Immune cell abundance and T cell receptor landscapes suggest new patient stratification strategies in head and neck squamous cell carcinoma.

Head and neck squamous cell carcinoma (HNSCC) is a molecularly and spatially heterogeneous disease frequently characterized by impairment of immunosurveillance mechanisms. Despite recent success with immunotherapy treatment, disease progression still occurs quickly after treatment in the majority of cases, suggesting the need to improve patient selection strategies. In the quest for biomarkers that may help inform response to checkpoint blockade, we characterized the tumour microenvironment (TME) of 162 HNSCC primary tumours of diverse aetiological and spatial origin, through gene expression and immunohistochemistry (IHC) profiling of relevant immune proteins, T cell receptor (TCR) repertoire analysis and whole-exome sequencing. We identified five HNSCC TME categories based on immune/stromal composition: (1) cytotoxic, (2) plasma cell rich, (3) dendritic cell rich, (4) macrophage rich and (5) immune-excluded. Remarkably, the cytotoxic and plasma cell rich subgroups exhibited a phenotype similar to tertiary lymphoid structures (TLS), which have been previously linked to immunotherapy response. We also found an increased richness of the TCR repertoire in these two subgroups and in never smokers. Mutational patterns evidencing APOBEC activity were enriched in the plasma cell high subgroup. Furthermore, specific signal propagation patterns within the Ras/ERK and PI3K/AKT pathways associated with distinct immune phenotypes. While traditionally CD8/CD3 T cell infiltration and immune checkpoint expression (e.g. PD-L1) have been used in the patient selection process for checkpoint blockade treatment, we suggest that additional biomarkers, such as TCR productive clonality, smoking history and TLS index, may have the ability to pull out potential responders to benefit from immunotherapeutic agents.