Poster #27 - Luisa Quezada Ziesse

MarginDetect reveals distinct clonal dynamics and its interactions within the microenvironment near the tumor margins in rare subtypes of Bladder Cancer

Luisa Quezada Ziesse, PhD student, Rutgers Cancer Institute of New Jersey Antara Biswas, PhD, Rutgers Cancer Institute of New Jersey Saum Ghodoussipour, MD, Rutgers Cancer Institute of New Jersey Subhajyoti De, PhD, Rutgers Cancer Institute of New Jersey

Rare subtypes of bladder carcinoma have a different prognosis compared to urothelial carcinoma, but the underlying determinants are not well understood. We develop MarginDetect, a novel computational genomics framework, to characterize the dynamics of tumor-stroma-immune interactions at tumor margins and interiors, and to identify aggressive subclones using spatial transcriptomics. We profiled bladder carcinoma samples from rare and common subtypes and identified active tumor-stroma margins by analyzing the pathways expressed within through machine learning techniques. We show that there are substantial subtype-specific differences in cell type composition, intercellular signaling, and cellular processes. These variations collectively suggest divergent mechanisms of microenvironment remodeling in bladder cancer subtypes. Differences in clonal phylogeny and spatial heterogeneity of major subclones between the samples suggested disparate clonal spatiotemporal dynamics and interactions with stromal and immune compartments, which were notably prominent at the tumor-stroma margins. Heterogeneity in the presence and architecture of tertiary lymphoid structures indicated differences in anti-tumor immunity. Cell-free DNA profiling from liquid biopsy supported the dominance of aggressive subclones and provided a means to track tumor progression non-invasively. We propose that our resources allow tracking the trajectory of neoplastic disease progression in bladder cancer subtypes and identifying aggressive malignancies.