples, (g) Fiber thickness increased in AA samples, with *: significant at FDR = 0.05, **: significant at FDR = 0.01.

Fig 4. Correlation Plots Among TWOMBLI Outputs in (a) Combined samples, (b) AA samples, (c) C samples.

Fig 5. a.) Mean FB ratios by sample. b) Mean FB ratios by racial group.