Hadron Therapy is an advanced radiotherapy technique for cancer treatment. It offers a better irradiation ballistic than conventional techniques and therefore requires appropriate quality assurance procedures. In this work, we upgraded the GEANT4-based GATE Monte Carlo platform in order to recalculate the TPS dose distributions in view of further benchmarking. In a first step, we selected an appropriate simulation environment (physics models and parameters) in order to produce accurate and efficient simulations. GATE simulations were validated using measurements and other Monte Carlo codes for depth-dose and transverse profiles. While a good agreement was found for depth-dose profiles, larger discrepancies were pointed out for transverse profiles. In a second step, we developed a modeling method to simulate active scanning beam delivery systems, which does not require to simulate the components of the treatment nozzle. The method has been successfully applied to an IBA proton therapy system and validated against measurements for complex treatment plans. Interfaces have also been developed in order to link DICOM RT ION PLAN and DICOM RT DOSE with GATE. Finally, we compared in a third step the TPS and Monte Carlo dose distributions in homogeneous and heterogeneous configurations. The beam models of both dose engines were in satisfactory agreement, allowing further evaluation of clinical treatment plans. A two-field prostate plan has been evaluated, showing a satisfactory agreement between the TPS and Monte Carlo, and demonstrating the novel capabilities of the platform for the evaluation of the TPS. To summarize, we selected an appropriate simulation environment for proton therapy, proposed a modeling method for active scanning systems and presented a method to compare the TPS and Monte Carlo dose distributions. All tools developed in GATE were or will be publicly released. A detailed validation stage of the system including absolute dosimetry is still necessary, in order to quantitatively evaluate its accuracy in various homogeneous and heterogeneous configurations. In this thesis, we have demonstrated that the GATE Monte Carlo platform is a good candidate for the simulation of active scanning delivery systems, allowing further TPS benchmarking. Moreover, the GATE platform also handles imaging applications, such as PET or prompt-gamma imaging towards online treatment monitoring and paves the way of interdisciplinary research advances.