Abstract 4280: Automatic identification of cancer cell-cell interfaces at the pixel level

Abstract

Abstract Deciphering cell-cell interactions is critical for understanding the role of the tumor immune microenvironment. Although multiplex imaging methods are increasingly used for spatial characterization of cell communities and tissue morphologies, quantification of the interface between contacting cells has been mostly unexplored. Here we design an unsupervised pixel-level clustering method that accurately detects the contacting interface between cells. First, we observed pixel-level clusters with cell type specific phenotypes including tumor, immune infiltrates and extracellular structures in two independent melanoma datasets using cytometry time-of-flight (CyTOF) imaging mass cytometry (IMC) images. More importantly, our pixel-level approach identified clusters with mixed phenotypes, which are demonstrated to be direct contacts between T cells and myeloid cells. Specifically, pixels in the interface clusters simultaneously express T cell (CD3+, CD4+ or CD8+) and myeloid cell (CD68+) phenotypic markers at comparable levels. Spatial mapping of the pixels in interface clusters shows that interface pixels are in the physical contact regions between T cells and myeloid cells. In addition, the abundance of T cell and myeloid cell interface clusters is correlated with positive response to immune checkpoint inhibitors (ICIs) (p<0.05). We also identified other clusters indicating interfaces between the stroma, tumor cells, and immune infiltrates. Our work enables reliable delineation of cell interactions at high resolution, and therefore can unravel new information about how cancer cells interact with their environment. Citation Format: Jie Zhou, Jan Martinek, Zichao Liu, Ali Foroughi Pour, Te-Chia Wu, Santhosh Sivajothi, Paul Robson, Karolina Palucka, Jeffrey Hsu-Min Chuang. Automatic identification of cancer cell-cell interfaces at the pixel level. [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 4280.