Abstract 3138: Topological analysis of spatial transcriptomics reveals different spatial interaction patterns in tumor microenvironment of lung cancer

Abstract

Abstract Background: As the tumor microenvironment (TME) consists of various cell types with complex spatial interaction, its spatial organization patterns affect response to immune-oncology treatment. Therefore, describing the spatial composition and interaction of cells in the tumor microenvironment (TME) is necessary. Here, we developed a tool, STopover, which adopts topological analysis in spatial transcriptomics to reveal cell-cell colocalization patterns in TME and capture the key components and niche of intercellular communication, and applied it to human lung cancer data. Methods: The spatial colocalization pattern of pairs of features was defined using topological data analysis with Morse filtration. The spatial network of spatial gene expression data was generated for each sample and then the patterns were summarized as connected components (CCs) based on the spatial distance between unit tissue regions and the persistence of each CC. The global and local extent of spatial overlap of a feature pair was calculated as Jaccard indices between extracted CC pairs. We applied STopover to 11 barcode-based spatial transcriptomic data of human lung adenocarcinoma with different PD-L1 expressions. Spatial mapping of cell types in TME was performed by CellDART. In addition, image-based spatial transcriptomic data of lung cancer were also used to find key spatial molecular interactions in TME using STopover. Results: First, STopover disclosed the distinct immune and stromal infiltration patterns in lung cancer tissues. Spatial colocalization of cancer cells and T-cells was heterogeneous and correlated with the immune-related markers such as MHC class I signature. The cancer types were clustered according to spatial colocalization patterns of various immune cells with cancer cells, which showed different infiltration patterns of immune cells. Moreover, STopover could estimate the top cell-cell interaction and emphasize the key locations based on the literature-supported ligand-receptor database. Conclusion: STopover is expected to account for significant spatial cell interactions in tumor-immune and tumor-stromal components and could be utilized as a platform to decipher the mechanisms underlying immune-oncology treatment response. Citation Format: Sungwoo Bae, Hyekyoung Lee, Kwon Joong Na, Dong Soo Lee, Hongyoon Choi, Young Tae Kim. Topological analysis of spatial transcriptomics reveals different spatial interaction patterns in tumor microenvironment of lung cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 3138.