Mapping immune activity in HPV-negative head and neck squamous cell carcinoma: a spatial multiomics analysis

Head and neck squamous cell carcinoma (HNSCC) exhibits low response rates to immunotherapies, with only about 15-25% of patients responding to monotherapy and 30-45% to combination therapy. This limited effectiveness is attributed to significant intertumor and intratumor heterogeneity, which affects the immunological activity of individual tumors and their regions, thereby influencing immunotherapy outcomes. Various biomarkers at the gene and protein expression levels have been identified to predict the response to immunotherapy in HNSCC.

In this study, we evaluated intertumor heterogeneity using a 27-gene expression signature to stratify tumors by their immunologic activity status. We investigated intertumor heterogeneity at the molecular and cellular level and further analyzed intratumor spatial heterogeneity within and across these subgroups by using spatial multiomics approaches.

Immunologically active tumors showed increased interferon-γ and interferon-α signaling and upregulation of major histocompatibility complex-I signaling and genes involved in antigen presentation. Chemokines such as CXCL8 and CXCL9, which are crucial for immune cell recruitment, were differentially regulated. The spatial analysis revealed that active tumors tended to show higher autocorrelation of homogeneous regions with immune cell infiltration compared with inactive tumors. Proximity measures showed an increased colocalization of immune cells, particularly CD8+ T cells, T helper cells, and regulatory T cells, near tumor cells in active tumors. Despite this high immune infiltration, HNSCC often has an immunosuppressive microenvironment, which we observed as a colocalization of programmed cell death protein-1+ (PD-1+) cytotoxic T cells and cytotoxic T cells, indicating regional differences in active and exhausted cell ratios. Furthermore, upregulation of JAK-STAT3 signaling in active tumors was potentially associated with immune evasion.

The spatial analysis at multiple omics levels allowed for a detailed investigation of molecular and cell type markers to further distinguish between immunologically active and immunosuppressive microenvironments and their spatial heterogeneity. Our study demonstrates that, besides gene expression signatures, cell colocalization signatures can infer immunological activity in HNSCC, thus predicting immunotherapy response.