High-Resolution Profiling of Head and Neck Squamous Cells Carcinoma Identifies Specific Biomarkers and Expression Subtypes of Clinically Relevant Vulnerabilities

Abstract

Background:: Head and neck squamous cell carcinoma (HNSC) is the seventh most common cancer worldwide. Although there are several options for the treatment of HNSC, there is still a lack of better biomarkers to accurately predict the response to treatment and thus be more able to correctly treat the therapeutic modality. Methods:: First, we typed cases from the TCGA-HNSC cohort into subtypes by a Bayesian non-negative matrix factorization (BayesNMF)-based consensus clustering approach. Subsequently, genomic and proteomic data from HNSC cell lines were integrated to identify biomarkers of response to targeted therapies and immunotherapies. Finally, associations between HNSC subtypes and CD8 T-cell-associated effector molecules, common immune checkpoint genes, were compared to assess the potential of HNSC subtypes as clinically predictive immune checkpoint blockade therapy. Results:: The 500 HNSC cases from TCGA were put through a consensus clustering approach to identify six HNSC expression subtypes. In addition, subtypes with unique proteomics and dependency profiles were defined based on HNSC cell line histology and proteomics data. Subtype 4 (S4) exhibits hyperproliferative and hyperimmune properties, and S4-associated cell lines show specific vulnerability to ADAT2, EIF5AL1, and PAK2. PD-L1 and CASP1 inhibitors have therapeutic potential in S4, and we have also demonstrated that S4 is more responsive to immune checkpoint blockade therapy. method: First, we typed cases from the TCGA HNSC cohort into subtypes by a Bayesian non-negative matrix factorization (BayesNMF)-based consensus clustering approach. Subsequently, genomic and proteomic data from HNSC cell lines were integrated to identify biomarkers of response to targeted therapies and immunotherapies. Finally, associations between HNSC subtypes and CD8 T-cell-associated effector molecules, common immune checkpoint genes, were compared to assess the potential of HNSC subtypes as clinically predictive immune checkpoint blockade therapy. Conclusion:: Overall, our HNSC typing approach identified robust tumor-expressing subtypes, and data from multiple screens also revealed subtype-specific biology and vulnerabilities. These HNSC expression subtypes and their biomarkers will help develop more effective therapeutic strategies. result: The 500 HNSC cases from TCGA were put through a consensus clustering approach to identify six HNSC expression subtypes. In addition, subtypes with unique proteomics and dependency profiles were defined based on HNSC cell line histology and proteomics data. Subtype 4 (S4) exhibits hyperproliferative and hyperimmune properties, and S4-associated cell lines show specific vulnerability to ADAT2, EIF5AL1, and PAK2. PD-L1 and CASP1 inhibitors have therapeutic potential in S4, and we have also demonstrated that S4 is more responsive to immune checkpoint blockade therapy.