Therapeutic antibodies that block the programmed death-ligand 1 (PD-L1)/programmed death-1 (PD-1) pathway have revolutionized immuno-oncology by inducing robust and durable responses in patients with various cancer including advanced non-small-cell lung cancer (NSCLC). However, these responses only occur in a subset of patients, even in case of PD-L1 overexpression. Elucidating the determinants of response and resistance but also of severe immune-mediated adverse events is key to improving outcomes and developing new treatment strategies. Biomarkers that predict immune checkpoint inhibitors efficacy and toxicity are urgently needed and could emerge from characterization of tumor microenvironment. The purpose of PREDICTION project is to elucidate response and toxicity predictive immunophenotypic signatures using a new in situ multiplexed strategy with imaging mass cytometry Hyperion. Patients treated with anti-PD-1 pembrolizumab will be selected on their response and toxicity profiles. Then, tumor samples will be analysed with Hyperion technology, allowing delineation of cell subpopulations and cell-cell interactions, highlighting tumor heterogeneity and to determine correlations between response and toxicity features. The number of co-analysable markers enables global vision on the same tissue section. A better understanding of the tumor microenvironment complex system will lead to discover new predictive biomarkers potentially transferable to current practice.