Integrative genomic analysis identifies clinically relevant subtypes of head and neck cancer characterized by hypoxia, T-cell infiltration, and EMT

Current classification of head and neck squamous cell carcinomas (HNSCC) based on anatomic site and clinical stage fails to capture biologic heterogeneity or adequately inform treatment. Here we use consensus clustering on 371 HNSCC in three cohorts, including a new cohort of 134 patients with locoregionally advanced HNSCC and 43% HPV(+) tumors to identify five HNSCC subtypes. Subtypes closely correlate with biologic processes (hypoxia, cytotoxic T-cell infiltration, copy number changes, EGFR/HER-ligand phenotype), survival, and HPV-status. Two biologically distinct HPV-associated subtypes are identified. Gene modules of candidate drivers characterize each subtype using the CONEXIC algorithm. The proposed 5-subtype classification provides a biologic basis for future clinical and preclinical studies and lays the groundwork for improved prognostication and therapy development in HNSCC. We use consensus clustering on 371 HNSCC in three cohorts, including a new cohort of 134 patients with locoregionally advanced HNSCC and 43% HPV(+) tumors to identify five HNSCC subtypes.