Abstract
Preventing cancer is vastly better than treating the disease in terms of a patient's quality of life and healthcare costs. Yet, to screen for chemopreventative drugs or evaluate interventions aimed at lowering cancer risk, quantitative readouts of risk are needed. In the breast and in other organs of epithelial origin, apical-basal polarity is key to homeostasis and is one of the first tissue characteristics lost during cancer initiation. Therefore, apical-basal polarity may be leveraged as an "architectural" determinant of cancer risk. A classic approach to quantify the localization of epithelial polarity markers is visual scoring at the microscope by trained investigators. This approach is time-intensive and limited to low throughput. To increase the speed, accuracy, and scoring volume, we developed an algorithm that essentially replaces the human eye to objectively quantify epithelial polarity in microscopy images of breast glandular units (acini). Acini in culture are identified based on a nuclear stain and the corresponding masks are divided into concentric terraces of equal width. This positional information is used to calculate radial intensity profiles (RP) of polarity markers. Profiles with a steep slope represent polarized structures, whereas more horizontal curves are indicative of non-polarized acini. To compare treatment effects, RP curves are integrated into summary values of polarity. We envision applications of this method for primary cancer prevention research with acini organoids, specifically (1) to screen for chemoprevention drugs, (2) for toxicological assessment of suspected carcinogens and pharmacological hit compounds, and (3) for personalized evaluation of cancer risk and risk-reducing interventions. The RadialProfiler algorithm developed for the MATLAB computing environment and for users without prior informatics knowledge is publicly available on the Open Science Framework (OSF).