Cancer is one of the leading causes of death across the world, which needs proper diagnosis and effective treatment. Over the last decade, immune-oncology therapies have achieved breakthrough records of survival rates for several types of cancer. However, multiple metabolic dysregulations in the tumor microenvironment promote immunosuppression. One of such is the high concentration of adenosine maintained due to the hydrolysis of ATP released by cells in conditions of hypoxia. Adenosine can bind to adenosine receptors, which are class A G protein-coupled receptors (GPCRs), and exerts an immunosuppressory response by acting on adenosine A2A and A2B receptors (A2AAR and A2BAR) expressed on effector T and NK cells. In this work, we present a comprehensive computational characterization a novel series of pyrimidinone derivatives that guided the development of a novel family of dual A2A/A2B antagonists, developed as potential lead candidates to improve lymphocyte activity, following our drug discovery program1. After a docking exploration on our refined 3D models of A2AAR and A2BAR, we performed systematic free energy perturbation (FEP) calculations with QligFEP2 to explore the structural determinants of the affinity shifts due to a number of chemical modifications on the common scaffold. The results support the conserved binding mode previously proposed for the parent pyrimidinone behind the design of A2BAR antagonists, while opening the door to an alternative binding orientation on the A2AAR, which might be relevant to explain the dual A2A/A2Bantagonism. Further explorations are guiding the optimization of these compounds to develop new drugs for improving cancer immunotherapy.