Abstract 3598: Classification of the collagen ECM in normal human tissue as a biomarker for future breast cancer incidence

Abstract

Abstract Because we do not know who will develop breast cancer, or who will have a relapse of cancer following surgery, there is an urgent need for the development of biomarkers that can be detected specifically in normal, nondiseased tissue. For 30 years there has been a search for the genetic markers that will predict risk of developing breast cancer, but even while family history is a known risk factor, only 10% of breast cancer risk can be specifically linked to a genetic mechanism (BRCA-1, BRCA-2). Furthermore, individuals who are at high risk due to genetics or family history do not always develop breast cancer. Breast density is another indicator of future incidence, as high-density tissue is correlated with a 4-6 fold increased risk of developing cancer. We hypothesized that a population analysis of the properties of the individual collagen fibers that comprise the extracellular matrix (ECM) in nondiseased breast tissue is a biomarker for the onset of disease. We used multiphoton second-harmonic generation (SHG) imaging to generate high-contrast images of the collagen ECM that can be gathered from common histologic slides that require no additional staining. Our goal was to determine the range of heterogeneity in the properties of the collagen matrix in normal samples. We imaged samples from 141 normal patient donors to comprise 4 different cohorts that define this ground truth state of the matrix. The cohorts were from donors who had never been diagnosed with breast cancer; who had previously been diagnosed with breast cancer but now were disease free; who had never been diagnosed with breast cancer at the time of tissue donation but later went on to develop breast cancer; and those who had donated tissue while in remission from an initial breast cancer diagnosis but later went on to have a recurrence of breast cancer. The SHG images of each patient sample were then analyzed using the curvelet transform-based ctFIRE and CurveAlign software platforms to generate data (fiber length, width, straightness, density, angle with respect to a boundary of epithelial ducts/lobules, and the relative alignment of fibers to each other) on each individual collagen fiber. Advanced statistical measurements and principal component analysis were used to classify the nature of each of the cohorts. Based on our measurements of the structure and organization of collagen fibers, we found that while measurements within a cohort were consistent, there were unique attributes of the collagen matrix that defined each cohort. Furthermore, we performed regression analysis of our measurements against the standard clinical features (age, race, BMI, etc.). Because our various cohorts delineate the nondiseased, involuted, precancer, and prerecurrence normal matrix, respectively, we feel that these data serve as a highly useful, novel classifier, one that describes the clinical impact of the timbre of the matrix in nondiseased women. Citation Format: Ryan J. Gigstad, Tianjie Wang, Yifei Liu, Menggang Yu, Yuming Liu, Adib Keikhosravi, Kevin W. Eliceiri, Patricia J. Keely, Matthew W. Conklin. Classification of the collagen ECM in normal human tissue as a biomarker for future breast cancer incidence [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 3598.