Abstract B027: Spatial transcriptomic profiling of the head and neck cancer tumor microenvironment to guide design of microphysiological systems for reporting treatment response

Abstract Background: Microphysiological systems (MPS), advanced 3D models that reproduce organ form and function, show promise for modeling patient-specific treatment outcomes in head and neck cancer (HNC) as they aim to recapitulate the architecture and function of the tumor microenvironment (TME). However, cell and extracellular matrix (ECM) composition within these environments is often decided arbitrarily and may struggle to mimic different microenvironments i.e. primary tumors or metastatic lymph nodes. To guide the cellular and physical composition of the TME in vitro we employed spatial transcriptomic profiling of a tissue microarray (TMA) from patients with HNC to aid in identification of ECM components and cell populations that can be translated into MPS to improve their fidelity to tumor tissue. Methods: Spatial transcriptomic profiling was performed using the GeoMx whole transcriptome atlas on a TMA containing matched primary and lymph node tissue from 44 patients with HNC. Each patient was analyzed in duplicate, with segmentation between tumor (pan-cytokeratin staining) and stroma (collagen staining) for a total of 366 areas of interest profiled. Data was analyzed using differential gene expression, gene set enrichment analysis and linear mixed effect models. MPS were constructed using cells isolated from HNC tumor tissue, comprising tumor epithelial cell spheroids surrounded by a hydrogel matrix containing fibroblasts and tumor infiltrating leukocytes. Molded lumens seeded with blood and lymphatic endothelial cells form vasculature. This MPS permits visualization of readouts such as tumor spheroid growth, migration and viability, angiogenesis, and treatment response modulated by different ECM and cell combinations. Results: 91% of the samples returned acceptable data with a gene detection rate of >5%. Differential gene expression analysis demonstrated that 68.9% of the expressed genes were shared between primary and lymph node tumors while 58.3% were shared between primary and lymph node stroma. Pathway analysis identified significant differences between primary and lymph node tumor in pathways associated with ECM organization and degradation and immune cell cytokine signaling. Identification of ECM proteins enriched in primary and lymph node stroma can guide modulation of biological hydrogels for MPS construction by including these additional proteins in the matrix. Deconvolution of cell type composition using CIBERSORTx also permits modulation of cell proportions in the MPS to more closely mimic tumor tissue. MPS representing primary or lymph node TME’s created with a standard collagen/fibronectin or enhanced guided matrix or standard vs guided cell ratio’s will be built and analyzed for metrics of cell growth, viability, angiogenesis and response to cisplatin. Conclusions: We performed spatial transcriptomic analysis of primary and lymph node tumor tissue from patients with HNC. These data are guiding the ECM and cellular composition of MPS to develop models that faithfully represent patient tumor tissue and response to treatment. Citation Format: Rene Welch Schwartz, Adeel Ahmed, Rong Hu, Randall J Kimple, Paul M Harari, Irene M Ong, Sheena C Kerr. Spatial transcriptomic profiling of the head and neck cancer tumor microenvironment to guide design of microphysiological systems for reporting treatment response [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Functional and Genomic Precision Medicine in Cancer: Different Perspectives, Common Goals; 2025 Mar 11-13; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2025;85(5 Suppl):Abstract nr B027.