Head and neck cancer transcriptomics: Profiling tumor biology for chemoradiotherapy response

Patients with head and neck squamous cell carcinoma (HNSCC) often present with advanced stage disease. This is often treated with cisplatin based chemoradiotherapy (CRT). A large proportion of HNSCC patients treated with CRT suffer from locoregional recurrences. Upfront identification of patients with poor treatment responses could guide alternative treatment choices. The multicenter DESIGN consortium was founded to find prognostic markers to identify these poor responders. As part of the DESIGN consortium, this thesis focusses on the identification of these biomarkers by using modern techniques such RNA-sequencing, bioinformatics and machine learning. We did this by focusing on several biological processes such as hypoxia (Chapter 2), epithelial-mesenchymal transition (EMT) (Chapter 3) and DNA-repair (Chapter 4). We showed that current hypoxia profiles are representing chronic hypoxia, but that acute hypoxia is also a prognostic biomarker. The impact of EMT in HNSCC is not well-known. By using EMT gene-expression profiles from other cancer types, we showed that EMT is an important biomarker in HNSCC. Using machine learning techniques we generated and validated a HNSCC-specific EMT model. Using similar techniques, we also generated and validated a DNA-repair crosslink repair model. This model is based on cell-line data and it’s sensitivity for Olaparib and Mitomycine C. Tumors with DNA repair defects, identified by this model, have a poor prognosis and tumor cells show a more invasive phenotype in vitro. The relation between previous investigated biomarkers and additional known biomarkers was analyzed in a large multivariable study (Chapter 5). This is all discussed in the final chapter.