Identification of lesion subtypes in biopsies of ductal carcinoma in situ of the breast using biomarker ratio imaging microscopy

Abstract

Although epidemiological studies propose aggressive and non-aggressive forms of ductal carcinoma in situ (DCIS), they cannot be identified with conventional histopathology. We now report a retrospective study of human biopsy samples using biomarker ratio imaging microscopy (BRIM). Using BRIM, micrographs of biomarkers whose expression correlates with breast cancer aggressiveness are divided by micrographs of biomarkers whose expression negatively correlates with aggressiveness to create computed micrographs reflecting aggressiveness. The biomarker pairs CD44/CD24, N-cadherin/E-cadherin, and CD74/CD59 stratified DCIS samples. BRIM identified subpopulations of DCIS lesions with ratiometric properties resembling either benign fibroadenoma or invasive carcinoma samples. Our work confirms the existence of distinct subpopulations of DCIS lesions, which will likely have utility in breast cancer research and clinical practice.

Figure 1. Illustration of BRIM.

A DCIS section was labeled with anti-CD44 (A) and anti-CD24 (B). (Panels A,B) were prepared identically. (Panel C) reveals intraductal CD44^hi/CD24^lo cells at high contrast. The white arrows identify a region of CD44^hi/CD24^lo cells that are included in the quantitative line profile analyses of (panels D–F). (Panels D–F) show quantitative line profile analyses (the line profile extends from the right to left hand sides of the image at the level of the arrow). Noise reduction and contrast enhancement are seen in the ratio image of (panel F). The pseudocolor image in (panel C) is scaled as indicated by the bar on the right side. (Distance scale range is shown on the lower left side of (panel A).

Figure 2. Stratification of DCIS samples using tissue sections stained with anti-CD74 and anti-CD59.

A broad range of stromal cell ratio values were observed for samples from five DCIS patients (panels A–E). To quantify these data, further image processing was performed. The ratio of each pixel in a sample of normal tissue is shown in the white plot of (panel F) whereas an identical plot of DCIS tissue is shown as black in this panel. This diagram shows that gray values of >130 (see arrow) are only found in the DCIS sample. (Panel G) shows pixels with gray values of >130 (color-coded red), which confirms that stromal cells account for high BRIM values. The ascending Pareto plots of (panels H–J) show electronic counts of CD74^hi/CD59^lo particles (ordinate) versus the data point index (abscissa). Each dot represents the BRIM score of one patient. Samples from simple fibroadenoma patients (H) display little or no signals (N = 16). DCIS samples (I) segregate into two populations: those scoring near the level of fibroadenomas and those samples scoring highly for the biomarker ratio pair (N = 23). High levels of statistical significance (P < 0.0001) are seen in comparing the low BRIM with high BRIM samples. The high BRIM scores can be seen for the DCIS population and IDC patient samples (J) (N = 26). Quantitative BRIM counts are shown at the top of (panels A–E). A ratio scale is given on the far right of (panel E) (20× objective).

Figure 3. A three-dimensional rendering of all BRIM data for DCIS patients.

Values of the parameters CD74/CD59 (z axis), CD44/CD24 (x axis) and N-cadherin/E-cadherin (y axis) are plotted for each DCIS patient. Some samples had BRIM values far from the origin (0, 0, 0). Of the samples near the origin, 5 had scores of <50 for CD74/CD59 and 0–2 for the other two parameters. Note that DCIS samples near the walls of the three-dimensional plot could be interpreted as false-negatives if only one BRIM parameter was measured.