Advancing Education, Research, and Quality of Care for the Head and Neck oncology patient.
Background: Compared to head and neck cutaneous malignancies, mucosal head and neck squamous cell carcinoma (HNSCC) are generally thought to be poorly responsive to immunotherapy. Drivers of immunotherapy resistance are not well understood. Single cell RNA sequencing (SCRNAseq) approaches allow for interrogation of the tumor and immune microenvironment, which may serve to elucidate resistance mechanisms. Published HNSCC SCRNAseq datasets include a limited number of patients and tumor types. In this study, we present a meta-analysis of existing data that allows comparison of HNSCC at the single cell level across a number of critical disease characteristics. We sought to explore whether head and neck cancers harbor significant immune populations and whether immune related differences exist between different sites and subsites within HNSCC.
Methods: Studies with single-cell data in HNSCC were included. Data sets not including tumor samples, adequate numbers of cells, or lacking available cell annotations were excluded. Data sets were integrated by RPCA method from R package Seurat to remove batch correction. Graph-based unsupervised clustering and UMAP were used to reveal cell clusters, which were then annotated based on known cell type specific marker genes. Wilcoxon signed-rank tests were used to detect differences between samples.
Results: Four datasets were included. A total of 383,843 cells were included in the analysis representing 41 unique tumors samples, 21 unique tumor-infiltrating lymphocyte (TIL) samples, 50 matched peripheral blood mononuclear cell samples, 7 metastatic lymph node samples, and 5 healthy tonsil samples. Samples included 34 HPV- oral cavity tumors, 3 HPV- laryngeal tumors, and 25 HPV+ oropharynx tumors.
Among oral cavity tumors, T/NK cells made up at least 25% of total cells in 14/18 samples and at least 50% of total cells in 7/18 samples. Among oropharynx tumors, T/NK cells made up at least 25% of total cells in 9/22 samples.
Compared to HPV+ oropharynx tumors, oral cavity HNSCC had significantly higher proportions of NK cells (p = 0.02), CD4 T cells (p = 0.008), CD8 T cells (p = 0.001) and lower proportions of myeloid cells (p = 0.0001) and epithelial cells (p = 0.001).
We then compared oral cavity subsites to evaluate for differences in immune infiltration and makeup. 21 of 34 had accompanying subsite information including 11 oral tongue, 4 mandibular alveolus, 3 buccal, and 3 floor of mouth tumors. After combining TIL samples with the CD45+ fraction of bulk tumor samples, there were no significant differences among the oral cavity subsites of NK cells (p = 0.73), CD4 T cells (p = 0.11), CD8 T cells (p = 0.56), or B cells (p = 0.98) between subsites.
Conclusions: Our results leverage existing datasets to create the largest HNSCC SCRNAseq cohort to date. These data reveal a relative immune cell abundance among HNSCC. Significant differences in immune compartments exist between HPV+ oropharynx and HPV- oral cavity tumors. No significant differences in immune fractions were observed between oral cavity subsites. Further analyses at the protein and spatial level are needed to investigate drivers of immunotherapy resistance among HNSCC.