The objective of the thesis was to decipher the molecular state of tumor infiltrating dendritic cell (DC) and their relation to the tumor microenvironment. By combining the analysis of human tumor samples by flow cytometry and RNA sequencing, of tumor secretome and of a large dataset of in vitro DC-Tcell interactions I obtained 2 main findings. First, we reported a novel classification of human activated DC, that are either “secretory” that is specialized in secreting cytokines and chemokines, or “helper” that is specialized at inducing the secretion of a broad range of T helper cytokines after cell co-culture. DC infiltrating inflamed human head and neck cancer matched the “secretory” phenotypic and transcriptomic signatures. Beyond this novel biological concept, this classification is of importance as a theoretical basis for adjuvant-based immunotherapy. Secondly, we showed that tumor inflammation was not the main prognostic factor for oral cavity cancer (OCC) patients, but that MMP2 and the presence of extra-nodal extension were independent predictors of reduced disease-specific survival. We could stratify OCC into 4 prognostic groups and showed that they had similar expected rates of response to immunotherapy. Our data may serve to design a biomarker-driven clinical trial proposing neoadjuvant chemotherapy or immunotherapy to high-risk patients, with the goal of reducing the percentage of OCC patients that will present with early and severe recurrences.