Traditional 2D in vitro and animal models are known to translate poorly the results from pre-clinical studies into the clinic, making discovering and developing new therapies a very slow and expensive process. New models, such as organs-on-chips, can provide more accurate results of human conditions. In this thesis, human tissue models were used to investigate how novel protein therapies reach and specifically kill tumor cells in complex environments. The results were combined with a mathematical model to predict the best dose for the therapies to kill tumor cells without affecting other healthy cells in the tissue. Low oxygen levels were also included in the system to mimic the environment of large solid tumors better. Better therapies can be developed through an improved understanding of the journey of therapeutic proteins in these complex models, ultimately achieving a faster and more effective translation to the patients.