Abstract 4876: Data-driven discovery of cancer-specific targets for hepatocellular carcinoma using spatial transcriptomics

Abstract

Abstract Background: Hepatocellular carcinoma (HCC) is frequently diagnosed in advanced stages, with limited options available for systemic treatment. Given the significant investment of time, the high costs, and the notable failure rates involved in developing new drugs, it is crucial to identify appropriate cancer-specific surface targets, especially aiming at antibody drug conjugates and radioligand therapy. This study aims to propose a pipeline utilizing spatial transcriptomics (ST) to prioritize potential cancer-specific surface targets for HCC. Methods: Visium ST samples were collected from fresh-frozen tissues of HCC patients undergoing hepatectomy. The spots with over 500 expressed genes and a mitochondrial gene percentage below 20% were selected. Counts were normalized with sctransform, and the tumor cell proportion in each spot was predicted using CellDART. Cancer region-specific genes were identified based on three criteria: high topological overlap with tumor cell distribution (using STopover), a high correlation coefficient with tumor cell proportion, or significantly higher expression in the tumor cell cluster compared to others in single-cell RNA-sequencing dataset. Genes identified through these methods were pooled and filtered based on average expression in the whole tissue using human normal cell atlas data and surface proteins. To select drug candidates with low normal organ toxcity, Tabula Sapiens datasets were used. Genes expressed in over 50% of cells and with an average expression above 1 in any critical organs were excluded. Results: One ST sample was obtained from each of the 7 HCC patients. The topology, correlation, and single-cell-based criteria resulted in 42, 15, and 7 genes, respectively, with only 1 gene (P4HB) overlapping across all three methods. After filtering genes with high expression in normal organs based on Tabula Sapiens, 25 genes remained. In the GEPIA databases, GPC3 exhibited the highest log fold change in RNA expression between HCC and normal liver (6.280), followed by TM4SF4 (2.156), EFNA1 (0.581), GJB1 (0.498), and IRS1 (0.391). To externally validate the results, 17 ST datasets from 5 patients provided by HCCDB were used. Among the selected genes, only GPC3 and EFNA1 had higher average expression in the tumor compared to other compartments (normal, stromal, and immune) in at least 3 out of 5 patients. Conclusion: The proposed platform prioritizes cancer-specific surface targets for HCC, focusing primarily on spatial gene expression profiles obtained from ST and gene expression data from external databases. This analytical tool can be further utilized to identify molecular targets for developing antibody drug conjugates or radioligand therapies, which are designed to specifically deliver therapeutic agents to the tumor. Citation Format: Sungwoo Bae, Dongjoo Lee, Daeseung Lee, Kwon Joong Na, Suk Kyun Hong, YoungRok Choi, Nam-Joon Yi, Kwang-Woong Lee, Kyung-Suk Suh, Hongyoon Choi. Data-driven discovery of cancer-specific targets for hepatocellular carcinoma using spatial transcriptomics [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4876.