Abstract 751: High-resolution and AI-enabled single-cell spatial transcriptomics and histopathology integrated to reveal tumor differentiation and immune exclusion in skin squamous cell carcinoma

Abstract Skin squamous cell carcinoma (SCC) is characterized by heterogeneity in differentiation states and immune exclusion within the tumor microenvironment (TME). Using the Bruker Spatial Biology CosMx® Whole Transcriptome (WTX) panel, which profiles approximately 19, 000 genes at single-cell resolution, we examined spatial gene expression in FFPE SCC sections. Individual single cell boundaries were defined utilizing a trained AI cell segmentation model. H&E staining on the same tissue provided histopathological context, enabling the integration of molecular and morphological findings. Combining CosMx WTX data with H&E staining, downstream AI models could be used to further predict cell types and transcriptomic signatures directly from H&E images. FFPE SCC samples were subjected to CosMx WTX imaging to generate spatially resolved transcriptomic data. H&E staining, performed after CosMx WTX, was aligned with transcriptomic data using custom software. Histopathological analysis identified keratin pearls, invasive fronts, and stromal compartments. Bioinformatics mapped gene expression signatures of keratinization, immune exclusion, and extracellular matrix remodeling, providing a detailed view of tumor architecture. Thousands of genes were mapped with high resolution, allowing in-depth exploration of the SCC TME. Cell typing identified spatially distinct cancer subpopulations and their relationship with stromal and immune components. Ligand-receptor analysis revealed spatial patterns of cell-cell communication, while proximity analyses highlighted unique immune interactions. Macrophages near cancer cells showed increased transcription of specific genes compared to distant macrophages, suggesting spatially restricted tumor-immune signaling. Integration with H&E staining revealed molecular correlates of keratinization, immune exclusion, and invasive front dynamics, bridging histopathology with spatial transcriptomics. This integration offers the potential to train AI models that predict tumor cell types and molecular profiles directly from H&E images, enabling faster and more accessible analysis. This study integrates single-cell spatial transcriptomics with histopathological features to reveal the molecular mechanisms of tumor differentiation, immune evasion, and stromal remodeling in SCC. The combination of spatial transcriptomics and H&E images holds potential for future AI models that enable non-invasive, real-time tumor profiling, providing a foundation for therapeutic strategies targeting SCC progression and immune suppression. Citation Format: Patrick Danaher, Michael Patrick, Shanshan He, Liang Zhang, Stefan Rogers, Michael Rhodes, Haiyan Zhai, Joseph Beechem. High-resolution and AI-enabled single-cell spatial transcriptomics and histopathology integrated to reveal tumor differentiation and immune exclusion in skin squamous cell carcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 751.