Women of African heritage suffer a higher breast cancer mortality compared to their European counterparts. Though the biologic basis for these disparities remains poorly defined, recent studies suggest definitive roles for biological variation in the gene expression pathways governing tumor behavior and alterations in the tumor microenvironment. The transcription factor Kaiso (ZBTB33) is a gene regulatory factor, found in both the nucleus and cytoplasm of breast cancer cells, that has been functionally linked to racial differences in survival outcome in several epithelial cancers. In this study we leverage machine learning and artificial intelligence to define functional linkages between Kaiso, autophagy and the immmune tumor microenvironment that contribute to racial differences in breast cancer survival. We accomplish this through application of machine learning and artificial intelligence to characterize the Kaiso dependent differences in spatial and topological features of the tumor microenvironment using multiplex immunofluorescent technologies to profile a unique breast cancer health disparities cohort (Specific Aim One). We then apply this technology to examine the impact of Kaiso disruption on autophagy and the immune tumor microenvironment using a murine orthotopic allograft model for Kaiso depletion in the presence and absence of pharmacologic blockade of autophagy (Specific Aim Two). We then perform a large-scale application of artificial intelligence and deep learning to profile the spatial and topological features of the tumor microenvironment in 901 racially diverse breast cancer specimens by multiplex immunohistochemistry to define the detailed role of Kaiso, autophagy and the tumor microenvironment in population-specific differences in breast cancer outcome (Specific Aim Three). Together with a closely integrated multi-disciplinary team of breast cancer pathologists, cancer biologists, computer scientists, biostatisticians, bioinformaticians and data scientists, we will define new prognostic and predictive biomarkers that link Kaiso to tumor progression, the immune tumor microenvironment, breast cancer outcome and how their association differs by race.
Elevated expression of the multifunctional transcriptional regulator, Kaiso (ZBTB33) is associated with worse breast cancer survival based on race. In this study we leverage machine learning and artificial intelligence to study a novel breast cancer health disparities cohort to define new functional linkages between Kaiso, autophagy and the tumor microenvironment that contribute to racially disparate breast cancer outcome.
