Head and neck squamous cell carcinoma (HNSCC) incidence rates are rising globally, mainly due to the consumption of carcinogen-containing products and HPV infections. However, treatment success especially in HPV-negative patients is limited [1]. Patients with recurrent and metastatic disease are often treated with anti-cancer immunotherapy targeting the PD-1 pathway. However, most patients do not respond to the treatment with checkpoint inhibitors despite being selected based on PD-L1 positivity in immunohistochemistry. This suggests that patient selection cannot rely entirely on PD-L1 alone due to alternative checkpoints or pathways that affect the immune response. To systematically profile the tumor-immune microenvironments of responders and non-responders with single-cell and spatial resolution, we selected a patient cohort of 90 individuals and applied multiplex immunofluorescence imaging of 25 markers to capture cellular phenotypes and their functional states. We employed machine learning based image analysis, which identified more than 1 million cells, covering various tumor and immune cell phenotypes, including T cells, B cells, macrophages as well as cancer cells and their states (epithelial, mesenchymal, proliferative). We identified distinct patterns of cellular organizations and interactions that significantly varied between responders and non-responders. Our data revealed strong differences in immune cell abundances and their spatial distributions especially at the tumor invasive front. In the future, we will combine these promising data with the deep visual proteomics workflow employing single-cell laser microdissection followed by deep mass-spectrometry based low-input tissue proteomics. By pooling phenotypically similar cells, we will link cellular neighborhood information to functional proteome sates. With this integrated, multimodal approach, we can combine the advantages of targeted and untargeted spatial proteomics, enabling us to identify potential new drug targets, prognostic biomarkers of treatment response and infer mechanisms of treatment resistance.

[1] Hashibe, Mia et al., Journal of the National Cancer Institute (2007)